Abstract:  The prediction of protein crystal packing arrangements and the de novo design of new crystal forms are notoriously challenging problems. The general form of both problems will require new algorithms to efficiently sample the possible crystalline configurations of proteins. To efficiently enumerate candidate crystal forms, we have adapted grid-based fast Fourier transform strategies first developed in the context of protein docking algorithms. To maximize performance of the grid-based, shape-complementarity scoring scheme, we have optimized parameters for the recognition of authentic protein crystal packing arrangements. At the same time, we have also used computational protein design approaches to re-engineer existing protein crystals. In so doing, we are engineering a new class of molecular scaffold to support programmed assembly of functional domains within precisely defined 3-D materials.

As an example application, we are engineering host-guest crystals for catalytic applications. When Nature needs to maximize enzymatic productivity, or to sequester toxic intermediates, it packages hundreds of copies of multiple enzymes into large compartments called biological microcompartments (BMCs). Key examples of such proteinaceous organelles include carboxysomes or metabolosomes.  We aim to recapitulate the functionality of these materials by engineering highly porous protein crystal scaffolds capable of adsorbing and organizing multiple enzymes. In essence, the proposed research will convert individual protein crystals into massively parallel assembly chassis capable of organizing billions of guest enzymes.