Today's mRNA vaccines must be kept at -20 to -80°C; the new formula stayed effective for two months at 37°C and a year at room temperature in mice.
Researchers at MIT used an AI algorithm to make mRNA vaccines stable enough to skip the -20 to -80°C storage they now need, which makes them hard to ship to places without freezers. The team, led by Ana Jaklenec and Robert Langer, reported the work in Nature Biotechnology.
The fragile mRNA travels inside lipid nanoparticles, fatty shells that carry it into cells, and the team set out to stabilize shells similar to those in Moderna's Covid-19 vaccine. It first spent months testing additives that had worked in its earlier projects, and none got the vaccines fully stable. So it built an algorithm with MIT's Computer Science and Artificial Intelligence Laboratory that learns from very few data points.
From nearly 50 FDA-approved additives, the researchers kept five of the most promising. The algorithm then proposed ratios of those five; the team checked two candidates at a time in cells and returned the results to the algorithm for its next proposal. After several rounds and a few weeks, one formula was ready for animal tests. Mina Konaković Luković, a co-author from the lab, says the algorithm got there in a handful of iterations, where such work usually means testing every option.
Dried and then stored for two months at 37°C or a year at room temperature, the vaccines produced immune responses in mice that matched those from a fresh formulation like Moderna's. The team also made solid patches of microscopic needles that dissolve in the skin. The patches delivered the vaccine without cold storage in rodents and in nonhuman primates, and drew an immune response similar to that from injections.
The same approach stabilized a formulation similar to Pfizer's. Once a carrier is stable, the researchers say, it could be adapted to any mRNA payload, including vaccines now being developed against cancer.