Polyglycerol-Based Lipids as a New Alternative to PEG in Lipid Nanoparticles for Improved Drug Delivery

The LNPs were formulated by mixing aqueous and organic phases via microfluidics. Negatively charged mRNA in the aqueous phase formed a stabilizing complex with ionizable cationic lipids in the organic phase. Remaining lipids in the organic phase self‐assembled around these complexes to form mRNA‐LNP systems. Finally, ultrafiltration centrifugation was used for buffer replacement to storage buffer, before measuring the size and PDI of the resulting LNP systems using DLS. Desirable nanoparticle diameters were in the 60–150 nm range, to facilitate cellular uptake and produce robust immune responses [23, 24]. Furthermore, a low PDI was desired as an indicator for a homogeneous population with a narrow size distribution. Size of nanoparticles may be decreased by the presence of a stealth lipid, resulting in steric hindrance during self‐assembly and reducing particle agglomeration.

In this study, microfluidic process flow conditions were optimized to render PEG‐LNPs with a mean particle size of 110.0 nm (PDI 0.045) (Table 2). With a chain length akin to PEG2000 (42 vs. 44 repeating units, respectively; Table 1), lPG resulted in the smallest LNPs of all tested lPG‐variations, followed by MeOlPG‐block‐lPG. Contrasting, MeOlPG and lPG‐block‐MeOlPG's shorter chain lengths provided limited stealth effects, forming significantly larger particles [25]. The differing chain length limits conclusions that can be drawn from comparing the two block copolymers, but may indicate that LNP size is independent of additional functional groups along the lPG backbone.

The comparatively larger size of lPG‐ versus PEG‐LNP systems has previously been observed with other PEG‐alternatives (such as pSar‐LNP systems) and is therefore not indicative of a limitation specific to our formulation [26, 27, 28]. Nonetheless, adjustments of production parameters may help reduce lPG‐LNP diameters. By varying microfluidic settings (e.g., microfluidic chip conformation, flow rate/ratio) or types of aqueous buffer, previous works have shown that a reduction in nanoparticle size is possible [4, 29]. Increasing molar amounts of stealth lipid can decrease LNP size but comes at the expense of reduced bioavailability, cellular uptake, and higher anti‐PEG antibody activation [11, 30, 31].

Zeta potential can provide additional insights for LNP formulations, as strong electrochemical charges (∼ ±30 mV) can elevate particle repulsion to improve colloidal stability, but also lead to reduced circulation half‐life and increased cellular toxicity [32, 33]. Zeta potential of tested lPG‐LNPs was between −4.10 (MeOlPG‐block‐lPG) and −6.78 mV (MeOlPG), whereas PEGylated LNPs had a zeta potential of −3.30 mV. As all formulations displayed a nearly neutral zeta potential, cellular toxicity due to excess charges is unlikely [34].

Colloidal storage stability was further assessed by DLS measurements of mRNA‐LNPs at 4°C, demonstrating negligible change in particle diameter over 3 weeks. lPG‐functionalization was able to provide equivalent polymeric stealth effects as PEGylation (Figure 2a, Table S1 Supporting Information). The sole exception was MeOlPG‐LNPs, growing in diameter and PDI during the measurement period (Figure 2b), which may be attributed to MeOlPG's comparatively shorter chain length providing less stealth effects leading to particle agglomeration at 4°C storage [35].

Encapsulation efficiency of mRNA was assessed by comparing unencapsulated (free) mRNA with encapsulated mRNA released after lipid membrane lysis with TritonX‐100. lPG‐LNPs had the highest encapsulation efficiency (93.4%), followed by lPG‐block‐MeOlPG (79.4%), MeOlPG‐block‐lPG (78.1%), and finally MeOlPG‐LNPs (61.9%). Following, lPG may form compact (albeit slightly larger) LNPs with comparable encapsulation efficiency as PEG‐LNPs (Table 2). While adding some methoxy groups in the studied block‐co‐polymers slightly reduced mRNA encapsulation, MeOlPG had an entirely methoxylated polymer backbone. This feature may contribute to agglomeration and the formation of large, loosely packed MeOlPG‐LNPs with mRNA breaching the particle surface to reduce encapsulation efficiency [31].

Cryo‐TEM imaging of eGFP mRNA LNP systems was performed immediately after microfluidic production (day 0) and after 4 weeks of storage in PBS at 4°C. At day 0, LNPs appeared as unilamellar vesicles in the nanometer range, consistent with prior DLS measurements (Figure 3). As expected, the presence of less definitive lamellar structures was limited due to the relatively high N/P ratio of 6 [36]. Notably, when PEG was replaced with lPG variations, LNP morphology featured a visible lipid bilayer shell rather than a monolayer, surrounding an amorphous electron‐dense core. This “solid core” morphology may result from an interior oil‐phase containing deprotonated ionizable lipids and mRNA [29, 35]. Despite these morphological differences between PEG‐ and lPG‐LNPs, the mRNA encapsulation ability was likely maintained based on previously described results.

After 4 weeks of storage, minor structural changes were observed for all LNP formulations, including thickening of the lipid monolayer and expulsion of mRNA. However, the presence of mainly intact lPG‐LNP systems several weeks after preparation supports previous stability data and may indicate sufficient stability for future therapeutic applications.

A major concern with PEGylated therapeutics is the activation of anti‐PEG antibodies upon administration, leading to accelerated blood clearance (ABC) and hypersensitivity reactions [37, 38]. To observe potential antigenicity and cross‐reactivity, we evaluated the anti‐inflammatory activity of lPG‐LNPs with anti‐PEG antibodies. Previous studies found good agreement between ELISA results using human serum anti‐PEG IgG and commercially available animal anti‐PEG IgG [39]. In the present assay, we used rabbit monoclonal IgG anti‐PEG antibodies, which selectively bind to the terminal methoxy groups of mPEG via Van der Waals and hydrogen bond interactions [40, 41]. Although all lPG‐variations contained terminal methoxy groups, their binding affinity to anti‐PEG antibodies was negligible. ELISA measurements revealed binding concentrations of 0.83 (lPG), 0.50 (MeOlPG), 0.68 (lPG‐block‐MeOlPG), and 0.66 ng mL⁻¹ (MeOlPG‐block‐lPG), respectively (Figure 4). In contrast, PEG‐LNPs at 2 mol% and 0.2 mol% displayed significantly elevated binding activity (1310.55 and 816.81 ng mL⁻¹, respectively). A 100‐fold dilution of PEG‐LNPs was required to reduce anti‐PEG antibody binding to levels comparable to those observed for lPG‐LNPs (0.58 ng mL⁻¹, p < 0.05).

This is a promising result, as mPEG has previously been associated with high anti‐PEG antibody titers compared to other PEG derivatives [22, 42]. Additionally, the C─C─O motif in the PEG backbone is known to promote antibody recognition via hydrophobic interactions, with 3–7 antibody binding sites per monomer [39]. Increased hydrophilicity of the C─C─O unit when substituting PEG with lPG may disrupt such interactions, and remains of interest for future assessments of the antigenicity of lPG‐LNPs [39].

Furthermore, other PEG alternatives such as pOx and pSar have been shown to evade recognition by anti‐PEG antibodies [26, 43]. In line with these findings, our observations show that lPG‐functionalized LNPs may also evade antibody recognition, suggesting a decreased likelihood of ABC effects upon repeat administration [44]. Taken together, these findings further support lPG as a promising PEG alternative for mRNA delivery.

To evaluate lPG‐LNPs for gene delivery applications, cell viability of HepG2 cells co‐incubated with mRNA lPG‐LNP systems was assessed in vitro using a CCK8‐assay (Figure 5). As lipid concentrations potentially deviated across formulations, cell viability was assessed using the mRNA‐concentration of each sample. This approach also facilitated consistency with subsequent transfection experiments, which were performed using equivalent mRNA doses. Given the near‐neutral zeta potential of each LNP formulation as well as the known biocompatibility of lPG, minimal cytotoxicity was expected [45, 46]. Cell viability was over 80% when co‐incubated with mRNA concentrations up to 1 µg mL⁻¹ in vitro, indicating non‐cytotoxic behaviour at the appropriate doses. Specifically, the highest cell viability at 1 µg mL⁻¹ mRNA was observed for lPG‐block‐MeOlPG LNPs (95.4%), comparable to PEG‐LNPs (95.6%). The lowest cell viability at 1 µg mL⁻¹ mRNA was associated with lPG‐LNPs (90.4%) and overall by MeOlPG‐LNPs (83.4% at 0.5 µg mL⁻¹). However, all observed differences across formulations and concentrations were statistically insignificant. These results suggest that modifications of the lPG backbone, such as increased polarity or variations in molecular weight, seemingly had a negligible effect on cell viability.

As cellular uptake and transfection efficiency have previously been described as dependent on PEG chain length or degree of PEGylation, we sought to investigate if similar factors in lPG‐LNPs influence delivery and intracellular mRNA release [30]. To model mRNA delivery, HepG2 cells were transfected with lPG‐LNPs encapsulating eGFP‐encoding mRNA. Subsequent fluorescence microscopy showed successful cellular uptake and translation of mRNA to express eGFP throughout cells (Figure 6a–f). These results indicated transfection ability was maintained when replacing PEG‐ with lPG‐variations in LNP formulations.

As HepG2 cells tend to aggregate and form clusters, a more detailed analysis was obtained by flow cytometry to quantify fluorescence of individual cells, as well as the proportion of eGFP‐expressing cells (Figure 7a). Median fluorescence intensity per cell (Figure 7b) ranged from 3068.0 to 3890.0 across all lPG‐ and PEG‐LNPs alike, independent of polymer length, functionalization, and mRNA dosage. Results indicate effective intracellular delivery and release of mRNA with all tested lPG‐variations, even at concentrations as low as 0.05 µg mL⁻¹ (corresponding to approximately 0.05 pg mRNA per cell for seeding densities of 10 000 cells per well). Previous studies suggest that secondary amine end groups, such as those found in pSar‐functionalized LNPs, can provide a slight positive charge to interact with the negatively charged cellular membrane and enhance mRNA delivery [28]. Given the highly functionalizable backbone of lPG, introducing similarly charged moieties may offer a promising strategy to improve cellular uptake.

While all LNP systems successfully delivered mRNA, delivery efficiency varied between formulations (Figure 7c). Around 43% of cells treated with lPG‐ or MeOlPG‐LNPs exhibited eGFP fluorescence. A dose‐dependent response was observed, with eGFP expression decreasing to around 30% at lower mRNA concentrations. Differences in methylation and side‐group polarity seemingly had a negligible impact, as indicated by the insignificant difference between these formulations. Notably, transfection efficiency was not impacted by the lower initial encapsulation efficiencies and increased particle diameters of MeOlPG‐LNPs.

Interestingly, adding block‐co‐polymers to LNPs slightly reduced transfection efficiency to around 40% at higher mRNA concentrations and around 25% at lower concentrations. Reduced transfection efficiency over 24 h may result from slower transfection rates when using block‐co‐polymers rather than homopolymers in LNP formulations. Although previous studies suggested a dependence on polymer chain length for transfection efficiency, such an effect was not observed—likely due to the modest difference in repeating units of the polymers tested (42 for lPG vs. 27 for MeOlPG), which may be insufficient to significantly influence cellular uptake [25].

The linear polyglycerol derivatives (Poly(ethoxy ethyl glycidyl ether)s (PEEGE) as precursors for both lPGs, and MeOlPGs) were prepared by living anionic polymerization with rac‐2,3‐bis(benzyloxy)propan‐1‐ol 1 as initiator via reaction with the respective oxirane monomer. Initiator 1 (0.25 g, 0.92 mmol) was dissolved in KOtBu in THF (0.20 g, 1.8 mmol in 5 mL THF) under argon atmosphere and heated to 50°C for 2 h to deprotonate the alcohol group. The generated t‐BuOH and solvent were removed in high vacuum. The remaining alcoholate initiator was completely dried, re‐dissolved in dry toluene, and heated to 80°C under argon atmosphere. Then, freshly distilled monomer was completely dried (either rac‐ethoxyethyl glycidyl ether (EEGE, 2, 3.0 mL, 21 mmol) for PEEGE, or rac‐glycidyl methyl ether (3, 1.6 mL, 18 mmol) for MeOlPG), was added to the alcoholate toluene solution and polymerized for 24 h at 80°C under argon atmosphere. The reaction was quenched by first cooling down to rt, followed by the addition of methyl iodide (1 mL) and further stirring at RT for 2 h. After this time, the mixture was concentrated under reduced pressure, and subsequently dried in vacuum. For purification, the obtained oil was dissolved in Et₂O and centrifuged to separate the insoluble salts. Further purification of the polymer was done by flash chromatography followed by dialysis in acetone (1 kDa MWCO, RC tubing) for 3 days. The dibenzyl‐functionalized PEEGE 2a and MeOlPG 3a were obtained as slightly yellow oils with 80% yield (Scheme 1).

Compound 2a¹H‐NMR: (500 MHz, CDCl₃, TMS): δ (ppm) = 1.19 (t, J = 7.04 Hz 128H, OCHCH₃), 1.28 (d, J = 5.24 Hz, 128H, CHCH₃), 3.74‐3.38 (m, 307H, H_a, H_b, H_c, OCH₂CH₃), 4.52 (s, 2H, CH_f), 4.72‐4.5 (m, 42H, CH_e, OCHCH₃) 7.34‐7.26 (m, 10H, Bn). M_n(NMR): 6676 g/mol, n = 42.

Compound 3a 1H‐NMR: (500 MHz, CDCl₃, TMS): δ (ppm) = 3.33 (s, 82H, OCH₃), 3.75‐3.36 (m, 140H, CHCH₃), 4.52 (s, 2H, CH_f), 4.66 (s, 2H, CH_e) 7.33‐7.26 (m, 10H, Bn). M_n(NMR): 2920 g/mol, n = 27.

The copolymerized polyglycerols (consisting of lPG and MeOlPG), were prepared by living anionic polymerization with rac‐2, 3‐bis(benzyloxy)propan‐1‐ol 1 as the initiator via reaction with the respective oxirane monomer. Initiator 1 (0.25 g, 0.92 mmol) was dissolved in 1 M KOtBu in THF (8.2 mL, 8.2 mmol) under argon atmosphere and heated to 50°C for 2 h. The generated t‐BuOH and solvent were removed in a vacuum. The remaining alcoholate initiator was completely dried, re‐dissolved in dry toluene, and heated to 80°C under an argon atmosphere. For copolymer 4a [Bn₂O‐PEEGE‐b‐MeOlPG‐OCH₃], freshly distilled monomer ethoxyethyl glycidyl ether (2.0 mL, 14 mmol) was added to the mixture and stirred for 12 h, followed by glycidyl methyl ether (1.0 mL, 12 mmol) for the second polymer unit of MeOlPG and further stirring for 24 h at 80°C under argon atmosphere. Similarly, for block copolymer 5a [Bn₂O‐MeOlPG‐b‐PEEGE‐OCH₃], glycidyl methyl ether (1.0 mL, 12 mmol) was added to the initiator alcoholate and stirred for 12 h, followed by the addition of ethoxyethyl glycidyl ether (2.0 mL, 14 mmol) for the second polymer unit of PEEGE to polymerize for a further 24 h at 80°C under argon atmosphere.

The reaction was quenched by the reaction of methyl iodide (1 mL), similarly to the synthesis of homopolymers. The mixture was left stirring for 1–2 h at rt, then concentrated under reduced pressure and subsequently dried in vacuum. For purification, the obtained oil was dissolved in Et₂O and centrifuged to separate the insoluble salts. Further purification of the polymer was done by flash chromatography and dialysis in acetone (1 kDa MWCO, RC tubing) for 3 days. The dibenzyl‐functionalized block copolymers 4a and 5a were obtained as slightly yellow oils (Scheme 2).

Compound 4a¹H‐NMR: (500 MHz, CDCl₃, TMS): δ (ppm) = 1.19 (t, J = 7.04 Hz 54H, OR₁(OCHCH₃)), 1.28 (m, 54H, CHCH₃), 3.33 (s, 40H, OR₂(OCH₃)), 3.70‐3.37 (m, 220H, H_a, H_b, H_c, H_d, H_e, H_f OR₁(OCH₂CH₃)), 4.52 (s, 2H, CH_f), 4.70‐4.52 (m, 20H, CH_g, OCHCH₃), 7.34‐7.26 (m, 10H, Ar). M_NMR : 4316, n_(EEGE) = 18, m_(OMe) = 13.

Compound 5a¹H‐NMR: (500 MHz, (CD₃)₂CO, TMS): δ (ppm) = 1.19 (t, J = 7.04 Hz 68H, OCHCH₃), 1.28 (m, 68H, CHCH₃), 3.33(s, 58H, OR₁(OCH₃)), 3.70‐3.37 (m, 271H, H_a, H_b, H_c, H_d, H_e, H_f OR₂(OCH₂CH₃)), 4.50 (s, 2H, CH_f), 4.66‐4.50 (m, 20H, CH_g, OCHCH₃), 7.34‐7.26 (m, 10H, Ar). M_NMR: 5429, n_(OMe) = 19, m_(EEGE) = 22.

The lPG dibenzyl ethers were deprotected via hydrogenation in Toluene with Pd/C (10% w/w) as a catalyst. Compounds 2a, 3a, 4a, and 5a were each dissolved in Toluene and flushed with argon three times. Then hydrogen gas was flushed into the solution and the reaction was left in hydrogen atmosphere at room temperature for 3–4 days. The mixture was filtered through Celite to remove the catalyst, and the filtrate was concentrated under reduced pressure. The diol‐functionalized (Polymer‐OCH₂CH(OH)CH₂OH) were obtained as colorless, viscous oils in quantitative yield after drying in high vacuum.

Compound 2b¹H‐NMR: (500 MHz, CDCl₃, TMS) 1.19 (t, 3H, OCHCH₃,), 1.28 (d, 3H, CHCH₃), 3.74‐3.38 (m, 7H, H_a, H_b, H_c, OCH₂CH₃), 4.72‐4.5 (m, 1H, CH_e, OCHCH₃)

Compound 3b¹H‐NMR: (500 MHz, CDCl₃, TMS): δ (ppm): 3.32 (s, 3H, OCH₃,) 3.67‐3.34(m, 5H, H_a, H_b, H_c,)

Compound 4b¹H‐NMR: (500 MHz, (CD₃)₂CO, TMS): δ (ppm) 1.19 (m, 3H, OR₁(OCHCH₃)), 1.23 (m, 3H, OR₁(CHCH₃), 3.33(s, 3H, OR₂(OCH₃)), 3.70‐3.37 (m, 12H,H_a, H_b, H_c, H_d, H_e, H_f OR₁(OCH₂CH₃)), 4.70‐4.64 (m, 1H, OCHCH₃).

Compound 5b¹H‐NMR: (500 MHz, CDCl₃, TMS): δ (ppm) 1.19 (m, 3H, OR₂(OCHCH₃)), 1.29 (m, 3H, OR₂(CHCH₃), 3.33(s, 3H, OR₂(OCH₃)), 3.74‐3.39 (m, 12H,H_a, H_b, H_c, H_d, H_e, H_f OR₁(OCH₂CH₃)), 4.70‐4.64 (m, 1H, OCHCH₃).

To install the lipid ester moieties, the diol‐functionalized lPG was dissolved in DCM under argon atmosphere and triethylamine (4 equiv.) was added at room temperature. The reaction mixture was stirred for 30 min and then cooled down to 0°C, upon which myristoyl chloride (2.4 equiv) was added dropwise. The mixture was stirred for 2 days at room temperature and then purified by flash column chromatography followed by dialysis in regenerated cellulose membrane with 1 kDa MWCO in acetone to yield the diester products. The methoxylated lPG amphiphile 3c was obtained as colorless oil (80% yield).

Compound 2c¹H‐NMR: (500 MHz, CDCl₃, TMS): δ (ppm) = 0.88‐0.86 (t, 6H, CH₃), 1.19 (t, m, 127H, OCHCH₃), 1.28 (m, 168H, CHCH₃, alkyl chain), 1.63‐1.55 (m, CH_e, CH_e″), 2.32‐2.25 (m, CH_d, CH_d″), 3.69‐3.42 (m, 307H,H_a, H_b, H_c, OCH₂CH₃), 4.72‐4.5 (m, 42H, OCHCH₃).

Compound 3c¹H‐NMR: (500 MHz, CDCl₃, TMS): δ (ppm) = 0.88‐0.86 (t, 6H, CH₃), 1.31‐1.27 (m, 40H, alkyl chain), 1.72‐1.54 (m, CH_e, CH_e″), 2.35‐2.25 (m, CH_d, CH_d″), 3.32 (s, 81H, OCH₃) 3.67‐3.34 (m, 140H, H_a, H_b, H_c,). M_n(NMR) = 2860 g/mol.

Compound 4c¹H‐NMR: (500 MHz, (CD₃)₂CO, TMS): δ (ppm) = 0.88‐0.86 (t, 6H, CH₃), 1.17 (m, 56H, OR₁(OCHCH₃)), 1.28 (m, 110H, OR₁(CHCH₃) alkyl chain), 1.66‐1.56 (m, CH_g, CH_g″), 2.39‐2.24 (m, CH_h, CH_h″), 3.32 (s, 38H, OR₂(OCH₃), 3.67‐3.40 (m, 203H, H_a, H_b, H_c, H_d, H_e, H_f OR₁(OCH₂CH₃)), 4.73‐4.69 (m, 18H, OR₁(OCHCH₃)).

Compound 5c¹H‐NMR: (500 MHz, CDCl₃, TMS): δ (ppm) = 0.88‐0.86 (t, 6H, CH₃), 1.17 (m, 58H, OR₂(OCHCH₃)), 1.28 (m, 98H, OR₂(CHCH₃) alkyl chain), 1.63‐1.52 (m, CH_g, CH_g″), 2.34‐2.23 (m, CH_h, CH_h″), 3.34 (s, 57H, OR₁(OCH₃)), 3.67‐3.40 (m, 278H, H_a, H_b, H_c, H_d, H_e, H_f OR₂(OCH₂CH₃)), 4.72‐4.67 (m, 19H, OR₁(OCHCH₃)).

To obtain the hydroxylated side chain amphiphiles, the EEGE‐protected side chains were deprotected. To a 0.05 mg mL⁻¹ solution of the PEEGE‐based compound in acetone, an aqueous solution of oxalic acid (2 equiv. per repeating unit of EEGE, as a solution in 3:1 acetone:water, to a concentration of 0.15 g oxalic acid mL⁻¹ water) was added dropwise under fast stirring at room temperature. After 3.5 h, the reaction was diluted with deionized H₂O and dialyzed against deionized H₂O (1 kDa MWCO, RC tubing) until a pH of 6 to 7 was reached. The solvent was removed under reduced pressure, affording the respective product 2d/4d/ 5d.

Compound 2d¹H‐NMR: (500 MHz, MeOD, TMS): δ (ppm) = 0.88‐0.86 (t, 6H, CH₃), 1.28 (m, 40H, alkyl chain), 1.63‐1.55 (m, CH_e, CH_e″), 2.32‐2.25 (m, CH_d, CH_d″), 3.69‐3.42 (m, 218H, H_a, H_b, H_c, OCH₂CH₃). M_n(NMR) = 3550 g/mol.

Compound 4d¹H‐NMR: (500 MHz, CD₃OD, TMS): δ (ppm) = 0.95‐0.93 (t, 6H,CH₃), 1,34‐1.26 (m, 40H, alkyl chain), 1.68‐1.64 (m, CH_g, CH_g″), 2.41‐2.37 (m, CH_h, CH_h″), 3.39 (s, 38H, OR₂(OCH₃)), 3.78‐3.47 (m, 167H, H_a, H_b, H_c, H_d, H_e, H_f,). M_n(NMR) = 3060 g/mol.

Compound 5d¹H‐NMR: (500 MHz, CD₃OD, TMS): δ (ppm) = 0.89‐0.87 (t, 6H,CH₃), 1,35‐1.28 (m, 40H, alkyl chain), 1.66‐1.52 (m, CH_g, CH_g″), 2.36‐2.33 (m, CH_h, CH_h″), 3.34 (s, 55H, OR₁(OCH₃), 3.78‐3.47 (m, 178H, H_a, H_b, H_c, H_d, H_e, H_f).

The size of LNPs was measured using dynamic light scattering (Zetasizer Ultra, Malvern Panalytical Ltd) in 1x PBS, reported as the z‐average mean particle diameter. Measurements were repeated at 7 day‐intervals for stability over time, with samples stored at 4°C. Zeta‐potential of 1x LNPs in PBS was measured on a Zetasizer Ultra device using folded capillary zeta cell cuvettes (Malvern Panalytical Ltd.).

A Quant‐iT Ribogreen RNA Assay was used to determine mRNA encapsulation efficiency. Briefly, LNPs were diluted 67‐fold in TE‐buffer. Subsequently, lipid membranes were lysed by adding 2% v/v Triton X‐100 to release encapsulated mRNA. Control samples containing unencapsulated mRNA were diluted with equivalent volumes of TE buffer. Samples were incubated for 10 min at 37°C, before diluting 1:1 with Ribogreen solution. Fluorescence was then quantified by a plate reader (Spark, TECAN) (excitation 485 nm; emission 528 nm), and encapsulation efficiency (EE) was calculated following Equation (): 1

Chloroform‐cleaned 300 mesh R1.2/1.3 holey carbon grids (Quantifoil, MicroTools GmbH, Jena, Germany) were hydrophilized by glow discharging in an EMSCOPE SC500 (10 mA, 60 s) before applying 4 µL aliquots of mRNA‐LNP solution to grids. Samples were vitrified by automatic blotting (blot force ‐11, blot time 3.5 s, wait time 3 s) and subsequent plunge freezing with an FEI Vitrobot Mark IV (ThermoFisher Scientific) using liquid ethane as cryogen. Grids were clipped with copper clip rings and C‐clips to obtain Autogrids. Vitrified samples were transferred to the autoloader of an FEI TALOS ARCTICA electron microscope (Thermo Fischer Scientific), equipped with a high‐brightness field‐emission gun (XFEG) operated at an acceleration voltage of 200 kV. Micrographs were acquired on a FEI Falcon 3 direct electron detector (Thermo Fischer Scientific) at normal magnifications of 3400, 4300, and 28 000, corresponding to calibrated pixel sizes of 31.3, 24.6, and 3.7 Å pixel⁻¹, respectively.

Binding of LNPs to anti‐PEG antibodies was measured using a competitive ELISA assay according to the manufacturer's instructions. Briefly, LNPs were diluted 1:1 with PEG‐BSA standard solution before transferring 50 µL to the provided 96‐well plate. The plate was incubated on a plate shaker (400 rpm) for 45 min at rt. Afterward, liquid was aspirated out of each well and the plate was washed using wash buffer three times. Subsequently, 100 µL TMB substrate was added to each well, before incubating on a plate shaker (400 rpm) in the dark for 15 min. Finally, 100 µL stop solution was added per well and mixed for 1 min on a plate shaker, before measuring optical density at 450 nm on a Tecan Spark plate reader.

HepG2 cells (Leibnitz Institute DSMZ – German Collection of Microorganisms and Cell Cultures GmbH) were cultured at 37°C and 5% CO₂ in high glucose Dulbecco's Modified Eagle Medium with GlutaMAX supplement (#31966‐021, Gibco/Thermo Fisher Scientific) containing 10% v/v fetal bovine serum (#10270‐106, Gibco/Thermo Fisher Scientific) and 1% v/v penicillin‐streptomycin (#P4333‐100ML, Sigma–Aldrich/Merck). Once cells reached 70% confluency, cells were ready for seeding in experiments.