Protein-protein interactions are ubiquitous throughout nature at many length and time scales—from transient interactions between individual proteins for signaling and electron transfer to the self-assembly over large distances of bacterial S-layer protein coats. Extensive research has been undertaken to attempt to mimic, interrogate, interrupt, or design protein-protein interactions, but natural protein-protein interactions often form as a result of many accumulated weak interactions over large, heterogeneous molecular surfaces, making them challenging to design. As a way to overcome these challenges, we have previously introduced methods of synthesizing protein-protein interactions with minimal surface design through the directional and highly controllable coordination of metal ions on protein surfaces.

The goal of this thesis work has been to expand the scope and functionality of these metal-directed protein and peptide interactions. First, we show that, like proteins, short peptides can be directed to fold and assemble in biologically relevant ways using coordination chemistry, while incorporating additional metal-based functionality on the peptide backbone. We then extend the scale of protein self-assembly to highly ordered, crystalline protein nanotubes with tunable diameters. Finally, we demonstrate the ability to assemble protein-DNA nanomaterials in a manner that, similar to what is observed in nature, relies on the sum of a number of weak interactions to form highly ordered protein-DNA arrays. Overall, we demonstrate the ability to use metal ions to coordinate interactions on scales as small as single protein-protein interactions to as large as micrometer scale arrays.

Here, we have expanded the functionality of these metal-directed protein and peptide interactions. First, we show that peptides can be directed to fold and assemble in biologically relevant ways using coordination chemistry, while incorporating additional metal-based functionality on the peptide backbone. We then extend the scale of assembly to highly ordered, crystalline protein nanotubes with tunable diameters. Finally, we demonstrate the ability to assemble protein-DNA nanomaterials in a manner that, similar to what is observed in nature, relies on the sum of a number of weak interactions to form highly ordered protein-DNA arrays. Overall, we demonstrate the ability to use metal ions to coordinate interactions on scales as small as single protein-protein interactions to as large as micrometer scale arrays.

Proteins, perhaps more than any other class of molecule, make life as we know it possible. The unrivaled champions of functional diversity, the 20+ different amino acids which comprise these polypeptide chains enable them to fold into sequence-specified structures (and structural ensembles) that physically encode their (structure-specified) functions. Yet, the vast majority of cellular proteins operate exclusively as self-assembled multi-subunit complexes, which combinatorially broadens the scope of their (shapes, structures, sizes, and thus) capabilities to encompass roles as molecular-scale catalysts, supramolecular nonequilibrium nanomachines, and cellular-scale motors and intercommunications. This makes proteins highly desirable for the bottom-up construction of new materials: they are perfectly monodisperse nanoscale building blocks, possessing well-defined molecular structure(s) (and thus function(s)) that can be facilely/robustly fine-tuned via routine genetic engineering procedures. As exemplified throughout biology, (synthetic) proteins possess the intrinsic potential to propagate and amplify functional structure (structural function) over multiple scales, promising tailor-made self-assembled macroscale architectures and devices with emergent properties. Towards this goal, this dissertation describes several specific examples of this general principle, whereby new physical phenomena spontaneously manifest within flexible 2D materials self-assembled from an engineered protein.

We first report a theoretical and experimental dissection (followed by exploitation) of the bulk-scale behavior of crystalline lattices constructed from a square-shaped tetrameric aldolase with a designed cysteine at its corners (C98RhuA). Upon oxidative self-assembly, the topology and flexibility of the intermolecular disulfide bonds enable the resulting architecture to undergo “breathing” motions that allow it to switch between porous and close-packed conformations without losing crystallinity. We characterized the free energy landscape associated with these dynamic motions, validated its accuracy against experiment, and determined that it is dominated by solvent reorganization entropy. We then exploited the lack of native intermolecular interactions to rationally engineer this landscape through the introduction of negatively charged glutamate residues at prescribed locations, which was then analogously characterized both experimentally and computationally. We further determined that this repulsion could be toggled via the chelation of Ca2+ ions, enabling controllable access to discrete conformational states through environmental conditions. This variant was subsequently utilized as a selective “gatekeeper” coating for a highly specific and sensitive chemical sensor for HCN.

Simultaneously, we characterized the self-assembly of C98RhuA at mineral interfaces via in situ AFM. We found that the use of a charged surface as a template for crystallization enabled the formation of three new crystal morphologies. We validated the observed crystal patterns against atomic structural models and used free-energy calculations to rationalize the formation of each crystal morphology. We also determined that C98RhuA possesses a sizeable macrodipole moment (~1200 D), which calculations indicated explained the preference for antiparallel packing of protein units (p4212 symmetry); this is the first time dipole-dipole interactions have been shown to control protein self-assembly. Finally, the alignment of dipole moments in surface-grown crystals was predicted them to be “electrets”. Indeed, MD simulations of both p4 symmetry (parallel packing) and p4212 protein crystals revealed that the former exhibits a conformation-dependent membrane voltage of ≤−100 mV (in excellent agreement with analytical calculations), thus delineating the connection between the individual building block and macroscale properties when the patterning of component orientations is controlled. Finally, preliminary computational investigations into the epitaxial growth of C98RhuA crystals on these surfaces are described.