| Summary: | In vitro models are often used both to characterize and test therapeutics for neurodevelopmental disorders (‘NDDs’). While in vitro models have extraordinary potential to develop therapies for patients, they have historically been confounded by absence of robust phenotypes and/or in vitro phenotypes that fail to translate from laboratory bench to bedside. Within this thesis work, we attempt to address three areas in which in vitro models may be improved – gene selection, model validation and identification of disease-relevant functional assays suited for therapeutic testing. Publicly available databases aggregating identified and annotated disease-causing variants for Mendelian diseases have rapidly expanded over the past two decades. Elucidating mechanisms of disease and developing therapies using in vivo model systems often is both time and cost intensive. Thus, determining which subsets of genes are more likely to generate addressable signals in a dish may lead to more effective drug development. In chapter 1, we identify a set of genes ideally suited for therapeutic inhibition. Specifically, we leverage the aforementioned large genetic databases to identify a set of genes likely to act through a gain-of-function mechanism that are both tolerant to loss-of-function mutations and in the druggable genome.
In chapter 2, we aim to characterize the degree of conservation of transcriptomic dysregulation between a human in vitro cortical organoid (‘hCOs’) model, and two mouse models of a severe neurodevelopmental disorder resulting from HNRNPU deficiency. Human model systems may improve upon animal models when human pathogenesis and patient phenotypes are divergent from animal models due to species-specific etiology. However, human model systems often lack the heterogeneity and cell-type specificity and maturity seen in primary fetal samples. Importantly, some mouse models of HNRNPU deficiency have muted phenotypes compared with human patients. We hypothesized that while there are distinctions between humans and mice with HNRNPU deficiency, there will be overlap in transcriptomic dysregulation between human and mouse models. In fact, we find 45-day-old HNRNPU+/- hCOs have consistent transcriptomic dysregulation to embryonic mouse models, but not to perinatal mice. Our findings suggest hCOs are a viable model for characterizing HNRNPU deficiency; however, such models may only be appropriate for elucidating a transcriptomic disease state at a specific developmental time period.
Functional assays for neurodevelopmental disorders can aid in understanding whether transcriptomic dysregulation is relevant to patient symptoms, as genomic findings may not always correlate to disease-relevant phenotypes. Further, relevant functional phenotypes can then be utilized for testing potential therapeutics. Importantly, seizures are commonly present in a significant subset of neurodevelopmental disorders and seizure phenotypes have been described as driven by aberrant synchrony in neuronal networks. Using a multielectrode array platform, investigators can use a variety of computational methods to quantify aspects of synchrony in vitro. In chapter 3a, we introduce topological approaches capable of identifying novel synchrony phenotypes in primary neuronal networks from mouse models of neurodevelopmental disorders. Certain mouse models will be confounded by species-specific pathogenesis and/or vastly different developmental timelines and fail to generalize to human patients, motivating the need for functionally active and physiologically relevant human in vitro models. In chapter 3b, we attempt to generate human networks with balanced levels of excitation and inhibition and find confounding lack of functional maturation of inhibitory neuronal subtypes in 90-day-old stem cell-derived neuronal networks. Future work generating in vitro human neuronal networks with functionally mature inhibitory neurons would complement the findings in chapters 1 and 2 and allow for more efficient therapeutic development strategies that may lead to improved patient outcomes.
|
|---|
