Algorithm for Optimized mRNA Design Improves Stability and Immunogenicity

Abstract: Messenger RNA (mRNA) vaccines are being used for COVID-19, but still suffer from the critical issue of mRNA instability and degradation, which is a major obstacle in the storage, distribution, and efficacy of the vaccine. Previous work showed that optimizing secondary structure stability lengthens mRNA half-life, which, together with optimal codons, increases protein expression. Therefore, a principled mRNA design algorithm must optimize both structural stability and codon usage to improve mRNA efficiency. However, due to synonymous codons, the mRNA design space is prohibitively large, e.g., there are $\sim!10^{632}$ mRNAs for the SARS-CoV-2 Spike protein, which poses insurmountable challenges to previous methods. Here we provide a surprisingly simple solution to this hard problem by reducing it to a classical problem in computational linguistics, where finding the optimal mRNA is akin to finding the most likely sentence among similar sounding alternatives. Our algorithm, named LinearDesign, takes only 11 minutes for the Spike protein, and can jointly optimize stability and codon usage. Experimentally, without chemical modification, our designs substantially improve mRNA half-life and protein expression in vitro, and dramatically increase antibody response by up to 23$\times$ in vivo, compared to the codon-optimized benchmark. Our work enables the exploration of highly stable and efficient designs that are previously unreachable and is a timely tool not only for vaccines but also for mRNA medicine encoding all therapeutic proteins (e.g., monoclonal antibodies and anti-cancer drugs).

An Analysis of LinearDesign: Algorithm for Optimized mRNA Design

The paper presents an innovative approach to tackling the challenge of designing mRNA sequences optimal for stability and immunogenic response using a novel algorithm, LinearDesign. This algorithm leverages concepts from computational linguistics and formal language theory, specifically adapting lattice parsing techniques traditionally used for word-sense disambiguation to the domain of mRNA design. In doing so, LinearDesign aims to enhance the stability of mRNA molecules, which are notoriously fragile and require careful handling and storage.

Theoretical Framework and Algorithm Design

LinearDesign addresses the dual objectives of mRNA design: enhancing stability and optimizing codon usage. The need for stability arises from the inherent chemical instability and risk of degradation associated with mRNA molecules, a significant obstacle in both the storage and efficacy of mRNA vaccines. LinearDesign seeks to overcome these challenges by optimizing the secondary structure of mRNA designs, creating sequences with improved thermodynamic stability as quantified by the minimum free energy (MFE). Codon optimization, measured by the Codon Adaptation Index (CAI), is also integral to the process as it affects translation efficiency and effectiveness.

By representing the design space as a deterministic finite-state automaton (DFA), LinearDesign provides a framework to encode the multitude of possibilities in mRNA sequence design, reflecting the many ways amino acids can be represented in nucleotide sequences. The algorithm reduces the search for an optimal sequence to the problem of identifying the optimal path through this lattice, akin to finding the grammatically correct sentence with the highest probability in language processing tasks.

Empirical Results: mRNA Stability and Immunogenicity

The paper presents empirical evidence supporting the efficacy of LinearDesign in generating mRNA sequences that significantly enhance protein expression and immunogenicity. Experimental assays confirmed that mRNA sequences designed using LinearDesign have enhanced chemical stability, with longer half-lives and increased resistance to degradation. For instance, mRNA molecules generated by this algorithm showed up to a 23-fold increase in immunogenic response compared to traditional codon-optimized sequences.

Moreover, the experimental results showcased higher levels of in vitro protein expression indicating that LinearDesign optimizes not only the physical and structural properties of mRNA molecules but also their functional outcomes. Five of the seven mRNA designs exhibited significantly higher protein expression levels than the baseline, which were verified via flow cytometry in HEK-293 cells.

Algorithmic Efficiency and Potential Applications

This approach's core advantage is its computational efficiency, with LinearDesign shown to scale quadratically with sequence length, thus significantly reducing computation time in comparison to previous methods. A linear-time approximation algorithm using beam search further enhances its practical application, making it feasible for designing long mRNA sequences efficiently.

The significance of these computational advancements lies in their potential application beyond vaccine design. The proposed framework can support the development of mRNA-based therapeutics, including monoclonal antibodies and anti-cancer drugs, by optimizing the stability and effectiveness of therapeutic mRNAs.

Conclusion and Future Prospects

LinearDesign represents a comprehensive, scalable tool that potentially opens new avenues for mRNA vaccine applications and other therapeutic designs. Its cross-disciplinary techniques, integrating principles from computational linguistics, provide a versatile approach that is further exemplified by its adaptability to varying genetic codes and nucleotide modifications.

Moving forward, the flexibility of the DFA framework demonstrated by LinearDesign can be expected to handle new challenges presented by emerging pathogens and genetic conditions with precision, potentially transforming the landscape of mRNA medicine. Further exploration will likely focus on refining the algorithm to incorporate additional biological constraints and respond robustly to evolving therapeutic needs.