Thesis Central

Plant root exudates shape the structure and function of the rhizosphere microbiome by providing chemical cues and substrates that influence microbial survival, colonization, and interaction with the host. These exudates can select for beneficial microbes that support plant growth or suppress immunity, making them critical in mediating plant-microbe interactions. This study focuses on two Arabidopsis thaliana (Arabidopsis) associated, immunity-suppressive bacterial strains, Dyella japonica MF79 (MF79) and Brevundimonas sp. MF374 (MF374), which use distinct genetic mechanisms to suppress root immune responses. To evaluate how the root exudate composition influences bacterial fitness, transposon mutant libraries of MF374 and MF79 were grown in exudates derived from Arabidopsis Col-0, fls2 mutant, and cyp79b2/b3 mutant lines. Genome-wide fitness profiling using random barcode transposon-site sequencing (RB-TnSeq) and Random Forest classification revealed minimal fitness changes in response to flg22 treatment, but strong genotype-specific fitness differences driven by root exudate composition. In MF374, candidate genes involved in osmotic regulation, cell wall synthesis, oxidative stress response, and translation were identified as major contributors to fitness in root exudate environments. Targeted gene deletions confirmed reduced fitness in immune-active or metabolite-rich exudates; however, root colonization assays demonstrated that these fitness effects did not always translate to differences in colonization capacity. This framework enabled the identification and functional assessment of bacterial genes important for exudate-mediated fitness and adaptation. The findings offer insight into how variation in root exudate composition influences the assembly, persistence, and adaptation of immunity-suppressive bacteria within the root microbiome and have broad implications for understanding plant health and developing strategies for microbiome-based agricultural applications.

Rising atmospheric CO₂ concentrations in the recent century has highlighted the urgent need for effective carbon capture technologies. Among these, direct air capture (DAC) offers a unique carbon-negative solution by extracting CO₂ directly from ambient air. Traditional DAC approaches using thermal or pressure swings are energy-intensive, prompting interest in moisture swing capture, where CO₂ is absorbed at low humidity and released at high humidity. This study explores the use of cation exchange resins (CERs) as a potential alternative to the commonly studied anion exchange resins (AERs) for moisture swing DAC. By testing CERs with aminophosphonic and iminodiacetic functional groups and comparing them to AERs with phosphate and carbonate counterions, the study investigates CO₂ capture capacities, capture/regeneration rates, and performance across multiple cycles. Results suggest that CERs go through a similar but marginally different mechanism as AERs, where functional groups are hydrolyzed to produce hydroxyl ions, which then react to capture CO2 in the form of bicarbonate. The captured CO2 is released at high humidity as bicarbonate, and the acid form of functional groups neutralizes. However, their CO₂ capture capacity and rates are significantly lower than those of AERs, largely due to differences in hydrolysis mechanisms and functional group pKa values. Nevertheless, CERs demonstrated stable performance over repeated cycles, suggesting potential for improvement and application in scenarios where water purity is a limiting factor. This research offers insights into the feasibility of CERs for low-energy, water-tolerant DAC systems and highlights pathways for further material optimization.

Biomolecular condensates are membraneless organelles inside living cells that primarily comprise proteins and nucleic acids. The thermodynamic process of liquid-liquid phase separation has been proposed as a primary driver of biomolecular condensation, and it is recognized that phase separation is maintained by networks of biomolecular interactions within these liquid droplets. Canonical examples of condensing biomolecules include prion-like low-complexity domains (LCDs) of proteins, and simulations of single-component LCD condensates have predicted the presence of small-world topologies in the interaction networks underlying condensate stability. Recent experimental and theoretical works have also demonstrated inhomogeneities in single-molecule conformation, orientation, and dynamics within biomolecular condensates. Here, we systematically characterize the molecular networks underlying both LCD condensates and condensates formed by generic associative heteropolymers. Further, we investigate the relationship between network topologies and single-molecule properties within condensates. To probe LCD condensates, we employ a chemically specific, coarse-grained model of disordered proteins designed to reproduce phase separation statistics. We generalize our findings by varying sequence hydrophobicities using a generic binary model of associative heteropolymers, dubbed the “hydrophobic–polar” (HP) model. In both model systems, we find persistent small-world topologies underlying single-component condensates. These topologies feature molecular “hubs” with high network betweenness centrality and molecular “cliques” that represent densely interacting clusters of biomolecules; distal cliques in condensate volumes all localize to phase interfaces and are bridged by elongated hubs that remain near condensate centers. Strikingly, we find that relationships between network connectivity and biomolecular structure and dynamics are governed by power laws. Our work demonstrates that inhomogeneous single-molecule behaviors within biomolecular condensates can be well predicted from condensate network connectivities. Furthermore, we find that network cliques have substantially longer lifetimes than molecular hubs, and that the motion of molecules within cliques is spatially constrained. Together, these results reveal a dynamic hub-clique architecture underlying condensates and suggest that the physicochemical characteristics and material properties of phase interfaces are critical to pathological gelation and fibrillization processes observed in condensate aging.

The increasing integration of technology with our infrastructure systems, such as transportation and communication networks, spotlights the importance of understanding how individual selfish behaviors impact collective social efficiency. This thesis investigates atomic selfish routing games, where discrete, heterogeneous agents compete for shared resources, such as in road traffic or internet routing cases. Classic results such as the Braess paradox show that individual rational behaviors can lead to socially poor outcomes. To address these inefficiencies, this research develops a bandit-based learning algorithm that enables agents to iteratively learn low-latency routes with minimal information while accounting for the computational complexity that comes with large strategy spaces. This algorithm can find near-Nash equilibrium for the selfish routing game even in cases with thousands of players and graphs over a thousand edges and vertices.

Empirical tests on real-world transportation network datasets show that carefully designed incentive mechanisms formed using a modest and targeted version of the marginal cost pricing tax can effectively mitigate social inefficiencies. Specifically, a small proportion of roads taxed combined with a small degree of taxation is enough to recover a significant majority of system-wide inefficiency. Inspired by Stackelberg games and the emergence of autonomous vehicle technology, this thesis also explores the impact of introducing altruistically coordinated agents, analogous to autonomous vehicles. Results show a clear linear relationship between the proportion of altruistic agents and efficiency gains, showing potential for congestion mitigation via introducing such agents.

TDP-43 is an RNA-binding protein pathophysiologically implicated in amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD). Experiments by Mann et al. have demonstrated doses of "bait oligonucleotides" with high affinity for TDP-43 can abolish neurotoxicity by dissolving pathological TDP-43 condensates in the cytosol [1]. To increase the odds of translational success for this technology, growing and diversifying the group of sequences known to bind TDP-43 with high affinity is desirable. Computational screening is a promising method for prudent identification of sequences to test experimentally, saving time and money. In this thesis, we explore the use of observables from all-atom molecular dynamics simulations to score the bound poses of TDP-43 to various RNA sequences. Coarse-grained unbinding simulations and AlphaFold uncertainty measurements were also examined. The results indicate that all-atom RMSF correlates the best with experimental Kd (Pearson approx. 0.7). AlphaFold also weakly correlated (Pearson approx. 0.5), while coarse-grained simulations possessed no correlation. Temperature replica exchange simulations suggested that vanilla simulations do not fully sample the range of bound conformations for RNA, which could interfere with correlation of observables and Kd. Notably, vanilla simulations were able to persistently model residue interactions of known experimental significance (pi stacking and hydrogen bonding between TDP-43 and RNA), which may have contributed signal to the RSMF correlation. Our results suggest the RMSF has potential for use in a pose scoring function, but that it is insufficient alone to predict Kd. Due to noise in experimental affinity values and poor sampling in vanilla simulations, more high-quality experimental Kd data is imperative, in order to build more complex MD-based scoring functions while avoiding overfitting.