Nanoparticles Can Target Protein Cavities - But Surface Chemistry Matters as Much as Size

Much of modern drug discovery works by finding small molecules that fit into pockets on protein surfaces, blocking or activating the protein's function. The strategy works well when the target pocket is deep and geometrically defined - the right shape to grip a small molecule tightly. It fails at the interfaces between proteins, where surfaces tend to be broad, shallow, and difficult to occupy with molecules small enough to be taken as drugs. These sites have been called "undruggable," and they include some of the most important protein interactions in biology, including many that drive cancer and infectious disease.

Nanoparticles - structures with at least one dimension between 1 and 100 nanometers - are large enough to contact these broad interfacial surfaces and potentially disrupt protein-protein interactions. But designing nanoparticles that do this selectively, binding one cavity rather than any protein surface they encounter, requires understanding what determines their binding preferences.

A study published in the Journal of the American Chemical Society, led by Professor LI Yang at the Shenzhen Institute of Advanced Technology of the Chinese Academy of Sciences, provides that understanding through a carefully designed comparison. The researchers used the SARS-CoV-2 spike protein trimer - a structurally well-characterized target - to directly compare two nanoparticle types with similar sizes but completely different surface chemistries.

Two Nanoparticles, Two Binding Sites

The researchers compared cerium dioxide nanoparticles (CeO2NPs) and gold nanoparticles (AuNPs). Both have been tested in various biological contexts, but they differ fundamentally in their surface chemistry. The CeO2NPs surface is decorated with cerium atoms that can coordinate - form bonds - with specific amino acid functional groups. The AuNPs surface carries charge distributions that favor electrostatic interactions.

Both nanoparticle types bound to the spike protein trimer and showed antiviral effects in laboratory tests. But they bound to completely different sites. CeO2NPs targeted the central cavity of the spike trimer, a region enriched in aspartic acid residues. The cerium atoms on the nanoparticle surface form coordination bonds with the carboxyl groups of aspartic acid, anchoring the nanoparticle in that specific pocket and blocking the spike protein from recognizing the ACE2 receptor - the human protein it uses to enter cells.

AuNPs bound elsewhere. Their preferred site was a set of lateral cavities near the S1/S2 cleavage region of the spike protein, an area enriched in arginine residues. The binding there occurs through electrostatic attraction and hydrogen bonding rather than coordination chemistry, and it interferes with how host proteases activate the spike protein for viral entry - a different step in the infection process than the one targeted by CeO2NPs.

Size Is Necessary but Not Sufficient

The key finding is that size alone cannot explain which cavity a nanoparticle binds. Both nanoparticle types were comparable in physical dimensions, yet they engaged completely different sites. The selectivity emerged from the match - or mismatch - between the nanoparticle's surface chemistry and the local amino acid environment within each cavity.

"Precise cavity recognition therefore required both geometric compatibility and surface chemical interactions," the authors write. This insight reframes the design challenge for therapeutic nanoparticles. Researchers cannot simply shrink a nanoparticle to fit a target cavity and expect selective binding. They also need to engineer the nanoparticle surface to chemically complement the specific amino acid residues lining that cavity.

A Framework for Targeting What Was Once Off-Limits

The practical implication is a more principled approach to designing nanoparticles that can reach protein-protein interfaces. Given a target cavity's amino acid composition - information available from protein crystal structures and computational modeling - researchers can now reason systematically about what nanoparticle surface chemistry is most likely to achieve selective binding. The SARS-CoV-2 spike protein served as a model system; the same logic applies to other viral proteins, cancer-related protein complexes, and other targets where small molecules have repeatedly failed.