UC San Diego

Investigations into the link between vitamin D and cardiovascular disease (CVD) have yielded inconsistent results. The vitamin D metabolite ratio (VMR), the ratio of 24,25(OH)2D to 25(OH)D, has shown stronger associations with fracture and mortality than 25(OH)D alone. Our study assessed the association between the VMR and CVD outcomes. We evaluated a cohort of 6,313 participants from the Multi-Ethnic Study of Atherosclerosis (MESA), without pre-existing CVD, over 15 years. Utilizing Cox regression, we examined the associations of both the VMR and 25-hydroxyvitamin D [25(OH)D] with various cardiovascular events. Over the study, 800 participants developed CVD, including conditions such as myocardial infarction, resuscitated cardiac arrest, stroke, coronary heart disease death, and stroke death. Heart failure (HF) was observed in 398 participants, and 413 experienced cardiovascular mortality. Models were adjusted for factors including demographics, lifestyle, clinical conditions and medications, biomarkers, and kidney function. Participants averaged 62 years (range 44-84), with 53% females. The mean (SD) 25(OH)D level was 22.7 (11.0) ng/mL, and the mean VMR was 15.2 (5.0). In fully adjusted models, a two-fold increase in VMR was associated with a 24% reduction in incident CVD (HR: 0.76, 95% CI: 0.65-0.88). However, there was no association between the VMR and HF (0.98, 0.78-1.24), or cardiovascular mortality (0.96, 0.77-1.21). 25(OH)D was not significantly associated with any CVD outcome. In a diverse cohort, VMR was significantly associated with reduced incident CVD, but not HF or cardiovascular mortality. The results suggest that VMR may provide greater insight into vitamin D metabolism, compared with 25(OH)D alone.

Parental engagement enhances academic outcomes. While numerous studies have examined the academic benefits of parental engagement among families of different sociocultural backgrounds, limited research has focused on Mexican-heritage families rearing autistic children. Mexican heritage (MH) families make up a growing number of students in K-12 and have increased in autism diagnoses. Utilizing Barton and colleagues' (2004) definition of parental engagement, this study explored parental engagement among Mexican heritage families rearing autistic children. Research questions examined (1) MH families' distribution of household and educational tasks, (2) MH families' perceptions of educational engagement, (3) MH parents' description of their educational engagement practices, and (4) MH household's enactment of educationally engaging practices in their daily lives. This dissertation study employed a convergent parallel design using a survey, focus groups interviews, individual interviews, and video recordings of parents engaging in educational activities with their children to investigate findings. Results from this study revealed that MH mothers are the main participants in caregiving (i.e. playing, feeding, bathing child) and educational (i.e. speaking to teachers, attending IEP meetings) activities. MH families described three distinct themes related to educational engagement, educational engagement tools, mother’s educational engagement traits, and an independent future. MH parents’ descriptions of practices included parent versus educator’s perceptions of educational engagement, childrearing practices, parent-educator partnerships, supportive environments, and the impacts of COVID-19 on child development. Lastly, enacted practices revealed key strategies such as mothers providing resources, attending to their child’s mood, and at-home support of educational concepts. Findings from this study may be used to modify parental engagement strategies of special education teachers in Mexican heritage communities, provide an understanding of academic engagement strategies in MH homes, and help adapt existing parent engagement practices to better fit those of the home environment.

Marine microbial communities are crucial to ecosystem function and productivity, but their spatial and temporal distributions are highly variable. Microbes exhibit unique environmental preferences, called niches, that drive observable distribution patterns across space and time. However, it’s not well understood how much and at what scales external biotic and abiotic influences, such as competition and dispersal affect microbial distributions. It's expected that microbial distributions are going to change as a result of anthropogenic climate changes, such as increase sea surface temperatures and increased water column stratification. However, current predictive models rely on some assumptions about niches, such as niche stability over time, that have not been broadly tested or observed. This thesis aims to elucidate the mechanisms that shape spatial and temporal variability in marine microbial niches across three distinct chapters. The first chapter asks how the laboratory expectations and observations of niches in the field compare for a globally important genus of cyanobacteria Prochlorococcus. The second chapter asks how temporal variability and dispersal shape microbial realized niches across a latitudinal gradient by utilizing a simplified metacommunity model. The final chapter asks if and how microbial niches have adapted to spatial and temporal environmental change in the California Current Ecosystem. Understanding the mechanisms behind microbial distributions can influence our mitigation and management of broader ecosystem changes such as food web dynamics and carbon export.

Interstellar dust grains play prominent roles in physical processes across astronomical scales and affect all astronomical observations to varying degrees. However, our understanding of how dust evolves within galaxies and across cosmic time is incomplete.We investigate the dust life cycle and make predictions for the evolution of galactic dust populations across cosmic time by developing dust evolution models and integrating them into cosmological galaxy simulations.

In Chapter 2, we present two separate dust evolution models coupled with the ``Feedback In Realistic Environment'' (FIRE) model for stellar feedback and ISM physics. These models incorporate the main mechanisms comprising the dust life cycle but differ in their treatment of dust chemical composition and gas-dust accretion based on recent, contrasting approaches in the galaxy formation community. We test and compare these models in an idealized Milky Way-mass galaxy and find that both produce reasonable galaxy-integrated dust populations and predict gas-dust accretion as the main dust growth mechanism. However, only a model that simultaneously incorporates a physically motivated gas-dust accretion routine and tracks the evolution of specific dust species can reproduce observed spatial dust variability within the Milky Way, in both amount and composition.

In Chapter 3, we present a suite of cosmological galaxy simulations of Milky Way to dwarf halo-mass galaxies. These simulations utilize the dust evolution model presented in Chapter 2, which tracks the evolution of specific dust species and incorporates a physically motivated dust growth routine. We find that gas-dust accretion is the dominant producer of dust mass for all but the most metal-poor galaxies and, in the case of the Milky Way, dominates for the majority of the galaxy's life.We also discover that the onset of rapid growth via gas-dust accretion differs between dust species, arising from differences in element abundances, dust physical properties, and life cycles. These differences can explain the variable dust population, in both amount and composition, in the MW, LMC, and SMC and highlight the importance of accurate gas-dust accretion modeling for individual dust species.

Parkinson’s Disease (PD) is an age-related neurodegenerative disease characterized by tremors, bradykinesia, and rigidity. While several pharmaceutical therapies have been developed to address PD symptoms, there are no preventative treatment or therapies available. To address this gap, Kainos Medicine, Inc. developed KM-819, a FAF-1 inhibitor currently in phase 2 clinical trial1,2. In order to understand the mechanistic effects of this drug in humans early in the disease course, this research utilizes an iPSC model derived from Gaucher Disease (GD) patients – a population shown to acquire PD at significantly higher rates than the general population3. Specifically, live-cell imaging techniques are used to explore the quantity and morphology of lysosomes and mitochondria in GD and healthy control (HC) cell lines, with and without KM-819 treatment. This model system has been explored previously; 24h treatment with 1-20µM KM-819 was found to increase lysosome size and motility, with the most significant improvement consistently occurring around 1µM4. In order to explore the efficacy of KM-819 at low dosages, cells were treated for 24h with 0.1 µM KM-819 and imaged using LysoTracker. Consistent with previous findings, lysosome size and quantity both increased, even at low dosage. Next, the mitochondrial system was explored as it is frequently implicated in PD. Initial experiments focused on establishing baseline measurements for untreated cells; morphological analysis indicated that GD mitochondria present with increased fragmentation and reduced network complexity compared to HC cells. Treating GD cells with 0.1µM of KM-819 for 24h showed significant reductions in mitochondria fragmentation and increased network complexity.

The human brain is underpinned by a massive cellular complexity. A diverse conglomerate of cells, over 100 billion of them, functionally interact to power the most uniquely human organ. Unfortunately, the brain often encounters difficulties, and these neurological disorders drive significant clinical challenges. Most neurological disorders have no consistently effective therapeutic treatments. The work of this dissertation has been conducted with a single goal in mind: to improve the understanding of the human brain, in turn enabling the development of effective therapeutics to treat neurological disorders. To accomplish this, we conducted method development to enable effective single-nucleus profiling of the human brain, outlined tools for analyzing this data, carefully selected targets that may drive functional improvements, and developed and tested therapeutics capable of changing the brain. Here, we have profiled human brains with Down syndrome and matched controls to identify microglial overactivation, and a unique transcription factor, RUNX1, that appear to drive memory deficits. Additionally, we show that a potential therapeutic, targeting RUNX1, can reverse certain aspects of this biology. This work establishes a foundation for drug discovery, utilizing single-nucleus RNA-sequencing data to guide target selection and providing conceptual proof that these efforts can yield efficacious therapeutics.

The physiological processes associated with microbiomes represent the biology of individual microbial cells and their interactions with other cells. Moreover, the interactions between microbes are not static, but represent dynamic processes that are subject to ecological and evolutionary changes. One of the main drivers of rapid evolution in microbial species is the process of horizontal gene transfer, often mediated by mobile genetic elements. Few systems or tools exist that allow the study of this process in microbiomes. In this thesis, we review experimental methods to study horizontal gene transfer, as well as catalog computational methods to sequence and filter for mobile elements, including integrative and conjugative elements, plasmids, and viruses. We next leverage long read DNA sequencing to lay the groundwork for a complex in vitro washed cheese rind microbiome model. With our microbiome model and computational methods, we are able to 1) probe natural microbiomes in kefirs and cheese rinds to better understand the breadth and novelty of mobile elements and 2) begin to understand how different community members play a role in the evolution of species and mobility of plasmids within a genus. Our work in the identification, characterization, and tracking of mobile elements in microbiomes enhances our understanding of the unique and niche-adaptive genes carried by these fermenting microbes. Finally, we explore the potential clinical and industrial applications of long read and proximity ligation. 

Despite the power of randomized smoothing, the standard performance of a smoothed deep reinforcement learning (DRL) agent exhibits a significant trade-off between robustness. To address this issue, we proposed new algorithms to train robust DRL agents while attaining both superior clean performance and robustness performance in discrete and continuous action space. Our proposed DS-DQN and AS-PPO outperform prior state-of-the-art robustly-trained agents in robust reward by $1.54\times$ on average. Moreover, a stronger adversarial attack for smoothed DRL agents is proposed, which is $1.45\times$ more effective in decreasing the rewards compared to existing PGD attacks for DRL agents.

One of the fundamental jobs of the brain is to transform stimuli from the external environment into flexible behavioral outputs. In mammals, thalamocortical circuits perform many of the functions that underlie this complex sensory processing. First-order (FO) thalamic nuclei, such as the dorsal lateral geniculate nucleus (dLGN), relay incoming signals to the cortex, which generates a percept and motor commands. This initial path from the FO thalamus to the cortex is well understood, but interactions between the cortex and higher-order (HO) nuclei, like the pulvinar, remain a mystery. Competing theories on the role of cortico-pulvino-cortical circuits remain unresolved. One model suggests that HO nuclei serve as relays for information transmission between cortical areas; while the alternative proposes a modulatory function promoted by reciprocal thalamocortical loops. Advances in viral tools for anatomical tracing and targeted perturbation of neuronal activity now allow us to test these hypotheses. This dissertation investigates the anatomical and functional relationship between the pulvinar and extrastriate cortex in the mouse in an attempt to understand the nature of higher-order thalamocortical interactions. In Chapter 1, we map the input/output relationships of distinct projection classes in the pulvinar. Using monosynaptic g-deleted rabies virus, we show that driving layer 5 cortical inputs to the pulvinar are organized as a feedforward, transthalamic relay. We also describe a broad network of modulatory layer 6 inputs which are biased towards reciprocal connections with the pulvinar. Bottom-up input from the superior colliculus (SC) targets every cortical pathway through the pulvinar. Chapter 2 investigates the functional contribution of a pulvinar → extrastriate pathway to visual activity in vivo in awake, passively viewing animals. We selectively target a single pulvinar projection population for optogenetic inactivation and compare the effects to inactivation of the FO pathway. Unlike FO thalamocortical input, which is necessary for sensory transmission, the HO input to cortex is not responsible for sensory responses. Instead, our results support a modulatory, excitatory contribution of the pulvinar to cortical activity. In summary, this study establishes a general framework for the anatomical organization of HO thalamocortical circuits, whereby the pulvinar provides a parallel path between cortical areas and a secondary route for bottom-up visual signals. Our physiological results highlight the folly in inferring circuit function from anatomy alone, however, as this transthalamic pathway does not drive visual activity under passive, head-fixed conditions. Instead, our findings describe a potential driving pathway between sensory cortices which might relay a non-sensory or context-dependent message. Additional functional studies that engage this circuit in active behavioral states will be necessary to solve the puzzle of the pulvinar.

Transportation, science, and technology have led to increased sedentary/inactive lifestyles and corresponding, increasing risks for chronic diseases. This culture of sitting is also prevalent in universities. I surveyed students in two courses and analyzed responses to quantitative and qualitative questions. I show 1) Most students sit in university 15-25 hours/week and 52% fall below the U.S. Health Standard’s exercise threshold. 2) Demographics of gender (p<0.001), years in university (p=0.047), and intercollegiate athlete status (p<0.001) explained 20.5% of the variance in exercise rates (Multiple Linear Regression). Exercise trends significantly drop during first-year students’ transition from high school to university (Friedman’sxi test p<0.001). 3) After learning the biology behind exercise, students in a comparison course increased exercise rates compared to the students from the control course which significantly decreased their exercise rates (Welch’s t-test p<0.001). 57% of students in the comparison course now meet the U.S. Health Standard’s exercise threshold and the content of the comparison course had a significant effect on the number of students that increased their exercise trends (qualitative results, McNemar’s Chi-Square test p<0.001). Qualitatively, students mentioned a changed health perspective, understanding of the benefits of exercise, exercise’s importance in aging, and the impact of watching the professors lead by example as passionate exercise advocators. Students expressed interest in taking stand breaks throughout class and integrating application of biology to personal health in all biology courses. These changes have potential to teach students how to connect to a healthy routine they can sustain for a lifetime.