AI Designs Virus-Like Protein Shells for Drug and Vaccine Delivery
An international research collaboration led by Professor Sangmin Lee of Pohang University of Science and Technology and Professor David Baker of the University of Washington has successfully engineered artificial protein nanocages that replicate the complex self-assembly principles of natural viruses. Published recently in Nature, the breakthrough introduces a novel design framework leveraging artificial intelligence to overcome longstanding limitations in single-component protein architecture. Traditional protein nanocage development has relied on computationally derived perfect symmetry, a constraint that severely restricts structural scale and complexity. Natural viruses, by contrast, achieve massive, hollow shells through quasisymmetry, a mechanism wherein identical proteins subtly adjust their local positioning to form continuous, curved surfaces. The research team applied this biological blueprint by identifying precise angular and curvature relationships between protein building blocks. Using the AI-based structure generation tool RFdiffusion, the researchers designed a trimeric unit capable of simultaneously occupying pentagonal and hexagonal coordination environments depending on its location within the growing assembly. This geometric flexibility allows the proteins to interlock at varying orientations, naturally curving into large spherical caps rather than flat sheets. Experimental validation through cryo-electron microscopy confirmed that the AI-designed proteins spontaneously self-assembled into hollow structures ranging from 70 to 220 nanometers in diameter. The smaller assemblies formed intricate spherical geometries, while the larger variants demonstrated successful scalability by more than threefold. Notably, the system required only a single, entirely synthetic protein component, eliminating the need to repurpose existing viral proteins. This advancement addresses a critical bottleneck in nanomedicine, positioning protein nanocages as a premier platform for next-generation therapeutic delivery. Their hollow interior can stably encapsulate pharmaceuticals, enzymes, and genetic payloads, while their external surface remains modifiable for targeted antigen presentation. Industry observers note that successful commercialization could revolutionize precision medicine, enabling highly controlled vaccine formulations and systemic drug transport with minimal off-target effects. The Korean Ministry of Science and ICT highlighted the simultaneous publication of two Nature papers as a rare academic milestone, with Professor Lee serving as corresponding author on the single-component study and co-author on a companion paper detailing two-component quasisymmetric cages. Moving forward, the research team plans to refine size uniformity by integrating internal scaffold proteins and nucleic acid templates. By demonstrating that subtle geometric tuning can dictate macroscopic molecular architecture, the study establishes a scalable, AI-driven pathway for engineering advanced biomaterials with transformative clinical applications.
