Buffer Specificity of Ionizable Lipid Nanoparticle Transfection Efficiency and Bulk Phase Transition

LNPs are composed of an external layer composed of PEG polymer with DSPC and MC3 lipids with an internal bulk phase composed of DLin-MC3-DMA and cholesterol. MC3 is an ionizable lipid with a pK_a of 6.44,³² which makes it pH-sensitive (FigureA,B). The LNPs are internalized inside living cells at pH ∼ 7 until reaching the endosome, where mRNA is released at pH ∼ 6–6.5 (FigureA). To study the buffer effects, three common buffers were chosen (FigureC,D).

According to the Henderson–Hasselbalch equation (pH = pK_a + log[A⁻]/[HA]), citrate, phosphate, and acetate with pK_a of 6.40, 7.22, and 4.80, respectively, cover the whole pH range of the LNP route from internalization (pH 7) to mRNA release (pH 6). Sodium citrate buffer has been used in cationic lipid design for siRNA delivery¹ and is by far the most used for LNP preparation at acidic pH. Sodium phosphate buffer was used with KCl and NaCl to store Pfizer’s mRNA vaccine.³⁰ Sodium acetate buffer has been used to dilute mRNA for LNP encapsulation³³ and as a dialysis buffer for ethanol removal after LNP preparation.³⁴

To investigate buffer-specific effects on transfection kinetics, we prepared eGFP-mRNA LNPs in citrate, phosphate, or acetate buffer as described previously,¹⁴ followed by dialysis into water. We performed live-cell imaging on single-cell arrays (LISCA) after transfection and measured protein expression in a time-resolved manner.³⁵ To this end, LNPs were preincubated in a cell culture medium supplemented with serum and transfected into the human liver carcinoma cell line (HuH7) seeded on a single-cell slide (FigureA).

Observation of the GFP expression kinetics of single cells resulted in a fluorescence trajectory for each cell (FigureB). Distinct differences between the buffers were apparent from these traces. Averaging all traces revealed the highest overall GFP level for LNPs prepared in citrate buffer followed by those prepared in phosphate buffer and the lowest expression from the acetate LNPs. The single-cell resolution allowed for the calculation of the total protein amount per cell expressed as the area under the curve (AUC) and the distinct onset of protein expression for every single cell (FigureC). Protein expression varied depending on the buffer in which the LNP was prepared, with the highest eGFP expression for the citrate, followed by phosphate, and lowest for the acetate buffer. FigureD shows the mean AUCs, in relative units normalized to the AUC in citrate, for all three ionizable lipids versus the mean expression onset. The buffer-specific effect appears to follow a conventional Hofmeister series with citrate > phosphate > acetate for the expression. The observed ordering in the onset times exhibits an inverse relation between the expression efficiency and expression onset time (FigureD). This confirms a correlation between fast onset and high protein expression levels as previously reported in time-resolved studies.³⁶ To rule out buffer-specific effects on LNP preparation, LNP encapsulation (FigureE) and particle size (Table S1↗) were measured as controls. Note the encapsulation efficiencies and sizes are not affected by buffers within the accuracy of the measurement. In the case of MC3 the eGFP-mRNA LNPs encapsulation efficiency in buffer, 50 mM at pH 3, were (94 ± 4)% in citrate, (79 ± 8)% in phosphate and (97 ± 2)% in acetate respectively.

In summary, these findings indicate that the underlying endosomal release mechanism but not the formulation is dependent on the LNP preparation buffer.

Next, we investigate the impact of buffers on the phase behavior of ionizable lipid/cholesterol bulk phases as a function of pH. These two-component bulk phases serve as a model for the inner core structure of LNPs, given that PEG and DSPC are restricted to the outer shell of the LNPs. As described in the methods section, the preparation of bulk phase samples consists of three dialysis steps to mimic LNP production by replacing the ethanol with buffer. The buffer (citrate, phosphate, or acetate) at the pH of choice comes into play in the third dialysis step. FigureA presents the SAXS scattering profiles for MC3/cholesterol/buffer phases in the presence of citrate, phosphate, and acetate buffers. In the following sections, we describe the pH-dependent structural transitions observed. As shown in FigureB decreasing the pH induces a phase transition from inverse micelles (L_II)³⁷ to inverse hexagonal phase (H_II).³⁷ The inverse micellar phase undergoes ordering transitions from a disorder L_II through close-packed P6₃/mcc^37,38 toward an inverse cubic Fd3m phase.³⁹ This order of inverse phases as a function of pH has been reported before for MC3/cholesterol in citrate buffer.¹⁰ Here we find that the same behavior is also found in phosphate and acetate buffer but with slightly shifted pH values of transitions. In citrate buffer, P6₃/mmc and Fd3m phases are formed at pH 7.0 and pH 6.5, respectively, with decreasing pH. The same transitions are observed in the presence of acetate, but at a lower pH (P6₃/mmc at pH 6.5 and Fd3m at pH 6.0). A bigger difference is observed for phosphate, resulting in the L_II phase being found in the region from pH 7.5 to 6.0 and a mixed phase P6₃/mmc + Fd3m at pH 5.5.

This inverse cubic-to-inverse hexagonal (Fd3m–H_II) transition occurs at pH= 6.0 for citrate and at pH = 5.0 for acetate and phosphate (FigureA–C). Remarkably the critical pH value is shifted by one pH unit for acetate and phosphate compared to that for citrate. To demonstrate that the transition from inverse micellar (L_II) to inverse hexagonal phase (H_II) is a universal characteristic of clinically relevant ionizable lipids we also present the phase behavior of ALC-315/cholesterol and SM-103/cholesterol for all three buffers. We observe that the specific buffer effect on the L_II–H_II transition follows the trend pH (citrate) > pH (phosphate) = pH (acetate). This behavior can be explained by the common conic lipid shape factor which promotes inverse micellar phases at neutral pH.

The dimensions of the bulk phase are characterized by the nearest neighbor distance (d_NN), determined by extracting the lattice constants from SAXS measurements (Table S3↗). d_NN is defined as the smallest distance between neighboring water cores or channels. Significant differences in d_NN as a function of pH are seen in FigureD. We observe a general trend of increasing d_NN with a decrease in pH throughout all lipids and buffers. This general trend is caused by the increasing charge at the ionizable lipid’s headgroup with decreasing pH. As described later in the theoretical section increasing electrostatic repulsion reduces the curvature of the lipid–water interface and causes an increasing lattice spacing. We find two distinct exceptions to this rule in the data. The first exception to this trend is the transition from Fd3m to H_II which is accompanied by a reduction of d_NN. This discontinuity is due to the packing symmetry of H_II in contrast to the face-centered cubic Fd3m symmetry. We will come back to the associated energies in the inverse micellar to inverse hexagonal transition in the section on the mechanism of the pH transition. The second exception is the anomalous behavior found in the case of the H_II phases of MC3 in acetate buffer, which appears to shrink with increasing protonation.

Charging of the lipid head groups, and consequently, the geometric packing parameter of the lipids is influenced by pH, temperature, and salt concentration. In the following, we examine these parameters in detail in the case of MC3 lipid. The previous experiments were carried out by using a fixed concentration of 50 mM for all buffers at different pH values. However, the charge (z) and the concentration (c) ratio between the acidic and the basic components of the buffer depends on pH (FigureC,D). For instance, at pH 5.5, the ionic strength for the different 50 mM buffers is 151 mM for citrate, 52 mM for phosphate, and 42 mM for acetate. To understand if the observed buffer specificity (Figure) could be ascribed to an ionic strength effect of the different 50 mM buffers, SAXS measurements were performed of the LNP bulk phases dialyzed in the presence of citrate and acetate at pH 5.5 and at a defined ionic strength (namely 160, 200, and 300 mM). pH 5.5 was chosen since all buffers showed a clearly ordered structure (Fd3m or H_II). Figure S1A↗ shows that, for citrate buffer, H_II is the only phase occurring at pH 5.5 for the three ionic strengths. In the case of acetate (Figure S1B↗), Fd3m and H_II phases coexist at all ionic strengths, but an increase in ionic strength results in an increase in the signal of the inverse hexagonal structure compared to the Fd3m phase. The results indicate that electrostatic interactions play a role in driving the pH-dependent structural transition. It is known for triethanolamine buffer that an increase of ionic strength results in a shift in the equilibrium of −NH⁺ ⇌N + H⁺ toward the left, corresponding to a shift to higher effective pH (that is lower H⁺ activity).⁴⁰ Similarly, in our system, an increase in ionic strength would enhance the protonation of the MC3 headgroup which would stabilize H_II with respect to the Fd3m phase. We also investigated the effect of temperature at 22 and 37°C under pH 5.5. It is important to examine whether the transitions of the inner bulk phase of LNPs, which we consider the driving mechanism of mRNA transfection, occur in the same pH range observed at lab temperature compared to body temperature. In Figure S1B↗, we find the temperature increase favors the phase transition from inverse hexagonal H_II to Fd3m + H_II in the case of citrate and from Fd3m + H_II to Fd3m for acetate. That is, the temperature increase has the opposite effect of an increase in ionic strength (Figure S1A↗), tending to stabilize F3dm and shifting the Fd3m-H_II transition to lower pH values. The temperature dependence is consistent with the trend we would expect from the van’t Hoff equation for the temperature dependence of an acid equilibrium.⁴¹ The −NH⁺ ⇌N + H⁺ equilibrium is pushed toward dissociation at higher temperatures, leaving the MC3 molecule uncharged. This can be interpreted as an equivalent to pushing the system toward the behavior that would be expected at a lower pH. In summary, the effect of increasing ionic strength is to favor the inverse hexagonal phase, i.e., shift the phase transition to higher pH values, while that of increasing temperature is to favor the inverse micellar phases, i.e. pushing the phase transition to lower pH.

In the following, we discuss the mechanism that drives the pH-dependent L_II–H_II transition in lipid mesophases. We then ask the question of why this mechanism is buffer-specific. Transfection efficiency measurements (FigureD) showed a buffer-specific effect that follows a (conventional) Hofmeister series with citrate > phosphate > acetate. In SAXS measurements, this ordering is also observed for the pH value of the MC3 bulk phase transitions Fd3m–H_II. The bulk phase transitions observed by changing pH and ionic strength involve the MC3 charge, but crucially also involve a change in the curvature and area per MC3 headgroup. The balance between these properties determines the phase transition pH. Then, buffer specificity introduces shifts to that transition pH via the specific adsorption of buffer ions at the water–lipid interface (FigureA). The formation of the charge on MC3 at low pH favors the H_II phase over the inverse micelle phases due to the higher mean curvature of the latter, which gives the inverse micelle phases a larger positive electrostatic energy. But at the same time, the curvature of the interface and area per headgroup affects the L_II–H_II transition via the elastic bending energy of the lipid layer,²⁸ favoring the inverse micelle phases over the inverse hexagonal phases.

In general, energy is required if the lipid monolayer bends away from its preferred curvature. The bending energy is characterized by bending moduli κ and κ_G, and may be quantified via a Helfrich bending energy, E_bend = κ(C₁ + C₂ – C₀)²/2 + κ_GC₁C₂, where C₁ = 1/R₁ and C₂ = 1/R₂ are the curvatures associated with the radii R₁ and R₂ in perpendicular directions along the surface of the lipid layer layer and C₀ is the so-called spontaneous total curvature due to the conical lipid shape. Although spherical inverse micelles with C₁ = C₂ have lower individual curvatures (larger radii, corresponding to larger d_NN values in the case of MC3), the cylinders of the inverse hexagonal phase have a lower total curvature (C₁ + C₂) because of the flat dimension along the axis of the cylinders with C₂ = 0 (see also cartoon in FigureA). Protonation of the ionizable lipid headgroup will decrease the spontaneous curvature and drive the system from L_II into the H_II phase. The exact transition point depends on the ratio κ_G/κ, i.e., the contribution of Gaussian curvature and mean curvature to the total bending energy.^42,43 The electrostatic energy due to the headgroup charge favors the formation of the inverse hexagonal phase at low pH, while the higher spontaneous curvature favors the formation of the inverse micelle phases at high pH. The balance between the two explains the broad trend of the MC3 phase transitions, with the transition point lying at a pH close to the MC3 pK_a inside the LNPs of 6.44.⁴⁴

To explain the buffer-specific order observed for bulk phase transitions, we propose a mechanism of specific ion adsorption with a consequent buffer-specific change in the area per headgroup. As illustrated in FigureA, we identify near (charged N group) and far (O ester group) moieties within the headgroup. The N-moiety remains in contact with the aqueous phase, whereas the O-moiety may not be in contact, depending on the headgroup orientation (FigureA). The strength of interactions of buffer ions with the lipid surface, including the N-moiety, follows the conventional series citrate > phosphate > acetate, confirmed by our MD simulations (FigureB). On the other hand, MD simulations (FigureC), and evaluation of London dispersion coefficients of ions (Table S4↗) with the O-moiety, suggest that adsorption at the O-moiety would be stronger for acetate than other buffer ions. Adsorption of acetate at the O-moiety requires the ion to penetrate deeper into the lipid phase and, therefore, is expected to lead to an increase in the area per headgroup. In this way, acetate stabilizes L_II, consistent with SAXS data. At the same time, the adsorption of negatively charged ions at the N-moiety is expected to reduce the repulsion between the positively charged MC3 headgroup and will therefore reduce the area per headgroup at low pH. Consequently, citrate anions with preferred adsorption at the (positively charged) N-moiety (FigureB) decrease the area per headgroup, increasing the bending modulus and shifting the Fd3m-H_II transition to a higher pH by stabilizing H_II. Acetate ions with preferred adsorption at the O-moiety, likely involving London dispersion forces, increase the area per headgroup, decreasing the bending modulus and shifting the Fd3m-H_II transition to a lower pH by stabilizing Fd3m. This mechanism, involving two types of specific interactions between the buffer ions and the lipid headgroups, in particular, with the positively charged N-moiety (via electrostatic interactions) and the neutral O-moiety (via nonelectrostatic dispersion forces), is in agreement with the results obtained by investigating the effect of ionic strength. Indeed, results in Figure S1↗, showing different SAXS patterns for the same ionic strengths obtained with citrate and acetate, confirm the occurrence of different mechanisms for the two buffers. The inverse hexagonal H_II phase observed for citrate at the three investigated ionic strengths suggests its preferential interaction with the positively charged MC3 headgroups. By contrast, the coexisting Fd3m and H_II phases, observed for acetate at the three ionic strengths, are consistent with a partial stabilization of the neutral form of the MC3 headgroup. In short, we can identify a direct buffer effect mediated via electrostatics and ion dispersion interactions and an indirect effect via a consequent change in the area per headgroup.

To provide evidence of our hypothesis of specific ion adsorption at the lipid/water interface, we performed all-atom MD simulations in explicit water. To ensure the correct protonation degree of the ionizable MC3 lipid, we chose an MC3/POPC monolayer system (FigureA) for which the area per lipid and protonation degree at pH 5.0 and 7.5 were determined consistently in previous work.⁴⁵ In the simulations, citrate buffer was represented by the trivalent citrate ion, phosphate buffer by HPO₄^2–, and acetate buffer by the monovalent acetate ion. The affinity of the ions in terms of the probability distribution perpendicular to the lipid/water interface at pH 5.0 and 7.5 is shown in FigureB,C. The main adsorption peak at z = 0 nm corresponds to the interactions of the ions with the nitrogen moiety and follows a conventional Hofmeister ordering: citrate > phosphate > acetate. A second minor adsorption peak at z = −2 nm appears for acetate and corresponds to the interaction with the oxygen moiety. This peak is absent for the other ions and significantly smaller compared to the main adsorption peak. Further insights into the adsorption behavior at the interface can be gained from the local radial ion distributions g(r) around the charged and uncharged MC3 molecules. FigureD,E shows g(r) for the buffer ions around the N and O moieties (O1 representing the carbonyl O-moiety and O2 representing the ester O-moiety). For neutral MC3, the distributions at N- and O-moieties are similar (FigureD).

For charged MC3 (FigureE), pronounced differences between the ions and the different binding sites can be observed. The results show that citrate ions have a higher affinity toward the N atom in positively charged MC3 compared with the other buffer ions (as indicated by the highest peak in g(r) at a radial distance r = 0.28 nm in FigureE, top). Phosphate ions have the second highest affinity for the N-moiety followed by acetate. In the case of O-moieties, only acetate ions adsorb, while citrate and phosphate are depleted (FigureE, bottom). The simulation snapshots reveal that citrate ions are located at the lipid/water interface, while the acetate ions penetrate deeper into the lipid phase (FigureF,G). In summary, the affinity of buffer ions toward a lipid layer containing MC3 follows the Hofmeister series: citrate > phosphate > acetate. However, the local distribution of the ions is much more complex. The adsorption of acetate and depletion of phosphate and citrate at the oxygen is likely the cause of ion-specific buffer effects.

DLin-MC3-DMA(O-(Z,Z,Z,Z-heptatriaconta-6,9,26,29-tetraem-19-yl)-4-(N,N-dimethylamino) butanoate; (MC3, 99%), ALC-315 ([(4-hydroxybutyl)azanediyl]di(hexane-6,1-diyl) bis(2-hexyldecanoate)), and SM-102 (9-Heptadecanyl 8-{(2-hydroxyethyl)[6-oxo-6-(undecyloxy)hexyl]amino}octanoate) were purchased by MedChemExpress. 1,2-distearoyl-sn-glycero-3-phosphocholine, (DSPC, 99%), 1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N-[amino(polyethylene glycol) (DSPE-PEG2000, 99%), cholesterol, sodium citrate dihydrate (99%), citric acid (99%), monobasic sodium phosphate (99%), disodium hydrogen phosphate (99%), hydrochloric acid (37%), sodium hydroxide (97%), sodium acetate (99%), acetic acid (99.8%), and citric acid (99.5%) were purchased by Avanti Polar lipid Sigma-Aldrich.

ARCA eGFP (Enhanced Green Fluorescent Protein) mRNA (APExBIO) was encapsulated in an LNP of four lipid components (DLin-MC3-DMA, DSPC, Cholesterol, and DMPE-PEG2000) in a 50:10:38.5:1.5 molar ratio. mRNA and buffer (pH 3) were prepared in aqueous solution (0.075 mg/mL) while the lipid components (1.11 mg/mL MC3, 0.27 mg/mL DSPC, 0.52 mg/mL cholesterol, and 0.15 mg/mL DSPE-PEG2k) were dissolved in ethanol. Concentrations were chosen to reach a final LNP concentration of 0.05 mg/mL mRNA concentration with an N/P ratio of 4 and a final buffer concentration of 50 mM. Microfluidic mixing was carried out with the NanoAssemblr Spark (Precision NanoSystems) at a volume ratio of aqueous:organic 2:1 ratio. Following mixing, the LNPs were incubated for 20 min at room temperature. To remove residual ethanol and buffers from the solution, LNPs were transferred to Slide-A-Lyzer MINI dialysis cups with 3.5 kDa molecular weight cutoff (ThermoFisher Scientific) and dialyzed into water for 18 h at room temperature. Size distribution was measured using a DynaPro NanoStar (Wyatt) DLS device. Encapsulation efficiency was assessed using the Quant-it RiboGreen RNA dye (Invitrogen).

Cells were cultured in RPMI 1640 Medium (ThermoFisher Scientific) supplemented with 10% (v/v) FBS (fetal bovine serum, ThermoFisher Scientific, no. 10270106), 5 mM HEPES (GibcoTM, Thermofisher Scientific, #15630080), and 1 mM Na-Pyruvate (GibcoTM, Thermofisher Scientific, #11360070) at 37°C, 5% CO₂. For live-cell imaging, microstructures were prepared to allow single-cell culture. Therefore, six-channel μ-slides (ibidi) coated with cell-repellent PVA were treated with PLPP (N-(4-[benzoyl]benzyl)-N,N,N-triethylammonium bromide) (enamine) in an agarose and calcium peroxide solution and selectively illuminated with UV light (365 nm). Selective illumination was facilitated with a photomask patterned with 20 × 20 μm squares and an 80 μm spacer. After illumination, the channels were rinsed with water and 0.5 M HCl. The resulting squares were coated with laminin by incubation in a 20 μg/mL laminin (BioLamina) working solution in PBS for 1 h at 37°C.

Application of cells in the medium for 1 h led to self-assembly in a single cell pattern, as depicted in FigureA. eGFP-mRNA-LNPs were diluted in RPMI medium supplemented with FCS to a final concentration of 1 ng/μL and incubated for 1 h at room temperature to allow the formation of a protein corona. Subsequently, the LNP solution was applied for 1 h and washed afterward with L15-medium without phenol red (ThermoFisher Scientific). Cells were transferred to the microscope (Nikon TI Eclipse) and imaged over 30 h with image acquisition every 10 min. Image analysis was then performed using our in-house Python-based software including segmentation and background correction based on Schwarzfischer et al.⁴⁶ to generate fluorescence trajectories.

MC3/cholesterol bulk phases were prepared under a range of pH conditions via three dialysis steps. First, the MC3 and cholesterol were dissolved in ethanol and mixed in a molar ratio of 3:1 (MC3: cholesterol) to a total lipid concentration of 56.1 mg/mL (MC3 46.7 mg/mL and cholesterol 9.4 mg/mL). The mixture was put into a dialysis cup with a molecular weight cut off of 3.5 kDa. Samples were first dialyzed against a 50 mM citrate buffer (pH 3) containing ethanol (at a volume ratio of 3:1, buffer: ethanol) for 48 h. In the second dialysis step, the sample was dialyzed against PBS (1 mM KH₂PO₄, 155 mM NaCl, 3 mM Na₂HPO₄ 0.7 H₂O, pH 7.4) for 48 h. In the third step, the sample was dialyzed against NaCl 150 mM and the buffer of choice (citrate, acetate, or phosphate 50 mM at pH 3.5, 4.0, 4.5, 5.0, 5.5, 6.0, 6.5, 7.0) with the required final pH for 48 h. The samples at different ionic strengths (I) were prepared at different acetate, phosphate, and citrate concentrations to reach I = 160 mM, I = 200 mM, and I = 300 mM including the background salt NaCl 150 mM. The supernatant was removed from the cup and the solid precipitates were extracted for characterization by SAXS.

Synchrotron small-angle X-ray scattering (SAXS) was carried out at the P12 EMBL BioSAXS and P62 SAXSMAT beamlines, PETRA III, DESY (Hamburg, Germany). The beamline instrumentation has been described previously.^47,48 Further SAXS experiments were performed using an internal instrument at LMU. The crystallographic space groups of the liquid crystalline phases were determined from the relative peak positions. All of the measurements at P12 and P62 were performed in quartz capillaries. The scattering data background was subtracted by measuring the empty capillaries.

The nearest neighbor distance was calculated for each mesophase using the peak positions with their respective Miller indices following the formula shown in Tables S2 and S3↗ (see the Supporting Information↗).¹⁴

Ion-specific adsorption simulation setup: To investigate the adsorption of acetate, phosphate, and citrate at the lipid/water interface, we used a monolayer setup. The lipid monolayer contained MC3 and POPC lipids in a 1:4 molar ratio and was constructed using the Men-Gen web server⁴⁹ following the published procedure of our previous work.⁴⁵ Each leaflet of the monolayer contained 200 lipids separated by a water column of approximately 15 nm. The reason for choosing this setup was that the protonation degree at two pH values was determined consistently from a combination of simulations and scattering experiments.⁴⁵ This was crucial since the degree of protonation is inherently difficult to predict since it depends on the local lipid environment and is not directly accessible from experiments. Specifically, the lipid monolayer at pH 5 has a protonation degree of 67.5% ionizable MC3. At pH 7.5, the protonation degree is 14.5%. Both values significantly deviate from simple theoretical predictions. For each pH value, the monolayers were simulated at a buffer concentration of 50 mM. Sodium acetate (CH₃COONa), sodium phosphate (Na₂HPO₄), and sodium citrate (Na₃C₆H₅O₇) were used in the experiments. Thus, a total of six simulations of the lipid monolayer were performed, covering two pH levels and three different buffers. Each simulation was repeated three times to improve the sampling statics.

The AMBER Lipid 17 force field was used to describe the POPC lipids. For the cationic and neutral MC3 molecules, we used recently developed force fields, which closely reproduce the experimental structure of lipid layers and are compatible with the AMBER force field family.⁵⁰ The TIP3P water model⁵¹ along with Mamatkulov–Schwierz⁵² force field parameters for Na⁺ ions was also used. The citrate ion was described using the force field parameters obtained from Wright⁵³ while the acetate and phosphate ion force field parameters were obtained from Kashefolgheta.⁵⁴ The combination rule for the anion–cation interactions was modified to avoid crystallization artifacts. The resulting force field parameters are available at “(https://git.rz.uni-augsburg.de/cbio-gitpub/force-fields-buffer-ions↗).”

All-atom molecular dynamic simulations were performed using the GROMACS⁵⁵ package (v-2024). A gradient descent algorithm was used to minimize the energy of the system. The simulations were performed in the NVT ensemble to ensure the correct area per lipid.⁵⁰ The temperature was maintained at 293.15 K using the velocity rescale thermostat with a time constant of 1.0 ps. The Lennard–Jones potential was cut off and shifted to zero at 1.2 nm. Short-ranged electrostatic interactions were cutoff at 1.2 nm, and the Particle Mesh Ewald (PME) method was used to evaluate long-range electrostatics. Hydrogen bonds were constrained using the LINCS algorithm, and a time step of 2 fs was used. Each production run was performed for 70 ns. The first 20 ns of the simulation were discarded to account for equilibration and the rest of the trajectory was analyzed using GROMACS inbuilt modules and MDAnalysis.⁵⁶ The trajectories were visualized, and snapshots were generated using visual molecular dynamics (VMD).⁵⁷

Area per headgroup of H_II and L_II phases: to obtain the area per headgroup, we used data from our previous work,¹⁴ where we simulated the inverse hexagonal (H_II) and inverse micellar (L_II) phases. The H_II phase consisted of fully protonated DLin-MC3-DMA (MC3) lipids combined with cholesterol in a 3:1 molar ratio, maintaining a water-to-lipid ratio (n_w) of 12 to match the experimental lattice spacing of 60 Å. The L_II phase consisted of fully uncharged MC3 lipids using also n_w = 12. Details on the calculation of the area per headgroup are provided in the Supporting Information↗.

Content submitted to preprint server bioRxiv, Authors: Cristina Carucci, Julian Philipp, Judith A. Müller Akhil Sudarsan, Ekaterina Kostyurina, Clement E. Blanchet, Nadine Schwierz, Drew F. Parsons, Andrea Salis, Joachim O. Rädler. Title: Buffer specificity of ionizable lipid nanoparticle transfection efficiency and bulk phase transition. 2025, DOI: 10.1101/2025.01.17.63350. Repository: bioRxiv, the preprint service for biology. 10.1101/2025.01.17.633509↗ (version accessed January 21, 2025).