Development of SNALPs Containing Both mRNA and siRNA for In Situ Molecular Reprogramming

Having validated potential targets, PEI was substituted for a SNALP-based system. The SNALP system was selected to circumvent toxicity/formulation issues as described for PEI and for translational relevance. The ionizable lipid Dlin-MC3-DMA was chosen due to its availability, published potency, and its clinical application.³⁰ Many alternate lipid systems have been proposed including lipidoid-based systems, these may offer improved transfection or loading.^31,32 The pDNA was substituted for mRNA (mOX40L) due to the clinical acceptability of mRNA and its proximity to translation. SNALPs were prepared using previously published lipids and formulation parameters optimized for mRNA delivery. The proposed scheme for the SNALP structure is shown in Scheme 1. We aimed to produce SNALPs with a size no larger than 200 nm and with maximized encapsulation efficiency (EE%). As shown in Table 1, the size of the SNALPs was only slightly affected by the nucleic acid payload and ranged from 143 to 149 nm. The SNALPs bore a net positive charge (~16 mV) when the measurement was carried out in citrate buffer (pH 4) diluted with water and this was unaffected by nucleic acid content. The positive charge could be attributed to the fact that at low pH the ionizable lipid is protonated. When buffer was exchanged for PBS (pH 7.4) the charge was near neutral. A high encapsulation efficiency was achieved in all cases: SNALPs incorporating siRNA had the lowest efficiency at ~81%, whereas the combination SNALP had the highest at ~93%. We speculate that the formulation parameters, including lipid composition and manufacturing methods, are the major size limiting factor. The use of a microfluidic-based system has produced smaller particle with greater loading and allows for future scale up.³³

Open in a separate window

Scheme 1

Development of a mRNA/siRNA Dual-Encapsulating SNALP Suitable for In Situ Molecular Reprogramming

This project describes the development of a SNALP system for co-encapsulation of mRNA and siRNA. The proposed formulation is composed of ionizable lipid, cholesterol, neutral lipid, and PEG ceramide surrounding the relevant nucleic, mRNA and/or siRNA, and acid pay load. A graphical representation is shown.

TEM revealed that SNALPs have an irregular shape with a complex internal structure. Visually, there was no discernible evidence of the different nucleic acids affecting SNALP morphology or formation (Figure ​Figure11A and Supplementary Figure 2.). This irregularity may be due to the fact that SNALPs were produced using a simple mixing method rather than the controlled micromixing of a microfluidic system known to produce regular particles.³⁴ As RiboGreen assay, used to calculate loading, cannot distinguish mRNA and siRNA, there was a concern that should there be preferential loading of a single type of RNA this would be undetected. To address this, the unloaded RNA in the SNALP preparation was digested with RNase H, SNALPs were disassociated with heparin and the nucleic acid contents run on an agarose gel (Figure ​Figure11B). Alongside the SNALP contents, unformulated nucleic acids at a comparable ratio to the preformulation ratio (50:50) were also run, as shown in Figure ​Figure11C. Although this is a semiquantitative assay, and cannot be used to accurately quantify RNA, a comparable intensity ratio of RNA bands (siRNA/mRNA) was obtained for the RNA mix (~1.98) and the dissociated SNALP (~1.62) (Figure ​Figure11D). This suggests that the starting ratio of nucleic acids is maintained following formulation and RiboGreen quantification of total RNA is a suitable method for both mRNA and siRNA for future studies. Additionally, it demonstrates that the RNA content of the SNALP can be protected from nuclease attack.

Open in a separate window

Figure 1

SNALPs have an irregular structure with evidence of internal concentric rings and both mRNA and siRNA are loaded into SNALPs with minimal interference between molecules. (A) SNALPs were formulated with siRNA, mRNA, or a combination of both RNA molecules as previously described. SNALPs were drop-cast on to a graphene grid and imaging was carried out using a Tecnai Osiris transmission electron microscopy. Images represent a single event representative of the wider field. (B) To confirm that both types of nucleic acid can be loaded into SNALPs and that there is minimal hindrance between either molecule, SNALPs coformulating mRNA, and siRNA were treated with RNase H (1 mg/mL) to degrade non-encapsulated/external RNA. The enzyme was inactivated with heat and EDTA (1.25 mM), the SNALP was dissociated by incubation with 10% (v/v) heparin. The RNA was purified with Monarch RNA Cleanup Kit and run on a 2% agarose gel at 225 V for 25 min. (C) Resulting gel image. Free siRNA and mRNA were run as size markers. A mix of the two free nucleic acids (Mix) corresponding to the starting ratio of nucleic acid (50:50) at a quantity equal to the amount obtained from the SNALP was run alongside the RNA extracted from the SNALP (dissociated SNALP). (D) The intensities of the bands for both Mix and dissociated SNALP were measured using imageJ software.

SNALPs Transfect B16F10 Cells and Result in Simultaneous Expression of OX40L and Silencing of PDL1 with Minimal Toxicity

To determine whether the SNALP formulation was able to transfect cells in vitro, B16F10 cells were incubated with SNALPs containing nucleic acids for 48 h. SNALPs containing siNeg, siPDL1, mOX40L, mOX40L–iNeg (coformulation), mOX40L–siPDL1 (coformulation), and siPDL1 + mOX40L (mixture of two SNALPs) were used. Representative flow cytometry plots are shown in Figure ​Figure22A; gates were drawn based on isotype controls. As shown in Figure ​Figure22A, B16F10 cells express PDL1 at moderate levels and do not express OX40L under normal conditions (untransfected). In all cases, following transfection, cells behave as a single homologous population. For the untransfected cells and for cells treated with SNALPs containing mOX40L, there were 66.40% (lower quadrants) and 56.82% (lower quadrants) of B16F10 viable cells that did not express PDL1, respectively. However, when cells were treated with SNALPs containing siPDL1 alone or in the presence of mRNA, a reduction in the PDL1+ population is observed. For SNALPs entrapping siPDL1, 95.90% of cells were negative for PDL1 (lower quadrant); for coformulated SNALPs and SNALP mixture, 82.27 and 92.89% of the viable cells were negative for PDL1, respectively. This qualitative shift indicates that there has been a silencing in the expression of this marker. Figure ​Figure22B shows the downregulation of PDL1 relative to the control in terms of MFI; in all cases, treatment of cells with siPDL1 containing SNALPs resulted in the reduction of PDL1 expression levels to 25% of the control. PDL1 downregulation was comparable whether siRNA was used in isolation (siPDL1) or formulated with mOX40L in two SNALPs (siPDL1 + mOX40L) or coformulated (siPDL1–mOX40L). The PDL1 downregulation was shown to be siRNA-specific, rather than a byproduct of transfection, as siNeg did not induce any downregulation (Supplementary Figure 3).

Open in a separate window

Figure 2

Dual-targeting SNALPs can efficiently transfect B16F10 melanoma cells in vitro and display minimal toxicity. B16F10 cells were cultured until 90% confluent before being pulsed with SNALP formulations (0.75 μg of each type of RNA) for 48 h at 37 °C. Cells were harvested and doubly stained with fluorescently labeled anti-mouse OX40L and PDL1 monoclonal antibodies. (A) Shows representative flow cytometry plots. The conditions are as follows: untransfected, siPDL1, mOX40L, mOX40L–siNeg (coformulation), mOX40L–siPDL1 (coformulation), and siPDL1 + mOX40L (mixture of two SNALPs). Quadrant gates were drawn based on isotype control antibody staining, percentage of cells in each quadrant is inset. (B) Shows the values obtained for PDL1 silencing, expressed as MFI percentage of control normalized to 100%. OX40L expression (MFI) is shown in (C). For all the graphs, error bars correspond to standard error of the mean (SEM). Significance was examined with one-way ANOVA multiple comparison test (Tukey’s); n = 3–8 repeats for each SNALP formulation. (D) To assess viability of B16F10 cells after being pulsed with SNALPs or RNA-free lipid particles an MTT assay was carried out. A 2-fold dilution series of test formulations was prepared and incubated with cells for 48 h at 37 °C. Error bars were drawn by standard error of the mean (SEM) average of n = 10, significance was tested with a two-way ANOVA: Sidak’s multiple comparison test. *,p < 0.05.

In terms of mOX40L transfection, as shown in Figure ​Figure22A untransfected cells and siPDL1 SNALPs-treated cells did not express OX40L. Upon transfection with SNALPs encapsulating mOX40L alone, coformulated with siNeg, coformulated with siPDL1, or treated with a mixture of SNALPs, 88.00, 96.70, 97.40, and 95.94% of the cells were induced to express OX40L. In terms of MFI, transfection of cells with mOX40L resulted in high expression of the protein and was independent of whether the mRNA was used in isolation or formulated with siPDL1 (Figure ​Figure22C). Simultaneous delivery of siRNA and mRNA was previously carried out in a lipid-based system using methods similar to those outlined in this manuscript. In this work it was found that the inclusion of mRNA or alternate polyanions aided the siRNA in silencing.³¹ In our system we observed comparable silencing in the presence or absence of mRNA. We speculate that this discrepancy maybe due to a number of factors. In the previous work, a lipidoid was used in place of an ionizable lipid. It is possible that the lipidoid binds to RNA with a higher affinity than ionizable lipid; thus, the release of siRNA is aided by the presence of a polyanion which reduces the affinity for siRNA by neutralizing some of the charge. Alternately, in our system, the RNA concentration was not titrated, and we may see more obviously an effect of the coloading at lower concentrations. Future studies may comprise of further optimization of nucleic acid ratios similar to work which has been carried out for constructs delivering multiple plasmids.³⁵

The viability of B16F10 melanoma cells after SNALPs transfection was assessed with a quantitative MTT assay. For this experiment, a 2-fold dilution series of lipid preparations ranging from 0.02 to 0.44 mM, with or without RNA was tested. As shown in Figure ​Figure22D, the viability of B16F10 cells was 100% for those cells that were transfected with SNALPs encapsulating RNA (+RNA) at a concentration of lipid that ranged from 0.02 to 0.1 mM. At 0.22 mM RNA-containing SNALPs, the B16F10 viability was 89.30%, whereas at a concentration of 0.44 mM, viability further decreased to 59.30%. The RNA-free lipid particles (−RNA) were more toxic than the formulated at the lowest dose (82.0% viability compared to 100%). At 0.05, 0.11, 0.22, and 0.44 mM, the viability was 77.10, 75.90, 59.50, and 46.13% respectively. Therefore, from this experiment, it can be concluded that the cell viability decreases as the concentration of the lipid rises and that the SNALPs encapsulating RNA are significantly less toxic than the lipids alone. As a reference, the concentration of SNALPs used during the transfection experiments was between 0.05 and 0.11 mM, depending on nucleic acid content, which is within the nontoxic range. Taken together, these data show that simultaneous upregulation of OX40L and downregulation of PDL1 can be achieved using a single SNALP system with no detectable interference and that SNALPs are relatively nontoxic on a cellular level. To validate the in vivo potential of SNALPs, B16F10 cells were also transfected with luciferase expression mRNA (mLuc) in the SNALP system in the presence of serum as previously described. The presence of serum reduced the expression of luciferase by 51% (Supplementary Figure 4); however, cells were still readily transfected resulting in the high expression of luciferase protein.

SNALPs Can Transfect a Macrophage Cell Line Resulting in Expression of OX40L and Activation

Macrophages comprise a large proportion of tumor-associated cells, representing up to 50% tumor weight.³⁶ Furthermore, they are highly phagocytic, thus representing an additional target for transfection following in situ administration. To model the effect of transfection on phagocytes/APCs, a J774 mouse macrophage cell line was selected. The J774 macrophage line was incubated with SNALPs containing either siPDL1–mOX40L or siNeg–mLuc. LPS was included as a positive control for macrophage activation. As shown in Figure ​Figure33A, treatment of J774 with siNeg–mLuc containing SNALPs or LPS resulted in a 2- or 4-fold upregulation of PDL1 respectively. Treatment of cells with siPDL1–mOX40L containing SNALPs negates this upregulation, and PDL1 levels remain comparable to the untreated control. Consistent with the results obtained for B16F10 cells, OX40L expression could be induced only by SNALPs containing siPDL1–mOX40L, though the relative MFI was lower than that obtained for B16F10 cells (Figure ​Figure33B). To test whether SNALPs can upregulate the expression of maturation markers, the relative expression levels of CD80 and CD86 were tested (Figure ​Figure33C,D). Cells receiving either SNALP formulation or LPS had a CD80 expression 1.75-fold higher than the untreated control and were not significantly different from each other. Likewise, CD86 was also upregulated by both SNALPs to a similar extent, though LPS was more potent in this regard. Combined, this suggests that siPDL1–mOX40L SNALPs may be able to activate macrophages while also inhibiting the upregulation of PDL1 and inducing expression of OX40L.

Open in a separate window

Figure 3

SNALPS can transfect J774 macrophage in vitro resulting in activation. To assess effect of SNALPs on macrophage/APC populations J774 cells at 100 000/per well were cultured in a 12-well plate prior to addition of SNALPs loaded with mOX40L–siPDL1 or mLuc–siNeg (1 μg/well) as a negative control. Lipopolysaccharide (LPS) (1 μg/mL) was added as a positive maturation control. Cells were harvested after 48 h at 37 °C and stained with fluorophore labeled anti mouse PDL1 (A) and OX40L (B) monoclonal antibodies or anti-mouse CD80 (C) or CD86 (D). Cells were acquired on FACs Calibur flow cytometer. To analyze staining, cells were first gated by FSC/SSC profile before the relevant marker was assessed. Error bars correspond to the SD statistical analysis was carried out using Mann–Whitney test. *, p < 0.05.

Intratumoral Administration of SNALPs Leads to Transfection within the Tumor

Following the positive in vitro transfection, the biodistribution and in vivo transfection was assessed. To perform this, SNALPs were formulated containing mLuc and a lipid intercalating dye (DiR) to measure transfection and distribution, respectively. As siRNA and mRNA are both active in the cytosol and as the in vitro data shows they do not hinder the activity of each other, mRNA was used in isolation as a measure of transfection. Two injection routes were tested, the more clinically acceptable intravenous (i.v.) route, following which the particles would reach the tumor through the enhanced permeation and retention effect (EPR), and the direct intratumoral (i.t.) route, which had given positive results in the preliminary study. Mice were imaged at 4 h postinjection based on previous reports on mRNA in vivo transfection.³⁷ Representative images following whole body imaging are shown in Figure ​Figure44A. Administration of SNALPs i.t. led to strong DiR and luciferase signal from the tumor area, with the two signals corresponding in both location and relative intensity. Bioluminescence signals could be detected from the liver region in the i.v. group when the mice were imaged ventrally but not dorsally. In contrast, there is minimal DiR signal detected in the same group. This is due to the inability of fluorescence to penetrate the animal from deeper tissues. Results obtained from ex vivo analysis of the individual organs were in general agreement with the whole body imaging results (Figure ​Figure44B). Using the i.t. injection route, both luciferase and DiR signals were detected in the tumor. Luciferase expression within the tumor followed the DiR distribution pattern. As expected, i.v. administration resulted in a more disseminated biodistribution with organs such as the intestine and lungs giving positive DiR signals over background. Whether this was due to organ deposition of SNALPs, or SNALPs remaining in the vasculature, it could not be determined. It should be noted that no luciferase transfection of these organs was detected. The organs giving the strongest DiR and luciferase signals following i.v. injection were the liver and the spleen. The signals obtained for luciferase and DiR were then normalized per organ and are shown in Figure ​Figure44C,D. Quantitative transfection data (Figure ​Figure44C), in agreement with ex vivo organ imaging, showed that luciferase signals were highest in the tumor of mice after i.t. injection of SNALPs, while no signals were detected after i.v. administration. The i.v. administration resulted in the highest signals in the liver and the spleen (p > 0.05). Organ biodistribution quantitative data (Figure ​Figure44D) further confirmed that SNALPs were retained in the tumor following i.t. injection and in liver/spleen following i.v. injection (liver > spleen, p < 0.05). We did not observe tumor targeting following i.v. administration which had been expected due to the EPR effect.³⁸ This could have perhaps been foreseen as SNALPs of a similar composition have been used to deliver RNA to the liver, mediated by Apo E targeting.^39,40 There are recent studies showing that the modification of the lipid components of the nanoparticle can result in selective organ targeting, for example, the inclusion of permanently charged cationic lipids (DOTAP) results in strong splenic targeting; this approach has not been tested for tumors but may make for an interesting future study.⁴¹ The shape of the particles has likewise been shown to impact tumor uptake following i.v. administration, with star-shaped particles showing a higher uptake in tumors.⁴² However, what we understand of the EPR effect and passive tumor targeting is undergoing a radical shift with a growing body of evidence suggesting that a particulate nature alone is unsuitable for tumor targeting in the clinic.⁴³ SNALP accumulation in the spleen, a secondary lymphoid tissue, following i.v. injection could potentially be used to enhance systemic immune responses. Improved tumor targeting may be observed following i.v. injection if a targeting moiety is added as has been described.⁴⁴ Due to this observation, we opted to continue with the local, intratumoral approach. Intratumoral therapy has been gaining prominence recently, with a number of high impact, preclinical studies having demonstrated efficacy.^14,45 Moreover, there are several clinical trials either underway or having been completed assessing the suitability of this route as a clinical option for delivering mRNA or pDNA.⁴⁶⁻⁴⁹ It is based on the premise that the tumor itself can serve as a “vaccine” (i.e., a source of antigen) should the i.t. immune stimulation be potent enough (so-called in situ vaccination).⁵⁰ While this is an unconventional route, should it prove efficacious and technically realistic, it represents a promising approach allowing for delivery of concentrated immunotherapeutics to the tumor site potentially increasing potency and circumventing systemic toxicity.

Open in a separate window

Figure 4

Biodistribution and in vivo expression of mLuc following administration of RNA-loaded, DiR-labeled SNALPs via two different routes. C57BL/6 (n = 4 per group) were implanted with 1 × 10⁶ B16F10. On day 13 postimplantation, mice were injected with SNALPs formulated with 1% DiR and containing 13 μg of mLuc per mouse either i.v. or i.t. in 100 or 50 μL volume, respectively. One mouse was left untreated to serve as a negative control. (A) At 4 h postinjection, whole body imaging was carried out to assess both luminescence and DiR fluorescence (ex.745 nm, em. 800 nm) using an IVIS Spectrum in vivo imaging system. (B) Following imaging, mice were sacrificed, and organs were extracted and imaged as described above. For each of the images a single representative is shown. Organ images were analyzed using Living image software; DOI were drawn around each organ manually and both average radiance for luciferase expression (C) and average radiant efficiency for DiR (D) were plotted. In each case, the mean ± SD of the group is shown. Statistical analysis was carried out using a Mann–Whitney test. ns, nonsignificant; *, p < 0.05.

Both Immune and Nonimmune Cells Contribute to Uptake of SNALPs following i.t. Administration

To test cellular distribution of SNALPs within the tumor and TDLN, mice bearing B16F10 tumors were i.t. injected with fluorescently labeled (DiD) SNALPs containing nonspecific RNA. After 24 h, cells obtained from tumors and TDLN were examined by flow cytometry. Interestingly, despite being undetectable using IVIS whole body or ex vivo organ imaging (data not shown), a substantial amount of SNALPs signal was detected in the TDLN with 74–75% of B220+ (B cells) and CD11c+ (APCs) cells showing positive association (Figure ​Figure55A,C). Only 27% of the CD3+ cell (T cells) population showed association with SNALPs. In this study, the MFI signal represents the number of SNALPs associated with a cell population on a per cell basis. As shown in Figure ​Figure55C, the CD11c+ cells had the highest MFI (mean 714), approximately 8-fold higher than B220+ cells (mean 90), suggesting that these cells had the highest affinity for SNALP association.

Open in a separate window

Figure 5

SNALPs are distributed to both immune and nonimmune cells in the tumor and TDLN following i.t. injection. C57/Bl6 (n = 4) were implanted with B16F10 cells subcutaneously. Once palpable, tumors formed SNALPs formulated with 1% DiD and containing 13 μg of siNeg were injected i.t., one mouse was left uninjected to serve as a control. At 24 h postinjection mice were culled, and TDLN and tumors were extracted. A single-cell suspension from each of the tissues was obtained by physical maceration. (A, C) Lymph node cells were stained with antimouse B220, CD3, or CD11c. In each case, the percentage of cells of the parent population positive for DiD SNALPs is shown in (A), and the DiD MFI of the whole cell population is shown in (C). (B, D) Cells obtained from tumors were stained with anti-mouse CD45 and PI. The CD45 positive and negative population positive for DiD SNALPs as a percentage of viable is shown in (B) the DiD MFI of the whole cell population is shown in (D). Error bars correspond to SD.

Among the viable cells extracted from the tumor, ~32% of the CD45+ (leukocytes) population was associated with SNALPs, and ~49% of the SNALPs was associated with other cell populations (CD45- including B16F10 cells) (Figure ​Figure55B). The intensities of the SNALPs uptake, expressed as the MFI, were 637 and 265, for the CD45+ and the CD45– populations, respectively (Figure ​Figure55D). The data combined suggest that in terms of cell proportions CD45– cells were responsible for the majority of the cellular SNALP association. However, they became associated with the SNALPs to a lesser extent than CD45+ cells. It could be speculated that this discrepancy maybe due to the relative abundance of cells, with CD45– cells outnumbering CD45+ cells.

Following i.t. administration, SNALPs were associated with both immune and nonimmune cells (CD45+ leukocytes vs CD45– cells), though it should be noted we did not establish whether transfection was achieved equally in both. There is an ongoing debate as to whether the tumor itself is the target of immune checkpoint blockade or whether it is the “host” immune cells.^51,52 The observation of SNALP uptake in CD45+ cells in the tumor microenvironment may indicate that SNALPs can also be used to target tumor-associated immune cells. Indeed, we have shown in vitro that SNALPs can transfect a macrophage cell line. To capitalize on this, SNALPs may also be developed to target tumor-associated macrophages delivering mRNA/siRNA to switch them from the nonprotective M2 to the protective M1 phenotype.⁵³

Synthesis of mRNA

OX40L mRNA (mOX40L) was synthesized from a mouse TNFSF4 ORF mammalian expression OX40L plasmid. Luciferase mRNA (mLuc) was synthesized from Luciferase-pcDNA3 plasmid. The plasmid was first linearized with XbaI restriction enzyme, and then the mRNA was synthesized according to HiScribe T7 ARCA mRNA Kit (with tailing) protocol. Finally, mRNA was purified by spin-column with Monarch RNA Clean up Kit. The concentration of mRNA was measured by NanoDrop One (Thermo Fisher Scientific). RNA was assessed by agarose electrophoresis gel on a 1% agarose gel. Agarose and 1× (10 mM) sodium borate (SB) buffer were microwaved until the agarose was completely dissolved, and the solution became clear. Formaldehyde (37%) was added to a final concentration of 0.62%. The gel was transferred into running tank filled with 1× SB buffer, and the Millennium RNA Markers ladder plus the RNA sample was mixed with Gel red and formamide before being loaded into the gel. Finally, the gel was run at 225 V for 40 min and was visualized by illumination under ultraviolet light (GelDoc, Bio-Rad).

Preparation of SNALPs

To make the SNALPs for each condition, an ethanolic lipid (Supplementary Table 1) and aqueous (Supplementary Table 2) phases were prepared in separate tubes. After prewarming them at 60 °C for 3 min, 10 μL of the lipid phase was introduced into the aqueous phase. The liquid solution was rapidly pipetted up and down for 20 s, vortexed for 10 s, and incubated at 60 °C for 30 s. The process was repeated until all the lipid phase had been added. They were incubated for 1 h at 40 °C and flushed with dry N₂ to remove residual ethanol. For in vivo studies, SNALP buffer was exchanged using Amicon Ultra centrifugal filters (0.5 mL) 30 kDa (Merck Millipore) in accordance with manufacturer’s protocol. SNALPs were resuspended in 200 μL of 4-(2-hydroxyethyl) piperazine-1-ethanesulfonic acid (HEPES) buffer (pH ~7). The final mass ratio of ionizable lipid to RNA was 10:1 w/w. To produce fluorescently labeled SNALPs, DiD or DiR at a final molar percentage of 1% was included in the lipid phase. SNALPs were loaded with either nonspecific siRNA (siNeg), siRNA specific to PDL1 (siPDL1), luciferase mRNA (mLuc), or OX40L mRNA (mOX40L) and are named accordingly. When multiple RNAs are combined in a single SNALP, the formulation is named with a dash between the two RNA constructs (e.g., siPDL1–mOX40L). In transfection experiments where multiple SNALPs with differing constructs are utilized, a plus sign is added between the two RNA construct names (e.g., siPDL1 + mOX40L).

Physico-Chemical Characterization of SNALPs

SNALPs encapsulating different types of RNA were characterized in terms of size (z-average), polydispersity, and surface charge (z-potential) utilizing dynamic light scattering (DLS). In brief, SNALPs in citrate buffer were diluted with deionized water (1:10 v/v) and added to a disposable plain folded capillary zeta cell. Measurements were obtained at 25 °C in triplicate using Zetasizer Nano ZS (Malvern Instruments).

Annular Dark-Field Scanning Transmission Electron Microscopy (ADF-STEM) Imaging

SNALP formulation (5 μL) was drop-cast on to a graphene TEM grid. Annular dark-field scanning transmission electron microscopy (ADF-STEM) was carried out on a Tecnai Osiris TEM operated at 200 kV. STEM images were acquired with 50 pA beam current.

Determination of Total RNA Encapsulation Efficiency (EE%)

The encapsulation efficiency (EE%) was indirectly quantified with the Quant-iT RiboGreen assay according to manufacturers’ protocol and as described previously.⁶¹ Ribogreen is an RNA-detection agent, and due to its membrane impermeability when added to a SNALP formulation, it stains only external, nonencapsulated RNA. When a permeabilization agent, such as Triton X-100, is included, Ribogreen can stain interior as well as exterior RNA (total RNA). SNALPs were incubated with PBS or PBS + 0.4% Triton X-100 for 20 min. RNA standards comprising serial dilutions of siRNA in either PBS or PBS + Triton X-100 were prepared as measurement references. Following incubation, RiboGreen was added to each sample, and the fluorescence was measured at ex./em. 485/520 nm with a plate reader (BMG LABTECH, FLUOstar Omega). RNA concentration was interpolated from the relevant standard curve of siRNA diluted in either PBS or PBS plus Triton X-100 accordingly. The quantity of loaded RNA was determined by subtracting the values obtained in PBS (external RNA) from PBS + Triton X-100 (total RNA) as shown below in (eq 1). Encapsulation efficiency was then established based on the amount of RNA encapsulated relative to the preformulation quantity (eq 2). It should be noted that this method detects all RNA and cannot differentiate mRNA from siRNA.

1

2

Semiquantitative Assessment of RNA Loading Proportions

Since RiboGreen assay does not discriminate mRNA from siRNA, agarose gel electrophoresis was used to establish mRNA/siRNA loading proportions. A SNALP preparation was incubated with RNase H (20 U/mL) to degrade the unloaded RNA, after 20 min incubation at 37 °C, the enzyme was inactivated by heat (20 min at 65 °C) and by the addition of EDTA to a final concentration of 1.25 mM. The SNALPs were then treated with heparin (10% v/v) to dissociate the particle; liberated RNA was purified with Monarch RNA Cleanup Kit as per manufacturer’s instructions. Finally, RNA was run on an 1% agarose gel at 225 V for 40 min as described above (see the “Synthesis of mRNA” section). Gel images were analyzed using ImageJ, and manual regions of interest (ROI) were drawn around each band. Band intensities calculated by ImageJ were used to calculate siRNA to mRNA ratio both in the initial mix and the dissociated SNALP. Comparable ratio values would suggest siRNA and mRNA have comparable affinities to SNALP.

Cell Culture

Mouse melanoma B16F10, CT26, and macrophage J774 cells were maintained in an incubator (Sanyo, MCO-17AIC) at 37 °C, with 5% CO₂ and with a relative humidity of 5%. The cells were cultured in RPMI-1640 medium supplemented with 10% FBS, 1% penicillin–streptomycin, and 1% GlutaMAX. Cells were treated with 0.05% trypsin–EDTA when they achieved 90% confluency and were passaged every 2–3 days.

In Vitro SNALP Transfection of B16F10 Cell Lines

B16F10 cells were seeded in a 12-well plate at a density of 120 000/well with 1 mL of fully supplemented RPMI-1640 medium 24 h before the transfection and incubated at 37 °C and 5% CO₂. To perform SNALP transfection, a volume of SNALPs corresponding to 0.75 μg of RNA was diluted with 1 mL of serum-free RPMI-1640. The SNALP containing media (1 mL) was added a confluent well of B16F10 cells in a 12-well plate, after 4 h, the wells were supplemented with FCS to a final concentration of 10% v/v and incubated for a further 48 h at 37 °C and 5% CO₂. Transfected cells were stained with anti-PDL1-PE, anti-OX40L-APC, or with their respective isotype antibodies for 20 min at 4 °C before being washed 3 times with PBS and finally resuspended in PBS. Cells were acquired using a FACS Calibur flow cytometry (BD Biosciences). Relative OX40L expression is plotted as the obtained MFI. PDL1 expression was expressed as a percentage Mean Fluorescence Intensity (MFI) values from untransfected control.

SNALP Transfection of J774 Macrophages

To assess J774 transfection and activation, cells were maintained and cultured as described for B16F10. A volume of SNALPs corresponding to 1 μg was diluted in 1 mL of serum-free tissue culture media and added to a single well of confluent cells. After 4 h of incubation, FCS was added to a final concentration of 10% v/v; after 48 h, cells were harvested. J774 cells were transfected with SNALPs containing either mOX40L–siPDL1 or mLuc–siNeg as a negative control. An additional group received lipopolysaccharide (LPS) to a final concentration of 1 μg/mL as a stimulation control. Cells were harvested and stained with anti-PDL1-PE and anti-OX40L-APC, as well as anti-CD80-PE and anti-CD86-PE as markers of activation. Data were analyzed using FlowJo software (Treestar); cells were regated based on their FSC/SSC profile before the marker of interest was assessed. Marker expression was presented using their respective MFI values.

In Vitro Cytotoxicity Using MTT Assay

To assess the cytotoxicity of SNALPs, a 96-well plate was seeded with B16F10 (6000 cells/well), and cells were incubated at 37 °C and 5% CO₂ for 24 h until 90% confluency was achieved. SNALPs encapsulating RNA were added at a range of dilutions starting at 0.44–0.02 mM and incubated in complete media at 37 °C and 5% CO₂ for 48 h. In parallel, cells were also incubated with lipids without nucleic acid to assess the toxicity of the transfection reagent. Then, the 96-well plate was incubated for 4 h with 120 μL/well MTT working solution (5 mg/mL of MTT solution diluted 1:5 with tissue culture media). Subsequently, DMSO was added into each well, and the plate was incubated for 5 min at 37 °C. Finally, absorbance at 570 nm was measured with the plate reader, and cell viability was calculated by calculating absorbance as a percentage of the untreated cells.

Mice

Animal experiments were carried out in female C57BL/6 or BALB/c mice (6–8 weeks old, Envigo). All the experiments involving animals were previously approved by the local ethical committee and with the approval of the United Kingdom Home Office license and in accordance with the UKCCCR Guidelines (1998).

SNALP Biodistribution and In Vivo Transfection

To assess organ biodistribution and in vivo expression of mRNA, naïve female C57BL/6 mice (n = 4 per group) were bilaterally implanted with 10⁶ B16F10 cells. On day 13 postimplantation, mice were injected intravenously (i.v.) or intratumorally (i.t.) with DiR-labeled SNALPs loaded with mLuc (13 μg per mouse) in either 100 or 50 μL of HEPES buffered saline, respectively. One mouse was left untreated as a background control. At 4 h after SNALP injection, mice were injected subcutaneously with luciferin before whole body luminescence and fluorescence (ex. 745 nm; em. 800 nm) imaging on an IVIS Spectrum in vivo imaging system (PerkinElmer). Following whole body imaging, mice were sacrificed, and individual organs (brain, heart, lungs, stomach, liver, kidneys, and intestine) were imaged for both luminescence and fluorescence as described for whole body imaging. Data were analyzed using Living Image software (PerkinElmer). Fluorescence and bioluminescence values for individual organs were obtained by manually drawing regions of interest (ROI) around each organ prior to analysis. Data are expressed as the average radiance obtained per ROI for luciferase signals and average radiance efficiency per ROI for fluorescence signals.

Cellular Distribution of SNALPs in Solid Tumors and Tumor-Draining Lymph Nodes

To further assess the cellular distribution of SNALPs, naive female C57BL/6 mice (n = 4) were implanted with 10⁶ B16F10 cells. Once palpable, tumors were i.t. injected with DiD-labeled SNALPs (n = 3) corresponding to 13 μg of RNA per dose per mouse. At 24 h postinjection, tumors and tumor-draining lymph nodes (TDLN, inguinal on tumor bearing flank) were excised, and a single-cell suspension was prepared by physically macerating tumors in PBS through a 70 μm cell strainer. Tumor cells were stained with antimouse CD45 (leukocyte marker) and PI as viability dye. Cells obtained from the TDLN were stained with fluorescently labeled anti-mouse CD11c, B220, and CD3 to identify DCs, B cells, and T cells, respectively. Finally, cells were analyzed by flow cytometry as previously described. Data are presented as both percentage of cells showing positive association with SNALPs and relative MFI of the respective populations.

In Vivo Immunomodulatory Activity of SNALP Constructs and Survival Studies

C57BL/6 mice (n = 8 per group) were implanted with B16F10 cells as previously described. Once tumors are palpable, SNALPs containing 13 μg total RNA, produced as described, were i.t. injected at days 5, 7, and 11 after tumor implantation. Tumor size and mouse weight were monitored until the terminal humane end point was reached (tumor diameter 15 mm) at which point mice were euthanized. Tumors and TDLN were extracted and processed to obtain a single-cell suspension. Tumor cells were stained with anti-mouse CD4 and CD8 antibodies. TDLN cells were stained with anti-mouse CD4, CD8, CD44 (an antigen memory marker), and CD69 (an early activation marker) antibodies to identify CD4+/CD8+ cells and their activation status. A fixed quantity of precision count beads was added to establish absolute cell numbers obtained. Cells were acquired on a FACs Calibur Flow Cytometer, and data analysis was carried out using FlowJo. For tumor cells, data are presented as total CD4/CD8+ cells number per mg of tumor. TDLN cell data is presented as CD4/CD8+ cell number per TDLN and CD44+/ CD69+ cells as a percentage of CD4 or CD8+ cell population.

To monitor the impact of the SNALP treatment on long-term mouse survival, C57/BL6 mice (n = 9–10 per group) were implanted with B10F10 cells as described above. Mice were injected i.t. with formulations containing either: 6.5 μg of siPDL1, 6.5 μg of mOX40L or the combinatory SNALP (6.5 μg of siPDL1 + 6.5 μg of mOX40L). Tumors were measured every other day, and mice were culled at their humane end point (tumor length ≥ 15 mm, weight loss ≥ 10% of pretreatment body weight, or visible signs of distress). Mice which cleared the tumor were rechallenged with B16F10 cells as described for the first implantation, a control group of aged-matched mice which had not been exposed to B16F10 (naïve) was also included. Tumor growth was monitored until the humane end point was reached. As an alternate model, CT26 colon carcinoma cells (1 × 10⁶ per mouse) were implanted into the lateral flank of BALB/c (n = 7–8 per group). Once tumors had reached ca. 5 mm in diameter, mice were injected i.t. with siPDL1–mOX40L SNALPs containing 13 μg of RNA or buffer. The injections were repeated, and tumor growth was monitored as described for B16F10 model.

A.A.W. is the grateful recipient of a Maplethorpe Fellowship. K.A.J. acknowledges funding from the British Council (Newton Fund, 337313), Wellcome Trust (WT103913), and the Cancer Research UK King’s Health Partners Centre at King’s College London. Financial support is acknowledged from the European Commission under the Horizon 2020 Programme, by means of Grant Agreement No. 731019 (EUSMI). Images were drawn on BioRender.com.