UC San Diego

Metalloproteins utilize a limited set of bioavailable metal ions to perform a wide range of crucial biological functions. As our understanding of the structural and biochemical principles of metalloprotein function has improved, it has become possible to engineer completely new metalloproteins. Such studies provide a test of what we think we know about metalloprotein structure-function relationships, and at times lead to unexpected insights that would not be revealed by exclusively studying natural systems. They also enable the design of metalloproteins with functions beyond what is found in nature, enabling breakthroughs in fields such as chemistry, biotechnology, and medicine. However, designing artificial metalloproteins that function on par with their natural counterparts remains a considerable challenge because of the vast number of possible combinations of amino acids in a protein’s sequence. Indeed, most metalloprotein engineering approaches are limited to the redesign of preexisting proteins. In principle, computational protein design tools allow a more expansive, unbiased search of protein sequence space, but in practice, computational metalloprotein design has been limited by the challenges associated with efficiently searching this space, accurately predicting protein structures, and modeling metal-ligand interactions. To overcome these limitations, we combine computational tools with an understanding of transition metal coordination chemistry and protein assembly to design metalloproteins with new-to-nature structures and functions. We begin by using computational protein design to stabilize the apo state of the hemeprotein cytochrome cb562 with a minimal number of mutations and demonstrate that the redesigned protein retains its ability to assemble into engineered oligomeric structures. We then develop a computational approach that allows us to design new, symmetric protein assemblies around predefined metal coordination geometries, using protein building blocks such as cytochrome cb562. This approach provides a high degree of accuracy in the design of metal primary coordination spheres, which we take advantage of to design Tet4, a C3-symmetric trimer with a coordinatively unsaturated Zn center with abiological catalytic activity. We further show that Tet4 has promiscuous metal binding and can be further engineered to introduce asymmetric coordination environments and a substrate binding pocket, thereby demonstrating that our computationally designed protein assemblies are promising starting points for metalloenzyme engineering efforts. Finally, we use recently developed, machine learning-based protein design tools to de novo design proteins with di-Cu and di-Rh metal centers inspired by synthetic organometallic complexes. These studies advance our ability to accurately design diverse, tunable protein scaffolds around metal centers of interest, potentially expanding the structure-function relationships we can investigate through design and leading to better artificial metalloenzymes.