Developing a dataset of tertiary structural motifs for metalloprotein design

The intricate relationship between metal ions andprotein scaffolds plays a major role in most processes that are fundamental to Life. Metal cofactorsare often essential for the proper folding of polypeptide chains or function of folded proteins.The birth and exponential growth of the PDB over the past 50 years has provided the scientificcommunity with an invaluable resource for the study of such relationships, allowing for exampleto unveil the coordination number and binding propensity of coordinating residues. It has also beenexploited by Machine Learning studies aimed at ab initio identifying and differentiatingstructural and functional metal sites. However, the full potential of protein ligands to bind anygiven metal cofactors and tune their chemistry has not been exhaustively explored by evolution. Tothis aim, the de novo design of metalloproteins by leveraging the wealth of knowledgecontained in the PDB represents a promising strategy for the discovery of new catalysts.Ithas been shown that the protein structural space is highly degenerate and can be recapitulatedwith a finite set of structural units, such as elements of tertiary structure calledTERMs (TERtiaryMotifs).1 Here, we build on this concept and report the developmentof a dataset of highly specialized metal-binding elements: MetalTERMs. Over 100’000 MetalTERMswere identified from sites in which the metal was bound only by protein residuesand water molecules. Subsequently, MetalTERMs were clustered according to their root-mean-square-deviation, the total number of residues and the number of non-contiguous segments.We find that the number of clusters rapidly drops with the increase in complexity of thetertiary arrangement, and that MetalTERMs composed of at most three segments can recapitulateabout 90% of the whole dataset. This would indicate that medium to long-range mutationshave most likely only a marginal effect on the metal coordination sphere, and would thereforecorroborate the well-established adoption of the miniaturization approach for designing metalloproteinsfrom scratch. Our analyses have also enabled us to identify new-to-nature combinationsof most recurring MetalTERMs, which could finally lead to the design of unprecedented catalysts. 1Mackenzie et al, PNAS, 2016, 113 (47), E7438-E7447