Introduction

Driven by the need for rapid production and scalable manufacturing during the COVID-19 pandemic in 2020, messenger RNA (mRNA) vaccines were rapidly developed and authorised in many jurisdictions across the globe. Consequently, mRNA-based therapies have emerged as a promising new class of biopharmaceuticals for the treatment of a wide variety of diseases, including cancers, infectious diseases, genetic disorders, and autoimmune diseases. However, mRNA is an extremely unstable molecule susceptible to cleavage, severely affecting its potency. Monitoring and ensuring its stability throughout manufacture, storage, and delivery is therefore critical.

Currently, a wide range of analytical tools are used for monitoring mRNA’s Critical Quality Attributes (CQAs), but whether they lack the required resolution or introduce unwanted bias to the measurement, these tools do not give an accurate summary of the most important quality characteristics of an mRNA therapy. As a result, there is a clear gap in the analytical landscape: the need for a high-resolution, unbiased analytical method which can accurately assess the most important CQAs of an mRNA therapy at once.

Oxford Nanopore Technologies’ Direct RNA sequencing (DRS) platform is a long-read form of Next Generation Sequencing which is capable of reading full-length native RNA molecules without prior reverse transcription or amplification. This technique may have the potential to address the limitations of current analytical methods by measuring multiple mRNA CQAs at once. However, Oxford Nanopore Sequencing has not, as of yet, been employed as an analytical tool in biopharmaceutical production and its own limitations as an analytical tool must be critically assessed.

The thesis investigates the potential of Oxford Nanopore Sequencing (ONS) as a high-resolution analytical tool for assessing mRNA stability. The three main objectives of this project are as follows:

1. Assess the stability of naked and lipid nanoparticle (LNP)-encapsulated eGFP mRNA subjected to accelerated thermal degradation studies (25°C, 35°C, and 50°C) and analyse the samples using Capillary Electrophoresis (CE) and ONS comparing the two methods.

2. Investigate whether mRNA fragmentation occurs randomly or not using ONS data.

3. Identify the limitations of ONS as a tool for mRNA stability testing. 

1.1. Introduction to mRNA Therapeutics

mRNA therapies are a class of biopharmaceutical products which utilise the host’s cellular machinery for translation to encode therapeutic proteins. The mechanism of action of mRNA therapeutics takes its inspiration from the central dogma of molecular biology, as postulated by Francis Crick, which states that genetic information stored in the sequences of nucleic acids can direct protein synthesis (Crick, 1970). The active ingredient of an mRNA therapy is a carefully engineered mRNA transcript carrying genetic information to instruct the host cell’s ribosomes to translate a protein of interest.

mRNA therapies may be designed to replace or supplement essential proteins such as enzymes in patients with genetic deficiencies in those enzymes, or they may be designed as an immunotherapy/vaccine against infectious diseases or cancers by encoding antigens which are recognised by the host’s immune system (Kim, 2022). In the case of mRNA vaccines, the translated protein requires further cellular processing before it can be recognised by the immune system.

In non-immune, nucleated cells, the antigen protein encoded by an mRNA vaccine is not required for a specific cell function and is therefore marked for degradation by the addition of the regulatory protein, ubiquitin, to specific amino acid residues. The ubiquitylated protein is then directed to the host cell’s proteasome where it is digested into short fragments. These fragments are transported to 7 the endoplasmic reticulum where they are loaded onto the class I major histocompatibility complex (MHC I) and displayed on the outer surface of the cell membrane (Wang and Maldonado, 2006). Once the peptide fragment is displayed on the outer surface of the cell membrane, it is known as an epitope and it may be recognised by the T-cell receptor (TCR) of a cytotoxic CD8+ T lymphocyte, also known as a Killer T cell. The Killer T cell destroys the cell, causing the release of the rest of the antigenic protein into the extracellular matrix (ECM) (Chaudhary et al., 2021). 

In certain immune cells, the antigenic protein is taken up from the surrounding ECM via endocytosis and digested into peptide fragments within the endosomes. Inside of the endosomes, the peptide fragments are loaded onto the class II major histocompatibility complex (MHC II) and displayed on the outer surface of the immune cell membrane. While displayed on the outer surface of the immune cell, the MHC II-peptide complex is capable of being recognised by the TCRs of CD4+ T lymphocytes, also known as Helper T cells. Helper T cells stimulate the activity of B lymphocytes, which are immune cells that secrete antibodies which will bind to the antigen protein and its fragments, interfering with the antigen’s activity. This response allows for the development of adaptive immunity (Chaudhary et al., 2021). These pathways are illustrated in Figure 1 below. 

Furthermore, the mRNA and the LNP themselves can act as adjuvants as they can activate different pattern recognition receptors in the cell to stimulate the production of pro-inflammatory cytokines and further trigger the host’s immune response (Kiaie et al., 2022). Once endocytosed and released from the LNP, mRNA can trigger the endosomal Toll-Like Receptors (TLRs) 3, 7, and 8 which are activated by dsRNA and ssRNA. These TLRs may also be activated by the lipids used in the LNP’s composition as seen in Figure 2. 

Pseudouridine and other modified bases may be used in the design of an mRNA therapeutic as a means of mitigating adverse innate immune system reactions within patients (Karikó et al., 2005). Most notably, this technology has been used in the development of the Pfizer/BioNTech and Moderna COVID-19 vaccines, named Comirnaty and Spikevax respectively, which have all of their uridine residues replaced by the modified base N1-methylpseudouridine which has improved chemical, immunological, and productive properties when compared with pseudouridine substitution (Morais et al., 2021). 

1.2. RNA’s Intrinsic Instability

Compared to DNA, RNA is an inherently unstable molecule. This is because of the presence of a reactive hydroxyl group at the 2′ position of the ribose sugar (Greis et al., 2022). Under physiological conditions, the 2′-OH can act as a nucleophile and attack the adjacent phosphate group within the polymer’s backbone, leading to spontaneous cleavage of the phosphodiester bond. This intramolecular transesterification leads to the formation of an unstable intermediate phosphorane species which is readily hydrolysed, breaking the RNA polymer. This process, also known as in-line hydrolytic cleavage, is accelerated by alkaline pH, divalent metal ions (e.g., Mg2+), and elevated temperature (Oivanen et al., 1998). The secondary structure of the RNA polymer can also affect the stability of the molecule, with double-stranded regions less prone to cleavage and digestion by nucleases (Wayment-Steele et al., 2021).

The cleavage of the RNA molecule has serious implications for the efficacy of mRNA therapeutics. Cleavage of the mRNA strand can directly impact the potency of the mRNA resulting in decreased levels of desired protein expression (Crommelin et al., 2021). It is for this reason that mRNA stability is considered a Critical Quality Attribute.

RNA’s intrinsic chemical lability is a major reason as to why mRNA requires protective delivery systems such as lipid nanoparticles (LNPs) for therapeutic applications.

1.3. Lipid Nanoparticle Formulation for mRNA Delivery 

mRNA is a large, negatively charged, and highly unstable molecule. Naked mRNA cannot easily cross cell membranes due to electrostatic repulsion and is rapidly degraded by ubiquitous extracellular RNases, leading to very short half-lives in circulation. Therefore, a defensive barrier must be formulated in order to protect the mRNA from RNases and itself while aiding its entry into cells. Lipid nanoparticles are the most widely used formulation being used today, with other formulations being developed as alternatives (Hou et al., 2021). 

Encapsulation in LNPs protects mRNA from enzymatic degradation, enhances its stability in the bloodstream, and facilitates cellular uptake through endocytosis. Furthermore, the ionisable lipids in LNPs promote endosomal escape, ensuring that the mRNA reaches the cytoplasm where it can be translated into protein. Without LNP encapsulation, systemic delivery of mRNA at therapeutically relevant levels would be ineffective. The method of mRNA encapsulation in LNP using microfluidics is shown below in Figure 4.

Four different lipids are required for the successful mRNA encapsulation: a cationic lipid, a PEG lipid, cholesterol, and a helper lipid. Ionisable lipids are essential components of lipid nanoparticles (LNPs) used for nucleic acid delivery because of their pH-dependent charge properties (Cullis and Hope, 2017). During formulation at acidic pH (~4–5), they are positively charged and electrostatically interact with negatively charged nucleic acids, enabling efficient encapsulation. At physiological pH (7.4), they remain largely neutral, reducing toxicity, nonspecific interactions, and immunogenicity compared to permanently cationic lipids. Once the LNPs are taken up into endosomes, the acidic environment causes the ionisable lipids to regain positive charge, destabilising the endosomal membrane and promoting release of the RNA cargo into the cytoplasm. In addition, ionisable lipids support LNP self-assembly and stability, allowing high encapsulation efficiency at relatively low doses, which improves therapeutic efficacy while minimising side effects.

Helper lipids are structural lipids that stabilise the LNP bilayer and help maintain its integrity during circulation. They often adopt a phospholipid form (like DSPC, 1,2-distearoyl-sn-glycero-3 phosphocholine) and support the formation of a lamellar phase, improving stability and fusion with cell membranes. Cholesterol intercalates between lipid tails in the LNP and provides rigidity and mechanical stability to the nanoparticle. It enhances packing density, improves particle stability in serum, and facilitates membrane fusion events that aid endosomal escape. PEG-lipids form a hydrophilic corona on the surface of LNPs that reduces aggregation, prevents opsonisation, and prolongs circulation time. However, their presence is transient because PEG-lipids gradually dissociate after administration, allowing the LNPs to interact with cells and deliver their cargo. 

1.4. Critical Quality Attributes for mRNA Drug Products

For mRNA therapeutics, several critical quality attributes (CQAs) must be carefully controlled to ensure the safety, efficacy, and consistency of the drug product (Camperi et al., 2025). Key CQAs include mRNA integrity (full-length vs. truncated transcripts), sequence identity, capping efficiency, poly(A) tail length, and the presence of impurities such as double-stranded RNA (dsRNA), which can trigger unwanted innate immune responses, as well as residual DNA template, protein contaminants, and residual solvents or enzymes from in vitro transcription.

Several analytical techniques have been developed for the measurement of each attribute. Currently, the most widely used tool for the measurement of mRNA stability is capillary electrophoresis. CE functions by having an electric field applied across a narrow capillary, through which negatively charged molecules can migrate. Different molecules exhibit different speeds of migration through the capillary based on their size and charge. Molecules may be detected using fluorescence or UV absorbance depending on the method (Camperi et al., 2025). However, while CE is a fast and cost effective way of measuring mRNA stability, it does not give any information about the location of an mRNA strand cleavage.

Chromatographical methods have also been applied to the characterisation of mRNA CQAs. These methods, when used in tandem with mass spectrometric approaches are capable of providing unparalleled insight into the structure of mRNA molecules allowing for the high-resolution assessment of mRNA CQAs. Several of these methods involve the digestion of mRNA using RNases and the measurements of the subsequent oligonucleotides (known as a bottom-up approach); this obviously compromises the stability of the analyte meaning that another measurement must be used to measure mRNA stability (Camperi et al., 2025).

Next Generation Sequencing techniques have been explored as possible methods of assessing CQAs during manufacture (Gunter et al., 2023). These methods include short-read methods, such as Illumina sequencing, and long-read methods such as PacBio and Oxford Nanopore. The problem with many of these methods is the possibility that they may introduce bias to the measurement because of the steps involved in sample preparation. Many of these methods involve reactions such as reverse transcription (RT) or polymerase chain reaction (PCR) in order to analyse the mRNA. The presence of these interfering reactions mean that the original mRNA strand is not being directly sequenced and so the results may not be entirely accurate.

It is clear that a high-resolution, sequencing method which can directly measure the mRNA’s most important CQAs without interfering with the analyte’s structure is needed.

1.5. Nanopore Sequencing for Simultaneous Measurement of Multiple CQAs

Direct RNA sequencing (DRS) using the Oxford Nanopore Sequencing platform offers the possibility of simultaneously interrogating multiple CQAs in a single experiment. DRS reads native polyadenylated RNA molecules directly, enabling high-resolution assessment of mRNA integrity and size distribution, verification of sequence fidelity, and measurement of poly(A) tail length without the need for reverse transcription or amplification (Gunter et al., 2023, Garalde et al., 2018).

Furthermore, off-target alignments can reveal the presence of contaminating host or process-derived RNAs. Nevertheless, certain CQAs, notably double-stranded RNA (dsRNA) impurities and residual DNA template from in vitro transcription, remain challenging to quantify using DRS alone, as these analytes are not efficiently captured by the poly(A)-dependent sequencing workflow. Thus, DRS provides a powerful complementary approach for CQA analysis of mRNA vaccines, consolidating several key assessments into a single workflow while highlighting the need for orthogonal methods for other CQAs such as detection of the presence of dsRNA and residual DNA (Camperi et al., 2025). 

The method begins with polyadenylated RNA, such as in vitro transcribed mRNA. Polyadenylation is required because the sequencing adapter carries an oligo(dT) overhang that specifically hybridises to the RNA’s poly(A) tail. This ensures directional ligation and provides an anchoring site for adapter attachment. After this hybridisation, a second adapter containing a motor protein is ligated to the RNA. The motor protein is essential because it controls the translocation of the RNA strand through the nanopore at a uniform speed, allowing the ionic current disruptions caused by each nucleotide to be recorded and converted into sequence data (Figure 5).

The use of adapters is therefore critical both for physically tethering the RNA to the sequencing complex and for regulating its movement through the pore. Together, these features enable DRS to provide full-length sequence information, poly(A) tail length estimates, and base modification detection directly from native RNA molecules (Garalde et al., 2018).

While DRS using Nanopore sequencing does include a reverse transcription step in its sample preparation protocol, this is done to stabilise the RNA strand as it passes through the nanopore. The native RNA is the analyte being sequenced directly and so bias introduced from reverse transcription is minimal.

As each nucleotide passes through the pore, it alters the ionic current flowing across the membrane in a sequence-dependent manner. These changes are recorded in real time as a continuous electrical trace, often referred to as a squiggle. The nanopore senses multiple nucleotides at once rather than a single nucleotide (typically 5-6 bases at a time) and so the current signal at each instance corresponds to a particular k-mer, or group of bases. The raw current signal is subsequently processed by basecalling algorithms such as Dorado, which employs recurrent neural networks or convolutional architectures trained on large datasets of known sequences (Garalde et al., 2018).

These model maps patterns in the squiggle to the most probable nucleotide sequence, also estimating per-base quality scores. For direct RNA, additional models can be applied to detect features like poly(A) tail length and base modifications by identifying systematic deviations in the current signal. Thus, the workflow converts the raw electrical signal into interpretable RNA sequence data in a single, real-time analysis pipeline providing as an output FASTQ files which contain information about the base identity and the quality score assigned to each of the basecalled bases (Figure 6).

The aim of this project is to evaluate the suitability of Oxford Nanopore Sequencing for assessing mRNA stability, using eGFP mRNA as a model transcript, and comparing its performance against capillary electrophoresis. Specifically, this thesis investigates: the stability of naked and LNP encapsulated mRNA under accelerated degradation conditions, whether or not ONS can provide mechanistic insight into mRNA degradation, and what limitations of the platform are holding it back from becoming a widespread analytical tool for monitoring mRNA CQAs.

It is the hypothesis of this study that both naked and LNP-encapsulated mRNA stability will decrease under the accelerated degradation conditions measured in this study and that Oxford Nanopore Sequencing will give higher resolution information on mRNA length and stability when compared with capillary electrophoresis.

An important matter to consider is whether or not the mRNA is fragmenting randomly. The high resolution information gained from the sequencing of purposefully degraded mRNA may allow us to answer this question by measuring the distribution of fragments. This information may be useful for designing stabilised mRNA products in the future by pinpointing and protecting regions which are prone to cleavage.