Unconventional Lysine-Type Lipid Assemblies Enable Efficient Antisense Oligonucleotide Delivery with Distinct Structural Features

Abstract

Background/Objectives: Antisense oligonucleotides (ASOs) hold great therapeutic potential due to their precise ability to modulate gene expression, particularly for treating genetic and neurological disorders. However, effective delivery of ASOs remains a major challenge. While most recent research focused on lipid nanoparticles (LNPs) as ASO carriers, alternative formulations, preparation methods and lipid compositions on delivery optimization are not fully explored. In this study, we investigated two types of formulations, lipoplexes (LPXes) and LNPs, prepared using lysine-type cationic lipids, K3C14 or K3C16, selected from a lysine-type lipid mini-library for their superior formulation stability and distinct cellular entry mechanisms. Methods: The physicochemical properties of the formulations were characterized using dynamic light scattering. Cytotoxicity was evaluated in spleen and liver cell lines. LPXes and LNPs were assessed for ASO delivery efficiency using an engineered HEK293 split-luciferase cell line, while immune response was evaluated in human peripheral blood mononuclear cells. Cryogenic electron microscopy (Cryo-EM) images were captured for structural analysis. Results: Lysine-type lipid mini-library screening identified lipids with either a hydrocarbon spacer K3 or C14 fatty acid tail exhibiting great stability and safety. Among the tested LPX and LNP formulations, the K3C16 lipoplex demonstrated ASO delivery efficiency and immune responses comparable to the benchmark SpikeVax LNP formulation. Notably, Cryo-EM imaging revealed novel structures that have not been reported previously; the K3C14 lipoplex formed a rouleaux-like structure, whereas the K3C16 lipoplex exhibited a lipid nanosheet-like structure, distinct from the conventional LNP structure. Conclusions: These results highlight the potential of an unconventional type of lipid assembly for efficient ASO delivery.

1. Introduction

Antisense oligonucleotides (ASOs) have emerged as promising therapeutic agents over the past decades, particularly in the treatment of neurological disorders and rare genetic metabolic disorders [1,2]. They consist of 12 to 30 short synthetic single-stranded RNA or DNA oligonucleotides. Inside cells, ASOs function by binding to specific target ribonucleic acid (RNA) sequences through Watson–Crick base pairing, enabling them to downregulate, upregulate, or modify genes, thereby altering the resulting protein products [3]. To date, more than a dozen ASOs have been approved for the treatment of various diseases, such as spinal muscular atrophy and Duchenne muscular dystrophy [4,5]. Because of their target specificity, ability to directly modulate genes at the RNA level, and potential to target previously undruggable pathways, ASOs remain a prominent therapeutic modality in drug discovery.

Despite being promising therapeutic agents, delivering ASOs remains a major hurdle to their clinical translation. In vivo, naked ASOs entering the bloodstream are rapidly cleared through the renal system and are vulnerable to nuclease degradation due to their inherent chemical instability. Furthermore, ASOs can trigger immune responses, leading to the production of anti-drug antibodies (ADAs). The negatively charged nature of ASOs also inhibits their ability to cross cellular membranes, resulting in low cellular uptake via the process of gymnosis and limited escape from the endo/lysosomal pathway required to modulate the target RNA [6,7]. Extensive efforts have been made to chemically modify the phosphate backbone or the 2′ ribose sugar ring to improve the drug-like properties of ASOs. Modifications such as phosphorothioate (PS) and 2′-O-methyl (2′O-Me) are commonly used, offering benefits including reduced renal clearance, decreased immune responses, increased binding affinity to RNA, and enhanced enzymatic stability. While these advancements have improved the pharmacokinetics of ASOs, challenges related to cellular penetration and targeted delivery persist, emphasizing the need for alternative delivery strategies [8,9].

Inspired by the success of messenger RNA (mRNA) and small interfering RNA (siRNA) vaccines, the use of lipid nanoparticles (LNPs), a well-established non-viral gene delivery vehicle, has become a primary focus of ASO delivery research [10,11]. These carriers provide several crucial functions: they offer physical encapsulation that effectively shields ASOs from serum nuclease degradation, enhance cellular entry, and enable strategies for targeted tissue delivery [12,13]. These properties collectively enhance delivery efficiency and therapeutic potency. While most ASO-LNP formulations rely on ionizable lipids, alternative lipid formulations such as permanent cationic lipids remain underexplored, with limited comparative studies available. Permanent cationic lipids retain their positive charge regardless of environmental pH changes, making them particularly well-suited for efficient gene encapsulation and cellular delivery [14]. However, their high cytotoxicity has long posed significant challenges to clinical application [15,16]. Recent studies offer promising alternatives, showing that substituting the quaternary ammonium headgroups with positively charged peptide-like or amino acid headgroups reduced cytotoxicity without compromising delivery efficiency in vitro [15,17,18,19]. These findings open new avenues for continued exploration into permanent cationic lipids as viable candidates for gene delivery systems. In our previous publications, we reported that amino acid-based lipids exhibit superior transfection efficiency and a more favorable safety profile compared to commercial Lipofectamine 2000 [18,19]. Building on these findings, this study focuses on evaluating alternative ASO lipid formulations using lysine-type lipids with permanent cationicity. Lysine, being a natural amino acid, offers advantages, as its derivative, polylysine, is widely applied in biomedical research due to its high biocompatibility [20].

In this study, we conducted a comprehensive in vitro evaluation using various cell lines and primary cells. First, a lysine lipid mini-library, comprising lipids with varying hydrocarbon spacers (K0, K3, K5, or K7) and fatty acid tails (C14, C16, and C18), was screened for safe and stable formulations. Cytotoxicity was assessed using spleen and liver cancer cell lines, relevant to lipid nanoparticle biodistribution. K3C14 and K3C16 nanoassemblies were identified as optimal candidates due to their low toxicity and high stability. In addition, these nanoassemblies were previously reported to exhibit distinct cellular entry mechanisms: K3C16 employed membrane fusion, while K3C14 utilized endocytosis [21]. Building on these findings, we evaluated K3C14 and K3C16 nanoassemblies as ASO delivery vehicles. Two formulation types were tested: LPXes composed entirely of K3C14 or K3C16 lipids, and LNPs containing 50% K3C14 or K3C16 and helper lipids, both encapsulating splice-switch ASOs (ssASOs). As a model system, ssASOs were used to induce the alternative splicing of a luciferase gene in engineered HEK293 cells to assess delivery efficiency [22]. Additionally, inflammatory cytokine responses of different ssASO formulations were evaluated using peripheral blood mononuclear cells (PBMCs). Importantly, using Cryo-EM, we uncovered novel nanoassembly structures and further investigated their structure–activity relationships, correlating structural features with delivery efficacy and immune modulation.

2. Materials and Methods

2.1. Preparation of Lysine Mini-Library

The lysine-type lipids were synthesized and provided by Dr. Takeoka’s lab at Waseda University (Tokyo, Japan). A total of 10 mg of lipids were hydrated in 2 mL of 20 mM HEPES buffer (pH 7.2) at room temperature with vigorous stirring overnight, resulting in a 5 mg/mL nanoassembly suspension. The diameter (nm) of the particles was measured using a Zetasizer Ultra purchased from Malvern Panalyctical (Malvern, UK) with a 90° backward scattering setting. Nanoassemblies larger than 100 nm were sonicated for 30–90 min to reduce their size. Following sonication, the particle size in diameter (nm), polydispersity index (PDI), and zeta potential (mV) were measured, and the values were recorded as day 0. At day 30, the physicochemical properties were re-measured to assess stability. Between day 0 and 30, the suspensions were visually inspected weekly for any signs of precipitation. If noticeable precipitation was observed, the nanoassembly is disqualified for stability study.

2.2. Preparation of Lipoplexes

Lipoplexes were prepared based on our previously established methods and consisted of a single type of lysine lipid (Table 1) [18]. In short, 10 mg of lipids were hydrated in 2 mL of 20 mM HEPES buffer (pH 7.2) at room temperature with vigorous stirring overnight, resulting in a 5 mg/mL suspension. For K3C14, no further extrusion was performed since the size was well below 100 d.nm. For K3C16, an extrusion was done using a 0.22 um filter from MilliporeSigma (Hayward, CA, USA) and a handheld mini-extruder from Avanti Polar Lipids (Alabaster, AL, USA) 11 times at room temperature. The splice-switch ASO ordered from IDT (Coralville, IA, USA) was resuspended in sterilized water resulting in a stock concentration of 10.3 mg/mL. A corresponding amount of ASO solution was added to the nanoassembly suspension to achieve a nitrogen-to-phosphate (N/P) ratio of 5. While the ASO solution was added to the nanoassembly suspension, a constant stirring was applied and continued for 10 min, allowing for the formation of ASO–nanoassembly complexes.

Table 1.

Compositions of the ssASO formulations.

Formulations	Cationic/Ionizable Lipid	DSPC	Cholesterol	DMG-PEG2K	ssASO (N/P)
K3C14_LPX	100% K3C14	0%			5
K3C16_LPX	100% K3C16
K3C14_LNP1	50% K3C14	10%	39.5%	0.5%
K3C14_LNP2	50% K3C14		38.5%	1.5%
K3C16_LNP1	50% K3C16		39.5%	0.5%
K3C16_LNP2	50% K3C16		38.5%	1.5%
SM102_LNP1	50% SM102		39.5%	0.5%
SM102_LNP2	50% SM102		38.5%	1.5%

2.3. Preparation of ASO-LNPs

Lipids, including 1,2-distearoyl-sn-glycero-3-phosphocholine (DSPC) and 1,2-dimyristoyl-rac-glycero-3-methoxypolyethylene glycol-2000 (DMG-PEG2000), were purchased from Avanti Polar Lipids (Alabaster, AL, USA). The ionizable lipid Heptadecan-9-yl 8-((2-hydroxyethyl) (6-oxO-6-(undecyloxy) hexyl) amino) octanoate) (SM-102) was procured from MedChemExpress (Monmouth Junction, NJ, USA), while cholesterol was purchased from MilliporeSigma (Hayward, CA, USA). A microfluidic approach was used to prepare ASO-LNP formulations. First, lipid and ASO solutions were independently prepared. For the lipid stock mixtures, ethanol (EtOH) stock solutions for each lipid were mixed at the desired molar ratio (Table 1) to a total lipid concentration of 6 mM. For the ASO solutions, ssASO powder was dissolved in 20 mM sodium citrate (pH 4.0) to the target concentration. EtOH and ASO solutions were combined at a volume ratio of 3:1 aqueous to EtOH, and LNPs targeting a nitrogen-to-phosphate ratio (N/P) of 5 were prepared using a microfluidic instrument (Ignite, Cytiva, Marlborough, MA, USA), at a total flow rate of 12 mL/min. Unencapsulated ASO and residual EtOH were removed by centrifugal ultrafiltration (2000× g, 30 min) using 10 kD MWCO centrifugal filters (Amicon, MilliporeSigma, Hayward, CA, USA).

2.4. Encapsulation Efficiency by OliGreen Assay

Total ASO concentration and encapsulation efficiency (EE%) were quantified using the Quant-iT™ OliGreen™ ssDNA Assay (Invitrogen, Waltham, MA, USA). Free (unencapsulated) ASO in the formulation buffer was calculated with a 0–3000 ng/mL calibration curve prepared in TE buffer (10 mM Tris-HCl and 1 mM EDTA; pH 7). Total ASO was calculated after particle lysis using TE + 0.1% (v/v) Triton X-100, with a separate 0–3000 ng/mL calibration curve in the same matrix. Samples and standards (100 µL of each) were mixed with an equal volume of OliGreen working solution in a 96-well black/clear bottom plate and incubated for 5 min at room temperature in the dark. Fluorescence (ex = 485 nm and em = 520 nm) was recorded on a BioTek Synergy HTX Multi-Mode Microplate Reader. Concentrations were interpolated from their respective linear standard curves, and EE% calculated using the following equation:

EE% = [(Total ASO − Free ASO)/Total ASO] × 100%

2.5. Characterization of Physicochemical Properties

Size in diameter (nm), polydispersity index (PDI) and zeta potential (mV) was measured using the Zetasizer Ultra from Malvern with the classic 90° backward scattering setting. Three measurements were performed per sample and the mean ± SD was reported.

2.6. HEK293 Splice-Switch Assay

HEK293 split-luciferase cells encoding genes that form active luciferase upon engagement of ssASO was from Genentech (South San Francisco, CA, USA). The cells were cultured in DMEM supplemented with 10% FBS, 1× NEAA, 2 mM L-Glutamine and 1% Penicillin–Streptomycin. A total of 20,000 HEK293 Split Luc cells were seeded in each well in a 96-well tissue culture treated on a clear-bottom white plate (Corning Inc., Corning, NY, USA). On the second day, nanoassemblies along with free ASO controls were diluted based on the ASO concentrations to 400 nM or serially diluted from 1600 nM in a 2-fold dilution scheme and added to the wells, achieving a final concentration of 200 nM or 800 nM in the dilution scheme. The treated cells were incubated for 3 days at 37 °C and 5% CO₂. On day 3, D-luciferin sodium salt powder from GoldBio (San Louis, MO, USA) was dissolved in PBS, making a stock solution of 15 mg/mL. The D-luciferin solution was then further diluted to a final concentration of 150 μg/mL in each well. The plate was shaken at room temperature for 10 min followed by luminescence measurement using a plate reader.

2.7. Cell Viability Assay

The cell viability assay follows the same procedure for cell seeding and treatment mentioned in the HEK293 splice-switch assay. HCC374 human spleen cancer cell line (CVCL_V727) was obtained from UT Southwestern (Dallas, TX, USA) and HepG2 human liver cancer cell line (CVCL_0027) was obtained from ATCC (Manassas, VA, USA). Both cell lines were seeded at a density of 25,000 cells per well overnight prior to nanoassembly addition. On the second day, 10 µM and 100 µM freshly made nanoassemblies were added and incubated for 24 h at 37 °C and 5% CO₂. At the end of the incubation period, 100 µL of CellTiter-Glo Cell Viability reagent (Promega, Madison, WI, USA) was added to each well, and the plates were incubated at room temperature on a shaker for 10 min. The relative light units (RLU) were measured using a GloMax reader (Promega, Madison, WI, USA). Percent viability was calculated using the following equation:

% Viability = (Treated RLU/Untreated RLU) × 100%

2.8. PBMC Cytokine Profiling

Whole blood was obtained from three healthy human donors that voluntarily consented to participating in Genentech’s Samples for Science blood donation program. Whole blood was diluted 1:1 with PBS and carefully loaded to SepMate™ tubes (STEMCELL Technologies, Vancouver, BC, Canada) for the in vitro isolation of Peripheral Blood Mononuclear Cells (PBMCs) preloaded with Ficoll–Paque PLUS (Cytiva, Marlborough, MA, USA). The samples were centrifuged at 1200× g for 10 min at room temperature and the plasma was collected to another tube. The mononuclear cells were seeded at a density of 100,000 cells per well in a 96-well round bottom plate and cultured in RPMI-1640 medium supplemented with 10% FBS, 2 mM L-Glutamine, and 1× Pen/Strep. The cells were primed with 0.1 µg/mL R848 (InvivoGen, San Diego, CA, USA) for 2 h. After the priming, nanoassemblies were added to the cells, resulting in a final concentration of 200 nM, normalized based on ASO concentration. An amount of 0.1 µg/mL R848 was maintained in the cell treatment. Free ASO at 200 nM was included as a control, and cells without treatment only exposed to 0.1 µg/mL R848 were used as negative controls. After 24 h incubation, the cells were harvested by centrifugation at 300× g for 8 min. The supernatants were transferred to a 96-well plate and kept at −80 °C prior to luminex analysis. For the luminex cytokine analysis, a panel of pro- and anti-inflammatory cytokines (TNF-α, IFN-α, IL-1β, IL-6, IL-8, IL-12, IL-10, and MCP-1) was quantified using the MILLIPLEX assay platform (MilliporeSigma, Hayward, CA, USA). Values below the lower limit of detection were reported as zero.

2.9. Cryo-EM Imaging

Quantifoil QF-Cu300 R1.2/1.3 grids coated with a 2 nm carbon support film (Ted Pella, Redding, CA, USA) were glow-discharged for 7 s at 20 mA using a GloQube system (Quorum Technologies, East Sussex, UK). A 3 µL aliquot of the LNP suspension was applied to each grid, which was subsequently plunge-frozen in liquid ethane using a Vitrobot Mark IV (Thermo Fisher Scientific, Waltham, MA, USA) operating at 4 °C and 95% relative humidity. Vitrification was performed at the native sample concentration with a blot force of −13, a wait time of 45 s, and a drain time of 0.5 s.

Cryo-EM data were collected on a Talos Arctica transmission electron microscope (Thermo Fisher Scientific, Waltham, MA, USA) operated at 200 keV and equipped with a Gatan K3 direct electron detector and a BioQuantum energy filter (slit width 20 eV). Automated data acquisition was performed in SerialEM over a defocus range of −2 to −3 µm, with a total cumulative electron exposure of 25 e⁻/Ų due to the sensitivity to electron dose. Image processing and visualization were performed using 3dmod.

2.10. Data Analysis and Statistics

Data were plotted and analyzed using GraphPad Prism version 10.2.2 (Boston, MA, USA). The correlation graph was analyzed using the correlation analysis with two-tailed p value setting and Pearson correlation coefficient reported. Values between ±0.50 and ±1 suggest a strong relationship. A minus number indicates a negative correlation.

3. Results and Discussions

3.1. Lysine-Type Mini-Library In Vitro Cytotoxicity and Stability

Cytotoxicity and formulation stability are two critical parameters for the development of a successful gene delivery system. Multiple factors, such as head group, lipid morphology, linkage and hydrophobic tail length, can influence the cytotoxicity and stability of the cationic nanoassemblies [14]. To select the nanoassemblies with low cytotoxicity and long-term stability, we constructed a lysine-type lipid mini-library composed of lipids with the same lysine (K) head group with varying lengths of hydrocarbon spacers (K0, K3, K5, or K7) and dual fatty acid tails (C14, C16, and C18), resulting in 12 lipids in total (Figure 1a). The lysine head group has an isoelectric point of 9.87 and retains a positive charge at physiological pH. The hydrocarbon spacers are positioned between the lysine head group and the glutamate backbone, which connects to the fatty acid tails. These spacers provide flexibility to the head group and create a buffering zone between the hydrophilic head group and the hydrophobic tails. For screening, lipid nanoassemblies were prepared by simple hydration followed by sonication to control particle size, except for the K3C14 nanoassembly, which had an initial size under 100 nm and did not require sonication. Most nanoassemblies subjected to 30 to 90 min of sonication showed a significant reduction in size to below 200 nm. However, some formulations, such as K5C16 and K5C18, displayed persistently larger particle sizes above 200 nm even after 90 min of sonication (Supplementary Figure S1). Interestingly, a trend of increasing particle size was observed as lipid tail lengths increased from C14 to C18 while maintaining a fixed hydrocarbon spacer length, except for the K0 group. However, no such trend was observed when the tail lengths were fixed and the spacer lengths increased (Figure 1b).

Figure 1.

Lysine-type lipid mini-library chemical structure, size and stability. (a) The chemical structure of lysine-type lipids includes varying spacer lengths (shown in blue; n = 0, 3, 5, and 7) and fatty acid tail lengths (shown in red; n = 13, 15, and 17). The hydrocarbon spacers are positioned between the lysine head group and the glutamate backbone, which connects to the fatty acid tails. The total lengths of the fatty acid tails correspond to C14, C16, and C18. (b) Average size (nm) of the lysine-type nanoassemblies measured at day 0 post-sonication. The average size was set threshold, gated at 200 nm and represented by the dotted line. Data are shown as mean ± SD from two technical repeats. (c) A subset of lysine-type nanoassemblies with average size less than 200 nm post-sonication was evaluated for stability for 30 days. The average size and PDI were set thresholds, gated at 200 nm and 0.4, respectively, as indicated by the dotted lines. Green symbol represents nanoassemblies that remained within the gated range after 30 days, while red symbol denotes the outliers. Data shown average size (nm) and PDI measured at day 0 and day 30.

For stability assessments, a subset of nanoassemblies (K3 and K7 series, as well as K5C14) was evaluated. Nanoassemblies lacking a hydrocarbon spacer (K0C14, K0C16, and K0C18) were excluded from stability testing because they gradually precipitated out of solution after sonication, underscoring the importance of a spacer between the hydrophilic head group and hydrophobic tails for maintaining stability. Similarly, K5C16 and K5C18 nanoassemblies, which exhibited poor assembly quality and particle sizes exceeding 200 nm, were also excluded. After 30 days of storage in 20 mM HEPES buffer (pH 7.2) at 4 °C, K3C14, K3C16, K5C14, K7C14, and K7C18 demonstrated particle size growth but maintained an average size below 200 nm (Figure 1c). Conversely, K3C18 and K7C16 exhibited size increases exceeding 200 nm. In terms of PDI, most nanoassemblies fall under a value of 0.4, except for K7C18, which showed greater heterogeneity, even though its average size is under 200 nm.

In vitro cytotoxicity was assessed using human spleen cancer cells (HCC374) and human liver cancer cells (HepG2), as these cell lines represent tissues where nanoparticles typically accumulate in vivo [23,24]. In spleen cells, no noticeable cytotoxicity was observed at concentrations up to 100 µM (Figure 2). In liver cells, no change in cell viability was observed at a lipid concentration of 10 µM. However, several nanoassemblies exhibited cytotoxicity at 100 µM following 24 h of incubation (Figure 2). These results suggest cell type-dependent cytotoxicity, likely attributable to differences in biological functions between the two cell types [25]. In addition, the nanoassemblies (K0C14, K0C16, K5C16, K5C18 and K7C18) displayed cytotoxicity also exhibited poor assembly quality.

Figure 2.

Cytotoxicity of lysine-type nanoassemblies. HCC374 spleen cells and HepG2 liver cells were treated with lysine-type nanoassemblies for 24 hours at concentrations of 10 and 100 µM, and the percentage of cell viability (% Viability) was reported as a measure of cytotoxicity. Light gray bars represent the cytotoxicity at 10 µM, black bars represent the cytotoxicity at 100 µM, and the untreated group is indicated by a green bar. Data are shown as mean ± SD from two technical repeats.

Collectively, the screening of cytotoxicity and formulation stability pointed to a clear trend that lysine-type lipids with either a 3-hydrocarbon spacer (K3), such as K3C14 and K3C16, or C14 dual fatty acid tails, such as K3C14, K5C14 and K7C14, are essential for achieving lower cytotoxicity and higher stability. Among them, K3C14 stood out due to its overlap of both a 3-hydrocarbon spacer and C14 tails, resulting in a desirable safety and stability profile. Although K3C16, K5C14, and K7C14 exhibited comparable stability and safety, K3C16 was prioritized due to its distinct membrane fusion-based cellular uptake mechanism, as was demonstrated in our previous research [21,26].

3.2. Physicochemical Properties of the ssASO Formulations

For formulation comparison, we prepared two types of formulations, LPXes and LNPs, each encapsulating the ssASO. The LPXes were composed of nanoassemblies made entirely of 100% K3C14 or K3C16 lipids, while the LNPs consisted of 50% K3C14 or K3C16, with the remaining helper lipids adopted from the SpikeVax LNP formulation: 10% DSPC, either 38.5% or 39.5% cholesterol, and 0.5% or 1.5% DMG-PEG2K (Table 1).

For formulation, LPXes were prepared using a simple hydration and extrusion method to first prepare the nanoassemblies, followed by incorporating the ssASO with constant stirring to allow for electrostatic association. For the mini-library screening, particle size was controlled via sonication to increase throughput. In this part of the study, size of the K3C16 nanoassembly was controlled using an extrusion method, as extrusion has been reported to have a higher reproducibility and scalability [27]. Conversely, for K3C14, a simple hydration method was sufficient to produce nanoparticles with a diameter below 100 nm. The LNP formulations, by contrast, were prepared using a microfluidic mixing method, which is a standardized technique for LNP preparation. To control the sizes of the lysine-type LNPs, two concentrations of DMG-PEG2K (0.5% and 1.5%) were evaluated in this study. Throughout the experiments, SM102_LNP1 and SM102_LNP2 encapsulating ssASOs were used as benchmarks, whereas K3C14 and K3C16 nanoassemblies without ssASOs served as negative controls. Prior studies have reported that the nitrogen-to-phosphate (N/P) ratio influences the structural organization of formulations, with an N/P ratio of 5 promoting more stable electrostatic interactions between the LNP and ASO [28]. In this study, we kept the N/P ratio of 5 for ASO-LNPs and ASO-LPXes.

Both K3C14-LPX and K3C16-LPX exhibited an increase in particle size upon ssASO incorporation. K3C14-LPX showed a more pronounced size increase from 29.7 nm to 118.3 nm, whereas K3C16-LPX increased from 147.4 nm to 212.6 nm (Figure 3; Supplementary Table S1). Size increase is generally expected for the formation of pDNA and nanoassembly complexes due to the large size of pDNA [29]; however, since ASOs are small oligonucleotides in the 10s nucleotide length, a large increase in size for K3C14-LPX suggests a significant alteration in lipid–ASO packing structures. Despite changes in size, the zeta potential of both LPXes remained in a similar range to those of the empty nanoassembly, indicating the N/P ratio of 5 does not affect the overall surface charges. For the lysine-type LNPs, an inverse relationship was observed between particle size and the concentration of DMG-PEG2K, with particle size decreasing as DMG-PEG2K increased from 0.5% to 1.5% (Figure 3). This result is not surprising, as previous reports have shown increasing amounts of PEG lipids can control the size of LNPs [11,30]. The overall PDI of the LNPs was approximately 0.2 or lower, indicating a relatively homogeneous particle population. The zeta potential of K3C14 and K3C16 LNPs was approximately half that of the LPXes, as only 50% of the cationic lipids were used in the LNP formulations. The positive controls SM102 LNP1 and LNP2 exhibited zeta potentials close to zero, reflecting the net neutral charge of the ionizable lipid at physiological pH. Despite the low zeta potential, they exhibited excellent stability, which is attributed to the steric stabilization provided by PEG lipids [31,32].

Figure 3.

Physicochemical properties of ssASO formulations. Size (nm), PDI and zeta potential (mV) of ssASO formulations were characterized using dynamic light scattering. K3C14_empty and K3C16_empty are nanoassemblies without ssASO encapsulation. SM102_LNP1 and LNP2 are benchmark controls. LNP1 contains 0.5% DMG-PEG2K and LNP2 contains 1.5% DMG-PEG2K. Data are shown as mean ± SD from two independent batches of preparations.

As an important part of the formulation characterization, we assessed the stability of the formulations. After storage at 4 °C for 30 days in 20 mM HEPES buffer at a pH of 7.2, no significant changes were observed in particle size, PDI, and zeta potential for any of the formulations (Supplementary Figure S2). Another key physicochemical property is encapsulation efficiency (EE%). All LPX and LNP formulations achieved EE% > 80%, reflecting high encapsulation efficiency despite the different preparation methods and lipid compositions (Figure 4). This high EE% suggests that the ASOs were effectively shielded. Unlike LNPs, which pack nucleic acids within the core, LPXes carry a good portion of the nucleic acids on the surface via electrostatic interactions, increasing their susceptibility to enzymatic digestion in vivo [33,34]. The shielding effect observed in the OliGreen assay, combined with the unchanging zeta potential, indicates that the negatively charged ASOs were effectively “neutralized”. This demonstrates that K3C14 and K3C16 LPX provide robust protection for ASOs, highlighting their capability to shield and stabilize the encapsulated molecules.

Figure 4.

Encapsulation efficiency of ssASO formulations. Encapsulation efficiency (EE%) was measured by the OliGreen assay, and all ssASO formulations achieved an EE% of over 80%. Data are shown as mean ± SD from three technical repeats.

Overall, LPXes composed of a single type of lysine lipid can be prepared through a simple hydration and/or extrusion method, followed by the addition of ASO. The extrusion method allows for the effective control of particle size and ensures a relatively monodisperse formulation. In contrast, LNPs are typically prepared using a microfluidic mixing method, which often require the adjustment of the flow rate and the optimization of the lipid composition to achieve the desired size [35]. Additionally, post-purification steps are essential for LNPs to remove residual organic solvents, which could pose cytotoxic risks to cells. Nonetheless, formulation stability and EE% are compatible between the lysine-type LPXes and LNPs. Considering these factors, lysine-type LPXes offer advantages in terms of simplicity of preparation.

3.3. In Vitro Delivery of Splice-Switch ASO

We evaluated the delivery efficiency of the different formulations using a HEK293 split-luciferase cell line that was engineered to express luciferase protein upon splice-switch ASO (ssASO) binding to pre-mRNA, which triggers alternative splicing. Dosed at a 200 nM ASO concentration, K3C14 and K3C16 LPXes demonstrated comparable delivery efficiency, as indicated by the relative light units generated after 3 days of incubation with HEK293 cells (Figure 5a). Within the LNP group, the benchmark SM-102 LNP2 exhibited the highest delivery efficiency, while the lysine-type LNPs demonstrated approximately half the efficiency (Figure 5b). A negative correlation was observed between the LNP size and delivery efficiency; higher DMG-PEG concentrations reduced the LNP size and, in turn, increased the delivery efficiency (Supplementary Figure S3). Using the same HEK293 split-luciferase cells, we also checked the cytotoxicity after 3 days of treatment. Most ssASO formulations showed no noticeable toxicity at a 200 nM ASO concentration, with the exception of K3C14 LPX, which displayed approximately 20% cytotoxicity (Supplementary Figure S4). In comparison, K3C14_empty (without ASO encapsulation) demonstrated only slightly reduced (~8%) cytotoxicity, suggesting that changes in the ASO-K3C14 lipid packing structure may influence cytotoxicity.

Figure 5.

In vitro delivery of splice switch ASO (ssASO). (a) HEK293 split-luciferase cells were treated with a 200 nM ASO concentration for 3 days, after which relative light units (RLU) were measured following the addition of D-luciferin. Within the LPX group, both K3C14 and K3C16 LPXes exhibited a comparable magnitude RLU. (b) In the LNP group, SM102_LNP2 showed the strongest luminescence signal. Overall, lysine-type LPXes showed enhanced luciferase expression compared to their LNP counterpart. K3C14_empty, K3C16_empty, free ASO and untreated group were negative controls. Data are shown as mean ± SD from three biological repeats.

Overall, both K3C14 and K3C16 LPXes achieved greater luciferase expression compared to their LNP counterparts, likely due to the higher ratio of cationic lipids, which enhances transfection efficiency [36]. While SM-102 LNP2 demonstrated the highest expression levels, the two LPX formulations showed comparable performance, closely approaching those levels. Notably, K3C16 LPX exhibited no cytotoxicity at the tested concentrations, presenting a favorable safety and efficacy profile as a candidate ASO delivery vehicle.

3.4. Cryo-EM Structure of the ssASO Formulations and Their Implications for In Vitro Delivery

Given variations in formulation methods, lipid compositions, and the accompanying changes in the physicochemical properties of LPXes, obtaining detailed structural information is essential for understanding the structure–activity relationship. To this end, we captured a series of Cryo-EM images of LPXes and LNPs encapsulating ssASOs.

The LPX group, composed entirely of the positively charged lysine lipids K3C14 or K3C16, revealed surprising structural features. K3C14 LPX exhibited a rouleaux-like structure, characterized by stacked flat disks resembling the aggregation of red blood cells (Figure 6; Supplementary Figure S5). These rouleaux-like structures ranged in length from 100 to 200 nm. Interestingly, K3C14 nanoassemblies without ssASOs displayed a bicelle-like structure, representing their native state before ssASO binding (Supplementary Figure S6). The formation of rouleaux in the presence of ssASOs can largely be attributed to the interaction between the negatively charged ssASOs and the positively charged bicelle-like K3C14 structures. This interaction resulted in the stacking of bicelles, forming multilaminars where ssASOs intercalate between the bicelles. This complex lipid–ASO structure explained the significant increase in size measured by DLS. Typically, bicelle formation involves a combination of long- and short-chain lipids and utilizes a thin-film hydration method [37]. Bicelles, with their discoidal shape, offer several advantages over spherical drug delivery vehicles, including enhanced cellular uptake due to greater contact with the cell surface and reduced energy requirements for membrane bending [38]. Importantly, the K3C14 bicelle structure is formed using a single lipid type through a simple hydration method without the need for additional purification steps, streamlining the fabrication process. Although promising for ASO delivery, further optimization in size and composition is necessary, as they exhibited a wide range of sizes and displayed moderate cytotoxicity.

Figure 6.

Cryo-EM imaging reveals the structure differences of the ssASO formulations. K3C14_LPX, upper panel left, showed a rouleaux-like structure. K3C16_LPX, upper panel right, showed a nanosheet like structure. K3C14_LNP2, lower panel left, showed a mixture of multilamellar vesicle (MLV) and multivesicular liposome (MVL) structures. K3C16_LNP2, lower panel right, showed a predominant unilamellar vesicle (ULV) structure. Scale bar is 100 nm.

In contrast, K3C16 LPX exhibited a nanosheet-like structure, characterized by fatty acid tails forming a hydrophobic sheet-like core and lysine hydrophilic head groups oriented outward, appearing as a dark membrane-like structure on Cryo-EM images. (Figure 6; Supplementary Figure S5). This native nanosheet structure remained intact even in the absence of ssASOs, as shown in Supplementary Figure S6. Interestingly, the incorporation of ssASOs introduced noticeable curvature to the nanosheet. This curved nanosheet configuration is advantageous for ASO delivery. In materials science, nanosheets are well-regarded for their exceptional drug loading capacity due to their large surface area and membrane-like architecture [39,40]. The curvature of the K3C16 nanosheet assembly may enhance its interactions with cellular membranes, facilitating efficient cargo transfer and making it well-suited for ASO delivery. Furthermore, observations of membrane fusion, direct cytosolic release of FITC–dextran, and rapid calcium influx with K3C16 nanoassemblies further support their distinct cellular uptake mechanism, which involves fusion and efficient drug release [21,26,41]. These unique properties collectively contribute to the superior ASO loading capacity and high delivery efficiency of K3C16 LPX observed in this study. Additionally, the unique structure and specialized cellular uptake pathways of K3C16 LPX may result in distinct organ-specific distribution compared to conventional LNP formulations, which are typically characterized by high liver biodistribution.

In the LNP group, K3C14 LNP2 displayed a combination of multilamellar vesicle (MLV) and multivesicular liposome (MVL) structures (Figure 6; Supplementary Figure S5). The MLVs exhibited a smaller size distribution of approximately 100 nm, whereas the MVLs had a broader and less homogeneous size distribution ranging from several hundred nm to below 100 nm. In contrast, K3C16 LNP2 primarily revealed unilamellar vesicle (ULV) structures with sizes between 50 and 100 nm (Figure 6; Supplementary Figure S5). The positive control, SM102 LNP2, demonstrated a densely packed, highly uniform internal structure with particle sizes ranging from 50 to 100 nm (Supplementary Figure S6). Despite differences in structural complexity, K3C14 LNP2 and K3C16 LNP2 showed similar maximum effects on ASO delivery. However, SM102 LNP2 outperformed both, exhibiting a threefold greater maximum effect in ASO delivery. According to previous studies, particle size, internal structural order, and ASO packing density are critical factors influencing delivery efficiency [42,43]. While all LNPs demonstrated comparable encapsulation efficiency, the disordered, multilamellar, or hollow-core structures in K3C14 and K3C16 LNP2 may have hindered efficient ASO release, resulting in reduced delivery efficiency compared to SM102 LNP2. These results suggest that further optimization of lipid compositions and flow rate ratios is necessary to refine and control the internal structure of lysine-type LNPs.

3.5. Cytokine Profiling

Cytokine profiling is a critical step in evaluating the potential undesired immune response triggered by nanoassemblies, ensuring safety and optimizing formulations. In this study, a panel of proinflammatory and anti-inflammatory cytokines and chemokines (TNF-α, IFN-α, IL-1β, IL-6, IL-8, IL-12, IL-10, and MCP-1) was analyzed following 24 h of exposure to ssASO formulations at a 200 nM ASO concentration in human peripheral blood mononuclear cells (PBMCs) (Figure 7). In one of the three donors, most ssASO formulations induced a 3-fold increase in IFN-α and MCP-1 levels. Interestingly, even the naked ssASO control elicited a 2-fold increase in IFN-α, suggesting that the naked ssASO itself contributes to the IFN-α response. Across all three donors, a consistent trend of increased IL-1β production was observed for most ssASO formulations, despite donor-to-donor variability in cytokine responses. This increase indicates an active inflammatory response, likely attributable to the lipid components of the formulations, as the naked ssASO control did not induce a similar effect. These findings are consistent with earlier reports that cationic and ionizable lipids are key drivers of IL-1β release [44,45]. Supporting this, formulations composed entirely of K3C14 or K3C16 lipids, without any helper lipids, elicited even higher levels of IL-1β release. Interestingly, ssASO-loaded formulations of K3C14 or K3C16 demonstrated reduced IL-1β and IFN-α production compared to their empty nanoassembly counterparts, despite exhibiting no change in zeta potential. This reduced immune response suggests that particle size and the ASO–lipid packing structure may influence cellular interactions, thereby affecting immune cell activation and the subsequent cytokine response [46]. Aside from IL-1β, IFN-α, and MCP-1, no significant increase was observed in other cytokines or chemokines. Collectively, these results suggest that the lysine-type lipids and ionizable lipids used in the formulations elicit a moderate immune response, a characteristic that is irrelevant of formulation types.

Figure 7.

PBMC cytokine profiling. Heatmap of cytokine secretion from human PBMCs treated with ssASO formulations (200 nM) or empty nanoassemblies. Results are compiled from three different PBMC donors and normalized to the untreated group. Slightly elevated IL-1β secretion was observed across all the ssASO formulations tested in 3 donors.

4. Conclusions

Developing a lipid-based delivery system for antisense oligonucleotides (ASOs) that is safe, stable, efficacious, and scalable presents significant challenges due to the complex interplay of factors required to achieve a delicate balance. In this study, we assessed lysine-type cationic nanoassemblies as potential candidates for efficient ASO delivery. A comprehensive screening of a lysine-type mini-library, involving cytotoxicity evaluations in HCC374 (spleen cancer) and HepG2 (liver cancer) cell lines and monitoring formulation stability over one month, identified lysine lipids K3C14, K3C16, K5C14, and K7C14 as top-performing candidates. These lipids, bearing either a 3-hydrocarbon spacer or 14-carbon fatty acid tails, exhibited superior safety profiles and long-term stability. Among these, we selected K3C14 and K3C16 for further investigation. We employed two preparation methods and different lipid compositions to characterize two groups of formulations—LPXes and LNPs—encapsulating splice-switch ASOs. Our findings revealed that, in LNPs, smaller particle sizes and a well-organized, densely packed core were critical for achieving higher delivery efficiency. For LPXes, composed entirely of monotype cationic lipids, Cryo-EM imaging revealed unique structural features. K3C14 LPX exhibited a distinctive rouleaux-like arrangement resembling stacked red blood cells, while K3C16 LPX formed nanosheet-like structures characterized by a hydrophobic core and a hydrophilic outer layer. These nanosheet and rouleaux-like structures may have potential differential biodistribution profiles in vivo. Although K3C14 LPX displayed intriguing structural attributes, it requires further optimization to improve uniformity, stability, and cytotoxicity for superior nucleic acid delivery. In contrast, K3C16 LPX demonstrated an ideal combination of properties, including a particle size around 200 nm, a PDI of ~0.2, and ASO delivery efficiency comparable to the benchmark SM102 LNP2 (SpikeVax) formulation. Additionally, cytokine profiling indicated that both lysine-based and ionizable lipid formulations elicited a similar moderate immune response. Collectively, this study highlights the potential of lysine-based cationic nanoassemblies, particularly K3C16 LPX, as an emerging candidate for ASO delivery and sets the stage for further investigation of its in vivo therapeutic potential.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/pharmaceutics18020228/s1, Table S1: Physicochemical properties of the ssASO formulations. Figure S1: Lysine-type lipid nanoassemblies size before and after sonication; Figure S2: ssASO formulations one-month stability at 4 °C; Figure S3: ssASO-LNP formulations size and potency correlation; Figure S4: Cytotoxicity of ssASO formulations; Figure S5: Cryo-EM imaging reveals the structure differences of the ssASO formulations; Figure S6: Cryo-EM images of empty K3C14 and K3C16 nanoassemblies and SM102 LNP2.

Author Contributions

J.H.—Conceptualization, Methodology, Investigation, Formal analysis, Data curation, Visualization, Validation, Supervision, Writing—original draft, and Writing—review and editing. W.S.-B. and A.R.—Methodology, Investigation, Formal analysis, Data curation, Visualization, Investigation, Validation, Writing—original draft, and Writing—review and editing. N.Z.—Methodology, Investigation Formal analysis, Visualization, Validation, Writing—original draft, and Writing—review and editing. W.J.—Methodology, Investigation, Formal analysis, Visualization, and Validation. S.S.—Methodology, Investigation, Formal analysis, and Validation. M.C.J.—Methodology, Investigation, Formal analysis, Visualization, and Validation. J.L., C.-W.Y., and S.T.—Conceptualization, Methodology, Investigation, Formal analysis, Data curation, Visualization, Validation, Supervision, Writing—original draft, and Writing—review and editing. All authors have read and agreed to the published version of the manuscript.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

All the data related to the results of the article will be made available by the authors without any restrictions.

Conflicts of Interest

Authors Jieyan He, Whitney Shatz-Binder, Alexandra Robles, Nanzhi Zang, Wei Jia, Matthew C. Johnson, Jing Li, and Chun-Wan Yen are employed by the company Genentech/Roche. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Funding Statement

The authors thank the Genentech’s Innovation Fund program for financial support. The authors thank the Cryo-EM Research Center (CEMRC) in the Department of Biochemistry at the University of Wisconsin-Madison for the cryo-EM image collection.