An international research team led by a Korean scientist has successfully designed large-scale protein structures that replicate the self-assembly principles of naturally occurring viruses using artificial intelligence (AI). This groundbreaking work was announced by the Ministry of Science and ICT (MSIT) and led by Prof. Sangmin Lee of the Department of Chemical Engineering at Pohang University of Science and Technology (POSTECH), in collaboration with Prof. David Baker of the University of Washington, who is a recipient of the 2024 Nobel Prize in Chemistry.

The research, titled "Design of one-component quasisymmetric protein nanocages," is published in Nature and highlights a design principle where a single protein component can form pentagonal and hexagonal arrangements to self-assemble into virus-like structures.

Protein Nanocages: A New Era in Drug Delivery

Protein nanocages have emerged as a promising material in the biomedical field, representing the next generation of drug delivery systems. These structures are hollow, nanometer-scale entities formed through spontaneous binding of multiple proteins, effectively capable of carrying drugs, genetic materials, and enzymes within their interior space. Furthermore, antigens can be attached to their outer shells to enhance immunogenicity.

Challenges with Traditional Design Approaches

Historically, design technologies have relied on computationally derived "perfect symmetric structures," which has limited the complexity and size of achievable structures utilizing a single protein building block. This limitation can hinder advancements in creating versatile biomedical applications.

Principles of Quasisymmetry in Protein Design

In contrast to conventional approaches, naturally occurring viruses utilize one type of protein, repeated several times, adjusting the protein's position and local environment to construct large structures. This natural principle, known as quasisymmetry, has now been successfully integrated into the design processes of artificial proteins.

The research team determined that expanding viral shell size greatly depends on the angles and curvature between protein building blocks. If proteins are arranged too flatly, the shell fails to close; conversely, when curvature is excessive, the structure oversizes. By engineering this balance precisely, it was possible for a single protein to adaptively occupy both pentagonal and hexagonal formations based on its assembly context.

Innovative Structural Design Using AI

A trimeric unit—a cluster of three proteins—served as the foundation for this design. The team utilized RFdiffusion, an AI-based protein structure generation tool, to develop new connecting frameworks that enable proteins to assemble in various orientations, akin to stacking interlocking building blocks at differing angles to create a dome-shaped shell rather than a flat sheet.

Experimental Validation Through Cryo-Electron Microscopy

The team successfully produced these engineered proteins utilizing E. coli, observing their morphology using cutting-edge cryo-electron microscopy. The results demonstrated the spontaneous assembly of spherical structures ranging from as small as 70 nm to as large as 220 nm, with configurations resembling intricate "nano-soccer balls" for the smallest and three times that for the largest.

Structure Size Appearance Examples
70 nm Intricate "nano-soccer ball" Smallest assembly
220 nm Large shell structure Largest assembly

Significance and Future Prospects

This pioneering study has captured the attention of the scientific community as it does not rely on existing viral proteins but instead, employs a single, entirely AI-generated artificial protein to construct large, virus-like structures. If commercialized, this technology holds the potential to unlock transformative applications in comprehensive areas such as:

  • Vaccine development and antigen presentation systems
  • Advanced gene delivery methods
  • Novel drug delivery platforms

Future research is expected to focus on achieving uniformity in size control using internal scaffold proteins or nucleic acid templates to enhance stability and performance.

“Viruses are the finest example in nature showing that perfect symmetry is not the only path to sophisticated molecular architecture.” – Prof. Sangmin Lee

Key Publications

Two significant studies were published concurrently in Nature:

  • Sangmin Lee et al, Design of one-component quasisymmetric protein nanocages (2026).
  • Shunzhi Wang et al, De novo design of quasisymmetric two-component protein cages (2026).

Prof. Lee's unique contributions include serving as the corresponding author for one paper while holding a co-authorship on another, a remarkable achievement in scientific literature.

In conclusion, this study exemplifies how advancements in artificial intelligence can significantly impact molecular design and nanotechnology, potentially reshaping future medical therapies and delivery systems.

For Further Information

To learn more about this innovative research, you can access the article through the following link: AI-designed protein unlocks virus-like shells that could reshape vaccine and drug delivery.