Like molecular tiles forming a curved shell, designed proteins can now self-assemble into large cages by combining pentagonal and hexagonal patterns, a strategy used by many natural viruses.

Two back-to-back Nature papers report a major advance in computational protein design: researchers have created large, virus-like protein cages from scratch by programming how symmetry breaks during self-assembly.

For decades, quasisymmetry has fascinated structural biologists as one of nature’s most efficient solutions for building large molecular shells. Many viruses use this principle to construct capsids from repeated protein subunits that occupy subtly different local environments, while related forms of symmetry breaking appear in fullerenes, clathrin coats, and other curved molecular assemblies. Now, these studies show that quasisymmetry can be achieved by computational protein design.

By tuning local curvature between protein building blocks, the teams created quasisymmetric cages ranging from tens to hundreds of nanometers in diameter, spanning the size range of many natural viruses. Together, the one- and two-component systems produced cages reaching more than 200 nanometers across, containing hundreds to thousands of protein subunits, and achieving molecular weights from roughly 2 to more than 50 megadaltons.

Make complex structures simple

Viruses face a remarkable architectural challenge: they must build large protective shells while keeping their genomes compact. Many solve this problem through quasisymmetry, in which repeated protein subunits occupy slightly different local environments. This allows viruses to build much larger capsids than would be possible with perfect symmetry alone.

Recreating this strategy by computational design has been difficult. A designed protein cage must be precise enough to assemble reproducibly, but flexible enough for the same or repeated building blocks to adapt to different positions in the final shell. The new studies show that this balance can be modularly engineered by designing local curvature or parameterizing interaction angles that allow pentagonal defects to emerge within otherwise hexagonal lattices, closing the structure into a sphere-like cage.

Spontaneous symmetry breaking

This work builds on a broader effort to move protein design beyond perfectly symmetric nanocages, but it takes a distinct route. Earlier Nature studies used pseudosymmetry, combining similar but non-identical protein components to build larger and more complex cages. The new studies focus on quasisymmetry, the architectural principle used by many viral capsids, where repeated building blocks occupy subtly different local environments. Rather than assigning each position to a different designed part, the researchers programmed local curvature and allowed symmetry breaking to emerge during assembly.

In the one-component study, the team designed protein building blocks whose local geometry lies between two limiting cases: a perfectly symmetric icosahedron and a flat hexagonal lattice. By precisely tuning the interaction-angle parameters, the same protein sequence could form pentagonal and hexagonal local environments within one closed cage. Electron microscopy confirmed designed cages from T = 3 to T = 36, and cryo-EM revealed how symmetry breaking arises at designed subunit interfaces.

“Viruses taught us that perfect symmetry is not the only way to build precise molecular architecture,” said Sangmin Lee, PhD, a co-lead author and former Baker Lab postdoctoral scholar who is now an Assistant Professor at Pohang University of Science and Technology. “A small change in local geometry can have a huge effect on the final assembly—It is like changing the angle between molecular tiles and watching a flat sheet become a dome.”

From static structures to active intracellular probes

The two-component design made the cages modular and functional. In this system, one protein component forms the vertices of the cage, while a second designed component forms the edges. Changing the geometry of the edge component acts like a molecular dial, allowing the researchers to control cage size across a wide range. The cages could also be functionally fused to additional protein domains for cargo recruitment, receptor-mediated uptake, intracellular tracking, and protein relocalization inside cells. In one set of experiments, nanobody-functionalized cages recruited GFP-labeled Cas9. In another, cages bearing designed endocytosis tags were internalized by cells. Fluorescent cages expressed inside mammalian cells also served as genetically encoded rheology probes, reporting how large particles move through the crowded cytoplasm.

“The exciting part of the two-component system is that it turns quasisymmetry into a modular design platform,” said Shunzhi Wang, PhD, a co-lead author and former Baker Lab postdoctoral scholar who is now an Assistant Professor at NYU Grossman School of Medicine. “By changing the geometry of one component, we can tune the size of the cage; by adding functional domains, we can begin to control what the cage carries, where it goes, and how it behaves inside cells.”

Key findings

As detailed in the two Nature studies, the researchers developed new computational strategies for designing quasisymmetric protein cages. Key advances include

• Programmable symmetry breaking: The studies show that local symmetry breaking can be designed into protein assemblies, allowing repeated protein building blocks to form both pentagonal and hexagonal environments within closed cages.
• Curvature-controlled cage size: By tuning local curvature, the researchers programmed cage sizes from tens to hundreds of nanometers. These designed cages reached sizes and molecular weights comparable to natural viral capsids, but were built from computationally designed protein components rather than evolved viral proteins.
• Structural details revealed by electron microscopy: Negative-stain EM, cryo-EM, and cryo-electron tomography confirmed the intended cage architectures and revealed how local structural changes allow closure.
• Intra cellular assembly of functional protein nanomaterials: The two-component cages could be functionalized for cargo loading, receptor-mediated cell uptake, intracellular tracking, and protein recruitment inside cells.

Broader implications

Together, the studies show that computational protein design is moving beyond small, perfectly symmetric objects toward programmable mesoscale materials—structures at the size scale of viruses, vesicles, organelles, and intracellular assemblies.

These cages could become starting points for new delivery vehicles, vaccine platforms, intracellular sensors, programmable scaffolds, or artificial compartments. More broadly, the work suggests that protein design can now engage with principles usually associated with biology and soft matter: curvature, frustration, topological defects, and emergent assembly.

“These papers show that protein design is beginning to capture some of the architectural principles that nature uses to build at very large scales,” said David Baker, PhD, Director of the Institute for Protein Design, “with direct implications for both structural and synthetic biology.”

Led by the UW Institute for Protein Design, NYU Langone Health, and POSTECH, this research included collaborators from the IPD Core R&D Labs; the Holt Lab at NYU Grossman School of Medicine; the DiMaio Lab at UW Biochemistry; and the Veesler Lab at UW Medicine.

This work was supported by the Howard Hughes Medical Institute, The Audacious Project at the Institute for Protein Design, the Burroughs Wellcome Fund, the Defense Threat Reduction Agency, the National Science Foundation, the National Institutes of Health, the National Research Foundation of Korea, the Korean Ministry of Science and ICT, the Gates Foundation, the Human Frontier Science Program, Microsoft, Spark Therapeutics, the Nordstrom Barrier Institute for Protein Design Directors Fund, the Hypothesis Fund, and U.S. Department of Energy-supported synchrotron and structural biology resources. All funders are listed in the papers.