Multivalent interaction is often used in molecular design and leads to engineered multivalent ligands with increased binding avidities toward target molecules. The resulting binding avidity relies critically on the rigid scaffold that joins multiple ligands as the scaffold controls the relative spatial positions and orientations toward target molecules. Currently, no general design rules exist to construct a simple and rigid DNA scaffold for properly joining multiple ligands. Herein, we report a crystal structure-guided strategy for the rational design of a rigid bivalent aptamer with precise control over spatial separation and orientation. Such a pre-organization allows the two aptamer moieties simultaneously to bind to the target protein at their native conformations. The bivalent aptamer binding has been extensively characterized, and an enhanced binding has been clearly observed. This strategy, we believe, could potentially be generally applicable to design multivalent aptamers.

As a result of evolutionary selection, the subunits of naturally occurring protein assemblies often fit together with substantial shape complementarity to generate architectures optimal for function in a manner not achievable by current design approaches. We describe a “top-down” reinforcement learning–based design approach that solves this problem using Monte Carlo tree search to sample protein conformers in the context of an overall architecture and specified functional constraints. Cryo–electron microscopy structures of the designed disk-shaped nanopores and ultracompact icosahedra are very close to the computational models. The icosohedra enable very-high-density display of immunogens and signaling molecules, which potentiates vaccine response and angiogenesis induction. Our approach enables the top-down design of complex protein nanomaterials with desired system properties and demonstrates the power of reinforcement learning in protein design.

Protein crystallization plays a central role in structural biology, with broad impact in pharmaceutical formulation, drug delivery, biosensing, and biocatalysis. Despite this importance, the process of protein crystallization remains poorly understood and highly empirical, with largely unpredictable crystal contacts, lattice packing arrangements, and space group preferences, and the programming of protein crystallization through precisely engineered sidechain-sidechain interactions across multiple protein-protein interfaces is an outstanding challenge. Here we develop a general computational approach to designing three-dimensional(3D) protein crystals with pre-specified lattice architectures at atomic accuracy that hierarchically constrains the overall degree of freedoms (DOFs) of the system. We use the approach to design three pairs of oligomers that can be individually purified, and upon mixing, spontaneously self-assemble into large 3D crystals (>100 micrometers). Small-angle X-ray scattering and X-ray crystallography show these crystals are nearly identical to the computational design models, with the design target F4132 and I432 space groups and closely corresponding overall architectures and protein-protein interfaces. The crystal unit cell dimensions can be systematically redesigned while retaining space group symmetry and overall architecture, and the crystals are both extremely porous and highly stable, enabling the robust scaffolding of inorganic nanoparticle arrays. Our approach thus enables the computational design of protein crystals with high accuracy, and since both structure and assembly are encoded in the primary sequence, provides a powerful new platform for biological material engineering.