How Storage Conditions Affect the Stability of Liquid mRNA Lipid Nanoparticles

A pH-sensitive cationic lipid, TOT-28, was synthesized as described in the Supplementary Information. MedChemExpress (Monmouth Junction, NJ 08852, USA) supplied the DLin-MC3-DMA (MC3) (catalog No. HY-112251) and ALC-0315 (ALC) (catalog No. HY-138170). Cholesterol (chol) was purchased from SIGMA Aldrich (St. Louis, MO, USA, catalog No. C8667-25G). The NOF Corporation (Tokyo, Japan) supplied 1,2-distearoyl-sn-glycero-3-phosphocholine (DSPC, catalog No. MC-8080) and PEG_2k-DMG (catalog No. GM-020). Ribogreen was purchased from Molecular Probes (Eugene, OR, USA, catalog No. R11491). MedChemExpress (Monmouth Junction, NJ, USA) supplied the 2-(p-toluidino)-6-napthalene sulfonic acid (TNS) (catalog No. HY-W250727). DCVJ was purchased from SIGMA Aldrich (St. Louis, MO, USA, catalog No. 72335). The 6-dodecanoyl-2-dimethylaminonaphthalene (Laurdan) was purchased from invitrogen (Eugene, Oregon, USA, catalog No. D250). Firefly luciferase (Fluc)-encoding mRNA (CleanCap, 5moU modified) was purchased from Trilink BioTechnologies (San Diego, CA, USA, catalog No. L-7202). An iLiNP microfluidic device was fabricated and used for microfluidic mixing, as previously described [33,34].

The mRNA-loaded LNPs were prepared, as previously described [34]. Briefly, an ethanol solution containing an ionizable lipid, DSPC, chol, and PEG_2k-DMG at Mol% ratios of 50/3/47/1.5, respectively, was prepared to reach a total lipid concentration of 8 mM. The FLuc-mRNA was dissolved in 50 mM of citrate buffer (pH 4.0). The lipid mixture was combined with a mRNA aqueous solution using a glass-based iLiNP device at a total flow rate of 2 or 5 mL/min with a flow-rate ratio (RNA to lipid) of 3. The nitrogen-to-phosphate ratio was adjusted to 6. Syringe pumps (Harvard apparatus, MA, USA) were used to control the flow rate of the solutions. The resultant LNP solution was then dialyzed against 20 mM of Tris-HCl buffer (pH7.0) containing 9% (w/v) sucrose using a Slide-A-Lyzer MINI Dialysis Device (molecular weight cut-off MWCO 10 kDa) (Thermo Fisher Scientific, Rockford, IL, USA, catalog No. A52977) for 2 to 24 h in order to reconstruct the LNP structure by neutralization. The LNPs were concentrated via ultrafiltration using an Amicon Ultra-15 unit (MWCO 100 kDa) (Merck Millipore Ltd., Carrigtwohill, Co. Cork, Ireland, UFC910096) as needed. The LNP solution was then dialyzed against citrate-Tris buffer (pH 5.0, 5.5, 6.0, 6.5, or 7.0) containing 9% (w/v) sucrose. Sucrose was used to make the storage buffer isotonic.

The ζ-average size, polydispersity index (PdI), and ζ-potential of the LNPs were measured using a Zetasizer Nano ZS ZEN3600 instrument (Malvern Instruments, Worcestershire, UK). To reduce the influence of storage buffer pH during ζ-potential measurement, all LNPs were diluted 250-fold with 10 mM HEPES buffer (pH 7.4). The encapsulation efficiency and total concentration of mRNA were measured by Ribogreen assay, as described previously [34].

The apparent pKa of the TOT-28-LNPs was determined by TNS assay. Briefly, 30 μM of LNP lipids and 6 μM of TNS were mixed in 50 μL of 20 mM citrate buffer, 20 mM sodium phosphate buffer, or 20 mM Tris-HCl buffer, containing 130 mM NaCl, at pH values ranging from 3.5 to 9.5 in a black 384-well plate by means of an epMotion 5070 (Eppendorf SE, Hamburg, Germany). Fluorescence (excitation and emission wavelength of 321 and 445 nm, respectively) was read using a microplate reader (Varioskan Lux, Thermo Fisher Scientific, Waltham, MA, USA) at 37 °C. The pKa value was determined as the pH at which the half-maximum fluorescence intensity was reached.

BALB/c mice (female, 5–6 weeks old) were sourced from Japan SLC (Shizuoka, Japan). Mice were maintained on a regular 12 h light/12 h dark cycle in a specific animal facility at Hokkaido University. Four to five mice were housed in each cage. The mice were fed a pelleted mouse diet (cat# 5053, LabDiet, Gray Summit, MO, USA) and water was available ad libitum.

Mice were injected with Fluc mRNA-loaded LNPs at a dose of 5 µg mRNA/mouse into the caudal thigh muscle (50 μL injection volume). At 20 h post-injection, bioluminescence imaging was performed using an in vivo imaging system (IVIS Lumina III, PerkinElmer, Waltham, MA, USA) 20 min after intraperitoneal injection of D-luciferin at a dose of 150 mg/kg under anesthesia with three types of mixed anesthetic agents: 0.3 mg/kg of medetomidine, 4.0 mg/kg of midazolam, and 5.0 mg/kg of butorphanol.

DCVJ labeling of the LNPs was performed by adding 0.5 mM of DCVJ dissolved in DMSO to the mRNA-loaded LNP solution (1 mM as a total lipid concentration) to yield a lipid: DCVJ molar ratio of approximately 400:1. For the fluorescence measurements, 200 μL of the DCVJ-labeled LNPs (1 mM and 2.5 μM as total lipid and DCVJ, respectively) were added to black 96-well plates. The plates were incubated for 30 min or more at either 4 or 25 °C and fluorescence spectra (excitation and emission wavelength of 440 nm and 460–650 nm with 2 nm interval, respectively) were then measured using a microplate reader (Varioskan Lux, Thermo Fisher Scientific, Waltham, MA, USA). To standardize the DCVJ concentration, the DCVJ-derived absorption spectra (350–550 nm with 2 nm interval) were also measured after solubilizing the LNPs by adding 20 μL of 10% sodium dodecyl sulfate (FUJIFILM Wako Pure Chemical Corporation, Osaka, Japan) followed by incubation at 60 °C for 10 min with gentle mixing at 700 rpm. To estimate the viscosity, R (fluorescence to absorption ratio) was calculated following a method described in the literature [35]. Specifically, the integral of the fluorescence spectrum from 460 to 650 nm was divided by the integral of the absorption spectrum from 400 to 510 nm.

Laurdan labeling of the LNPs was performed by adding 1 mM of laurdan dissolved in DMSO to the lipid mixture (8 mM as a total lipid concentration) in ethanol to yield a lipid/laurdan molar ratio of 500:1 before formulation. For the fluorescence measurements, 200 μL of the laurdan-labeled LNPs (0.1 mM as a total lipid concentration, 0.2 μM as laurdan) was added to black 96-well plates. The plates were incubated for 30 min or more at either 4 or 25 °C and fluorescence was then measured using a microplate reader (Varioskan Lux, Thermo Fisher Scientific, Waltham, MA, USA). The samples were excited at 340 nm and the fluorescence spectra from 360 to 600 nm was obtained. To estimate the hydration degree of the LNP membrane, the generalized polarization (GP) value was calculated using the following formula:GP = (I₄₄₀ − I₄₉₀)/(I₄₄₀ + I₄₉₀),(1)

In that formula, I₄₄₀ and I₄₉₀ are the fluorescence intensities at emission wavelengths of 440 and 490 nm, respectively.

Capillary electrophoresis was performed using an QIAxcel Connect capillary gel electrophoresis system (QIAGEN, Hilden, Germany). The sample preparation was as follows. First, 100 µg mRNA/mL of mRNA-LNP samples was diluted to 5 ng/µL in QX RNA HS Dilution buffer^®. The mRNA-LNP sample was then added to an equal volume of 5% (w/v) tritonX-100 (SIGMA-ALDRICH, St. Luis, MO, USA) in QX RNA HS Dilution buffer^®. The mixture was incubated at room temperature for 10 min or more. One µL of the mixture was mixed with 1 µL of QX RNA Denaturation buffer^®, heated at 70 °C for 2 min, and then chilled at 4 °C for at least 5 min using a thermal cycler (MiniAmp Thermal Cycler, Thermo Fisher Scientific,Waltham, MA, USA). The samples were diluted with 8 µL of QX RNA Booster buffer^® and then transferred to a PCR tube for capillary electrophoresis analysis. The electropherogram for each injection was transferred to QIAxcel Screengel 2.0 software and Excel for analysis. The data calculation was based on the area under the curve of the sample line and above the base line. To define the peak derived from fragmented mRNA, we defined the mRNA fragment area as the region from A to B. The total mRNA area was defined as the region from A to reach the final retention time of measurement. The percentage of fragmented mRNA was calculated as the ratio of fragmented mRNA/total mRNA.

A: The retention time elapsed at 3 times the half-width of the half maxima of the internal standard marker peak.

B: The retention time was faster than 3 times the half-width at the half maxima of the intact mRNA peak.

To measure the TOT-28 in TOT-28-LNPs, the LNP samples were diluted to 4 mM of total lipid using a 9% (w/v) sucrose solution. Sixty µL of the diluted samples were extracted with 240 µL of acetonitrile/isopropanol (v/v 1:2). Following centrifugation (4 °C, 15,000× g, 10 min), the supernatant was filtered, and the filtrate was analyzed using an LCMS-2050 (Shimazu Corporation, Kyoto, Japan) with an ELSD-LT III detector (Shimazu Corporation). Separation was carried out using a Shim-pack Arata C18 Column (pore size: 120 Å; particle size: 5 µm; inner diameter: 50 mm; length: 2.0 mm) at a column temperature of 60 °C. Mobile phase A comprised 5 mM of ammonium acetate and mobile phase B consisted of isopropanol/acetonitrile (2:1) and 5 mM of ammonium acetate. Separation was accomplished using a step gradient with an initial 0 min hold at 60% of B, a 2 min gradient from 60 to 75% of B, a 4 min gradient from 75 to 98% of B, and a hold at 98% of B, which was delivered at a flow rate of 0.2 mL/min with an injection volume of 5 µL.

Freshly prepared mRNA-loaded TOT-28-LNPs exposed to the described pH and temperature conditions were added to Quantifoil R1.2/1.3 Cu 300 grids at 4 or 25 °C, and frozen in liquid ethane using FEI Vitrobot MarkIV. The grids were stored in liquid nitrogen until used for imaging. Glacios Cryo TEM (Thermo Fisher Scientific Inc., Waltham, MA, USA) was used to obtain the images. The instrument was operated at 200 kV. Samples were imaged at a 73,000× magnification.

Results are expressed as the mean ± the standard deviation (SD) of independent experiments. For in vitro experiments, technically independent experiments were conducted using LNP prepared each time. For comparisons between multiple groups, two-way analysis of variance (ANOVA) was performed followed by a Tukey post hoc test. Group size, definition of center, and dispersion and precision measures are also noted in the figure legends as appropriate. All graphs were prepared using GraphPad Prism software version 10.1.2 (San Diego, CA, USA).

TOT-28 consists of three oleic acid-derived hydrophobic scaffolds and a piperidine N-ethylpropionate structure as the hydrophilic head group (Figure 1A). The piperidine structure was adopted to minimize adduct formation followed by mRNA inactivation [21]. The N-ethylpropionate moiety was introduced to monitor the microenvironment around the hydrophilic region of ionizable lipids in mRNA-LNPs under different storage conditions. The N-ethylpropionate moiety is susceptible to hydrolysis under both acidic and basic conditions. Its hydrolysis susceptibility is affected by the degree of hydration around the ester bond. A parental TOT-28 and its hydrolyzed product could be quantified following separation by reverse-phase high-performance liquid chromatography (HPLC) based on the difference in hydrophobicity. Therefore, TOT-28 has a suitable structure for revealing the differences in the microenvironment around the hydrophilic group of ionizable lipids under various storage conditions.

TOT-28-LNPs were assembled by vigorously mixing TOT-28, cholesterol (Chol), 1,2-distearoyl-sn-glycero-3-phophocholine (DSPC), and 1,2-dimyritoyl-rac-glycero-3-methoxypolyethylene glycol-2000 (PEG_2k-DMG) at a fixed molar ratio with Firefly luciferase (FLuc) mRNA using a microfluidic device followed by purification with dialysis (Figure 1B). The TOT-28-LNPs showed an apparent acid dissociation constant (pK_a) value of 5.97 (Figure 1C). The physicochemical properties of the TOT-28-LNPs during storage at different values of pH (ranging from 5.0 to 7.0) and temperature (4 or 25 °C) were monitored (Figure 2). The diameters of the fresh TOT-28-LNPs were 123 to 133 nm (Figure 2A). The polydispersity index (PDI) of these fresh TOT-28-LNPs ranged from 0.17 to 0.23 (Figure 2B). The diameter and PDI gradually increased over the storage period independently of storage temperature and pH. The ζ-potentials of these fresh TOT-28-LNPs were −5.7 to −6.9 mV (Figure 2C). At 4 °C, the ζ-potential slightly decreased to −7.3 mV at the final timepoint at pH 5.0. On the other hand, the ζ-potential dropped significantly to a range of −10.2 to −11.4 mV at other values of pH. Furthermore, the ζ-potential decreased dramatically to within a range of −13.8 to −15.5 mV regardless of the pH values at 25 °C. Regardless of pH value, the encapsulation efficiency (EE) was constant over the storage period without depending on the storage conditions (Figure 2D). The EE of TOT-28-LNPs at pH 5.0 and 5.5 were approximately 95%, which was approximately 5% higher than that at higher pH.

The degree of the hydrolysis of an ester bond in the hydrophilic region of TOT-28 was monitored using a HPLC setup equipped with an evaporative light scattering detector (LC/ELSD) over the storage period (Figure 3). Retention time for the peak derived from hydrolyzed TOT-28 was 12.6 min, which was clearly separated from that of the parental TOT-28 observed at 13.5 min (Figure 3A). These peaks were assigned based on mass spectrometry (MS) analysis following consistent LC conditions. Under these LC/ELSD conditions, other potential hydrolyzed TOT-28 products lacking a hydrophobic scaffold were not detected throughout the experiments, which indicates that the ester bond in the hydrophilic region is highly susceptible to hydrolysis compared with that in the scaffold, as expected. The peak derived from the parental TOT-28 product was expressed as the relative percentage of the total peaks derived from all TOT-28 samples (i.e., sum of the parental and the hydrolyzed product) (Figure 3B,C). At 25 °C, the hydrolytic rate of the TOT-28 increased as the pH dropped to within a range of pH 6.0 or less and the parental TOT-28 was reduced to 70% after 4 weeks of storage at pH 5.0 (Figure 3C). By contrast, more than 94% of the TOT-28 was maintained intact at 4 °C after 8 weeks of storage across the range of tested pH. Interestingly, less of the TOT-28 was hydrolyzed at lower pH at 4 °C, which could indicate a potentially opposite tendency from that observed at 25 °C.

The degree of mRNA degradation in the TOT-28-LNPs was monitored during storage by means of capillary electrophoresis (Figure 4). The % of fragmented mRNA in the TOT-28-LNPs was expressed as a relative percentage of the total peaks (i.e., sum of both intact and fragmented mRNA), which was approximately 7% in all fresh LNPs (Figure 4B,C). Whereas only a slight increase to below 10% was observed at 4 °C over 8 weeks (Figure 4B), a clear increase of up to 30% was observed at 25 °C over 4 weeks (Figure 4C). The degree of increase at higher pH tended to become larger at 25 °C.

Based on the above observation that storage temperature changed the effect of pH on the ester hydrolysis of TOT-28, both hydration degree and microviscosity of the fresh TOT-28-LNPs at different pH (5.0 and 7.0) and temperature (4 and 25 °C) were measured to understand the impact of the storage conditions on the physicochemical microenvironment of the lipid membranes. The LNPs consisting of commercially available ionizable lipids DLin-MC3-DMA (MC3) and ALC-0315 (ALC) were also used for comparison. The hydration degree of the LNPs was measured using laurdan and is expressed using common generalized polarization (GP) values. Under all conditions, TOT-28 had a higher GP value and was more hydrophobic than other ionizable lipids. Regardless of the differences in ionizable lipids, the GP values dropped at higher temperature and lower pH, indicating a more disordered lipid bilayer in acidic environments and with higher temperatures (Figure 5A,B). However, the decrease in the GP values of the TOT-28-LNPs caused by a drop in pH was clearly lower than that for others. The ratios of GP values at pH 5.0 and at pH 7.0 were 0.95 (TOT-28), 0.85 (MC3), and 0.85 (ALC) at 4 °C compared with 0.84 (TOT-28), 0.68 (MC3), and 0.76 (ALC) at 25 °C, respectively. These results indicate that the hydration of TOT-28-LNPs was less affected by pH change compared with that of others. In addition, the degree of hydration measured by laurdan and the degree of hydrolysis of TOT-28 did not match, which was thought to be due to the fact that the localization of the fluorescent indicator and specific LNP lipids (ionizable lipids in this case) did not match.

The microviscosity of LNPs was measured using 9-(2,2-dicyanovinyl) julolidine (DCVJ) and is expressed as the R value, which is the integrated fluorescence emission-to-absorbance ratio. Regardless of the differences in ionizable lipids, the R values at 4 °C were higher than those at 25 °C under the same pH (Figure 5C,D). The TOT-28-LNPs showed the highest R values under all storage conditions. Interestingly, the trends in microviscosity between pH values of 5.0 and 7.0 were clearly different between the three ionizable lipids. The MC3-LNPs showed a lower microviscosity at pH 5.0, whereas the ALC-LNPs showed a slightly higher microviscosity at pH 5.0 for both temperatures. Furthermore, the TOT-28-LNPs showed a decreasing trend with a decrease in the pH at 4 °C and the opposite trend at 25 °C. These results show that differences in the chemical structures of ionizable lipids not only alter the degree of microviscosity between lipids, but the responses to changes in storage temperature and pH are also affected. It is difficult to clearly explain why the change in microviscosity according to storage conditions differs depending on the ionizable lipid, but it may reflect differences in the internal structure of LNPs due to differences in the three-dimensional structure of ionizable lipids. In any case, the fact that ionizable lipids exhibit different changes in physicochemical properties of LNPs depending on storage conditions emphasizes the risk of a uniform understanding of LNP properties and the importance of individual verification. Next, we observed the particle structure of TOT-28-LNPs exposed to different pH and temperature conditions using cryo-TEM (Figure 5E–H). The TOT-28-LNPs exhibited a distinct lamellar structure in the outer layer and an amorphous structure in the core at pH 7 (Figure 5E,F). This is consistent with the core–shell model proposed by Arteta et al., which features a shell layer composed primarily of DSPC [25]. Interestingly, at pH 5, which is lower than the apparent pKa of TOT-28, numerous dot-like high-electron-density nanostructures with diameters of approximately 3–5 nm were observed within the LNPs (Figure 5G,H). This is highly similar to the "electron-dense gritty texture of LNP cores reflecting lipid-bound, condensed-state mRNA mingled with the excess of ionizable lipids" suggested by Szebeni [36], and "the high-electron-density mottles associated with the bodies of the LNPs" observed by Brader [37]. Furthermore, TOT-28-LNPs contain a greater number of dot-like high-electron-density nanostructures compared with the aforementioned previous studies that examined Comirnaty or LNPs containing Moderna's proprietary ionizable lipids, which may be attributed to differences in the three-dimensional structure of the ionizable lipids. In addition, at pH 5, the lamellar structure of the LNP surface is either masked or has disappeared, which suggests the redistribution of protonated TOT-28 to the surface. These observations suggest that the pH of the external aqueous phase significantly affects the structure of internal mRNA, the arrangement of ionizable lipids, and the microenvironment surrounding each molecule. No obvious structural changes in LNPs were observed at each pH in response to storage temperature. However, the distinct LNP structures at different pH values could account for the opposite signs of microviscosity shift in response to pH changes at temperatures between 4 and 25 °C, and further detailed investigation is desirable.

To reveal the impact of storage conditions on the in vivo performance of LNPs, bioluminescent levels at the injection site (caudal thigh muscle) were quantified 20 hrs following an intramuscular injection of TOT-28-LNPs following storage under indicated conditions and time periods. Total flux of the fresh TOT-28-LNPs at different levels of pH varied from 4 × 10⁷ to 12 × 10⁷ p/s with the lowest level recorded at pH 5.0 (Figure 6A). At 4 °C, the gene expression level of the LNPs stored at pH 5.0 was maintained for up to 8 weeks whereas that stored at pH 5.5 or above dropped to approximately 50% of the initial levels (Figure 6C). At 25 °C, on the other hand, gene expression levels were maintained for 2 weeks but clearly dropped to less than 50% of the initial levels at 4 weeks irrespective of the storage pH (Figure 6C), probably due to hydrolysis of both TOT-28 and mRNA.