CodonBERT large language model for mRNA vaccines.

mRNA-based vaccines and therapeutics are gaining popularity and usage across a wide range of conditions. One of the critical issues when designing such mRNAs is sequence optimization. Even small proteins or peptides can be encoded by an enormously large number of mRNAs. The actual mRNA sequence can have a large impact on several properties, including expression, stability, immunogenicity, and more. To enable the selection of an optimal sequence, we developed CodonBERT, a large language model (LLM) for mRNAs. Unlike prior models, CodonBERT uses codons as inputs, which enables it to learn better representations. CodonBERT was trained using more than 10 million mRNA sequences from a diverse set of organisms. The resulting model captures important biological concepts. CodonBERT can also be extended to perform prediction tasks for various mRNA properties. CodonBERT outperforms previous mRNA prediction methods, including on a new flu vaccine data set.

Pan-SARS neutralizing responses after third boost vaccination in non-human primate immunogenicity model.

The emergence of SARS-CoV-2 variants, especially Beta and Delta, has raised concerns about the reduced protection from previous infection or vaccination based on the original Wuhan-Hu-1 (D614) virus. To identify promising regimens for inducing neutralizing titers towards new variants, we evaluated monovalent and bivalent mRNA vaccines either as primary vaccination or as a booster in nonhuman primates (NHPs). Two mRNA vaccines, D614-based MRT5500 and Beta-based MRT5500ß, tested in sequential regimens or as a bivalent combination in naïve NHPs produced modest neutralizing titers to heterologous variants. However, when mRNA vaccines were administered as a booster to pre-immune NHPs, we observed a robust increase in neutralizing titers with expanded breadth towards all tested variants, and notably SARS-CoV-1. The breadth of the neutralizing response was independent of vaccine sequence or modality, as we further showed either MRT5500 or recombinant subunit Spike protein (with adjuvant) can serve as boosters to induce broadly neutralizing antibodies in the NHPs primed with MRT5500. The data support the notion that a third vaccination is key to boosting existing titers and improving the breadth of antibodies to address variants of concern, including those with an E484K mutation in the Receptor Binding Domain (RBD) (Beta, Gamma).

Immunogenicity and efficacy of mRNA COVID-19 vaccine MRT5500 in preclinical animal models.

Emergency use authorization of COVID vaccines has brought hope to mitigate pandemic of coronavirus disease 2019 (COVID-19). However, there remains a need for additional effective vaccines to meet the global demand and address the potential new viral variants. mRNA technologies offer an expeditious path alternative to traditional vaccine approaches. Here we describe the efforts to utilize an mRNA platform for rational design and evaluations of mRNA vaccine candidates based on the spike (S) glycoprotein of SARS-CoV-2. Several mRNA constructs of S-protein, including wild type, a pre-fusion stabilized mutant (2P), a furin cleavage-site mutant (GSAS) and a double mutant form (2P/GSAS), as well as others, were tested in animal models for their capacity to elicit neutralizing antibodies (nAbs). The lead 2P/GSAS candidate was further assessed in dose-ranging studies in mice and Cynomolgus macaques, and for efficacy in a Syrian golden hamster model. The selected 2P/GSAS vaccine formulation, designated MRT5500, elicited potent nAbs as measured in neutralization assays in all three preclinical models and more importantly, protected against SARS-CoV-2-induced weight loss and lung pathology in hamsters. In addition, MRT5500 elicited TH1-biased responses in both mouse and non-human primate (NHP), thus alleviating a hypothetical concern of potential vaccine-associated enhanced respiratory diseases known associated with TH2-biased responses. These data position MRT5500 as a viable vaccine candidate for entering clinical development.