Text

theses

Despite significant advances in the field of computational protein modeling and design, the prediction of de novo metal-coordination and supramolecular assembly remains a largely unexplored area of bottom-up design. The computational tools and methods outlined in this dissertation are intended to reduce design complexity, promote a generalizable design framework, and lay the groundwork for the development of de novo metal-coordination and supramolecular assembly.

Thirty percent of proteins in Nature, by estimation, contain metal binding sites and these exhibit a diverse array of structural and functional utility. Among them, multi-nuclear metal clusters perform the most exquisite chemistry such as water oxidation, hydrogenation, and nitrogen fixation. In order to harness the catalytic potential of multi-nuclear metal clusters, we propose a general method for the design of multi-nuclear metalloprotein protein precursors, one that exploits the benefits of symmetric coordination and polydentate non-canonical amino acid derivatives. We have developed a computational searching algorithm (SyPRIS) to locate within a library of structurally determined symmetric protein oligomers a constellation of backbone atoms with a geometry compatible with a desired metal cluster. SyPRIS is shown to have 100% accuracy in the prediction of the native metal-binding sites of known symmetrically coordinated metal ions at the interface of oligomeric proteins (C2 and C3). Furthermore, in a crossmatch study of the benchmark structures, more than 1000 novel metal binding sites with native-like scores are predicted, suggesting the utility of SyPRIS for the incorporation of non-native amino acid coordination of a desired complex.

In order to complement the benefits that symmetry offers for reducing design complexity, we sought to expand the palette of available biocompatible non-native amino acid derivatives. A two-step synthesis provides a high-metal affinity bioconjugatable unit, 2,2’-(ethene-1,1-diyl)bis(1-methyl-1H-imidazole) or BMIE. Direct attachement of BMIE occurs by thiol-selective conjugate addition on the surface of a carboxypeptidase G2 variant (S203C). Additionally, we find that BMIE adducts can bind an assortment of divalent metal ions (Co, Ni, Cu, and Zn) in various bi- and tri-dentate tetragonal coordination geometries. Non-BMIE coordinated positions of the copper-bound modified protein display lability in the presence of several counter ligands (H2O/OH, tris, and phenanthroline), which highlights the potential for future catalytic applications. The site-selective modification of proteins for high-affinity metal-binding, combined with the ease of adduct formation and metalation make BMIE an attractive tool to augment multi-nuclear metalloprotein design.

The design of protein-based assembly is a burgeoning field with applications in biomedicine and bioremediation. However, topologies have been limited to integer-dimensions. Additionally, many questions remain with respect to the impact of protein anisotropy, colocalization on catalytic pathways, and the effect of kinetics on self-assembly. In order to address these unexplored questions, we developed a general design method and computational tools for fusion-mediated protein assembly directed by symmetry. To show that our design method could be extended to any set of symmetric proteins, we chose two members of the atrazine degradation pathway that are known to be symmetric: AtzA (D3) and AtzC (D2). The computational algorithm aligns protein oligomers along a shared symmetry axis (C2), and generates an ensemble of protein-protein interfaces by translating and rotating about the shared symmetry axis. In order to reach fractional dimensional topology, we introduced controlled stochasticity along the shared symmetry axis by accepting fusion-domain-added sequences that could adopt multiple energetically favorable members of our protein-protein interface ensemble. We developed a coarse-grained simulation that allowed us to analyze emergent topological patterns based on Boltzmann-weighted probabilities of the stochastic rotations about the shared symmetry axis. A comparison of our simulations to cryo-tomography data of the protein assembly show excellent agreement with the number of binding partners and fractal dimension. We show that co-assembly of enzyme pathway members increases pathway efficiency, but not significantly more than the control “globular” assembly with 10xGSS extended linkers. However, the fractal performed significantly better than the control in sequestration of an IgG antibody, implicating channel porosity size as a key design consideration for future pathway co-assembly. The computational programs and simulations shared in this dissertation should enable the bottom-up design of symmetry-driven self-assembly, and the prediction of emergent topological properties as a result.

Collectively, these studies should enable future efforts aimed at uncovering the fundamental design principles for functional metalloproteins and responsive supramolecular assembly with chemistry and functionality beyond those found in Nature.

ETD_10910

Ph.D.

Includes bibliographical references

doi:10.7282/t3-9snp-c291

ETD doctoral

The author owns the copyright to this work.