Aberrant expression of DUX4 in human muscle causes Facioscapulohumeral Muscular dystrophy (FSHD) which affects about 1 in 8,333 individuals worldwide, yet the mechanism by which DUX4 causes muscle wasting is unknown. The DUX4 gene is unique to humans and transgenic animal models have largely failed to exhibit the dystrophic phenotype and endogenous molecular changes characteristic of FSHD. Therefore, studies of DUX4 signaling in human muscle have been limited to patient biopsies and cultures of myogenic cells in vitro. There are no therapies that target the mechanism of disease, thus there is a pressing need for studies of DUX4 in mature human muscle. We have developed a method to xenograft human-derived muscle precursor cells, isolated from patients with FSHD and controls, into the tibialis anterior of immune-deficient mice to form xenografts of pure human muscle tissue. The FSHD xenografts are robust, mature, and well organized. Human myofibers are innervated and associate with human satellite cells, making the xenografts structurally comparable to intact human skeletal muscle. The FSHD but not control xenografts express DUX4 and DUX4 gene targets and have 4q35 methylation profiles typical of FSHD. The FSHD grafts also display a novel biomarker of FSHD, SLC34A2, measurable by immunofluorescence, which will provide a quantifiable metric for future therapeutic studies. We have described several modifications to the engraftment strategy that can be used to answer complex mechanistic and functional questions regarding the FSHD pathophysiology. Finally, we report promising data from a collaborative study with Fulcrum Therapeutics in which we were able to repress the DUX4-signaling pathway by administering a targeted small molecule therapy to FSHD grafts. Ours is the first scalable and reproducible in vivo model of FSHD muscle. Future studies will include those aimed at continuing to delineate the molecular pathogenesis of FSHD, performing functional tests to define the pathophysiology of FSHD, and testing new and exciting therapeutic strategies aimed at reducing the DUX4 program in human FSHD xenografts.

Enzymes are biological catalysts that enable life by accelerating specific reactions. As new infectious diseases are discovered, the enzymes produced by these organisms require characterization. In some cases, they can be inhibited to prevent disease; in other cases, they can be coopted or altered to serve as diagnostic or therapeutic tools. In this work, we explore the mechanisms of several enzymes involved in infection and immunity. In the first major theme, we characterize the mechanisms of fosfomycin resistance proteins from Escherichia coli (FosA3) and Klebsiella pneumoniae (FosAKP), which degrade the antibiotic fosfomycin. Although the active sites of FosA enzymes are well conserved, differences in activity between enzymes may be due to the presence of an allosteric site at the dimer interface of these enzymes. We solved crystal structures with a novel small molecule FosA inhibitor, termed ANY1, which binds with high affinity to the active site and acts as a competitive inhibitor of fosfomycin binding. By inhibiting FosA, ANY1 potentiates the effects of fosfomycin in several priority Gram-negative pathogens, and may serve as a suitable lead candidate for structure-guided drug design and pre-clinical development. In the second major theme, we describe the mechanisms of enzymes with activity on the carbohydrates of IgG antibodies. EndoS and EndoS2 are enzymes produced by Streptococcus pyogenes, which remove the conserved glycan on Asn297 of IgG to evade the host immune system. We discovered that these enzymes share a conserved mechanism of substrate recognition, binding primarily to the core and α(1,3) antenna of the IgG glycan. EndoS2 recognizes a more diverse set of glycans through differences in the both the active site and a coevolved carbohydrate binding module. We then describe the structure and function of AlfC, an α(1,6)-specific fucosidase from Lactobacillus casei with activity on the core fucose of antibodies. We identified D200 and D242 as the most likely catalytic residues, and provide a structural basis for the remarkable specificity of this enzyme and transfucosidase mutants. Finally, we describe a novel application for these carbohydrate-active enzymes in the directed evolution of antibodies based on yeast display.

Background. Memory consolidation occurs during sleep, providing an opportunity to enhance upper extremity (UE) function in people with residual impairments post-stroke. Targeted memory reactivation (TMR) has been used to enhance this process, which involves pairing auditory cues with task performance and subsequent cue replay during sleep. TMR application during sleep leads to increased task-related brain network connectivity and behavioral performance in healthy young adults. Yet it remains unknown whether TMR can enhance sensorimotor performance in individuals with stroke. Methods. Healthy younger and older adults and individuals with chronic stroke were trained on a non-dominant (or non-paretic) UE throwing task before a period of waking or sleeping consolidation, with some receiving TMR throughout the consolidation period. Study 1 involved the use of TMR throughout the first two slow wave sleep periods over a full night of sleep with young adults. Studies 2, 3, and 4 investigated whether TMR throughout a one-hour nap was sufficient to influence sensorimotor performance in young adults, older adults, and people with a history of stroke, respectively. Results. All studies found that TMR application during sleep enhanced sensorimotor performance. In addition, TMR during wake did not influence sensorimotor performance (Studies 1 and 2), and enhanced performance of a cognitive aspect of the trained task (Study 2). Additional generalization and transfer tests helped to support the hypothesis that TMR enhanced a task-specific motor program, as improvements were seen within the trained task but not un-trained, but similar tasks. Lastly, sleep alone appears to stabilize sensorimotor performance variability, but this process demonstrates an age-related decline. Conclusion. This dissertation has shown that the use of TMR during sleep is a useful method for enhancing sensorimotor performance in healthy young and old adults, as well as individuals with a history of stroke. Future research may lead to an adjunct to traditional physical rehabilitation protocols.

In the study of learning and decision-making in animals and humans, the field of Reinforcement Learning (RL) offers powerful ideas and tools for exploring the control mechanisms that underlie behavior. In this dissertation, we use RL to examine the questions of (i) how rats represent information about a complex, changing, task; (ii) what are the relevant variables underlying their decision-making processes; and (iii) whether those variables are encoded in the firing rates of neurons in the orbitofrontal cortex (OFC). We addressed these questions by making inquiries across three levels of understanding: computational theory, algorithmic representation, and physical implementation. Within this tri-level framework, we hypothesize that the subjects are engaged in a form of approximately optimal adaptive control. This involves their tracking critical, task-relevant, features of their environment as these features change, and then making appropriate choices accordingly. Two classes of RL algorithms were constructed, embodying different structural assumptions. One class of so-called return-based algorithms is based on elaborations of a standard Q-learning algorithm. The other, novel, class of income-based algorithms is based on a rather weaker notion of action-outcome contingency. Both classes of algorithm were parametrized and other factors were included such as perseveration. We t the algorithms to behavioral data from subjects using complexity-controlled empirical Bayesian inference. Internal variables from our algorithms were then used to predict neural ring rates of OFC neurons that were recorded as subjects performed the task. Linear regression, randomization testing, and false discovery rate analysis were used to determine statistically significant correlations between the predictors and neural activity. We found that income-class algorithms augmented with perseveration offered the best predictions of behavior. For the least restrictive statistical test (linear regression, p < 0.05), as many as 24% of the neurons were significantly correlated with variables associated with the best-fitting algorithm. By contrast, for our most restrictive test (randomized false discovery rate < 0.05), only 3% of the neurons passed as significant for one or more of our predictor variables. Other forms of neuronal dependence were apparent, including neurons that appeared to change their computational function dynamically depending on the state of the task.

Molecular dynamics (MD) simulations have been widely applied to study biomolecular systems. While MD simulations have been used in a variety of applications, the accuracy of the results depends strongly on the force field (FF) used. The commonly used FFs are additive or non-polarizable FFs. However, one limitation of the additive FFs is the lack of electronic polarizability to treat molecules. To overcome this, an attractive approach is to introduce the explicit treatment of electronic polarizability into the potential energy function known as polarizable FFs. In Chapter 1, an overview of the CHARMM empirical FF as well as the development and application of the polarizable CHARMM FF based on the classical Drude oscillator is presented. As noncovalent halogen bonding interactions between halogenated ligands and proteins are important in drug design, a well-parametrized polarizable FF for halogen containing compounds is required to perform more accurate modeling of halogenated molecules. During the course of this development, a significant contribution of halogen-hydrogen bond donor (X-HBD) interactions to ligand-protein complexes was uncovered in addition to halogen bond (XB) interactions (Chapter 2). In Chapter 3, the development of halogen polarizable FF to both aliphatic and aromatic systems using halogenated ethane and benzene model compounds is illustrated, with emphasis on optimizing both X-HBD and XB interactions as discussed in Chapter 2. To assure good reproduction of X-HBD interactions with proteins, further optimization of atom pair-specific Lennard-Jones (LJ) parameters is required for both the polarizable and non-polarizable halogen FFs (Chapter 4). Finally, Chapter 5 presents the current optimization of polarizable protein FF to yield a more accurate description of the polarization response in simulations. The resulting protein FF, referred to as Drude-2019, will be more applicable in the study of biomolecular systems. Overall, the advances made in this dissertation will facilitate important discoveries of a range of chemical and biological phenomena.