Introduction

Messenger RNA (mRNA)-based vaccines have emerged as a viable option for preventive and therapeutic applications, especially in the COVID-19 post-pandemic era1,2,3,4,5,6. The mRNA sequence serves as a template for the translation of a specific protein, which eventually induces both innate and adaptive immunity7. Compared to traditional vaccines, mRNA-based vaccines can be developed in a faster and cost-efficient manner8,9. Cell-free in vitro transcription ensures scalability of the manufacturing process and flexibility of the mRNA sequence1,10,11,12. mRNA vaccines are often formulated with lipid nanoparticles (LNPs), which protect the mRNA from nuclease degradation and ensure mRNA uptake by the cell, thus facilitating expression of the desired proteins1,3,13,14.

Inherent instability of mRNA molecules poses a real challenge in storage and transport of mRNA vaccines2,15,16. In particular, the hydroxyl group present at the 2′-position of ribose makes RNA susceptible to acid-, base-, and metal-catalyzed hydrolysis17,18. Once degraded, mRNA may no longer be translated correctly into the target protein, leading to a loss in vaccine efficacy2,16. Thus, mRNA integrity is considered as a critical quality attribute (CQA) of the mRNA Drug Substance (DS) and Drug Product (DP)16,18,19,20. Another important consideration is that hydrolysis occurs preferentially in the single-stranded regions of mRNA. Fortunately, in-vitro transcribed RNA molecules tend to form higher-order structures and multimerize through various intermolecular interactions, such as base-pairing, which helps in stabilization. Indeed, Zhang et al. demonstrated that increasing secondary structure can extend the half-life of mRNA2,21.

While developing strategies to optimize the mRNA sequence to limit hydrolysis is critical, it is also important to identify species that could affect the stability and efficacy of nucleic acid-based products18,20,22. Developing novel analytical methodologies that can rapidly monitor mRNA integrity and purity is essential to sustain the rapid production and successful deployment of mRNA vaccines16,22. Currently, analytical techniques, such as agarose gel electrophoresis, capillary gel electrophoresis (CGE), high performance liquid chromatography (HPLC), long-read sequencing, and quantitative polymerase chain reaction (qPCR) are used in the industry to assess quality and integrity of mRNA-based products12,19,23. In this study, we first compared the advantages and limitations of three common analytical methodologies in monitoring mRNA integrity:

  • capillary gel electrophoresis with fluorescent detection (FA-Fluo) using the automated Fragment Analyzer system.

  • capillary gel electrophoresis with UV detection (CGE-UV) using PA800 plus Capillary Electrophoresis system.

  • ion-pairing reversed-phase HPLC (IP-RPLC) with UV detection using Vanquish device.

Second, we investigated the hypothesis that an additional peak present in CGE-UV chromatogram could be attributed to mRNA oligomerization21,24. Additional analyses were conducted using asymmetric flow field flow fractionation (A4F) and size exclusion chromatography (SEC), enabling the collection and analysis of fractions represented by different peaks of interest on the chromatogram. Finally, we conducted in vitro protein expression assays with a DS containing different proportions of mRNA oligomers to determine the efficiency of translation, confirm the identity of the translated product, and assess its reactogenicity.

Methods

CGE-UV analysis

Experiments were carried out on a PA800 plus capillary electrophoresis system (SCIEX, Framingham, MA, USA), equipped with diode array detectors and driven by a 32 Karat system. Bare fused silica DNA capillaries of 100 μm inner diameter × 40 cm total length (30 cm to the detector) were purchased from SCIEX. Temperature of the capillary cartridge was set at 25 °C. The capillary was rinsed for 1.5 min at 20 psi with a 1x capillary conditioning solution and then rinsed for 2 min at 20 psi with an RNA separation gel. The 5x capillary conditioning solution and RNA separation gel were purchased from Agilent (Santa Clara, CA, USA). The injection was performed in electrokinetic mode in reverse polarity under − 2 kV for 30 s. Separation was performed in reverse polarity under − 6 kV for 45 min. UV detection was performed at 254 nm.

Samples were prepared by adding 10 µL of mRNA (1 mg/mL) into 114 µL of 3 M urea solution. 2 µL of 25-mer internal standard was added and samples were then heated at 70 °C for 5 min for partial denaturation, cooled at 5 °C, and analyzed. The 25-mer internal standard (5′-ACGUACGUACGUACGUACGUACGUA-3′) was purchased from Integrated DNA Technologies (Coralville, IA, USA).

FA-Fluo analysis

Experiments were carried out on a Fragment Analyzer system (Agilent). System was equipped with a fluorescent detector and driven by Fragment Analyzer Controller Software (Agilent). Capillaries used were 12-capillary arrays of 33 cm (Agilent). The intercalating dye was diluted 10 times into RNA separation gel, the capillary conditioning and the inlet buffer were diluted 5 times into RNase-free water. RNA separation gel, intercalating dye, 5x capillary conditioning solution, 5x dsRNA inlet buffer, and capillary storage solution were purchased from Agilent. The injection was performed in electrokinetic mode in reverse polarity under − 5 kV for 4 s. Separation was performed in reverse polarity under − 8 kV for 40 min.

Samples were prepared by adding 2 µL of mRNA (0.1 mg/mL) into 22 µL of RNA diluent maker purchased from Agilent. The mixtures were then heated at 70 °C for 2 min for denaturation, cooled at 5 °C, and analyzed.

mRNA extraction from LNP before CGE-UV and FA-Fluo analysis

Before conducting the CGE-UV and FA-Fluo analyses, mRNAs were extracted from LNPs using the following process. To disrupt the LNPs and precipitate the mRNA, 200 µL of mRNA-LNP and 800 µL of 60 mM ammonium acetate in isopropanol were added to the tube, briefly vortexed, and centrifuged at 14,000 g for 5 min at 4 °C. Supernatant was discarded, and pellet was washed with 1 mL of isopropanol and then centrifuged at 14,000 g for 5 min at 4 °C. Supernatant was discarded, and the pellet was dried for 5 min using a SpeedVac vacuum concentrator (Thermo Scientific). Finally, to solubilize the mRNA pellet, 200 µL of RNase-free water was added and mixed at room temperature. Isopropanol and ammonium acetate were purchased from Sigma-Aldrich Merck (Darmstadt, Germany).

IP-RPLC analysis

Experiments were carried out on a Vanquish HPLC System (Thermo Fisher Scientific, Waltham, MA, USA), equipped with a diode array detector and driven by the Chromeleon™ Chromatography Data System software (Thermo Fisher). DNAPac™ reverse phase column (Thermo Fisher), with 4-µm particles and dimensions of 2.1 × 100 mm was used. Column temperature was 70 °C and the flow rate was 0.35 mL/min. Mobile phase A consisted of dibutylammonium acetate (50 mM; TCI Chemicals, Chennai, TN, India) and triethylammonium acetate (100 mM; Sigma-Aldrich, St. Louis, MO, USA). Mobile phase B consisted of 49.5% acetonitrile (Fisher Chemical, Pittsburg, PA, USA), dibutylammonium acetate (50 mM), and triethylammonium acetate (100 mM) and mobile phase C consisted of 40% isopropanol (Fisher Chemical, Pittsburg, PA, USA), 40% acetonitrile, dibutylammonium acetate (50 mM), and triethylammonium acetate (100 mM).

Separation was accomplished by gradient with an initial 1.5 min hold at 25% B, a 3.0 min gradient from 25 to 50% B, a 14.5 min gradient from 50 to 56% B, a 5.5 min gradient and hold at 100% B, a 0.5 min gradient from 0 to 100% C, a 4 min hold at 100% C, a 0.5-minute gradient from 100 to 0% C, and a 4 min hold at 25% B. Samples of approximately 1 µg of mRNA were injected. UV detection was performed at 260 nm. Samples were prepared by diluting 10 µL of mRNA Drug Product (1 mg/mL) into 90 µL of 0.5% solution of Brij™ 58 (Sigma Aldrich, St. Louis, MO, USA) in 100 mM Tris-HCl (Invitrogen, Waltham, MA, USA).

Quantification of mRNA integrity

Integrity of mRNA products (DS and DP) was calculated using the relative peak area method for CGE-UV, FA-Fluo, and IP-RPLC. Integrity was defined as:

$$\:\text{\%}\:mRNA\:integrity\:=\frac{{A}_{\:mRNA\:intact}\:}{{A}_{mRNA\:intact}\:+{\:A}_{Fragments\:}}\:\times\:100$$

Where, A denotes the peak area of the respective species in the chromatogram.

A4F-MALS analysis

An online Eclipse A4F system (Wyatt Technologies, Santa Barbara, CA, USA) connected to a 1260 Infinity II diode array detector (Agilent Technologies, Waldbronn, Germany), a DAWN multiangle light scattering (MALS) detector (Wyatt Technologies), and a Wyatt Optilab Refractive Index detector were used to carry out the measurements.

Analysis of mRNA samples was performed using a long channel and a 10 kDa regenerated cellulose membrane. Mobile phase consisted of isotonic PBS 1X pH 7.4. A 400 μm spacer was employed, with an injection flow rate of 0.2 mL/min injection flow and a detector flow rate of 0.5 mL/min. Focusing was carried out at a flow rate of 0.75 mL/min (linear profile) for 4 min. Then 10 µL of undiluted mRNA (1 mg/mL) samples were injected. Injection volumes were adjusted for dilution and urea addition studies to ensure consistent quantities across different runs. UV detection was set at 260 nm. As quality control, A4F performance was checked by injecting 20 µL of 2 mg/mL of bovine serum albumin. Instrument control, data acquisition, and data processing were enabled by VISION RUN (Version 3, Wyatt Technologies) and Astra (Version 8.1, Wyatt Technologies).

SEC-MALS analysis

SEC measurements were performed on a 1260 Infinity II liquid chromatography system (Agilent), a Wyatt DAWN MALS detector, and a Wyatt Optilab refractive index detector (Wyatt Technologies).

Separations were realized on a 7.8 mm × 300 mm, 5 μm WTC050S5 column with 500 Å pore size (Wyatt Technologies). The mobile phase, consisting of PBS 1X filtered through a 0.1 μm filter, was delivered at an isocratic flow rate of 0.5 mL/min, with column temperature maintained at 25 °C. 10 µL of undiluted mRNA samples (1 mg/mL) was injected. Injection volumes were adjusted for dilution and urea addition studies to ensure the same concentration was injected. UV detection was set at 260 nm. Instrument control, data acquisition, and data processing were performed using both VISION RUN and Astra.

Flow cytometry analysis

In vitro protein expression was performed using HEK293T/17 cell line (ATCC CRL-11268). Cells were seeded in Dulbecco’s Modified Eagle Medium (DMEM; Gibco™, Waltham, MA, USA) with 5% fetal calf serum (FCS) in 96-well plates at 100,000 cells per well and samples were transfected simultaneously. Samples were then diluted in Opti-MEM™ (Gibco™) to reach a concentration of 4 µg/mL, then an 8 steps, 2-fold serial dilution was performed. Subsequently, samples were mixed with an equal volume of Lipofectamine 3000™ (Invitrogen) diluted according to the supplier’s instructions and incubated at room temperature for 15 min. 50 µL of each dilution was added in triplicate. After 24 h of incubation at 37 °C, expression of proteins at cell surface was analyzed.

Cells were washed in Cell staining Buffer (BioLegend, San Diego, CA, USA), stained with a proprietary primary specific antibody conjugated with Pacific Blue fluorochrome, and analyzed by flow cytometry. Data was acquired on a NovoCyte Advanteon using NovoExpress software (Agilent). Median fluorescence intensity and percentage of positive cells were acquired and multiplied to obtain the integrated median fluorescence intensity.

Western blot analysis

In vitro protein expression was performed using HEK293T/17 (CRL-11268; ATCC, Manassas, VA, USA). Cells were seeded in Dulbecco’s DMEM with 5% FCS in 12-well plates at 300,000 cells per well. After incubation for 24 h at 37 °C and 5% CO2, medium was replaced with 1 mL Opti-MEM™ per well. Samples were diluted in Opti-MEM™ to reach a concentration of 40 µg/mL, mixed with an equal volume of Lipofectamine 3000™ diluted according to the supplier’s instructions, and incubated at room temperature for 15 min. 100 µL of each dilution was added per well. After 24 h of incubation at 37 °C, cells were lysed using 150 µL of CelLytic M lysis buffer (Sigma Aldrich) per well. Cell lysates were mixed with an equal volume of Laemmli 2xC (Bio-Rad, Hercules, CA, USA) and then denatured for 5 min at 100 °C.

Samples were subsequently loaded on a Mini-PROTEAN TGX 4–15% Stain-Free gel for migration and transferred to a nitrocellulose membrane using the Trans-Blot Turbo System (Bio-Rad). After 1 h blocking with PBS 1xC – Tween 20 0,05% and 5% w/v skimmed milk, membranes were washed with PBS 1xC – Tween 20 0,05%. Protein detection was performed using a proprietary antibody conjugated with Horse Radish Peroxidase (HRP) diluted in PBS 1xC – Tween 20 0,05% and 1% w/v skimmed milk for 1 h. HRP detection was performed using the SuperSignal™ West Pico PLUS kit (Thermo Fisher) according to supplier instructions on a CHEMIDOC MP Imaging System (Bio-Rad). MagicMark™ ladder from Thermo Fisher was used to estimate the protein size.

mRNA reactogenicity: IRF and NF- κB response studies

THP-1 Dual (InvivoGen, San Diego, CA, USA) reporter cell line was used to assess the reactogenicity of naked mRNA. This monocyte cell line contains two reporter genes: a secreted alkaline phosphatase (SEAP) induced by the NF-κB pathway, and a secreted luciferase induced by IRF (Interferon Response Factor) pathway. THP-1 Dual cells were cultured in RPMI-1480 medium, supplemented with 10% of fetal bovine serum, 2 mM L-glutamine, 100 U/mL of penicillin/streptomycin, 100 µg/mL of normocin, 100 µg/mL of Zeocin® (copper-chelated phleomycin D1, InvivoGen), and 10 µg/mL of blasticidin in a CO2 incubator with saturating humidity and 5% CO2.

One million of THP-1 Dual cells were exposed to 1 µg mRNA by electroporation using Expert STx (Maxcyte, Rockville, MD, USA) following manufacturer’s instructions. After electroporation, cells were plated into 24-well plates with complete medium. 24 h post-electroporation, both reporter proteins were measured in the cell culture supernatant using QUANTI-Blue™ Solution (Invivogen), a SEAP detection reagent, and QUANTI-Luc™ 4 Lucia/Gaussia (Invivogen), a luciferase detection reagent. Quantification was performed with EnSight™ Multimode Plate Reader (PerkinElmer, Waltham, MA, USA), measuring luminescence and optical density (OD) at 620 nm for luciferase and SEAP, respectively.

Results

Comparison of analytical techniques to assess mRNA integrity

Analytical methods used in this study have different underlying principles. Although each method allows separation of mRNA molecules of different lengths, as shown by the separation of mRNA ladder samples (Supplementary Material; Figure S1), they may differ in resolution and offer different sets of advantages and disadvantages (Table 1). In CGE (FA-Fluo and CGE-UV), negatively charged mRNA molecules are attracted by the anode. Liquid gel matrix in the capillary acts as a molecular sieve, allowing smaller molecules to migrate faster and enabling size-based separation of mRNA. Separation of mRNA molecules is monitored by UV or fluorescence spectrometers in CGE-UV and FA-Fluo methods, respectively. FA-Fluo method is more sensitive because of the use of fluorescent dyes. Additionally, FA-Fluo device is easy to operate and ensures rapid analysis owing to the multiple parallel capillaries (12 to 96). However, use of fluorescent dyes also introduces the possibility of non-specific bias as fluorescent detection can depend on the ability of fluorescent dyes to bind uniformly to different mRNA strands. In contrast, CGE-UV method does not encounter this potential detection issue, as it detects mRNA through its intrinsic absorption of UV light. However, the device used for CGE-UV (PA800 plus) is more complex, necessitating additional training, and data analysis time is longer. Finally, both FA-Fluo and CGE-UV methods are sensitive to matrix effects due to the electrokinetic injection used and involve the use of carcinogenic, mutagenic, and reprotoxic chemicals in the separation gel. However, hydrodynamic injection could limit the sensitivity to matrix effects.

Table 1 Advantages and disadvantages of the three analytical methods.
Full size table

Separation mechanism of IP-RPLC method involves two types of interactions: electrostatic interactions using ion-pairing (IP) and hydrophilic/hydrophobic interactions using reversed-phase (RP) chromatography. These interactions allow the separation of molecules based on their charge and their hydrophobicity, respectively. Compared to other two methods, sample preparation is easier with IP-RPLC. Another valuable advantage of HPLC method is that it does not require prior extraction of mRNA from lipid nanoparticles (LNPs), which is time consuming16. Indeed, in IP-RPLC, direct deformulation is achieved thanks to the high temperature of the column, use of organic solvents in the mobile phases, and the addition of a small quantity of active surfactant during sample preparation. Skipping mRNA extraction from LNPs makes the analytical process faster and minimizes the risk of mRNA loss. Finally, intermediate precision studies were conducted for CGE-UV and IP-RPLC methods (Table S1). IP-RPLC demonstrated lower day-to-day variability compared to CGE-UV. Indeed, the 95% one-sided upper confidence limit of the intermediate precision obtained is 4.1% for CGE-UV and 1.2% for IP-RPLC. This indicates that IP-RPLC offers improved precision for mRNA integrity assessment.

Accelerated stability study

Accelerated stability studies at 5 °C and 25 °C were conducted for both DS and DP from day 0 (D0) through day 15 (D15) and monitored by three analytical methods. It is important to note that these DPs were specifically formulated without excipients for research purposes and do not represent final vaccine products. In all three methods, loss of mRNA integrity can be observed through a decrease of the main peak area, which represents intact mRNA. (Fig. 1, Figure S2, Table S2 and S3). For DS, lower retention time peak(s) in the chromatogram, attributed to mRNA fragments, increased slightly over the study period, highlighting further degradation of the mRNA molecule. Compared to FA-Fluo, IP-RPLC indicated slightly lower mRNA integrity (85% vs. 80% relative peak area at D0, respectively), probably due to the better resolution of HPLC, allowing better separation of the main peak from fragments (Table 2).

Fig. 1
figure 1

Accelerated stability study of mRNA DS (single, mRNA-based DS vaccine) monitored by: (A) FA-Fluo, (B) CGE-UV, and (C) IP-RPLC.

Full size image
Table 2 Accelerated stability study DS at 25 °C.
Full size table

Interestingly, an additional peak, representing approximately 15% of relative peak area, was present at higher retention time of the CGE-UV electropherogram, thus indicating the presence of a species approximately twice the size of the mRNA. This peak indicated possible oligomerization of the mRNA, specifically the formation of mRNA dimers in the samples21,24. Presence of this peak further decreased the relative peak area of the main peak in CGE-UV chromatogram (68% at D0). Area of the peak hypothetically associated with dimers did not show notable change considering the variability of the method (17% at D0 to 15% at D15) (Table 2). Notably, when we combine the relative peak area of the dimer and the main peak in the CGE-UV electropherogram, the results closely align with the relative peak area of the main peak obtained from the FA-Fluo analysis.

LNP-encapsulated mRNA (i.e., DP) contained a fraction of untailed mRNA molecules (i.e., mRNA without a poly(A) tail) which appeared as a shoulder to the main peak in CGE-UV and FA-Fluo chromatograms. Presence of untailed mRNA molecules can be attributed to a defect in enzymatic tailing during the synthesis of DS. This untailed mRNA peak appears more distinctively in IP-RPLC (Fig. 2), confirming better resolution of the method compared to CGE and FA methods. As observed in the accelerated stability study of DS, the main peak of DP decreased while the fragments peak increased over time.

Fig. 2
figure 2

Accelerated stability study of mRNA DP (monovalent DP monitored by: (A) FA-Fluo, (B) CGE-UV, and (C) IP-RPLC. Abs, absorbance; CGE, capillary gel electrophoresis; DP, Drug Product; DS, Drug Substance; Fluo, fluorescence; mRNA, messenger ribonucleic acid; nt, nucleotides; RFU, relative fluorescence unit; IP-RPLC, reversed-phase ion-pair high-performance liquid chromatography; T0, initial observation; TnD, observation on nth day; UV, ultraviolet.

Full size image

In IP-RPLC chromatogram, relative area of the main peak drastically decreased from 68% at D0 to 2% at D15 at 25 °C (Table 3). Yet, this sharp decrease was not accompanied by a corresponding increase of the fragments peak. A new peak at a higher retention time increased, which could be attributed to the mRNA-lipid adducts10. Control experiments with injection of the individual lipid components confirmed that these peaks were specific to mRNA-lipid adducts and not attributable to free lipids (Figure S3). mRNA-lipid adducts are products of reactions between mRNAs and the LNP ionizable lipids, leading to the loss of mRNA activity. mRNA-lipid adducts can be observed in HPLC as this method separates molecules based on their hydrophobicity. They likely cannot be observed in CGE because lipids are small, and their binding probably does not sufficiently impact the size and charge of mRNAs enough to lead to a retention time shift. The ability of HPLC to detect mRNA-lipid adducts formation during the accelerated stability study resulted into lower relative peak area of mRNA main peak compared to those observed in the FA-Fluo and CGE-UV methods.

Table 3 Accelerated stability study of DP at 25 °C.
Full size table

Comparison of DS and DP integrity

Accelerated stability studies demonstrated a faster loss of mRNA main peak area in DP compared to DS. When monitored using IP-RPLC, mRNA main peak area of DS decreased from 80% at D0 to 70% at D15, while the main peak area of DP dropped from 68% to 2% over the same period. Although this difference is mainly attributed to the presence of mRNA-lipid adducts at the DP stage, other degradation processes are also occurring. This is evidenced by the faster loss of the mRNA main peak in the DP compared to the DS, observed with both CGE-UV and FA-Fluo methods. To confirm this trend, an affiliated system was studied where stability of a mRNA-based DS was compared to the corresponding DP (i.e., same DS encapsulated in LNPs). CGE-UV data confirmed the trends observed in the initial system. For DS, mRNA main peak area decreased from 84% to 80% between D0 and D15, whereas it dropped from 84% to 47% for the DP over the same time interval (Table S4 and Table S5). These results confirm that, without excipients to stabilize the LNPs, degradation of mRNA is faster at DP stage, when the mRNA is encapsulated. This illustrates that, although encapsulation into LNPs protects mRNAs from nucleases-mediated degradation, it is unable to provide protection against other types of degradation such as catalyzed hydrolysis.

Characterization of the mRNA dimer

In order to investigate and explore the nature of the potential dimer peak observed in CGE-UV chromatograms, and confirm that it is not an artifact, two orthogonal chromatographic methods were deployed: A4F-MALS and SEC-MALS.

A4F-MALS confirmed the presence of two species having molecular weight of 650 kDa and 1300 kDa with radius of gyration values of 18 nm and 30 nm (Fig. 3A), respectively, thus confirming the presence of an mRNA dimer. SEC-MALS data also confirmed the presence of a second species twice as large as the single stranded mRNA (Fig. 3B).

Fig. 3
figure 3

Identification of mRNA dimer using (A) A4F-MALS and (B) SEC-MALS techniques. Abs, absorbance; CGE, capillary gel electrophoresis; DP, Drug Product; DS, Drug Substance; Fluo, fluorescence; LNP, lipid nanoparticle; mRNA, messenger ribonucleic acid; nt, nucleotides; RFU, relative fluorescence unit; IP-RPLC, reversed-phase ion-pair high-performance liquid chromatography; T0, initial observation; TnD, observation on nth day; UV, ultraviolet.

Full size image

Both species separated using SEC-MALS were collected, lyophilized, and purified using QIAGEN technology for further characterization. After fraction collection, concentrations of mRNA in the two fractions were 1.4 mg/mL and 1.1 mg/mL, respectively (Figure S4). CGE-UV analyses further confirmed that the successful collection of the fractions (Figure S5).

Nature of the dimer was further investigated by monitoring the effect of adding urea denaturing solution and raising the temperature on the dimer peak in A4F-MALS. Addition of 3 M urea in 1 mM sodium citrate solution to the dimer fraction and heating at 70 °C for 5 min resulted in nearly complete disappearance of the dimer peak (mass fraction: 21% to 1%) along with a proportionate increase in monomer peak (mass fraction: 78.6% to 98.4%) (Figure S6). This observation establishes that addition of urea in sodium citrate solution and heating facilitates dissociation of dimers into monomeric mRNA molecules, indicating weak bonding between the monomers, probably via interactions between limited number of base-pairs. These results obtained using A4F were further confirmed by CGE analyses. It appears that, sodium citrate solution plays a crucial role, as urea addition alone does not eliminate the dimer peak. Heat alone is also insufficient because, once the temperature drops, denatured mRNA may renature itself to find its most stable state.

Expression and reactogenicity of the mRNA dimer

In vitro translation evaluated based on the integrated median fluorescence intensity showed similar translation efficiency for the two fractions (monomer and dimer). Flow cytometry analysis of the monomer fraction, dimer fraction, and DS (i.e., monomer + dimer) showed no noteworthy difference in quantitative protein expression, indicating that both monomer and dimer expressed the protein of interest at comparable levels (Fig. 4A). This finding was further investigated by western blot analysis which demonstrated that protein expressed from either monomer or dimer mRNA were identical in size (Fig. 4B and Figure S7).

Fig. 4
figure 4

Properties of the protein expressed from monomer and dimer fractions. (A) Flow cytometric analysis, (B) Western blot analysis (this is a cropped version of the original blot image; the uncropped original blot is presented in Supplementary Figure S7), and (C) Reactogenicity studies. DP, Drug Product; DS, Drug Substance; kDa, kilo Dalton; Mw, molecular weight; Rg, radius of gyration. DS, Drug Substance; Dim, dimer; IFN, interferon; Mon, monomer; NF-κB, nuclear factor kappa B; THP-1, human leukemia monocytic cell line.

Full size image

Finally, reactogenicity studies were performed in THP-1 Dual cells as nucleic acid having extensive base-pairings, longer than 40 nucleotides, could potentially be recognized as double-stranded RNA by the innate immune system25. No notable impact of the presence of dimerized mRNA was found on the reactogenicity assay as similar IFN and NF- κB responses were observed for monomer and dimer fractions when the secreted reporter proteins were measured (Fig. 4C).

Discussion

mRNA integrity is one of the key CQA of mRNA-based Drug Products as degradation of this inherently fragile class of molecules could affect the efficacy of the product16,18,20. Therefore, robust and reliable analytical methods are necessary to standardize the assessment of mRNA integrity. Three methods tested in this study differ in some of their working principles and hence provide complementary information19.

FA-Fluo, CGE-UV, and IP-RPLC analyses of DS and DP showed the presence of intact mRNA and mRNA fragments. In all three methods, mRNA degradation resulted in a decrease in intact mRNA peak area. All three methods also detected the presence of untailed mRNA, upon analysis of a sample with poor enzymatic tailing. Compared to CGE-UV and FA-Fluo methods, IP-RPLC method provided better resolution of both the mRNA fragments and untailed mRNA peaks. Moreover, only IP-RPLC method was able to detect the presence of mRNA-LNP adducts, which resulted in a rapid decrease of mRNA main peak area at DP stage. These adducts are a known class of mRNA impurities that are difficult to detect using CGE-based methods10. Alternatively, FA-Fluo proved to be the fastest method due to high level of automation, especially at DS stage when mRNA extraction from LNP is not required. At DP stage, wherein mRNA extraction is time consuming, HPLC can be more efficient by saving sample handling time. Notably, loss of integrity was faster at DP stage compared to DS stage, partly because the experiment was carried out on DP formulated for research purpose without excipients to stabilize the LNPs. It is important to acknowledge that using DPs without excipients does not fully represent the behavior and stability of final vaccine products. However, these non-stabilized formulations are particularly valuable for comparing analytical methods, allowing for accelerated evaluation of method performance and sensitivity to degradation products.

CGE-UV results suggested the presence of mRNA dimers in both DS and DP samples. This was further confirmed using two orthogonal chromatographic methods. Both A4F-MALS and SEC-MALS detected a second species with twice the molar mass and size of the mRNA. SEC-MALS also facilitated collection of both species.

When the isolated, lyophilized, and purified fractions were subjected to in vitro translation, both monomer and dimer fractions showed similar translation efficiency. Flow cytometry and western blot analyses confirmed the expression of the protein of interest from both monomeric and dimeric mRNA. Protein expressed from the monomer and dimer fractions also showed no difference in reactogenicity in THP-1 Dual cells. Moreover, addition of 3 M urea in citrate solution and applying heat to the dimer fraction resulted in dissociation of the dimers into mRNA monomers. Furthermore, the dimer peak was not observed in IP-RPLC, likely due to the high column temperature (70 °C) and the use of organic solvents and hydrophobic stationary phases, causing dimers to dissociate into monomers.

These observations have two important implications. Firstly, dissociation of mRNA dimers under denaturing conditions suggests an equilibrium between the monomeric and dimeric mRNA species. It also indicates that the monomeric units are likely paired via non-covalent interactions, such as base pairing, similar to those found in the secondary structure of mRNA2,21. Secondly, both monomeric and dimeric mRNA appear to express the target protein and do not trigger innate immune response. Therefore, dimers could be considered as another conformational form of the original monomeric mRNA, resembling the secondary structure of mRNAs involving noncovalent interactions. Thus, the presence of dimers in the DS or DP may not imply an actual loss of mRNA integrity. However, further studies would be required to confirm the equivalence of dimeric and monomeric forms in other biological contexts, particularly regarding potential immunogenic or pharmacokinetic differences in vivo.

Although these 15-day studies were not designed to be representative of long-term stability, but rather to support analytical development, they have proven effective in identifying impurity profiles that were later observed in extended stability studies. However, the kinetics of fragment formation and adduct creation can vary significantly depending on the formulation, emphasizing the need for formulation-specific stability assessments.

Conclusion

This study highlights the need for robust analytical methods to expand the understanding of mRNA degradation mechanisms, optimize formulation stability, and maximize vaccine efficacy. All three analytical methods described in this study were successfully used to monitor the integrity of mRNA-based DS and DP. Compared to electrophoretic methods, IP-RPLC provided better resolution. CGE-UV analysis detected formation of mRNA dimers in DS and DP, which was confirmed using complementary analyses using two orthogonal techniques. Notably, at the same concentration, in vitro translation of the dimer led to same level of protein expression as for the monomeric mRNA. Furthermore, neither mRNA monomer nor the dimer triggered an innate immune response. Thus, the dimer may be considered as intact mRNA. IP-RPLC is the only method that allows separation of mRNA-lipid adducts and based on the literature, mRNA-lipid adducts do not appear to express proteins of interest. Furthermore, IP-RPLC is the only method for which extraction of the mRNA from the LNPs was not required prior to analysis.