[Paper Review] Algorithm for Optimized mRNA Design Improves Stability and Immunogenicity

He Zhang, Liang Zhang+12 more|arXiv (Cornell University)|Apr 21, 2020

RNA and protein synthesis mechanisms45 references20 citations

TL;DR

This paper introduces LinearDesign, a novel algorithm that optimizes mRNA sequences by jointly maximizing secondary structure stability and codon usage efficiency, reducing the search space from ∼10^632 candidates to a tractable problem via a computational linguistics analogy. The method achieves up to 23× higher antibody responses in vivo and significantly improved mRNA half-life and protein expression without chemical modifications.

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$ imes$ 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).

Motivation & Objective

• Address the critical challenge of mRNA instability and degradation that limits vaccine storage, distribution, and efficacy.
• Overcome the exponentially large search space (∼10^632 possible sequences for SARS-CoV-2 Spike) in designing optimal mRNA sequences.
• Simultaneously optimize both structural stability (via minimum free energy) and codon usage (via CAI) to enhance translation efficiency.
• Develop a scalable, fast algorithm that enables experimental validation of highly stable and efficient mRNA designs previously unreachable by brute-force or heuristic methods.

Proposed method

• Map the mRNA design problem to a classical problem in computational linguistics—finding the most likely sentence among phonetically similar alternatives—thereby transforming the search into a dynamic programming problem.
• Use dynamic programming to efficiently compute the optimal mRNA sequence that maximizes both thermodynamic stability (minimum folding free energy, MFE) and codon adaptation index (CAI).
• Model the coding region as a sequence of synonymous codons, with each position weighted by codon preference and local structural stability.
• Leverage the ViennaRNA package (RNAfold) to predict secondary structures and folding free energies (MFE) for in silico evaluation.
• Integrate UTRs (5’ and 3’) into full-length mRNA structure prediction to assess inter-UTR and coding region base pairing, minimizing destabilizing interactions.
• Validate designs experimentally via in vitro half-life assays and in vivo immunogenicity testing in mouse models.

Experimental results

Research questions

• RQ1 Can a principled algorithm jointly optimize mRNA secondary structure stability and codon usage to improve protein expression and immunogenicity?
• RQ2 Is it possible to reduce the intractable search space of synonymous mRNA sequences (∼10^632 for Spike) to a computationally feasible problem?
• RQ3 Can the algorithm generate designs that outperform codon-optimized benchmarks in both in vitro stability and in vivo immunogenicity without chemical modifications?
• RQ4 How do UTR interactions affect the structural stability and functional performance of full-length mRNA constructs?
• RQ5 To what extent can structural stability (MFE) and codon usage (CAI) be balanced to maximize overall mRNA efficiency?

Key findings

• LinearDesign designs the SARS-CoV-2 Spike mRNA in only 11 minutes, enabling rapid exploration of previously unreachable sequence space.
• Without chemical modifications, LinearDesign-generated mRNAs showed significantly improved in vitro mRNA half-life and protein expression compared to the codon-optimized benchmark.
• In vivo, LinearDesign-induced antibody responses were up to 23× higher than those from the codon-optimized benchmark sequence.
• The algorithm-generated sequences (e.g., sequence a) achieved a minimum folding free energy (MFE) of -2,287.3 kcal/mol, substantially more stable than the wildtype (-966.7 kcal/mol) and baseline (-1,244.4 kcal/mol).
• Sequences with higher structural stability (lower MFE) and optimized codon usage (CAI > 0.9) formed fewer and smaller base pairs with UTRs, reducing structural interference.
• The MFE-optimized sequence achieved -2,486.7 kcal/mol, the most stable of all tested sequences, and outperformed even commercial vaccine sequences (e.g., Moderna’s mRNA-1273, BioNTech’s BNT-162b2) in structural stability.