Scientists Map the Analytical Toolbox for Quality-Checking mRNA Drugs

Messenger RNA therapeutics have moved from laboratory curiosity to commercial reality, but the quality control machinery behind them has not kept pace with the science. A new review argues that the industry’s habit of running one analytical test per quality attribute is a costly bottleneck, and it lays out a roadmap for consolidating those tests into multi-attribute methods, or MAMs, that could measure several critical quality attributes in a single experiment.

The review, published in Discover Pharmaceutical Sciences by researchers including Kyle J. Tynan, Karndeep Singh, Trina Mouchahoir, Mark S. Lowenthal and Karen W. Phinney, surveys the electrophoretic, chromatographic, sequencing-based and mass spectrometry-based techniques commonly used to characterize mRNA drug substance. The authors compare each approach with respect to the number of quality attributes it can report, throughput, preprocessing requirements and total analysis time, and then critically assess which ones could realistically be integrated into a MAM platform.

The stakes are considerable. To be approved as safe and effective, mRNA manufacturers must demonstrate that they have met specifications on a battery of critical quality attributes, including identity, length, concentration, intactness and purity. Today, most analytical techniques evaluate only a single attribute per experiment, which means many different methods must be run in sequence. That inflates analysis time, instrumentation requirements and operational cost, and those costs ultimately flow through to the final drug product.

The authors put a number on the burden. If a manufacturer used the fastest available technique for each attribute, not necessarily the most accurate one, it would require a minimum of 44 hours of analyst time, and that estimate excludes the various types of equipment and training needed for each experiment. Consolidating the many assays needed to evaluate in vitro transcribed, or IVT, mRNA into a single assay or a handful of assays would significantly reduce time, training and equipment, lowering costs for manufacturers and for patients.

The review grounds its argument in the biology. Messenger RNA for human therapeutic applications consists of five key regions: the 5′-cap, the 5′-untranslated region, an open reading frame that encodes the target protein, the 3′-untranslated region, and the 3′-poly(A) tail, which is typically 15 to 250 adenosine residues long and enhances protein translation and stability. Each region plays a distinct role in ensuring translation and stability, and each can be compromised during manufacturing.

That manufacturing process, in vitro transcription, is the most common technique for producing target mRNA sequences. mRNA is synthesized from a DNA template in three main steps: initiation of transcription by an RNA polymerase at the T7 promoter of the DNA template, elongation of the RNA transcript, and termination of synthesis. The process leaves behind a characteristic set of impurities: unincorporated nucleotides, remaining plasmid template, and residual T7 polymerase. During elongation, the T7 polymerase can also produce double-stranded RNA, an immunogenic byproduct that must be quantified because it triggers an immune response. After IVT, a 5′-capping enzyme completes the synthesis.

For identity, the gold standard is sequence confirmation, but the review shows how expensive that certainty can be. Reverse transcription-polymerase chain reaction underpins both Sanger sequencing and high-throughput sequencing, yet RT-PCR alone does not provide sequence information, and its heavy preprocessing makes it a poor MAM candidate. Sanger sequencing takes up to five hours per run and relies on stochastic base addition, which the authors judge infeasible for a MAM. High-throughput sequencing generates cDNA libraries that must be fragmented, adapter-ligated and digitally aligned. Nanopore sequencing stands out: it reads the RNA directly rather than a cDNA copy, can distinguish various mRNA modifications, and can determine RNA integrity, plasmid template, antisense sequences and poly(A) tail length. The authors call it the most suitable sequencing platform for a MAM, but note that a minimum of 24 hours of experiment time is required, and data analysis time varies from 24 hours to two weeks depending on computing power.

Because sequencing is laborious, identity is often confirmed more cheaply by length. Gel electrophoresis separates transcripts by size through a polyacrylamide or agarose matrix, and with a length-calibration ladder the mRNA length, and thus its identity, can be determined in one to two hours, particularly in capillary format. Its primary drawback is limited resolution, with length determination errors ranging from 0.1 percent to 5 percent. The review highlights micelle capillary electrophoresis, which replaces traditional gels with entangled micelle networks, significantly reducing viscosity and run times while achieving effective length-based separation. In the micelle tagging variant, nucleic acid probes with alkane tails tether the mRNA to the micelle network, increasing resolution and enabling sequence-specific determination. The authors judge MCE and MTE the most MAM-suited electrophoretic methods, since they can rapidly determine length and intactness and, via a standard curve, concentration, though the micelle network must be freshly synthesized each day and probe development is labor intensive.

Concentration, which directly determines the administered dose, is most commonly measured by UV-VIS spectroscopy at 260 and 280 nanometers using the Beer-Lambert law, with an A260/A280 ratio of roughly 2.0 indicating a pure sample. The technique is straightforward but provides limited information: it does not assess identity, intactness or aggregation. Quantitative PCR and digital PCR offer absolute or relative quantification but take between two and fifteen hours, are labor and resource intensive, and require the mRNA to be in the range of 75 to 150 nucleotides, whereas a single nucleotide produces a signal in UV spectroscopy. Digital PCR eliminates the need for a calibration standard by partitioning samples before amplification, but variations in droplet formation can reduce accuracy.

Purity attributes dominate the rest of the survey. Intactness, the proportion of full-length transcripts carrying the 5′-cap, the full intended sequence and the 3′ poly(A) tail, can be visualized on gels but not accurately quantified from band smearing; the 3′:5′ RT-qPCR assay, which compares amplification at the two ends of the transcript, depends heavily on careful primer design and its preprocessing prevents MAM adaptation. For 5′ capping efficiency, the review catalogs fluorescent cap analogs, RNase H cleavage followed by capillary gel electrophoresis, quantitative splinted ligation RT-PCR, ion-paired reverse-phase HPLC, LC-MS, the 5′ CapQ assay, surface plasmon resonance and NMR, most of which demand extensive preprocessing or product-specific probes. Poly(A) tail length measurement by TAIL-seq, mTAIL-seq, PAL-seq and the family of PAT assays all require chemical modification, ligation and PCR steps that rule them out as MAM platforms. For double-stranded RNA, immunoblotting with the J2 monoclonal antibody is semi-qualitative and of low sensitivity, ELISA is prone to variability, and lateral flow strips, though fast, suffer from low sensitivity, with detection limits of 2 nanograms per milliliter for uridine-containing dsRNA and 70 nanograms per milliliter for m1psi-containing dsRNA. An RP-HPLC method developed by Rosa and colleagues is quantitative and improves detection limits, and the authors note it may be possible after optimization to determine dsRNA and poly(A) tail length in one experiment. Residual plasmid template is measured by qPCR, residual T7 polymerase by ELISA or an automated capillary Simple Western, and aggregation by size exclusion chromatography or dynamic light scattering, the latter taking roughly five minutes with automatic data analysis.

Two existing platforms show what consolidation looks like in practice. VAX-seq uses long-read nanopore sequencing followed by bioinformatic analysis to assess mRNA sequence and length, poly(A) tail length, integrity and unincorporated nucleotides in a single run, mapping reads to the reference plasmid and, if the sequence fails to match, aligning against the production strain genome to quantify plasmid contamination. Pfizer’s oligonucleotide mapping method digests the sample with RNase T1, resolves the fragments by ion-paired reverse-phase HPLC, and detects them by UV absorbance and tandem mass spectrometry, achieving 100 percent sequence coverage along with poly(A) tail length and 5′ capping efficiency in one experiment; the Pfizer MAM requires three days to acquire sequence, poly(A) tail length and capping efficiency, consolidating what would otherwise be three different techniques into one. The drawback is an extensive data set and extremely long processing time.

The review’s conclusion is measured rather than triumphant. At present, MAM technologies can couple only a subset of quality attributes and often require large datasets and extensive data processing, and the total time to perform a MAM can exceed that of measuring each attribute individually. Even so, a single consolidated assay offers reduced training requirements, lower capital investment and simplified method lifecycle management. The authors consider it unlikely that all attributes can be determined in a single experiment, and instead recommend identifying attributes that can be readily coupled, such as mRNA length, concentration and intactness within a single RP-HPLC method given appropriate standard curves. Approaches relying on chemical modification or enzymatic digestion inherently limit the number of assessable attributes and should be avoided. Purity-related impurities such as unincorporated nucleotides, residual plasmid DNA and residual T7 polymerase are attractive MAM targets because they readily separate using conventional electrophoretic or chromatographic techniques, but after purification they typically fall below the detection limits of standard UV detection, so the authors emphasize signal amplification through fluorescence-based detection and intercalating dyes as a viable path forward. As demand for mRNA therapeutics expands, they argue, robust mRNA-specific MAM platforms will be crucial for efficient manufacturing, product quality and lower patient cost.

Subject of Research: Analytical techniques for characterizing critical quality attributes of mRNA drug substance and their suitability as multi-attribute methods

Article Title: A review of analytical techniques for characterization of mRNA drug substance in therapeutics and their suitability as a multi-attribute method

Article References: Tynan, K. J., Singh, K., Mouchahoir, T., Lowenthal, M. S., & Phinney, K. W. (2026). A review of analytical techniques for characterization of mRNA drug substance in therapeutics and their suitability as a multi-attribute method. Discover Pharmaceutical Sciences, 2(1), Article 19. https://doi.org/10.1007/s44395-026-00058-0

Image Credits: AI Generated

DOI: 10.1007/s44395-026-00058-0

Keywords: mRNA therapeutics, multi-attribute method, critical quality attributes, in vitro transcription, lipid nanoparticles, quality control, mass spectrometry, capillary electrophoresis, sequencing, poly(A) tail, dsRNA impurities, LC-MS