New Formulation Could Help RNA Vaccines Withstand Higher Temperatures

Researchers at MIT have developed a new formulation that could make RNA vaccines significantly more resistant to heat, potentially reducing their dependence on ultracold storage. The approach uses artificial intelligence to optimize the lipid nanoparticles that protect and deliver messenger RNA, or mRNA, inside the body.

The researchers found that the modified vaccines could remain stable at room temperature for up to a year and at nearly 100 degrees Fahrenheit (38 degrees Celsius) for as long as two months. The findings could eventually make RNA vaccines easier to store and distribute, particularly in regions where reliable cold-storage infrastructure is limited.

AI Helps Improve RNA Vaccine Stability

RNA vaccines gained widespread attention during the COVID-19 pandemic and are now being developed for a range of other applications, including vaccines and treatments targeting cancer and infectious diseases.

One major challenge is the fragility of RNA molecules. To prevent degradation and help RNA enter cells, vaccines typically package the genetic material inside lipid nanoparticles, known as LNPs.

Even with this protection, existing RNA-LNP vaccines generally require storage at temperatures ranging from about -4 to -112 degrees Fahrenheit (-20 to -80 degrees Celsius). Those requirements can complicate transportation and distribution, particularly in areas without extensive cold-chain infrastructure.

MIT researchers sought to determine whether formulations similar to those used in FDA-approved COVID-19 vaccines could be made more resistant to heat.

Vaccines Remained Stable at Elevated Temperatures

Using an AI-based algorithm, the researchers adjusted the composition of the lipid nanoparticle formulations and identified combinations capable of maintaining stability under warmer conditions.

The resulting vaccines remained stable at room temperature for up to one year. They also maintained stability for two months when stored at approximately 100 degrees Fahrenheit.

In experiments involving mice, COVID-19 vaccines formulated with the heat-resistant particles produced an immune response comparable to that generated by an RNA vaccine similar to one developed by Moderna.

The algorithm also reduced the number of laboratory experiments required to identify promising formulations.

“The real beauty of this algorithm is that we can use it with small data sets,” said Ana Jaklenec, a principal investigator in MIT’s Koch Institute for Integrative Cancer Research. “It’s really hard to run thousands of experiments, so this algorithm allows us to more easily achieve formulations with features that we want—in this case, stability.”

Jaklenec and Robert Langer, the David H. Koch Institute Professor, are senior authors of the research paper published in Nature Biotechnology. Graduate student Jinbi Tian and postdoctoral researcher Khanh Tran are the lead authors.

Researchers Turn to Machine Learning

Scientists have previously attempted to improve RNA vaccine stability by adding excipients, including sugars, salts and polymers, to lipid nanoparticle formulations.

Jaklenec and Langer had previously developed polymer-stabilized LNPs capable of tolerating higher temperatures. However, those particles differed somewhat from the FDA-approved formulations used in the Moderna and Pfizer COVID-19 vaccines.

For the latest research, the team focused on improving the stability of formulations more closely related to those already used in approved vaccines.

Initially, researchers tested excipients that had been successful in previous experiments, but the approach produced limited results.

“We were trying to use and screen excipients that we’ve previously used successfully to stabilize LNPs, but it just wasn’t working. It was really frustrating for the team,” Jaklenec said.

The researchers then collaborated with scientists at MIT’s Computer Science and Artificial Intelligence Laboratory, or CSAIL, to develop a machine-learning algorithm capable of making predictions using relatively small experimental data sets.

“We’d used our algorithms for various automated experimental design applications before, but never on a biological problem like vaccine stability,” said Mina Konaković Luković, an assistant professor of electrical engineering and computer science at MIT’s CSAIL and an author of the study. “It was surprising to see how quickly the algorithm converged on a stable formulation—getting there in just a handful of iterations, rather than the exhaustive search that would normally be required.”

Heat-Tolerant Vaccines Could Expand Distribution Options

Improving temperature stability could simplify the transportation and storage of RNA vaccines while reducing dependence on specialized freezers.

More stable formulations could also support alternative delivery systems, including microneedle patches. These patches contain hundreds of tiny, vaccine-filled needles that dissolve after being applied to the skin, releasing the vaccine without conventional injections.

The findings suggest that combining machine learning with formulation research could accelerate the development of RNA vaccines designed to withstand a wider range of storage conditions. If the approach proves effective in further testing, it could make mRNA vaccines more practical to distribute in places where maintaining an ultracold supply chain remains difficult.