Effect of the ratio of betamethasone to TNF-α siRNA co-encapsulated in solid lipid nanoparticles on the acute proinflammatory activity of the nanoparticles

Abstract

There is evidence that encapsulating glucocorticoids into nucleic acid-containing nanoparticles reduces the inflammatory toxicities of the nanoparticles. Herein, using betamethasone acetate (BA), a glucocorticoid, and a solid lipid nanoparticle formulation of siRNA, we confirmed that co-encapsulating BA into the siRNA solid lipid nanoparticles significantly reduced the proinflammatory activity of the siRNA nanoparticles in a mouse model. Using TNF-α siRNA, we then showed that the BA and TNF-α siRNA co-encapsulated into the solid lipid nanoparticles acted as a dual anti-inflammatory and synergistically reduced TNF-α release by mouse macrophages in culture following stimulation with lipopolysaccharide, as compared to solid lipid nanoparticles encapsulated with TNF-α siRNA or BA alone. Importantly, upon studying the effect of the ratio of BA and TNF-α siRNA on the proinflammatory activity of the resultant nanoparticles, we identified that BA and TNF-α siRNA co-encapsulated solid lipid nanoparticles prepared with a BA to TNF-α siRNA weight ratio of 2:1 induced the lowest proinflammatory cytokine production by macrophages in culture. This result was in comparison to nanoparticles prepared with BA to TNF-α siRNA ratios both higher and lower than 2:1 (i.e. 4:1, 1:1, and 0.5:1), and is likely due to differences in molecular interactions among the various components in the BA and TNF-α-siRNA co-encapsulated solid lipid nanoparticles at these ratios. Encapsulating glucocorticoids into siRNA-nanoparticles represents a viable strategy to reduce the proinflammatory activity of the nanoparticles; however, the ratio of the glucocorticoid to siRNA in the nanoparticles requires optimization.

1. Introduction

RNA interference (RNAi) is a biological process that regulates gene activity through the binding of small interfering RNA (siRNA) to complementary mRNA.^{1, 2} These double-stranded RNA sequences are 20 to 30 nucleotides long and have emerged in recent years as promising therapeutic agents, with the first RNAi-based therapy, Alnylam’s Onpattro® (patisiran), gaining approval by the U.S. FDA in 2018.^{3, 4} However, there are a number of challenges to the successful delivery of siRNA, especially when intravenously administered. In systemic circulation, siRNA is rapidly degraded by serum nucleases and quickly eliminated through glomerular filtration. Furthermore, siRNA, particularly longer sequences (i.e. greater than 30 bp) and those that have certain motifs, can activate the innate immune system.^{5, 6} Nanoparticle delivery systems, including lipid-based nanoparticles, have been implemented to overcome these limitations and facilitate successful delivery of siRNA to the cytosol of target cells by prolonging systemic circulation and improving cellular uptake of siRNA.^7–9 However, the design and composition of these nanoparticles must be carefully considered. In addition to the innate immune response that may be caused by siRNA, certain materials used to formulate the nanoparticles may also be immunogenic.¹⁰ For example, cationic lipids, which can improve siRNA encapsulation in nanoparticles and its internalization by target cells, may induce strong type I and type II interferons, as well as activate signal transducer and activator of transcription 1 (STAT1).¹¹ Data from published studies have shown that administration of a glucocorticoid immediately preceding the administration of siRNA-containing solid lipid nanoparticles can significantly reduce the acute inflammatory response induced by the siRNA formulation.^{12, 13} There are also previous reports that encapsulating a glucocorticoid, such as dexamethasone, into nucleic acid-containing liposomes and nanoparticles, including siRNA-containing nanoparticles, reduces the inflammatory activity of the resultant nucleic acid formulations.^{14, 15} For example, Yoon et al. (2016) reported that HEI-OC1 cells treated with polymeric nanoparticles co-encapsulated with dexamethasone and plasmid DNA induced significantly lower expression of inflammatory cytokines (IL-1β, IL-12, and IFN-γ) than dexamethasone-free nanoparticles.¹⁵ The authors also co-encapsulated dexamethasone and siRNA into polymeric nanoparticles, but did not report the effect of dexamethasone on the siRNA-nanoparticles’ ability to induce inflammatory cytokines.¹⁵

Previously, we developed a solid lipid nanoparticle formulation of TNF-α siRNA to treat chronic inflammatory diseases such as rheumatoid arthritis.¹⁶ This formulation, which makes use of a cationic lipid to complex/condense siRNA, was effective in treating mice with collagen-antibody induced arthritis that was unresponsive to methotrexate.¹⁶ However, when tested in healthy mice, these nanoparticles, at a high dose (i.e. 200 mg nanoparticles per kg), induced inflammatory cytokines (e.g. IL-6, MCP-1) in mouse plasma samples (Aldayel & Cui, unpublished data). In the present study, we tested the feasibility of encapsulating BA into the siRNA solid lipid nanoparticles to reduce their proinflammatory activity both in vitro and in vivo. We then examined the effect of the ratio of the BA to TNF-α siRNA in the nanoparticles on their proinflammatory activity and identified an optimal BA to TNF-α siRNA ratio. Surprisingly, the BA and TNF-α siRNA co-encapsulated solid lipid nanoparticles prepared with the highest BA to TNF-α siRNA ratio were not the least proinflammatory. Finally, we also evaluated the combined effect of BA and TNF-α siRNA in the solid lipid nanoparticles on reducing TNF-α release by mouse macrophages in culture following stimulation with lipopolysaccharide.

2. Materials and Methods

2.1. Materials, cells, and animals

Lecithin (refined, from soybean) was from Alfa Aesar (Ward Hill, MA). Cholesterol, lipopolysaccharide (LPS, from Salmonella enterica), and tetrahydrofuran (THF, HPLC grade) were from Sigma-Aldrich (St. Louis, MO). The 1,2-dioleoyl-3-trimethylammonium-propane (chloride salt) powder (DOTAP, >99%) was from Avanti Polar Lipids (Alabaster, AL). PEG2000-hydrazone-stearate (PHC) was synthesized using a previously published method.¹⁷ Fluorescently labeled siRNA (BLOCK-iT™ Fluorescent Oligo), control siRNA (Stealth RNAi™ siRNA Negative Control, Med GC), and diethyl pyrocarbonate (DEPC)-treated water were from Invitrogen (Carlsbad, CA). TNF-α siRNA was synthesized by Integrated DNA Technologies, Inc. (Coralville, IA). BA and prednisolone were from TCI America (Portland, OR). ELISA kits (IL-6, MCP-1, TNF-α) were from BioLegend (San Diego, CA). Ultrafiltration devices (Amicon, 100 K molecular weight cutoff (MWCO)) were from MilliporeSigma (Burlington, MA). Dialysis devices (Float-A-Lyzer® G2, 50 K MWCO) were from Spectrum Labs (Waltham, MA). All solvents were of HPLC grade and from Fisher Chemical (Waltham, MA).

Mouse J774A.1 macrophage cells were from the American Type Culture Collection (Manassas, VA) and grown in Dulbecco’s Modified Eagle Media (DMEM) from Gibco (Grand Island, NY) supplemented with 10% (v/v) fetal bovine serum (FBS) and penicillin (100 U/ml)/streptomycin (100 μg/ml), both also from Gibco.

Mice (female, BALB/c, 20 to 25 grams) were from Charles River Laboratories (Wilmington, MA). Animal studies were conducted following the U.S. National Research Council Guidelines for Care and Use of Laboratory Animals. The animal protocol was approved by the Institutional Animal Care and Use Committee at The University of Texas at Austin (UT Austin).

2.2. Preparation and characterization of solid lipid nanoparticles

2.2.1. Preparation of betamethasone acetate (BA) and siRNA co-encapsulated solid lipid nanoparticles (BA-siRNA-SLNs)

Solid lipid nanoparticles containing siRNA and BA were prepared as previously described with modifications.¹⁶ Briefly, DOTAP (2.5 mg in 680 μl of chloroform) was added dropwise to an siRNA in RNAse-free water solution (450 μl, 15 μg in a buffer containing 10 mM Tris–HCl and 1 mM EDTA in water, pH 7.5) under stirring. The DOTAP-siRNA mixture was briefly sonicated in a water bath sonicator and allowed to stir at room temperature for 30 min to 1 h, at which point 1.36 ml of methanol was added to create a clear solution. This solution was stirred at room temperature for an additional 30 min to 1 h. After this second period of stirring, 680 μl of chloroform was added to the solution to extract the DOTAP-siRNA complexes and separate the aqueous phase, which was discarded. Lecithin and cholesterol (3.2 mg and 1.6 mg, respectively, in 200 μl of chloroform), PHC (2 mg in 100 μl of chloroform), and BA (at various amounts in 100 μl of chloroform) were added to the remaining organic layer containing the DOTAP-siRNA complexes, mixed, and dried under nitrogen gas. The dried mixture was dissolved using tetrahydrofuran (THF, 500 μl) and added dropwise to RNAse-free water (5 ml) while stirring to form nanoparticles through nanoprecipitation. Solvent was evaporated through stirring under a chemical hood at room temperature for a minimum of 6 h. Once formed, the nanoparticles were collected by ultrafiltration, washed, and re-suspended to a desired concentration using DEPC-treated water or phosphate-buffered saline (PBS, pH 7.4, 10 mM). The size, polydispersity index, and zeta potential of the resultant nanoparticles were measured using a Malvern Zetasizer Nano ZS (Westborough, MA).

2.2.2. Determination of encapsulation efficiency of BA and siRNA in the nanoparticles

Using the filtrate collected during the ultrafiltration process, the amount of BA and siRNA within the nanoparticles was indirectly determined. To quantify the encapsulation efficiency of siRNA, nanoparticles were prepared using fluorescently-labeled siRNA (BLOCK-iT™ Fluorescent Oligo). A standard curve was prepared and the amount of siRNA within the collected filtrate was determined using a Synergy HT microplate reader (BioTek Instruments, Winooski, VT). This was subtracted from the total amount of siRNA added to the formulation to calculate the amount and percentage of siRNA contained in the nanoparticles. The amount of BA in the collected filtrate was quantified by mixing the filtrate with an equal volume of methanol, which was then analyzed with a reversed-phase high performance liquid chromatography (HPLC) method. An Agilent 1260 Infinity LC with an Agilent ZORBAX Eclipse Plus C₁₈ column (5 μm, 4.6 mm × 150 mm) was used for chromatographic separation. The column temperature was set at 30°C with a flow rate of 1.5 ml/min. The mobile phase consisted of acetonitrile and water (40:60, v/v). Similarly, the amount of BA in the filtrate was subtracted from the total amount of BA added during preparation and used to calculate the amount of BA in the nanoparticles and the encapsulation efficiency.

2.2.3. Transmission electron microscopy

Morphological characterization of BA-siRNA-SLNs was performed using transmission electron microscopy (TEM). One drop of nanoparticle suspension (5 μl) was applied to carbon-coated 400 mesh grids after activation, either stained with 2% uranyl acetate (pH 4.5) or left unstained, and allowed to dry prior to examination using an FEI Tecnai Transmission Electron Microscope (FEI, Hillsboro, OR).

2.2.4. In vitro stability of BA-siRNA-SLNs in a simulated biological medium

The stability of BA-siRNA-SLNs in a simulated biological medium was assessed by adding aliquots of the nanoparticle suspension to 10% FBS in PBS (10 mM, pH 7.4) and allowing the mixture to incubate at 37°C and 150 rpm in a MaxQ™ 5000 Floor Shaker Incubator (Thermo Fisher Scientific) over time. The size of the nanoparticles was measured immediately after mixing with the medium, and then after 2, 4, 6, and 17 h of incubation using a Malvern Zetasizer Nano ZS. The BA-siRNA-SLNs were prepared with a BA to siRNA (Medium GC negative control) ratio of 4:1 (w/w).

2.2.5. In vitro release of BA and siRNA from solid lipid nanoparticles

The release of siRNA and BA from BA-siRNA-SLNs was evaluated using dialysis devices (Float-A-Lyzer®, 50 K MWCO, 5 ml). Nanoparticles collected using ultrafiltration were re-suspended with PBS (10 mM, pH 7.4) to 5 ml, which was then added inside the dialysis device. As a control, fluorescent siRNA (BLOCK-iT™ Fluorescent Oligo) and BA were added together at equivalent amounts to those contained within the BA-siRNA-SLNs (60 μg or less of BA) to 5 ml of PBS and placed inside a separate dialysis device. For these studies, BA was diluted either from a 60 mg/ml solution in dimethyl sulfoxide (DMSO) or from a 6 μg/mL solution in PBS. Each device was placed in a 50 ml conical tube containing 25 ml of PBS (10 mM, pH 7.4) as release media to maintain sink conditions and placed in a MaxQ™ 5000 Floor Shaker Incubator at 37°C and 150 rpm. The reported water solubility of BA is 16.4 μg/ml.¹⁸ Samples were collected in triplicate at designated time points and analyzed. To quantify siRNA release, samples were read using a Synergy HT microplate reader. Ethyl acetate was used to extract BA from the release samples, with prednisolone (10 μg) added as an internal standard. Extracts were dried under nitrogen, re-dissolved using a mixture of methanol and water (1:1) and analyzed by reversed-phase HPLC using the method described above.

2.3. Differential scanning calorimetry and X-ray diffraction

BA-TNF-α-siRNA-SLNs prepared with a BA to siRNA weight ratio of 0:1, 2:1, or 4:1 were dried using a Labconco FreeZone freeze dryer (Labconco, Kansas City, MO) without adding any cryoprotectant. Differential scanning calorimetry was performed on the nanoparticles and individual formulation components (i.e. lecithin, cholesterol, BA) using a Q20 DSC (TA Instruments, New Castle, DE). A heating rate of 10°C/min was used with a temperature range of 0°C–240°C. A sealed empty aluminum pan was used as a reference. X-ray diffraction (XRD) analysis was done using an R-Axis Spider with a Cu sealed tube source and a large, image plate detector (Rigaku, The Woodlands, TX). All XRD patterns were collected with a step size of 0.01 and 1 second/step over a 2θ range of 2 to 45.

2.4. In vitro effect of BA-TNF-α-siRNA-SLNs on TNF-α cytokine release by macrophages

To assess the effect of co-encapsulating BA with TNF-α siRNA in the solid lipid nanoparticles on their ability to reduce TNF-α release by macrophages, J774A.1 mouse macrophages were seeded into a 96-well plate (7,000 cells per well) in DMEM. After overnight incubation, the cells were treated with BA-TNF-α-siRNA-SLNs (prepared at a BA to siRNA ratio of 4:1), TNF-α-siRNA-SLNs, or BA-SLNs (prepared with an equivalent amount of BA as in the BA-TNF-α-siRNA-SLNs). Cells were treated with the nanoparticles (300 ng/well of siRNA and 900 ng/well of BA) overnight (37°C, 5% CO₂) and then stimulated with LPS (100 ng/ml) for 24 h. As controls, cells were left untreated (i.e. only stimulated with LPS) or untreated and unstimulated with LPS. Cell culture medium was analyzed for TNF-α content using ELISA, and MTT assay was performed to measure cell viability.

2.5. In vitro optimization of BA to siRNA ratio in BA-TNF-α-siRNA-SLNs

To determine the minimum amount of BA needed in the nanoparticle formulation relative to the amount of TNF-α siRNA for inhibition of proinflammatory cytokine production, BA-siRNA-SLNs were prepared using a constant amount of TNF-α siRNA but varying amounts of BA (i.e. BA to siRNA ratios of 4:1, 2:1, 1:1, 0.5:1, or 0:1). J774A.1 cells were seeded into a 96-well plate (20,000 cells per well) in complete DMEM and, after overnight incubation (37°C, 5% CO₂), treated with BA-TNF-α-siRNA-SLNs (siRNA concentration, 2 μg/well) for 4 h. As controls, cells were stimulated with LPS (100 ng/ml) or left untreated and unstimulated. Cytokine levels in culture medium were assessed using ELISA, and MTT assay was performed to measure cell viability.

2.6. In vivo study of BA-siRNA-SLNs to reduce acute inflammatory response

To evaluate the effect of BA in the solid lipid nanoparticles on reducing acute proinflammatory cytokine production, healthy mice were intravenously injected into the tail vein with one of the following, at a dose of 0.5 mg/kg siRNA: BA-siRNA-SLNs (prepared with Med GC Negative Control siRNA or TNF-α siRNA), siRNA-SLNs (prepared with Med GC Negative Control siRNA or TNF-α siRNA), or left untreated. As a control, one group of mice receiving siRNA-SLNs was also intraperitoneally injected with betamethasone sodium phosphate (BSP, at a molar equivalent dose to the BA in the BA-siRNA-SLNs) immediately before intravenous injection of the siRNA-SLNs. Mice were sacrificed at 2 h post-injection, and plasma samples were collected for cytokine analysis using ELISA.

2.7. Statistical analysis

Statistical analyses for all studies, except the in vivo studies, were performed using a single-factor ANOVA followed by a two-sided t-test, with a p value < 0.05 considered significant. Analyses of cytokine levels in the two in vivo studies were conducted by performing a Kruskal–Wallis test by ranks, followed by a Mann-Whitney U test. A Grubb’s test was also applied to the in vivo data. All statistical analyses were performed using Excel (Microsoft, Redmond, WA) or GraphPad Prism (San Diego, CA).

3. Results and Discussion

3.1. Encapsulation of BA into siRNA-containing solid lipid nanoparticles (siRNA-SLNs) reduces the acute inflammatory response induced by the siRNA-SLNs in healthy mice

Data from previously published studies have demonstrated that administration of glucocorticoids prior to the administration of siRNA-containing solid lipid nanoparticles can successfully reduce the acute immunogenic response induced by the nanoparticles.^{12, 13} For example, in a study by Abrams et al., co-treatment with dexamethasone when administering siRNA-containing solid lipid nanoparticles reduced proinflammatory cytokine production by healthy mice in a dose-dependent manner, even at high doses of siRNA (e.g. 9 mg of nanoparticles/kg).¹² Though successful in limiting the acute inflammatory response, this approach has critical toxicity considerations regarding long-term systemic exposure to glucocorticoids.¹⁹ Premedication with dexamethasone has also been used in clinical trials to reduce acute inflammatory reactions to siRNA-solid lipid nanoparticles; however, in this study, dexamethasone administration was infrequent due to the monthly dosing schedule of the nanoparticles. For therapies where more frequent administration is required, this strategy may pose a significant risk to patients.²⁰ As such, we investigated the potential of encapsulating betamethasone acetate (BA), a glucocorticoid, directly into the siRNA-containing solid lipid nanoparticles previously developed in our laboratory to suppress their ability to induce acute proinflammatory responses.¹⁶ Yoon et al. (2016) previously co-encapsulated dexamethasone and plasmid or siRNA into polymeric nanoparticles. Here, they showed that HEI-OC1 cells treated with polymeric nanoparticles co-encapsulated with dexamethasone and plasmid DNA induced significantly lower expression of inflammatory cytokines than dexamethasone-free nanoparticles,¹⁵ although the authors did not report the inflammatory activity of the nanoparticles co-encapsulated with dexamethasone and siRNA.¹⁵

In the present study, using BA as a glucocorticoid and the siRNA-SLNs developed in our laboratory, we studied the effect of co-encapsulating BA in siRNA-SLNs on the proinflammatory activity of the nanoparticles. Based on preliminary studies, we determined that the maximum amount of BA that could be encapsulated into our solid lipid nanoparticle formulation was approximately three times the amount of siRNA currently encapsulated in the formulation (i.e. approximately 45 μg BA to 15 μg siRNA), with BA added during preparation of the formulation at an amount four times that of siRNA (i.e. 60 μg BA and 15 μg Stealth RNAi™ siRNA Negative Control, Med GC). Therefore, we proceeded with our initial studies using a 4 to 1 (w/w) BA to siRNA ratio for preparation of our nanoparticles, yielding a final ratio of 3 to 1 (w/w) BA to siRNA, which has been previously demonstrated as effective at reducing acute proinflammatory responses to siRNA-containing lipid nanoparticles.¹² The inclusion of BA in the siRNA-SLNs did not significantly affect the encapsulation efficiency of siRNA, nor the particle size, polydispersity index, and zeta potential of the resultant nanoparticles (Table 1). Shown in Figure 1A is a representative TEM image of the BA-siRNA-SLNs. At the BA to siRNA ratio used, the encapsulation efficiency of BA in the resultant solid lipid nanoparticles was 76.0 ± 0.7%. The burst release of BA from solid lipid nanoparticles was minimal (~5%), followed by sustained slow release over one week (Figure 1B). Furthermore, the inclusion of BA in the nanoparticles did not alter the release profile of siRNA (Figure 1B), i.e. minimal burst release as previously reported.¹⁶ In vivo, it is expected that the release profiles of BA and siRNA from the BA-siRNA-SLNs will likely be different from the in vitro profiles due to differences in the release medium (i.e. PBS vs. blood and interstitial fluid), as well as the presence of enzymes to help degrade and/or metabolize the lipids in the nanoparticles. The resultant BA-siRNA-SLNs were did not significantly aggregate in a simulated biological medium (i.e. 10% FBS in PBS, pH 7.4, 10 mM) at 37°C (Supplemental Figure S1), indicating that the nanoparticles would not extensively aggregate upon i.v. injection as well.

Table 1.

Particle size (diameter), polydispersity index (PDI), and encapsulation efficiency of BA-siRNA-SLNs and siRNA-SLNs prepared with either BLOCK-iT™ Fluorescent Oligo siRNA or Medium GC Negative Control siRNA. BA was added at a weight ratio of 4 to 1 with siRNA. As a control, BA-SLNs were prepared with an amount of BA equivalent to that in the BA-siRNA-SLNs. Data are mean ± S.D. (n = 6–21 for size and PDI, 4–5 for zeta potential, and 3 for encapsulation efficiency). Statistical analyses did not reveal any significant difference among the nanoparticles in size, zeta potential, and encapsultion efficiency (p > 0.05).

	Size (nm)	PDI	Zeta potential (mV)	% siRNA encapsulation	% BA encapsulation
BA-siRNA-SLNs	132.8 ± 24.3	0.23 ± 0.06	50.3 ± 2.9	93.6 ± 2.6%	76.0 ± 0.7%
siRNA-SLNs	120.4 ± 29.8	0.22 ± 0.07	41.9 ± 20	94.7 ± 2.6%
BA-SLNs	107.4 ± 28.8	0.22 ± 0.017	42 ± 19.9		79.6 ± 3.5%

Figure 1.

A representative TEM image of BA-siRNA-SLNs prepared with a BA to siRNA ratio of 4:1 (i.e. 4:1 BA-siRNA-SLNs) (A) and the in vitro release profiles of BA and siRNA from the 4:1 BA-siRNA-SLNs in PBS (pH 7.4, 10 mM) (B). In A, the SLNs were unstained. In B, the siRNA was fluorescently-labeled, and the BA in the Free BA group was diluted from a BA in DMSO solution. Data are mean ± S.D. (n = 3).

To evaluate the ability of BA in the BA-siRNA-SLNs to reduce the production of proinflammatory cytokines following administration of the siRNA-containing solid lipid nanoparticles, healthy, female BALB/c mice were i.v. injected with BA-siRNA-SLNs or siRNA-SLNs or were left untreated. The nanoparticles were prepared with Med GC Negative Control siRNA, and the ratio of BA to siRNA in the BA-siRNA-SLNs during preparation of the nanoparticles was 4 to 1 (w/w), with a final encapsulated ratio of 3 to 1 (w/w). As an additional control, one group of mice were intraperitoneally injected with betamethasone sodium phosphate (BSP, a betamethasone salt with higher water solubility than BA) at a molar equivalent dose to the BA contained within the BA-siRNA-SLNs immediately before i.v. administration of the siRNA-SLNs. Mice were sacrificed 2 h after injection and plasma was collected to determine proinflammatory cytokine levels based on previously reported data from Abrams et al. that showed inflammatory cytokine levels in mouse plasma samples were highest 2 h after injection of an siRNA-containing lipid nanoparticle formulation, as compared to 6 h or 24 h.^{12, 13} As shown in Figure 2, the plasma collected from mice injected with the BA-siRNA-SLNs had significantly lower levels of IL-6, MCP-1, and TNF-α than that from mice injected with the BA-free siRNA-SLNs. Additionally, there were no significant differences in the plasma levels of proinflammatory cytokines between mice treated with BA-siRNA-SLNs and mice treated with BSP immediately prior to the BA-free siRNA-SLNs, demonstrating that the BA in our BA-siRNA-SLNs was as effective at reducing acute immune response as pre-treatment with betamethasone. Our approach of encapsulating the glucocorticoid into the siRNA-SLNs has advantages over pretreating with a separate injection of glucocorticoid. For example, only a single administration is needed for the BA-siRNA-SLNs, as compared to injections of glucocorticoid and nanoparticles separately. Additionally, it is expected that excluding the BA that is released from the BA-siRNA-SLNs before the nanoparticles are taken up by cells, the BA remaining in the siRNA-SLNs will mainly affect cells that take up the SLNs, thus avoiding or minimizing its effect on other cells and tissues.

Figure 2.

Proinflammatory cytokine levels, (A) IL-6, (B) MCP-1, (C) TNF-α, in the plasma samples from healthy mice receiving either 4:1 BA-siRNA-SLNs or siRNA-SLNs (with or without co-administration of betamethasone sodium phosphate, BSP). BA was added at a ratio of 4 to 1 with siRNA, which was the Medium GC negative control siRNA. All measurements were normalized to the untreated group. Data are presented as mean ± S.D. (n = 5 for BSP + siRNA-SLNs treated group, n = 9 for untreated group, n = 13 for BA-siRNA-SLNs and siRNA-SLNs treated groups; * p <0.05, ** p <0.01, *** p <0.001).

3.2. Solid lipid nanoparticles containing both BA and TNF-α siRNA are more effective than solid lipid nanoparticles containing only BA or TNF-α siRNA at reducing TNF-α production by macrophages

The above studies were completed using a non-functional negative control siRNA. To evaluate the effect of including BA in siRNA-SLNs prepared with functional siRNA, we prepared BA-siRNA-SLNs with TNF-α siRNA and evaluated their ability to influence TNF-α production by macrophages in culture. To do this, we prepared solid lipid nanoparticles containing BA and TNF-α siRNA (BA to TNF-α siRNA ratio, 4:1), TNF-α siRNA only, or BA only, and treated J774A.1 mouse macrophages with them. Cells were stimulated with LPS to induce the production of TNF-α. As shown in Figure 3, both BA-SLNs and TNF-α-siRNA-SLNs significantly reduced TNF-α production by macrophages stimulated with LPS, and the effects of the BA-SLNs and the TNF-α-siRNA-SLNs were not significantly different from one another (p = 0.398). However, the solid lipid nanoparticles containing both BA and TNF-α siRNA resulted in a significant decrease in TNF-α release as compared to nanoparticles containing either entity alone (Figure 3), to a level that was not significantly different from that of healthy cells. In fact, the coefficient of drug interaction (CDI) between the BA and TNF-α siRNA was 0.702, indicating that BA and TNF-α siRNA co-encapsulated in the solid lipid nanoparticles had a synergistic effect on reducing TNF-α release by the macrophages,²¹ which may be attributed to the two different mechanisms through which BA and TNF-α siRNA work to reduce TNF-α production (i.e. transcription vs. translation).^{1, 22} Glucocorticoids such as betamethasone are known to reduce TNF-α production by affecting gene expression at the transcriptional level through complexation with the cytosolic glucocorticoid receptor and subsequent binding to glucocorticoid responsive elements (GRE) in the promoter region of DNA sequences and/or interaction with other transcription factors, such as NF-κB.^{22, 23}

Figure 3.

Comparison of solid lipid nanoparticles containing BA alone, TNF-α siRNA alone, or both BA and TNF-α siRNA in their ability to inhibit TNF-α production by J774A.1 mouse macrophages stimulated with LPS. Cytokine levels were assessed with ELISA. Data are presented as mean ± S.D. (n = 8, *** p <0.001).

3.3. Optimization of the BA to TNF-α siRNA ratio in solid lipid nanoparticles

It is known that the composition of an siRNA sequence plays a critical role in its inherent immunogenicity.²⁴ In a study by Sioud, TNF-α siRNA sequences were shown to be particularly immunogenic as compared to a number of other siRNAs.²⁵ Our choice of a 4 to 1 BA to siRNA weight ratio in the above studies was in part based on previous reports for a lipid nanoparticle formulation containing an unknown siRNA sequence and the glucocorticoid was administered separately from the siRNA nanoparticles.^{12, 13} Due to its immunosuppressive properties, the amount of BA incorporated into the formulation should be kept at the lowest amount that sufficiently reduces the acute inflammatory response. Therefore, we sought to identify the optimal ratio of BA to TNF-α siRNA necessary to reduce acute proinflammatory cytokine production by the TNF-α-siRNA-solid lipid nanoparticle formulation. To do this, solid lipid nanoparticles were prepared using various weight ratios of BA to siRNA. As mentioned above, data from a pilot study showed that it was not possible to prepare BA-encapsulated siRNA-solid lipid nanoparticles at a BA to siRNA ratio higher than 4 to 1 (w/w) without reducing the amount of siRNA in the nanoparticles, likely due to limitations on the encapsulation efficiency of BA in the solid lipid nanoparticles. Furthermore, data in Figure 3 demonstrated that solid lipid nanoparticles prepared with a BA to TNF-α siRNA ratio of 4 to 1 were effective at reducing TNF-α production by macrophages in culture. Therefore, BA to TNF-α siRNA weight ratios equal to and lower than 4 to 1 were used to prepare BA-TNF-α-siRNA-SLNs, and their ability to reduce inflammatory cytokine production by J774A.1 macrophages in culture was evaluated. The physical properties of the various BA-TNF-α-siRNA-SLNs are shown in Table 2. All nanoparticles were similar in size, polydispersity index, and zeta potential. As shown in Figure 4, the 2 to 1 BA to siRNA ratio reduced the production of IL-6 and TNF-α by J774A.1 macrophages to the lowest levels, which were not significantly different from healthy, unstimulated cells. For all other ratios tested, IL-6 and TNF-α levels in culture medium were significantly higher than that in healthy, unstimulated cells (Figure 4). Therefore, BA-TNF-α-siRNA-SLNs prepared with BA and TNF-α siRNA at a weight ratio of 2 to 1 were used for further studies in healthy mice to identify the extent to which the BA in the TNF-α-siRNA-SLNs can inhibit the acute proinflammatory activity of the TNF-α-siRNA-SLNs. As previously done with BA-siRNA-SLNs prepared with a BA to siRNA ratio of 4:1, the BA-TNF-α-siRNA-SLNs prepared with a BA to TNF-α siRNA ratio of 2:1 were further characterized. Shown in Figure 5A is a representative TEM image of the BA-TNF-α-siRNA-SLNs, and the in vitro release profiles of BA and TNF-α siRNA from the nanoparticles are shown in Figure 5B. The nanoparticles were largely spherical, and as expected, appeared smaller than the particle size measured by dynamic light scattering (Table 2). Again, the release of TNF-α siRNA from the nanoparticles was slow and limited, but the release of BA from the nanoparticles was much faster and more extensive.

Table 2.

Size (diameter), polydispersity index, and zeta potential of BA-TNF-α-siRNA-SLNs prepared at differing BA to siRNA weight ratios. Data are mean ± S.D. (n = 26–39 for size and PDI, 6–17 for zeta potential). Statistical analyses did not reveal any significant difference among the nanoparticles in size and zeta potential (p > 0.05).

BA to TNF-α-siRNA ratio (w/w)	Size (nm)	PDI	Zeta Potential (mV)
4 to 1	118 ± 24.9	0.2 ± 0.05	41.1 ± 17.6
2 to 1	115 ± 27.2	0.21 ± 0.06	40.9 ± 17.6
1 to 1	109.7 ± 20.3	0.22 ± 0.06	39.4 ± 18.8
0.5 to 1	110.7 ± 19.8	0.24 ± 0.075	45.2 ± 15.6
0 to 1	114.8 ± 22.1	0.2 ± 0.09	37.3 ± 16.6

Figure 4.

Comparison of solid lipid nanoparticles prepared with different ratios of BA to TNF-α siRNA in their ability to stimulate proinflammatory cytokine production by J774A.1 mouse macrophages. (A) IL-6 cytokine levels corrected to healthy, untreated cells (designated by the lower dashed line), (B) TNF-α levels corrected to healthy, untreated cells (designated by lower dashed line). The upper dashed lines represent the levels of cytokines released by J774A.1 cells stimulated with LPS. Data are mean ± S.D. (n > 3, * p <0.05, ** p <0.01, *** p <0.001 as compared to LPS stimulated cells).

Figure 5.

A representative TEM image of 2:1 BA-TNF-α-siRNA-SLNs (A) and in vitro release of betamethasone (BA) and siRNA (fluorescently-labeled BLOCK-iT™ Fluorescent Oligo) from 2:1 BA- siRNA-SLNs in PBS (pH 7.4, 10 mM) (B). In A, the SLNs were stained with uranyl acetate. In B, the BA in the Free BA group was diluted from a BA in PBS solution, and data are mean ± S.D. (n = 3).

To evaluate the effectiveness of co-encapsulating BA and TNF-α siRNA in solid lipid nanoparticles at reducing the proinflammatory activity of the solid lipid nanoparticles, BA-TNF-α-siRNA-SLNs prepared at a ratio of 2:1 were i.v. injected into healthy BALB/c mice. Plasma was collected 2 h after the injection and analyzed for levels of proinflammatory cytokines (i.e. IL-6, MCP-1, and TNF-α), which are presented in Figure 6. Mice in control groups were left untreated or i.v. injected with TNF-α-siRNA-SLNs containing an equivalent dose of TNF-α siRNA only. The cytokine levels in the plasma of mice treated with the BA-TNF-α-siRNA-SLNs are similar to healthy, untreated mice (not significant for MCP-1). For mice treated with the TNF-α-siRNA-SLNs, a large variation in the cytokine levels was seen (Figure 6). Though not significantly different, the mean cytokine response in the BA-TNF-α-siRNA-SLNs-treated mice was lower than that in mice treated with the TNF-α-siRNA-SLNs (1.8-fold for IL-6 and 2.3-fold for MCP-1). Plasma TNF-α levels were also assessed in all treatment groups, but no notable differences were seen (data not shown), possibly due in part to the therapeutic effect of the TNF-α siRNA. Overall, it appeared that the acute proinflammatory cytokine responses of healthy BALB/c mice to our TNF-α siRNA-solid lipid nanoparticles were highly variable when assessed at 2 h. However, encapsulation of BA into the TNF-α siRNA-solid lipid nanoparticles rendered them less proinflammatory and less variable, further confirming the feasibility of reducing acute inflammatory adverse events induced by siRNA-solid lipid nanoparticles by encapsulating a potent glucocorticoid, such as betamethasone, into the nanoparticles. We expect that the same strategy would be effective for other siRNA nanoparticle formulations and for glucocorticoids other than betamethasone. However, for different nanoparticles with different siRNA sequences, when an alternative glucocorticoid is used (i.e. a glucocorticoid other than betamethasone acetate), the ratio of the glucocorticoid to the siRNA will likely require optimization to identify the minimum effective amount of glucocorticoid needed in the specific nanoparticles.

Figure 6.

Proinflammatory cytokine levels in plasma from healthy mice receiving either 2:1 BA-TNF-α-siRNA-SLNs or TNF-α-siRNA-SLNs. Data are mean ± 95% confidence intervals (n = 9 for untreated group and 2:1 BA-TNF-α-siRNA-SLNs treated group, n = 12 for TNF-α-siRNA-SLNs treated group). The n.s. means not significant.

3.4. Potential mechanism underlying the optimal weight ratio of BA and TNF-a siRNA in the BA-siRNA-SLNs

In Figure 4, we found that solid lipid nanoparticles prepared with BA and TNF-α siRNA at a 2 to 1 weight ratio resulted in the lowest levels of cytokine induction, even lower than those prepared with a BA and TNF-α siRNA ratio of 4 to 1. However, based on the mechanism of glucocorticoids, we anticipated that a higher ratio of BA to siRNA (e.g. 4 to 1) would be more effective. Therefore, we sought to identify the potential mechanism(s) through which the solid lipid nanoparticles prepared with a relatively lower BA to siRNA ratio (i.e. 2 to 1) outperformed those prepared with a higher BA to siRNA ratio (i.e. 4 to 1) in our in vitro screening experiment.

Because of its structural composition, we hypothesized that the interactions of BA with other lipid components in the formulation (i.e. lecithin, cholesterol, DOTAP, and the stearoyl group in PHC) may be different depending on the BA to siRNA ratio and therefore affect the release of BA from the nanoparticles. In vitro, burst-released BA would potentially be available for uptake (i.e. by simple diffusion) and subsequent therapeutic effect faster than BA released from the nanoparticles following endocytosis and lysosomal escape. To determine if early release of BA from the nanoparticle contributed to the results from our in vitro screening assay, we compared the in vitro release profiles of BA from BA-TNF-α-siRNA-SLNs prepared at a BA to TNF-α siRNA ratio of 4 to 1 and 2 to 1. The TEM images of the nanoparticles prepared at 4 to 1 weight ratio of BA to TNF-α siRNA are shown in Supplemental Figure S2. No notable differences were seen between the morphologies of the BA-TNF-α-siRNA-SLNs prepared with a BA to TNF-α siRNA ratio of 4:1 and 2:1. Similar to our initial release study, nanoparticles that contained the same amount of TNF-α siRNA, but different amounts of BA, were re-suspended in release medium and placed inside a 50 KDa MWCO dialysis device which was then placed in 25 ml of release medium. At designated time points, the release medium was collected and analyzed for BA content. As shown in Figure 7, the amounts of BA released by the BA-TNF-α-siRNA-SLNs prepared at 4 to 1 weight ratio of BA to TNF-α siRNA and that prepared at the 2 to 1 ratio did not differ significantly within the first 4 h, which is the exact time the BA-TNF-α-siRNA-SLNs were incubated with the J774A.1 macrophages when we identified the optimal ratio of BA to TNF-α siRNA in the solid lipid nanoparticles (Figure 4). However, the rate at which BA was released (i.e. the percent of total BA released over time) from the BA-TNF-α-siRNA-SLNs prepared with a BA and TNF-α siRNA ratio of 2 to 1 was higher than that from the BA-TNF-α-siRNA-SLNs prepared with a BA and TNF-α siRNA ratio of 4 to 1, as the former nanoparticles initially contained a lower total amount of BA. Therefore, the mechanism through which the BA-TNF-α-siRNA-SLNs prepared with a BA and TNF-α siRNA ratio of 2 to 1 induced the lowest level of proinflammatory cytokine release by J774A.1 macrophages in culture is likely not solely related to the amount of BA released during the 4 h of incubation time. Of course, in cell culture medium and in the presence of the J774A.1 macrophages, the release of BA from the BA-TNF-α-siRNA-SLNs may potentially be different than that seen in the release medium used. Nonetheless, the differences in the release rates of BA from BA-TNF-α-siRNA-SLNs prepared with BA to TNF-α siRNA ratios of 4 to 1 and 2 to 1 indicate that there may be alterations in the molecular interactions between BA and other components of the solid lipid nanoparticles.

Figure 7.

In vitro release of betamethasone (BA) from BA-TNF-α-siRNA-SLNs in pH 7.4 PBS up to 4 h. Nanoparticles were prepared using BA to TNF-α siRNA ratios of 2 to 1 or 4 to 1. Data are mean ± S.D. (n = 3).

To investigate this further, we also performed DSC and XRD to evaluate BA-TNF-α siRNA-SLNs prepared with a BA to TNF-α siRNA ratio of 4 to 1, 2 to 1, and 0 to 1. Analysis of the DSC curves revealed notable differences between the BA-TNF-α-siRNA-SLN preparations in the 110 to 200 °C range (Figure 8A), near the melting point of cholesterol and lecithin (Figure 8B), indicating that there indeed were differences in the molecular interactions among the various components in the BA-TNF-α-siRNA-SLNs prepared at different ratios of BA to siRNA. Of particular note, the peak near 150 °C, which corresponds to the melting point of cholesterol, shifted left and diminished as the amount of BA was increased in the formulation. Due to the structural similarities between BA and cholesterol, this finding could indicate a change in either the encapsulation or molecular interaction of BA and cholesterol in the SLN formulation as the amount of BA increased.

Figure 8.

DSC curves of TNF-α solid lipid nanoparticles prepared using BA to siRNA weight ratios of 4 to 1, 2 to 1, 0 to 1 (A), as well as those of bulk BA, cholesterol, and lecithin (B). The experiment was repeated twice with a similar trend.

We were unable to detect other apparent differences in the DSC curves (Figure 8) as well as the X-ray diffractograms of the various BA-TNF-α-siRNA-SLNs (Supplemental Figure S3). However, it is important to note that the absence of specific peaks does not rule out the possibility that crystalline material may be present, as detection of crystalline material within the nanoparticles may not be feasible due to the relatively lower content of BA in the preparation.^{26, 27} In addition, the XRD may be limited in its ability to detect changes in the state of BA at low concentrations and unique peaks may be lost in the baseline or superimposed upon by other materials in the formulation, especially considering that the BA-TNF-α-siRNA-SLNs are a complex system containing up to six different components.²⁶ Other more sensitive techniques are likely needed to fully understand the molecular interactions among the various components in the BA-TNF-α-siRNA-SLNs. However, our data clearly showed that encapsulation of BA, a glucocorticoid, into siRNA-solid lipid nanoparticles significantly reduced the acute proinflammatory activity of the siRNA and/or nanoparticles, and importantly, that the weight ratio of the BA and siRNA used to prepare the nanoparticles had a significant effect on this activity.

As mentioned earlier, it is expected that this strategy of encapsulating glucocorticoids into siRNA carriers will be applicable to other siRNA nanoparticle formulations, and that choice of glucocorticoid is not limited to betamethasone. It is worth emphasizing that the exact amount of glucocorticoid needed to suppress the proinflammatory activity of different siRNA sequences in a different carrier or delivery system may not be the same. The approach of co-encapsulating siRNA with a glucocorticoid is also not exclusive to other approaches intended to reduce the proinflammatory activity of siRNA, such as the utilization of siRNA with reduced immunostimulatory properties and the selection of less immunogenic carrier materials.

4. Conclusion

The potential for siRNA and/or its delivery systems to induce acute inflammatory responses is a significant hurdle to the development of siRNA therapeutics. Herein we confirmed that co-encapsulation of betamethasone acetate (BA), a glucocorticoid, and siRNA into solid lipid nanoparticles successfully inhibited the proinflammatory activity of the nanoparticles. Using TNF-α siRNA, we then showed that the BA and TNF-α siRNA co-encapsulated into the solid lipid nanoparticles acted synergistically to reduce TNF-α production by mouse macrophages in culture. Importantly, we discovered that the ratio of the BA to TNF-α siRNA in the solid lipid nanoparticles needs to be optimized, as we found that nanoparticles prepared with a higher ratio of glucocorticoid to siRNA were not the least proinflammatory. As such, the use of a different glucocorticoid, siRNA sequence, or nanoparticle formulation likely requires optimization to identify the optimal glucocorticoid to siRNA ratio that results in a formulation that contains the lowest amount of the glucocorticoid, but has the least inflammatory toxicities.