Foldit is a citizen science game developed at the University of Washington that allows people to predict and design protein structures for use in scientific research. The game challenges ordinary people to map the complex energy landscape of a protein sequence to find the lowest energy structure as determined by the protein modeling software Rosetta. The lowest energy structure is the one most likely to fold. While Rosetta’s software uses random sampling to predict a protein’s structure from its sequence, Foldit takes advantage of people’s intuition about things like avoiding steric clashes and large cavities to produce structures that can rival or exceed the accuracy of what Rosetta’s software can produce. However, there are still some flaws in Foldit that can cause a large fraction of player designs to not fold correctly. One of the biggest issues concerns an excess of buried unsatisfied polar atoms in player designs caused by imperfect modelling of the electrical potential energy of these atoms in the game. These atoms can prevent a protein from folding by eliminating the energy gap between folded and unfolded states. We have introduced a score penalty which penalizes designs with buried polar atoms. We are using existing scientific software to determine these atoms, but we have had to modify the relevant code to make it fast enough to be playable while retaining accuracy. We have also tested our calculations against known ways of calculating these atoms, and against previous Foldit puzzles to see if adding the penalty improves designs. We hope to even test this filter on some recent Coronavirus binder puzzles to get better solutions. An increase in design quality would mean that such a score penalty is warranted and it would allow for Foldit structures to be more useful across a range of protein design applications.

When all goes according to plan, newly synthesized proteins within cells fold down an energetic funnel into a functional, minimal energy configuration. If a protein does not fold properly, it is both energetically unfavorable and nonfunctional, often with hydrophobic parts exposed to the aqueous environment. This creates the potential for misfolded proteins to form insoluble aggregates, which are toxic to cells. These aggregates crowd the cellular environment and impair cellular functions, which on a single cell scale, leads to cell death, and on a larger organism scale, leads to diseases like Alzheimer's, Parkinson's, and Huntington's. To deal with this problem, cells have protein quality control (PQC) systems. PQC is composed of two classes of enzymes: chaperones that help proteins to fold properly and ubiquitin-protein ligases that tag misfolded proteins with ubiquitin, leading to degradation in the proteasome. Previous studies concluded that chaperones are required for protein degradation. In this study, we investigated yeast ubiquitin-protein ligase San1, which often requires chaperones, but can recognize substrates independently. The ubiquitin-protein ligase San1 recognizes patches of hydrophobicity on misfolded proteins and is able to accurately tag them for destruction in the proteasome. Primarily through the use of cycloheximide-chase degradation assays and fluorescent microscopy, I worked within our team and found that Hsp70 chaperone dependence is variable along a spectrum of independent to dependent. By studying the interactions of the folding and degradation enzymes, our lab is gaining new insights into the coordination of PQC pathways

Small Heat Shock Proteins (sHSPs) play crucial roles in protein homeostasis, the state of maintaining steady internal cellular conditions, despite changes to the cellular environment. As the “molecular life rafts” of the cell, sHSPs target partially misfolded proteins during stress and without consuming energy, prevent toxic aggregation. Mutations or malfunctions of the sHSP, HSPB5 (B5), in humans are associated with neurodegenerative diseases such as Alzheimer's and Alexander's disease, as well as cancers, myopathies, and cataracts. As one of the ten sHSPs encoded in the human genome, B5 contains structural elements (the building block) common to all sHSPs: a highly conserved alpha-crystallin domain (ACD), that is flanked by variable, less conserved, N- and C-terminal regions (NTR and CTR, respectively). Previous studies determined that interactions between building blocks occur between a three amino acid region known as the I-X-I motif (a “knob”) in the CTR, and a hydrophobic groove (“hole”) in the neighboring block, similar to how two pieces of Lego come together to build a larger structure. IXI motifs are believed to be an important structural element, and interestingly, in B5, there is an additional IXI motif in the NTR, but its role is unknown. I hypothesize that IXI motifs compete for binding into the hydrophobic groove, indicating the “knob” into “hole” interaction is loose, where the other “knob” can bind the “hole” when one leaves. Other binding partners of sHSPs such as Bag3 utilize the IXI motif to bind and regulate sHSPs. My goal is to compare binding of three IXI peptides (NTR-IXI-B5, NTR-IXI-B8, and Bag3) to the groove through NMR peptide titrations to determine which peptide (B5 or B8) in competition with Bag3 has the strongest binding affinity to the groove. These results will assist in demystifying the function of the NTR and in directing future sHSP studies.

 Host restriction factors are components of the innate immune system that play a vital role in inhibiting viral infection. One such example is a family of antiviral proteins, APOBEC3s, that inhibit retroviruses such as HIV by hypermutating the genome through cytidine deaminase activity. There are seven members of the APOBEC3 (A3) family found in primates, A3A-A3H. Each of these APOBEC3s vary in their protein expression levels and antiviral activity Antiviral activity of these proteins depends on expression levels and ability to be packaged in budding virions. A3C, in particular, is of interest because it is highly expressed and packaged but lacks potent antiviral activity when compared to its A3 counterparts. Moreover, when we examined A3C expression across primate evolution, we found that many primates encoded only unstable versions of A3C. Preliminary data, on the other hand, suggests an alternative role of A3C in modulating the effects of the more potent A3 proteins, A3D, A3F, and A3G. Coimmunoprecipitation assays suggest that A3C interacts with other A3 proteins, and subsequently decreases the packaging of the potent A3s into budding virions thus inhibiting their antiviral activity. Future experiments include removing endogenous A3C from cells to determine if the antiviral activity of other A3’s increases in the absence of any A3C. These data suggest that A3C has evolved to fine-tune the amount of antiviral activity of other A3 family members, thus allowing for better control of these potential mutagenic proteins.

The link between protein sequence and structure is not always apparent. The dogma is that sequence determines structure, but it is not clear how very different sequences can give rise to the same structure. Here, we employ high temperature molecular dynamics unfolding simulations to probe the pathways and specific interactions that direct the folding and unfolding of the SH3 domain, a family of small proteins consisting of two β-sheets arranged to form a barrel. SH3 domain proteins are involved in various functions including protein binding, cell signaling, and nucleic acid modification. The SH3 metafold consists of 753 proteins with the same structure but varied sequence and function. To investigate the relationship between sequence and structure, we selected 17 SH3 proteins with an average pairwise sequence identity of only 27%. Six unfolding simulations were performed for each protein and unfolding transition states were determined, revealing two unfolding/folding pathways. Transition states were also expressed as mathematical graphs of contacts between chemical groups, and three positions in the transition state structure were consistently more connected to the rest of the graph than other nearby positions. These positions represent a folding hub connecting different portions of the structure in the transition state. Analysis of the multiple sequence alignment and covariation also highlighted positions with high conservation due to packing constraints and long-range contacts. This study demonstrates that the SH3 domain can fold through two distinct pathways, but certain folding/unfolding characteristics are conserved independent of sequence and unfolding pathway. By identifying similar interactions, we demonstrate how different sequences can have the same influence on folding pathway and final structure.

Cannabinoids, the main constituents of Cannabis, are a class of highly abused compounds of which, Δ-9-tetrahydrocannabinol (THC) is the primary psychoactive molecule. Despite the wide use of THC, its metabolism in humans is still in need of greater understanding. THC is metabolized to 11-OH-THC and sequentially to 11-COOH-THC by cytochrome P450 enzymes (CYPs) 2C9 and 2C19. 11-COOH-THC is then further conjugated to form 11-COOH-THC-glucuronide by UDP-glucuronosyltransferases (UGTs) 1A1 and 1A3. THC and its subsequent metabolites have been shown to bind to liver-type fatty acid binding protein (FABP1). FABPs are intracellular lipid binding proteins (iLBPs) that regulate the homeostasis of their endogenous ligands by solubilizing these hydrophobic compounds in the cytosol. Knockout of FABP1 in mouse hepatocytes was shown to decrease the formation and clearance of 11-OH-THC while the metabolism of 11-COOH-THC appeared to be unaffected. The goal of the current investigation is to translate these results into the human liver. Previously, our lab expressed and purified human FABPs to test their effect on THC metabolism in incubation assays with human liver microsomes (HLMs) and recombinant enzymes. After initiating the THC reaction with the CYP cofactor, NADPH, the 11-OH-THC product was extracted and quantified using LC-MS/MS. We found that both FABP and albumin changed the metabolic rate of THC in an enzyme specific manner. Because 11-OH-THC formation was altered in the presence of FABPs compared to the HSA control, we extend this method to continue our investigation with 11-COOH-THC metabolism. Considering that UGTs are on the luminal rather than the cytosolic side of the endoplasmic reticulum and that 11-COOH-THC has greater water solubility, we expect to observe enzyme and substrate specific effects of FABP. 11-COOH-THC and 11-COOH-THC-glucuronide are biological markers of THC metabolism so understanding this metabolic pathway is important for developing better methods of characterizing THC use in humans.

Excessive reactive oxygen species (ROS) production in cells induces oxidative stress causing a variety of cell damage that relates to the pathophysiology of more than 200 types of diseases, including Alzheimer's disease and cardiovascular lesion. Many molecules in the ROS family also act as secondary messengers in a series of intracellular signaling mechanisms, such as the neural regulation of opioid addiction. Time-resolved measurements of ROS dynamics is critical to the development of therapeutics for related diseases. The genetically encoded fluorescence indicator (GEFI) can be used as an efficient tool for real-time ROS detection. However, the current ROS fluorescent probes have many downsides hindering their performance, such as the low signal-to-noise ratio, slow responding kinetics, low sensitivity, in vivo incompatibility, and limited potential to target subcellular structures. Thus, we engineer a novel fluorescent protein probe for ROS detection based on the structure-guided protein designing principles and high-throughput variant library screening technique. We first used protein structures to guide mutations for improved allosteric coupling between the ROS binding domain and the fluorescent reporter protein, as well as for strengthening the network of hydrogen bonds surrounding the ROS-binding residues. We also implement a novel engineering platform to optimize the efficiency of screening a large number of mutated variants. Thus far, we have generated multiple sensor variants that demonstrate better fluorescent performance and faster response times to direct ROS stimulation compared to the existing ROS protein probes. Our aim is to monitor the real-time ROS production in the signaling mechanism of behaving mice and the ROS-induced oxidative stress in the pathological neuron and cardiomyocyte cell lines to validate the in vivo and in vitro performance of our sensor.

RNA-based therapeutics have attracted significant interest as a promising alternative to traditional cancer therapy methods. Coupled with other technologies such as the application of nanocarriers, RNA-based treatments have shown the potential to regulate gene expression in tumor cells with high efficacy and reduced safety risks. Despite these advantages, delivering sufficient amounts of therapeutic cargo while escaping from the proteolytic environment of the endosome has remained a long-standing challenge. This project focuses on addressing this issue by evolving the I53-50-v4 (V4), a self-assembling synthetic protein nanoparticle engineered to encapsulate its own mRNA, to increase RNA packaging and prevent degradation upon endocytosis. With previous data suggesting a decrease in mRNA encapsulation with an increase in mRNA length, additional designs were generated to further validate this finding. Furthermore, different variations of de novo pH-responsive trimers for endosomal escape were fused to the exterior of the V4 and tested for RNA packaging. Using computational tools, algorithms were also developed to model the porosity of the nanoparticle and to label residues for cationic mutations. Protein nanoparticles were expressed and purified via Immobilized Metal Affinity Chromatography (IMAC) and SEC (Size Exclusion Chromatography). Constructs were then analyzed in their mRNA encapsulation levels through RT-qPCR. Success of this project would demonstrate an increase in encapsulation levels with a decrease in mRNA length. As the phenotype of each nanoparticle design is spatially linked to its genotype, the top performing designs can also be selected and further evolved using both experimental and computational approaches. This will allow the development of a novel assay for screening desired proteins and narrow the gap towards achieving an efficacious RNA-delivery system.