Protein structure is critical for understanding biological processes. Accordingly, proteins with similar structures are expected to carry out similar functions, leading to their classification based on structure. However, structure-based protein classifications can be computationally expensive and require the determination or prediction of structures for comparison. I evaluated whether this process could be accelerated by using the programs Foldseek and ProtProfileMD to generate linear 3Di sequences encoding structural information from protein structures (Foldseek) or amino acid sequences alone (ProtProfileMD). I used maximum likelihood phylogenetic analyses to determine the sensitivity of Foldseek and ProtProfileMD in detecting homology for a family of pathogenic Mammarenavirus glycoproteins. I found that Foldseek and ProtProfileMD are both highly sensitive to remote homology and can recreate the phylogenetic tree of the Mammarenavirus genus using Protein Data Bank structures (Foldseek) or amino acid sequences alone (ProtProfileMD). I computationally generated variants of one pathogenic Mammarenavirus species, Machupo, and used maximum likelihood phylogenetic analyzes to find differences in amino acid and Foldseek 3Di sequences to generate predictions on sequence features that may contribute to glycoprotein internalization and virus infectivity. I am now working to evaluate the function of predicted Machupo variants in order to assess structural determinants of receptor binding and anti-viral neutralization.

Strained cyclic intermediates, such as arynes and strained cyclic alkynes, serve as valuable building blocks for the synthesis of complex molecules. Nitrogen-containing versions of these intermediates are particularly desirable, as nitrogen-containing heterocycles are prominent in bioactive natural products and pharmaceutical agents, but their synthesis can pose many difficulties. This presentation explores the generation and trapping of seven-membered azacyclic allenes. Synthetic access to these intermediates will be described along with results from representative [4+2], (3+2), and [2+2] cycloaddition trapping reactions. Structural differences between six- and seven-membered azacyclic allenes, as well as the origins of regio- and stereoselectivities of reactions will also be discussed.

This abstract has been withheld from publication.

With escalating atmospheric CO2 levels and continued reliance on fossil fuels, there is an increased demand for sustainable methods for energy conversion and carbon capture. Nature has many powerful examples of catalysts that perform these processes efficiently under mild conditions. Metalloenzymes like [NiFe]-hydrogenases (H2ases) catalyze the reversible hydrogen evolution reaction (HER) at negligible overpotentials, while carbon monoxide dehydrogenase and acetyl-coenzyme A synthase (ACS) play key roles in the Wood-Ljundahl pathway, one of nature’s primary CO2 fixation routes, where CO2 is converted to acetyl-CoA. Despite their exceptional reactivity, these enzymes are large, complex, and highly oxygen-sensitive, making them difficult to study and apply practically. To address this, the Shafaat lab has developed nickel-substituted rubredoxin (NiRd) as a robust protein-based H2ase mimic. Prior work has shown a specific primary sphere cysteine-to-serine mutation in NiRd turns off HER activity, yet yields a similar electron paramagnetic resonance spectrum, or a similar electronic structure, as ACS once in its reduced form. This suggests the possibility of switching functionality from H2ase-like to ACS-like activity. Probing substrate binding and reactivity across a range of small molecules using spectroscopic methods can provide insight into how the local protein environment controls metal reactivity. Understanding that relationship can inform enzyme-inspired catalysts for sustainable fuel generation and carbon capture.

This abstract has been withheld from publication.