Visual proteomics is a technique that combines the structural power of cryogenic electron microscopy (cryoEM) with the analytical power of mass spectrometry-based proteomics to uncover the identities of unknown species in complex mixtures. The emerging field of visual proteomics is uniquely positioned to better understand the exoproteome of Clostridium thermocellum (Ct), an anaerobic organism known for its ability to efficiently degrade biomass. With this approach I helped more precisely determine the molecular composition of cellulosomes in the Ct cellular media, and provide information for engineering biomass degradation. Given a newly determined cryoEM reconstruction of an unknown filament from Ct exoproteome at 4.07Å resolution from pooled fractions of concentrated Ct cellular media, I facilitated the application of visual proteomics to determine its identity. To achieve this, proteomic data were systematically assessed by fitting AlphaFold models of proteomic hits into the observed density of the filament structure. Therefore, using single-particle cryo-EM and bottom-up mass spectrometry, I have helped identify and characterize unknown filaments isolated from the growth media of Ct. Guided by the proteomic data, I correctly assigned the identity of these filaments; their identity was further confirmed through biochemical experiments.

Accumulation of proteins such as beta-amyloid has been described as a key observation of some neurodegenerative diseases, including Alzheimer's disease. These proteins may be present extracellularly in sufficient concentrations to impact electrical conductance, which may impact neurotransmission. This study investigates the electron conductivity of bovine serum albumin (BSA) protein mounted on self-assembled monolayers of 1-meta, 9-meta, & 1-para carboxylic acid carboranethiols, using the well-established molecular-junction setup with eutectic gallium-indium (EGaIn) as the moldable electrode contact. Structural changes in the adsorbed protein layer were monitored by x-ray diffraction & contact angle, while changes in the solvated protein were monitored by circular dichroism. Both high negative & high positive voltages were associated with high conductance through the protein preparations. -500mV was associated with -13nA, while +500mV was associated with 22nA on average. At low voltages, the tested protein (BSA) showed consistently diminished conductance compared to the results at high voltage, across both positive & negative current. These results will enhance our understanding of the resistance of protein and the role structural changes play in varying that resistance.

Directed evolution (DE) has emerged as a powerful strategy in enzyme engineering, enabling iterative mutation of proteins with modest initial activity to generate highly active and stereoselective catalysts. However, similar to natural evolution, DE frequently introduces a tradeoff between catalytic activity and thermostability, often yielding enzymes with reduced melting temperatures that limit their practical utility. Addressing this limitation typically requires large-scale sequence modifications rather than single-point mutations. Inverse folding model pipelines, such as GeoEvoBuilder and StabilizeIT, provide a potential solution by preserving active site geometry while enabling extensive sequence redesign to improve protein stability. In addition, these models have demonstrated the ability to enhance enzyme promiscuity, thereby expanding substrate scope. In this study, the DE enzyme HATR-5 was selected due to its significantly reduced melting temperature relative to its parent enzyme and its narrow substrate specificity toward select olefins. We investigate whether inverse folding approaches can be applied post-DE to improve thermostability while maintaining or enhancing catalytic generality. This work aims to evaluate the utility of inverse folding as a complementary strategy to directed evolution for increasing the robustness and applicability of engineered enzymes.

Small proteins (<100kDa) are difficult to visualize using cryo-electron microscopy (cryoEM), limiting 3D analysis of their roles in disease and metabolism. Previous strategies attempted to overcome this by attaching small proteins to larger scaffolds via binder intermediates, but such structures faced low binding affinity and cost issues. This project leverages latest AI protein design tools—RFDiffusion, ProteinMPNN, and AlphaFold—to engineer de novo binders that precisely target small cancer proteins. Binders were then genetically fused to an icosahedral nanoparticle and expressed through E.coli protein purification, creating a stable imaging platform that can improve high resolution cryoEM analysis of small cancer-related proteins. This supports the integration of highly efficient AI-protein design tools to generate binders capable of resolving structural biology challenges, further benefitting therapeutic research.