Electrostatic adsorption of polyanions onto lipid nanoparticles controls uptake, trafficking, and transfection of RNA and DNA therapies

LNPs Can Be Stably Layered with Biologically Relevant Polyanions.

To test the impact of polyanion structure on the performance of LLNPs, we layered LNPs with a library of four carboxylated polyanions: hyaluronate (HA), poly-L-aspartate (PLD), poly-L-glutamate (PLE), and poly-acrylate (PAA) (Fig. 1A). These biopolymers were chosen for their therapeutic promise, given that they have previously modulated NP uptake by specific cancer and primary cells. HA has facilitated NP tumor targeting, since it binds the CD44 receptor over-expressed in several cancers; PLD has further improved uptake in ovarian cancer lines (22–24). PLE and PAA have, respectively, up-regulated and down-regulated NP uptake across panels of ovarian cancer and primary immune cells (22). The polyanions varied in monomer structures, sampling biologically relevant saccharides (HA), uniformly repeating poly(amino acid)s (PLD, PLE), and hydrocarbon-based acrylate backbones (PAA). Polyanion molecular weights were 20, 14, 15, and 15 kDa, respectively. To test whether layering was modular across LNPs, we formulated a set of LNP cores, varying in component lipids, lipid ratios, and nucleic acid cargos (either mRNA or pDNA). Formulations contained commercially available ionizable cationic lipids used in prior applications, with either single valency (DLin-MC3-DMA, DLin-KC2-DMA, and ALC-0315) or multiple valency (C12-200 and cKK-E12) (SI Appendix, Table S1) (2, 27–30). LNPs were suspended in acidic conditions to express positive surface charge, then incubated in equal volumes of each of the four polyanions (Fig. 1B). Stable layering was defined as a state achieving complete surface charge conversion from >30 to <−30 mV, with diameters below 200 nm as measured by dynamic light scattering (DLS). Optimal polyanion concentrations were determined empirically through titration (SI Appendix, Fig. S1). Relative to unlayered LNPs, LLNP diameters increased by 5 to 50 nm (Fig. 1C and SI Appendix, Table S2). Variations in sizes are attributed to both polyelectrolyte secondary interactions and hydration effects. For example, HA, which caused the largest diameter increases, can strongly self-associate through hydrogen bonding, leading to a thicker adsorbed layer; its swelling may also be due to high water association (31, 32). Layering LNPs induced complete charge reversal, from unlayered LNP zeta potentials of +35 ± 1 mV to layered zeta potentials of −23 ± 10 to −36 ± 17 mV (Fig. 1D). The magnitude of charge reversal depended on the weight equivalents of polyanion used for layering. Nucleic acid encapsulation efficiency did not change significantly upon the addition of surface layers, suggesting that the polyanions do not displace encapsulated cargo (Fig. 1E). Furthermore, LLNPs retain internal core structure, as confirmed by transmission electron micrography (TEM) (Fig. 1F).

Surface Polyanion Alters Apparent pKa of LNPs.

We hypothesized that the addition of a charged polyanion with acidic carboxylic groups may change the apparent pKa of the LLNP. Prior work optimizing ionizable lipids has demonstrated that the apparent pKa of the LNP impacts its transfection potency, immunogenicity, and targeting in vivo (18, 28, 33–35). We therefore quantified the impact of surface polyanions on the apparent pKa of LLNPs. To illustrate trends in pKa, mRNA-LNPs with ionizable lipid ALC-0315, which has a reported pKa of 6.5, were layered with the four polyanions. ALC-0315 was chosen as a representative lipid given its monovalency and use in the Pfizer/BioNTech COVID-19 vaccine (1). Apparent pKa was determined using 2-p-toluidinonaphthalene-6-sulfonic acid (TNS), an anionic fluorophore that emits signal when in hydrophobic environments, such as cationic lipid membranes, through previously established binding assays (35). Briefly, pKa was reported as the pH at which the normalized TNS fluorescence, a proxy for LNP protonation, was 50% of the maximum detected TNS fluorescence. While unlayered LNPs exhibited a pKa of 6.5 (95% CI [6.4, 6.6]), the addition of PLD, PLE, and PAA decreased the apparent pKa to 4.7 (95% CI [4.4, 4.9]), 5.0 (95% CI [4.7, 5.3]), and 4.4 (95% CI [3.3, 4.9]), respectively (Fig. 1G). The acidified pKa of LLNPs suggests that the proximity and concentration of the deposited polyanion’s functional groups—in this case, acidified carboxyl groups—may modulate the pKa of the LNP (Fig. 1G). Notably, HA-LLNPs exhibited a pKa of 6.5 (95% CI [6.3, 6.6]) similar to unlayered LNPs, though HA itself has a reported pKa ranging from 3 to 4 (36, 37). Furthermore, the pKa curves of LLNPs showed a buffering zone, within which TNS fluorescence plateaued, between pH 3.5 to 4, 3.5 to 4.5, and 3.5 to 5 for PLD, PLE, and PAA, respectively. The diffusion of TNS, an anionic molecule, through the outer polyanion layer into the core may be driven by hydrophobicity gradients. The presence of two ionizable species—the polyanion and the ionizable lipid—may create a multi-step equilibrium.

Surface Polyanion Reduces LNP Surface Protein Adsorption.

Given that LNP extrahepatic delivery is hindered by adsorption of liver-targeting endogenous proteins, we aimed to quantify the impact of the surface polyanion on protein adsorption (10). To test the impact of the surface polyanion on LNP stability in biological conditions, both unlayered and LLNPs were incubated in either mouse plasma or water at 37 °C and then washed via centrifugation. DLS measurements indicated that unlayered LNPs in mouse plasma were 131 ± 6 nm larger than LNPs in water, likely due to the adsorption of plasma proteins. By contrast, PLD-LLNPs displayed a size increase of 21 ± 15 nm; PAA-LLNPs displayed a size increase of 9 ± 6 nm (SI Appendix, Fig. S2A). This improved stability may be attributed to the polyanion’s ability to mitigate protein adsorption. In a separate study, unlayered and layered LNPs containing DOPE-Cy5 were similarly incubated in mouse plasma for 1 h at 37 °C and washed via centrifugation. Total adsorbed plasma protein content was quantified with a BCA assay and normalized by Cy5 signal. Within this study, LLNPs adsorbed fourfold less total plasma protein than did unlayered LNPs (Fig. 1H). The zeta potentials of all LLNPs tested were between −29.3 ± 0.9 and −38.7 ± 2.0 mV; therefore, differences in adsorbed protein content are not attributed to any major differences in surface charge (SI Appendix, Table S3). Differences in adsorbed protein profiles have been shown to influence biodistribution and transfection in vivo (38). Preliminary characterization of adsorbed protein profiles was visualized with SDS-PAGE (SI Appendix, Fig. S2B); however, further proteomics analysis of the proteins preferentially bound to each layer will provide greater insights into LLNPs’ cellular interactions in future work.

Surface Polyanion Alters LNP Transfection of mRNA and pDNA.

To quantify the impact of LNP surface chemistry on transfection, different cell lines were treated with unlayered and layered LNPs containing luciferase-encoding mRNA. To isolate the impact of the polyanion, LNP core formulations were held constant in a 25/52.5/16/2.5 ratio of ALC-0315/cholesterol/DOPE/DMG-PEG-2000 and then layered with each polyanion. Cell lines included HEK293T, adherent epithelial cells; RAW 264.7, highly phagocytic macrophages; and EL4 and Jurkats, T lymphoma cells with low uptake rates. To assess the impact of the surface layer on rates of uptake and internalization, cells were treated with LNPs for either 4 h or 24 h. For the shorter treatment, cells were washed at 4 h and incubated for an additional 20 h in fresh media before readout, to allow time for detectable protein translation.

Comparisons of transfection after either 4 h or 24 h treatment suggested that polyanion coatings alter LNP uptake or RNA trafficking and release in a cell-specific manner (Fig. 2 A and B). Perhaps the most apparent trends are selective cell transfection based on the outer layer; for example, HA-LLNPs show significant uptake or association with each cell line examined in this study, but selective transfection in RAW 264.7 macrophages. PLE-LLNPs appear to drive much more transfection in HEK293T and Jurkat cells over EL4s and macrophages at longer timeframes. This behavior is particularly notable because macrophages tend to take up NPs at high rates compared to other cells; however, PLE-LLNPs do not show significant cell uptake or transfection in macrophages. After a 24 h treatment, PLD-LLNPs generated comparable or greater luciferase expression than unlayered LNPs, outperforming other layered formulations and providing strong transfection across these cell lines independent of cell association (Fig. 2B). On the other hand, PAA, which is a synthetic and highly charged polyanion, shows significantly lower uptake and relatively low transfection across each cell line. These observations contrast sharply with the unlayered LNPs, which exhibit entirely different trends in association and transfection. These results indicate that the polyanion layer coats the LNP surface and can meaningfully alter the trafficking, and ultimately the transfection, in different cell types.

There are also meaningful differences in the apparent kinetics of association and transfection of the NPs. In HEK293Ts, for example, PLE-LLNPs treated for 4 h induced significantly lower luciferase levels than did unlayered LNPs, which increased to much higher relative levels after 24 h treatment, suggesting that this polyanion may direct trafficking in a manner that leads to slower mRNA delivery. Conversely, PAA-LLNPs after 4-h treatment generated comparable luciferase expression to unlayered LNPs; however, after 24-h treatment, PAA-LLNPs showed reduced luciferase levels relative to unlayered LNPs, suggesting that this layer may undergo trafficking that leads to either rapid uptake and clearance, degradation, or endosomal trapping within the first hour. mRNA delivery kinetics varied by cell type; in Jurkats, for example, PLD- and PLE-LLNPs treated for 4 h exhibited an increased luciferase expression over unlayered LNPs; these differences among groups decreased after 24 h treatment. In RAW 264.7 s, HA-LLNPs treated for 4 h transfected comparably to unlayered LNPs, but transfection was greatly reduced relative to other groups at 24 h. Trends in LLNP mRNA transfection were also assessed in two cancer cell lines via flow cytometry, using LNP cores of the same lipid composition, containing EGFP-expressing mRNA (SI Appendix, Fig. S3). LOXIMVI melanoma cells exhibited similar trends in LLNP transfection; HepG2 liver carcinoma cells exhibited less specificity for polyanion outer layers.

To assess the effect of layering on a different core, HA, PLE, and PAA were layered onto LNP formulations in a 50/38.5/1.5/10 ratio of DLin-MC3-DMA/cholesterol/DMG-PEG-2000/DSPC. Similar trends as those seen with the original cores were observed at 4-h and 24-h treatments across cell lines (SI Appendix, Fig. S4). These data suggest that the LNP surface chemistry and resulting interactions are influenced by the polyanion outer layer coating, beyond the lipid composition.

To characterize LNP cellular association, cells were treated with LLNPs containing 0.5 mol % Cy5-labeled DOPE for either 4 h or 24 h; cellular Cy5 fluorescence was measured on a Tecan plate reader (Fig. 2 C and D). Notably, HA-LLNPs generated the highest Cy5 signal across cell lines at both incubations. PAA-LLNPs associated least across cell lines that similarly demonstrated reduced luciferase expression. Interestingly, though PLD-LLNPs generated the greatest luciferase expression, they associated moderately across cell lines. The discrepancy between cell association or uptake and transfection trends suggests that polyanions may further modulate the internal trafficking and RNA release.

To test the impact of polyanion adsorption on transfection of other nucleic acid cargos, we formulated ALC-0315 LLNPs containing GFP-encoding pDNA, layered with each of the four polyanions. pDNA delivery requires different intracellular release and trafficking profiles than mRNA delivery, since the former requires nuclear translocation for effective transfection. HEK293T, EL4, and HepG2 liver carcinoma cells were treated with LNPs for 4 h, then washed and cultured for an additional 20 h; GFP-positive cells were assessed via flow cytometry (Fig. 2E). Plasmid delivery across cell lines was enhanced most notably by HA-LLNPs, which more than doubled transfection rates. HA has previously improved lipoplex-mediated pDNA transfection, potentially due to intracellular interactions associated with its glycosaminoglycan structure (25). PLD- and PLE-LLNPs decreased pDNA-LLNP transfection in both EL4s and HEK293Ts. Transfection rates in HepG2s, which divide and intake material rapidly, were not significantly impacted by PAA-, PLD-, or PLE-LLNPs.

Intracellularly, LLNP Polyanions Traffic with pDNA Cargo.

To examine the impact of the polyanion on intracellular trafficking of pDNA LLNPs, confocal microscopy was conducted on EL4s and HEK293Ts treated with unlayered or layered LNPs for 4 h. These cell lines were chosen as representative adherent and suspension cultures. To facilitate co-tracking of lipid, nucleic acid, and polyanion, DOPE-Cy5 labeled LNPs were loaded with MFP488-labeled pDNA and layered with BDP 558/568-labeled polyanions.

After 4 h treatment, EL4s internalized both unlayered and layered LNPs; polyanions were taken up along with core lipids and pDNA (Fig. 3A). Across all groups, MFP488-pDNA signal appeared diffuse throughout the cytoplasm. EL4s treated with unlayered and HA-LLNPs additionally displayed some punctate pDNA signals co-localized with lipid signals; these NPs may be trapped in trafficking endosomes or lysosomes. pDNA nuclear localization was minimal, expected given the low rates of pDNA transfection in these cells. Though all groups exhibited some pDNA colocalization with the cell membrane, colocalization appeared strongest in cells treated with PLE- and PAA-LLNPs. In past studies, PLE has demonstrated localization to extracellular surfaces in cancer cell lines; similar cellular interactions may promote extracellular localization in EL4 lymphoma cells (22, 39). DOPE signal appeared punctate within the cytoplasm and cell membrane. The detected DOPE signal was significantly weaker than that of pDNA or polyanion; this may partially be due to the relative infrequency of labeled DOPE molecules in NP composition. Polyanion signal in LLNPs appeared diffuse in the cytoplasm, similar to pDNA dispersal. The relative positions of all three components suggest that polyanions and pDNA translocate through the cell in more similar patterns than do LNP lipids (Fig. 3 B and C). HEK293Ts exhibited similar internalization patterns; in addition, the three labeled LNP components were visualized trafficking between cells (SI Appendix, Fig. S5).

To further assess the mechanisms of uptake and intracellular trafficking of LLNPs, EL4s were treated with unlayered or layered LNPs with MFP488-labeled pDNA and BDP555/568-labeled polyanions. In this study, HA was not labeled. After either 4 or 24 h, cells were stained for each of clathrin, caveolin, or lysosomes, then imaged (SI Appendix, Figs. S6–S12). At 4 h, HA-, PLE-, and PAA-LLNP DNA localized more with clathrin markers than with caveolin; minimal localization was observed with caveolin (SI Appendix, Figs. S6, S7, and S9). In comparison, cells treated with unlayered LNPs showed variable degrees of co-localization of LNP DNA with clathrin; co-localization with caveolin was minimal. Prior work has observed clathrin-dependent uptake of HA-layered NPs; by contrast, LNPs are thought to traffic by macropinocytosis (22).

Quantification of co-localization of LNP DNA and LAMP1 at 4 h suggested a trend that LLNP cargo is less associated with lysosomes than unlayered LNP cargo, though statistical significance was not achieved (SI Appendix, Figs. S8 and S9). Interactions between the surface polyanion and cellular surface markers therefore modulate uptake rates and mechanisms.

PLE and HA Moderately Improve Hepatic and Splenic Transfection.

Next, we evaluated the impact of the surface polyanion on LNP biodistribution and transfection in healthy female C57BL/6 mice injected intravenously (i.v.). LNPs with each of the four polyanions induced no toxicity 72 h after dosing, as measured by liver enzymes, serum protein, and weight changes (SI Appendix, Fig. S13) (40, 41). Lower blood urea nitrogen (BUN) levels across groups, relative to literature-determined ranges, may be attributed to variability in vendor strains; however, no significant difference in BUN levels was observed between mice treated with LLNPs and mice treated with saline.

To study biodistribution and transfection, mice were dosed with unlayered and layered ALC-0315 LNPs at 0.3 mg/kg luciferase-encoding mRNA. Four hours after injection, the heart, lungs, liver, kidneys, and spleen were imaged for luciferase expression ex vivo. PLE-LLNPs increased liver and spleen luciferase expression 1.8-fold and 2.5-fold, respectively, over unlayered LNPs (Fig. 4A). HA-LLNPs increased spleen luciferase expression 2.3-fold over unlayered LNPs; however, this formulation did not increase liver transfection (Fig. 4A). In contrast to in vitro trends, PLD-LLNPs showed minimally improved or decreased transfection across organs; PAA-LLNPs performed similarly to unlayered LNPs in all organs studied. No significant differences from unlayered LNPs were detected in the heart or lungs for any polyanion. Comparing the ratio of weight-normalized flux in the spleen and liver, we observed that HA-LLNPs showed higher preferential localization to the spleen than did other layers (Fig. 4 B and E). In comparison, PLE-LLNPs increased expression in both liver and spleen.

For a longer-term biodistribution study, we analyzed organs for luciferase expression 24 h after transfection. At 24 h, HA-LLNPs showed reduced luciferase expression across all measured organs (Fig. 4 C and F). All other LLNPs performed similarly to unlayered LNPs, suggesting that the outer polyanion confers the greatest benefit at earlier times (Fig. 4 C and D and SI Appendix, Fig. S14). Differences in transfection of LLNPs decrease over time, potentially due to stripping or degradation of the outer layer over time.

To compare trends in in vivo transfection across cores, 4-h organ-level transfection studies were conducted with LLNPs in which the core was composed of a 50/38.5/1.5/10 ratio of DLin-MC3-DMA/cholesterol/DMG-PEG-2000/DSPC (SI Appendix, Fig. S7); PLD was excluded given minimal differences between PLD-LLNPs and unlayered LNPs in prior studies. Similar trends were observed by swapping the LLNP core; however, most notably, HA-LLNPs with the alternate ionizable lipid decreased spleen luciferase expression 2.3-fold relative to unlayered LNPs (SI Appendix, Fig. S15). The prominent difference in splenic performance may be attributed to synergistic impacts of both the polyanion and core helper lipids, though additional studies are required to isolate the impact of each component.

To evaluate the pharmacokinetics in vivo, mice were dosed i.v. with Cy5-labeled LNPs. Blood retention times were evaluated over a 24-h period post-administration by quantifying the Cy5 signal in the blood serum (SI Appendix, Fig. S16). Minimal differences in residence time were observed across LLNPs.

Materials.

Dilinoleylmethyl-4-dimethylaminobutyrate (DLin-MC3-DMA), N,N-dimethyl-2,2-di-(9Z,12Z)-9,12-octadecadien-1-yl-1,3-dioxolane-4-ethanamine (DLin-KC2-DMA), and 1,1′-((2-(4-(2-((2-(bis(2-hydroxydodecyl)amino) ethyl) (2-hydroxydodecyl)amino)ethyl)piperazin-1-yl)ethyl)azanediyl)bis(dodecan-2-ol) (C12-200) were purchased from MedChemExpress. 1,2-dimyristoyl-rac-glycero-3-methoxypolyethylene glycol-2000 (DMG-PEG-2000) and 3,6-bis[4-[bis (2-hydroxydodecyl)amino]butyl]-2,5-piperazinedione (cKK-E12) was purchased from Cayman Chemical Company. Cholesterol, 1,2-distearoyl-sn-glycero-3-phosphocholine (DSPC), 1,2-dioleoyl-sn-glycero-3-phosphocholine (18:1 (Δ9-Cis) PC), 1,2-dioleyol-sn-glycero-3-phosphoethanolamine (18:1 (Δ9-Cis) PE), 1,2-dioleyol-sn-glycero-3-phosphoethanolamine-N-(Cyanine-5) (18:1 Cy5 PE), and 6-((2-hexyldecanoyl)oxy)-N-(6-((2-hexyldecanoyl)oxy)hexyl)-N-(4-hydroxybutyl)hexan-1-aminium (ALC-0315) were purchased from Avanti Polar Lipids.

CleanCap Firefly Luciferase mRNA (5-methoxyuridine) was purchased from TriLink Biotechnologies. The Label IT Nucleic Acid Labeling Kit (MFP488) was purchased from Mirus Bio.

Sodium hyaluronate was purchased from Lifecore Technologies. Polyacrylic acid, sodium salt (PAA, 15 kDa) solution, N-hydroxysuccinimide ester (NHS), and N-(3-dimethylaminopropyl)-N-ethylcarbodiimide hydrochloride (EDC) were purchased from Sigma Aldrich. Poly-L-glutamic acid sodium salt (PLE100, 15 kDa) and poly-L-aspartic acid, sodium salt (PLD100, 14 kDa) were purchased from Alamanda Polymers. MWCO dialysis membranes were purchased from Repligen. BDP-558/568-amine was purchased from Lumiprobe.

Methods.