Theoretical basis for stabilizing messenger RNA through secondary structure design

Abstract

RNA hydrolysis presents problems in manufacturing, long-term storage, world-wide delivery and in vivo stability of messenger RNA (mRNA)-based vaccines and therapeutics. A largely unexplored strategy to reduce mRNA hydrolysis is to redesign RNAs to form double-stranded regions, which are protected from in-line cleavage and enzymatic degradation, while coding for the same proteins. The amount of stabilization that this strategy can deliver and the most effective algorithmic approach to achieve stabilization remain poorly understood. Here, we present simple calculations for estimating RNA stability against hydrolysis, and a model that links the average unpaired probability of an mRNA, or AUP, to its overall hydrolysis rate. To characterize the stabilization achievable through structure design, we compare AUP optimization by conventional mRNA design methods to results from more computationally sophisticated algorithms and crowdsourcing through the OpenVaccine challenge on the Eterna platform. We find that rational design on Eterna and the more sophisticated algorithms lead to constructs with low AUP, which we term 'superfolder' mRNAs. These designs exhibit a wide diversity of sequence and structure features that may be desirable for translation, biophysical size, and immunogenicity. Furthermore, their folding is robust to temperature, computer modeling method, choice of flanking untranslated regions, and changes in target protein sequence, as illustrated by rapid redesign of superfolder mRNAs for B.1.351, P.1 and B.1.1.7 variants of the prefusion-stabilized SARS-CoV-2 spike protein. Increases in in vitro mRNA half-life by at least two-fold appear immediately achievable.

Figure 1.

(A) Hydrolysis of the phosphodiester bond in the RNA backbone bond. This mechanism proceeds via an ‘inline attack’ backbone conformation, depicted in (B): the attacking 2′-hydroxyl group is in line with the phosphate group and the leaving 5′ oxygen. (C) Sequence schematics of all mRNA design challenges in this work. (D) mRNAs designed by conventional means for therapeutics are prone to hydrolysis in regions that have high probability of being unpaired (shown in yellow, left panel). A design for an mRNA vaccine encoding the prefusion-stabilized SARS-CoV-2 full spike protein (S-2P) dramatically reduces the probability of being unpaired throughout the molecule (purple, right panel).

Figure 2.

(A) Enumerating all possible coding sequences for small tag proteins reveals that the coding sequence with the lowest energy for its MFE structure (blue star) is not always the same as the coding sequence with the lowest AUP (red star). (B) MFE structures for mRNA solutions for the HA tag that minimize AUP, minimize free energy, and maximize codon adaptation index (CAI), with nucleotides colored by their computed probability of being unpaired.

Figure 3.

Sequences designed rationally by participants during Eterna's OpenVaccine challenge result in the lowest AUP values for mRNAs encoding a variety of model proteins used for studying translation and as model vaccines, ranging in length from 144 nucleotides (the Multi-epitope Vaccine) to 855 nucleotides (eGFP + degron (35)). (A) Force-directed graph visualization of MFE structures predicted for sequences with lowest AUP value from each design source, colored by the computed probability of each nucleotide being unpaired. (B) While ΔG(MFE) and AUP are correlated, the design with the lowest AUP is not the same as the design with the lowest ΔG(MFE). Starred points indicate the design for each design strategy with the lowest AUP value, calculated with ViennaRNA.

Figure 4.

mRNA designs with low AUP (A) have codon adaptation index values consistent with high translation efficiency, (B) show a range of values for the probability the first 14 nucleotides of the coding sequence is unpaired (AUP^init,14), and (C) do not have helices longer than 33 nts, suggesting they are unlikely to raise an innate immune response that would shut down cellular mRNA translation. (D) Eterna designs show structural diversity as characterized by the maximum ladder distance (MLD), the longest path of contiguous helices present in the minimum free energy (MFE) structure of the molecule. (E) MFE structures predicted in the ViennaRNA structure prediction package are depicted for designs with a variety of MLD values, indicating similarly stabilized stems for a range of topologies.

Figure 5.

Design of stabilized mRNAs for the SARS-CoV-2 full spike protein achieve the same degree of stabilization as in smaller mRNA design challenges. (A) Solutions voted upon by the Eterna community show a diversity of structures while maintaining low AUP values. The solutions with the lowest AUP are structurally similar to and were derived from the ΔG(MFE) optimal structure from LinearDesign. (B) AUP values from different design methods are consistent across different mRNA lengths. A two-fold increase in lifetime is predicted by changing from a ‘Standard’ design method (methods that do not stabilize structure) to a design method that increases structure.

Figure 6.

Low AUP solutions as superfolder mRNAs. Predicted stabilization derived from a low AUP solution is robust to (A) when calculated at higher temperatures, (B) when calculated in other folding algorithms, (C) in the presence of added UTRs and (D) small variations in protein sequence.