A research team at KAIST has developed a promising new intranasal antiviral platform that uses advanced AI engineering to block rapidly mutating respiratory viruses, including influenza and COVID-19, at their earliest point of entry. The innovation could mark an important step forward in the global fight against respiratory infections, particularly as vaccines alone cannot fully protect against viruses that evolve quickly and in many different forms.
At the core of the new platform is a re-engineered version of interferon-lambda, a naturally occurring immune protein that plays a central role in stopping viral infections in the respiratory tract. Interferon-lambda has long been considered a promising therapeutic candidate, but its clinical usefulness has been limited. When administered in the nose, it breaks down rapidly due to heat, mucus, enzymes and the natural clearing mechanisms of the nasal mucosa. This means that its antiviral effects disappear before it can offer meaningful protection.
AI-driven protein
To address these limitations, KAIST researchers leveraged AI-driven protein design tools to fundamentally redesign the structure of interferon-lambda. By transforming flexible loop regions into more rigid helix structures, the team dramatically increased the protein’s stability. Additional surface engineering steps made the protein less prone to aggregation, while glycoengineering, adding protective sugar chains, further strengthened resilience. The result is a version of interferon-lambda that can withstand temperatures up to 50°C for two weeks, an enormous improvement over the original molecule.
This engineered stability is crucial for practical use, particularly in settings where cold-chain infrastructure is limited. According to the researchers, creating a protein that remains potent without refrigeration could open the door to wider access in low-resource regions. An important advantage during global outbreaks of new or mutant viruses.
But stability alone is not enough for an effective mucosal therapy. The protein also needs to remain in the nasal cavity long enough to exert its protective effects. To achieve this, the KAIST team encapsulated the redesigned interferon-lambda in nanoliposomes. These are tiny lipid-based vesicles commonly used in drug delivery. They then coated these nanoliposomes with low-molecular-weight chitosan, a biocompatible material known for its strong mucoadhesive properties.
This combined approach allows the antiviral compound to spread effectively through the nasal mucus and adhere to the mucosal surface for extended periods. In animal models infected with influenza, the treatment reduced viral levels in the nasal cavity by over 85%, demonstrating a powerful inhibitory effect.
Universal antiviral tool
Beyond influenza, the platform is designed as a universal antiviral tool. Because interferon-lambda activates broad innate immune responses rather than targeting a specific virus, it may help protect against a wide range of existing and emerging respiratory pathogens. This includes viruses like SARS-CoV-2, but also future variants that may be difficult to address quickly with conventional vaccines.
Professor Ho Min Kim of KAIST emphasizes that combining AI-based protein design with advanced drug-delivery engineering allowed the team to overcome two major barriers at once: the protein’s structural fragility and its short retention time in the nose. “This platform, which is stable at high temperatures and stays in the mucosa for a long time, is an innovative technology that can be used even in developing countries lacking strict cold-chain infrastructure,” he says. He also highlights the scalability of the technology, noting its potential for developing new antiviral drugs and mucosal vaccines.
The work exemplifies how digital technologies, including AI protein design, are reshaping biomedical innovation. By accelerating the redesign of complex biological molecules and optimizing them for real-world use, AI is enabling new therapeutic strategies that were previously not feasible. For respiratory diseases that spread quickly and evolve unpredictably, such advances may prove transformative.
While further testing, including human clinical trials, is needed, the early findings signal an important direction for the future of infectious-disease prevention: fast-acting, easy-to-administer mucosal therapies engineered with AI and optimized for global use.
Nasal nanomedication
Last month, researchers reported that they succeeded in developing a non-invasive nanomedicine approach that offers new hope for treating glioblastoma, an aggressive and typically fatal brain cancer. The strategy uses spherical nucleic acids (SNAs), nanostructures containing therapeutic DNA fragments, administered as simple nasal drops. This enables the treatment to bypass the blood–brain barrier and activate immune pathways that glioblastoma normally evades.
The nanotherapy specifically targets the STING signalling pathway, which triggers immune activation when abnormal DNA is detected. Traditional STING activators degrade quickly and must be injected directly into the tumour, an invasive process. By contrast, intranasal SNAs reached the brain efficiently in mouse models, activated immune cells around the tumour and avoided harmful effects elsewhere in the body.
The treatment became even more effective when combined with agents that stimulate T cells: tumours were fully eliminated after one or two doses, and long-term immune protection was achieved. Researchers caution that STING activation alone will not cure glioblastoma, but the approach marks a major step toward safer, more potent therapies for brain tumours and other cancers resistant to immunotherapy.
