Proteins populate structural ensembles. Defining these ensembles and understanding the role of the interconversions between structures is a grand challenge of structural biology. My work addresses that challenge through the development and application of new methods to reveal sparsely populated structures. Quantitative electron-density map interpretation, implemented in Ringer, provides an objective, systematic method to identify previously undiscovered alternate side chain substates that mediate conformational transitions in proteins.

Next, I applied these methods to study the role of the interconversions of an enzyme, the human proline isomerase CypA, between two conformations during its catalytic cycle. Using the dual strategies of ambient-temperature X-ray crystallographic data collection and automated electron-density sampling, I defined the previously undiscovered minor state as a network of alternate side chain conformations. A conservative mutation outside the active site inverts the equilibrium between the substates and causes large, parallel reductions in the conformational interconversion rates and the catalytic rate.

The temperature dependent differences in electron density observed with CypA led me to critically examine the assumption that crystal freezing does not significantly bias protein structure. I found extensive remodeling of the crystal lattice upon freezing. Crystal freezing also leads to improved packing through reduction of small voids and a reduction in protein volume. I used real-space electron density sampling to show that these voids can be transiently populated by alternate conformations in the room temperature ensemble. This work shows how crystal freezing biases our understanding of protein packing and can lead to differences in the spatial distribution of the dynamic features of protein side chains.

These studies highlight the importance of conformational diversity in protein function. By looking beyond the model and into the map, we can find that polysteric regions often populate conformations that resemble the structures populated along reaction or evolutionary trajectories. Thus, understanding polysterism yields insights into where a protein might visit during its reaction cycle and where it has been during its evolution.

The growing prevalence of multidrug resistant pathogens is drastically diminishing our supply of effective antibiotics, in turn increasing the risk associated with bacterial infections. A contributing factor to this issue is how rapidly bacteria adapt to restore the resistant phenotype. Regardless of the strength of the antibiotic concoction, the capability of evolved resistance will always remain a lingering threat. This highlights an important question: How can we design antibiotics that are more resilient to evolved resistance? I have studied this question in the context of group A streptogramin (SA) analogs, antibiotics which bind the peptidyl transferase center (PTC) of the bacterial ribosome and are deactivated by resistance enzymes like VatA, which adds an acetyl group to the macrocycle and prevents ribosomal binding.In Chapter 1, I explore how we can overcome VatA-mediated resistance by leveraging newly developed techniques in the Seiple lab that can create fully synthetic SA analogs with great diversity. Using a combination of cryogenic electron microscopy (Cryo-EM) and molecular biology techniques, we have developed a pipeline to optimize and identify new SA antibiotics that overcome resistance mediated by wild type VatA while not compromising ribosomal inhibition. Our iterative design strategy led us to identify several SA analogs that overcome resistance mediated by wild type VatA and show improved ribosomal binding. Our best performer, analog 47, was shown in a murine thigh model of infection to be 20X more effective at reducing the bacterial load of VatA-expressing S. aureus than flopristin, a comparative semisynthetic SA derivative. Crystallography shows that analog 47 reshuffles the binding site of VatA, which may explain the reduced kcat/Km of VatA at acetylating this analog compared to flopristin. Finally, energy calculations show that binding of 47 in the ribosome is of lower energy than in VatA. Despite these encouraging results, it is important to remember that, given a selective pressure, bacteria will evolve to restore their ability to grow and proliferate without hindrance in the new environment. In our case, this means that treatment of analog 47 on wild type VatA-expressing bacteria would encourage the directed evolution of VatA variants that restore the ability to deactivate the new antibiotic. Therefore, in Chapter 2, I endeavor to explore the effects of single amino acid mutations in VatA on the survival of bacteria grown in the presence of our best SA analogs and their stepwise scaffolds using Deep Mutational Scanning (DMS) and Next-Generation Sequencing (NGS). Our results revealed several strong gain of function mutations in the C-terminal domain, a region distal to the active site. We hypothesize that mutations in this region, which shows poor sequence conservation to other related Vat enzymes, likely contribute to the stability and folding of VatA. Both of these chapters highlight our efforts to improve the durability of SA antibiotics. As demonstrated in Chapter 1, success in improving SA antibiotics’ performance can be further magnified by administering them with their binding partner: group B streptogramins (SB). When used together, SA and SB act synergistically and induce a bactericidal effect. In Chapter 3, I discuss recent work exploring the potential for these two antibiotics to perform click chemistry on the ribosome and form a covalently linked molecule. Throughout my graduate career, I have also implemented Cryo-EM in various collaborations to study the interactions between analogs of other antibiotic classes and the E. coli ribosome. I highlight my contributions to these works in Chapter 4.

The large scale motions that are necessary for the function of many proteins are controlled by atomic level fluctuations. Understanding what causes these fluctuations and how they are translated to the larger motions is essential for understanding the basic mechanism of many biological systems. The ability to control the fluctuations in a predictable way would significantly improve our rational drug and protein design efforts. But, often the important fluctuation that control conformational of proteins are not known. This is in part due to lack of experimental techniques that bridge these scales. Furthermore, computational efforts to meet this need are a work in progress. This thesis describes efforts to understand the mechanistic basis of large-scale motions in different biological systems by developing and applying new tools for the measurement of conformational dynamics. We probed the conformational dynamics of the enzyme CypA using X-ray free electron lasers (XFELs) to validate a known allosteric pathway of the enzyme with this radiation damage free technique. We studied differences in conformation dynamics and allosteric ligand accessibility in αI-domain containing integrins using NMR and room temperature crystallography. By comparing the conformational heterogeneity of two homologous integrins, we have found evidence that the conformational landscape critically influences their ability to bind allosteric modulators. Finally, we visualized the solvent in the influenza M2 proton channel using XFELs in order to understand its mechanism of proton conduction. Taken together, our work highlights the essential role that conformational heterogeneity plays in the function of disparate biological systems, including the atomic-level motions that enable allosteric control of enzyme activity in CypA, the dependence of ligand binding on the conformational heterogeneity of LFA-1, and the solvent conformational heterogeneity that allows proton conduction through the M2 channel. This work is a significant advance towards a mechanistic understanding of the basis of conformational dynamics in the systems described, and will enable future work to manipulate these biological systems to fight disease and improve human health.

Ubiquitin is essential for eukaryotic life and varies in only 3 amino acid positions between yeast and humans. However, recent deep sequencing studies indicate that ubiquitin is highly tolerant to single mutations. We hypothesized that this tolerance would be reduced by chemically induced physiologic perturbations. To test this hypothesis, a class of first year UCSF graduate students employed deep mutational scanning to determine the fitness landscape of all possible single residue mutations in the presence of seventeen different small molecule perturbations. These perturbations uncover 'shared sensitized positions' localized to areas around the hydrophobic patch and the C-terminus. For the important residue Lys 63, we found sensitization in eight conditions and increased tolerance mutation in two conditions. In total, our experiments have uncovered a highly sensitizing condition for every position in ubiquitin except Ser57 and Gln62. Our data show how chemical stresses can reduce buffering effects in the ubiquitin proteasome system. Finally, this study demonstrates the potential of lab-based interdisciplinary graduate curriculum.

Protein folding converts a disordered polymer to a globular structure, reducing many conformational degrees of freedom and incurring a significant conformational entropy penalty. However, residual conformational entropy is retained in a protein’s folded native state. Subtle changes between positions within the native state, mostly from sidechains, alters residual conformational entropy, leading to differences in binding affinity and allosteric communication. While NMR has provided measurements of conformational entropy, these measurements do not provide information on where this entropy is coming from, such as if this is coming from a sidechain moving harmonically or anharmonically. However, we can take advantage of the fact that X-ray crystallography and CryoEM experimental data capture the conformational ensemble allowing us to measure the motion of residues and their atomistic structure. This provides an unparalleled platform to answer how, where, and why conformational entropy. The first chapter of this thesis presents the improvements to the qFit algorithm. This algorithm allows for the automated modeling of multiple conformations per residue across a protein for high resolution X-ray crystallography or cryo-EM. We present algorithm improvements including the ability to run the program on a laptop. This algorithm was the basis for much of the future work of my thesis. The second chapter contains the findings of the relationship between conformational heterogeneity and ligand binding. Using qFit, we identified the changes in conformational heterogeneity between matched bound and unbound high resolution X-ray structures. We identified a reciprocal relationship upon ligand binding where as binding site residues become more rigid, distant residues become more flexible, indicating an entropic compensation. The third chapter contains my future outlook on the questions and techniques to probe conformational entropy mechanism. This chapter includes how to integrate new modeling techniques to understand how different motions of residues lead to differences in conformational entropy.

Proteins are fascinating machines, and the investigation of their structures has led to many insights in biology and drug discovery. Proteins are not static, and their dynamics and their sub-populated states are often relevant to their natural function or regulation. During my PhD, I worked on several projects revolving around the concept that by modeling multiple conformations from structural data, and therefore capturing the flexibility of protein structure, that enhanced insight could be gained from the structural data. First, I was involved in a review that described the state of the field in what was known about the extent of protein dynamics and the state of the art for measuring and modeling the states sampled by the protein. From there, I solved the structure of several ubiquitin mutants that undergone design and selection to alter their function. By collecting high resolution room temperature crystal structures, we revealed the extent of change in conformational heterogeneity across the directed evolution of these mutants. Finally, due to advances in technology and algorithms, we moved on to modeling minor states that were the result of a ligand binding event at less than full occupancy. I was involved in the fragment screening experiment for the protein PTP1B, where we assessed the ligandability of the protein via both cryo and room-temperature data collection.