The mechanistic target of rapamycin (mTOR) is a kinase found in two multi-protein complexes, mTOR complex 1 (mTORC1) and 2 (mTORC2). These complexes are integral parts of an evolutionarily conserved signaling pathway known as the mTOR pathway. mTORC1 and mTORC2 are controlled by an array of intra and extracellular stimuli via different upstream regulators. The two complexes are known to exert distinct functions by phosphorylating downstream targets and ultimately orchestrate cell growth and metabolism. Aberrant mTOR activity is associated with numerous diseases, with particularly profound impact on the nervous system. Specifically, mutations in genes encoding for mTOR regulators result in a collection of neurodevelopmental disorders known as mTORopathies.Tuberous Sclerosis Complex (TSC) is one of the most well characterized mTORopathies. TSC is caused by mutations in the TSC1 or TSC2 genes, which encode proteins that negatively regulate mTORC1 signaling. Current therapeutic strategies focus on rapamycin and its analogs that are inhibitors of mTORC1. However, several studies have shown that chronic rapamycin inhibits both mTORC1 and mTORC2 in a cell-type specific manner, raising the possibility that mTORC2 suppression might also exert therapeutic benefits in TSC. This idea has been corroborated by some studies showing that mTORC2 is involved in cellular processes that are altered in TSC, such as myelination and mGluR-dependent synaptic long-term depression. Most recently a study showed that mTORC2 suppression can provide therapeutic benefits in other mTORopathies. It is currently unknown which mTOR complex is most relevant for TSC-related brain phenotypes. To model TSC we used in vitro systems of primary hippocampal cultures where we examined postnatal loss of Tsc1 and we also used the Emx1-Cre mouse line to conditionally delete Tsc1 embryonically from forebrain excitatory neurons. To investigate which mTOR complex is responsible for TSC neurologic manifestations we used genetic strategies to target Raptor and Rictor and selectively reduce mTORC1 or mTORC2 activity respectively. Interestingly, our study revealed that the two complexes regulate each other’s activity and loss of either Raptor or Rictor affects the signaling of both complexes. As it has been previously shown loss of Tsc1 results in increased mTORC1 activity and decreased mTORC2. We found that reduction of Raptor, but not Rictor, rebalances both mTORC1 and mTORC2 signaling and improves the morphology of Tsc1 knock-out neurons in vitro. We also observed that Raptor reduction in vivo, was sufficient to prevent several neurologic phenotypes in a mouse model of TSC, including mTORC1 hyperactivity, neuronal hypertrophy, demyelination, network hyperactivity and premature mortality. Finally, we examined Raptor manipulation as a therapeutic strategy by postnatally injecting shRNA in mice with embryonic loss of Tsc1. We found that postnatal Raptor downregulation can rescue both cell and non-cell autonomous mechanisms including mTORC1 hyperactivity, neuronal hypertrophy, and myelination. We also found that shRptor can significantly extend survival and improve the overall development of Tsc1-KO mice. Notably, when we examined the effects of Raptor manipulation as therapeutic strategy for TSC-related seizure activity we found that downregulation of Raptor did not improve the phenotype. Interestingly, neither rapamycin treatment was able to rescue this phenotype suggesting that this seizure like activity in our in vitro model, cannot be rescued via mTORC1 suppression after it has been established. Overall, this thesis provides novel insights in the regulation and function of the mTOR pathway in neurons. We identify mTORC1 as the complex that drives neurologic manifestations in mouse models of TSC and propose that Raptor manipulation could be a promising therapeutic strategy for TSC and potentially other mTORopathies. We have also established an in vitro model where we can study TSC related seizure-like activity. This model reveals that cell-autonomous changes that drive neuronal hyperactivity due to Tsc1 loss can become mTORC1 independent over time. Together this data generates new insights that will aid in understanding more in depth the molecular mechanisms that drive TSC neurologic manifestations and provide important information for the development of novel preventative and therapeutic strategies.

mTORC1 is a central signaling hub that integrates intra- and extracellular signals to regulate a variety of cellular metabolic processes. Mutations in regulators of mTORC1 lead to neurodevelopmental disorders associated with autism, which is characterized by repetitive, inflexible behaviors. These behaviors likely result from alterations in striatal circuits that control motor learning and habit formation. However, the consequences of mTORC1 dysregulation on striatal neuron physiology are largely unknown. To investigate this, I deleted the mTORC1 negative regulator Tsc1 postnatally and prenatally from identified striatonigral and striatopallidal neurons and examined how cell autonomous upregulation of mTORC1 activity affects their morphology and physiology. I find that postnatal loss of Tsc1 increases the excitability of striatonigral, but not striatopallidal, neurons and selectively enhances corticostriatal synaptic transmission. I find that prenatal deletion of Tsc1 also enhances excitability of striatonigral, but not striatopallidal, neurons, and that this occurs through an increase in cortico-striatonigral synaptic transmission. I also observe that a loss of endocannabinoid-LTD drives enhanced cortical excitation of striatonigral neurons, and that this is associated with improved motor routine learning.

To investigate the molecular mechanism driving these synaptic changes, I performed translating ribosome affinity purification to determine gene expression changes in mice with loss of Tsc1 from either striatonigral or striatopallidal neurons. I find that Tsc1 deletion results in numerous translational differences in striatal neurons, and that many of these differences are specific to either striatonigral or striatopallidal neurons. Further, I extend my findings in the Tsc1 model to the Cntnap2-/- ASD mouse model, and find that corticostriatal transmission and motor routine learning are also enhanced by deletion of Cntnap2. These findings highlight the critical role of corticostriatal transmission in ASD mouse models, with implications for striatonigral pathway dysfunction in neuropsychiatric disease.

Midbrain dopamine (DA) neurons in the substantia nigra pars compacta (SNc) and ventral tegmental area (VTA) project throughout the brain to modulate a diverse set of behaviors and brain states. Although their effects are widespread, they represent a very small fraction of neurons. Dopamine neurons rely on their diverse inputs, outputs, physiology, and gene expression in order to maintain their broad behavioral influence. To better understand the different roles dopaminergic projections play in the brain, the circuits need to be studied in isolation as adjacent neurons and circuits often mediate distinct behaviors. Because of their close anatomical proximity, novel tools and methods must be devised in order to precisely dissect dopaminergic sub-circuits. New tool development will depend on identifying the underlying genetic signature of dopaminergic subpopulations to allow consistent, reproducible labeling of distinct populations. Gene expression analysis will also reveal unique genes expressed in DA neurons that could lend functional importance to already established dopaminergic circuits.

Here we used single-cell RNA-sequencing to determine the unique gene expression signature of DA neurons and identify genes that define distinct dopaminergic subpopulations. We used retrograde tracing, electrophysiology, and disease models to show that genetically defined populations are functionally distinct with defined projections, unique physiological characteristics, and selective sparing in a Parkinson’s disease model. Through this work, we defined two circuits that arise from populations in the ventromedial VTA marked by their expression of Neurod6 and Grp and determined that single genes are not sufficient to label dopaminergic subtypes. Because of this, we developed a Flp-recombinase dopaminergic reporter mouse (DAT-Flp). Using this mouse, we were able to use intersectional genetic tools to selectively label Neurod6-DA neurons to enable future investigations into the role that these neurons play in dopaminergic functions.

Autism spectrum disorder (ASD) is a neurodevelopmental disorder characterized by persistent deficits in social communication and interaction, and the presentation of restricted repetitive patterns of behavior. The prevalence of ASD has increased steadily in the last 40 years, with a recent study from the Center for Disease Control (CDC) indicating that roughly 1 in 36 children in America is diagnosed with ASD. The disorder has a strong genetic component, and there have now been over 100 high confidence ASD risk genes identified through genetic sequencing of those with ASD. As improved genetic tools allow for these genes to be identified, genetic access in animal models has also improved, allowing for the development of many mouse models of ASD that harbor mutations in these identified genes. ASD risk genes vary greatly in the types of proteins for which they encode, including ion channels, neurotransmitter receptors, cell adhesion molecules, and machinery implicated in all aspects of transcription and translation. Despite this genetic heterogeneity, the major diagnostic criteria of ASD still fit within the two symptom domains listed above. Given this shared symptomology, much work in ASD research has focused on identifying brain regions that may be commonly impacted across a range of underlying genetic alterations. The basal ganglia, in particular the striatum, the primary input center of the basal ganglia, has arisen as one such brain region. We believe that dysfunction of the striatum, given its known role in action selection, motor learning, and habit formation, may be particularly implicated in the restricted, repetitive behavior domain of ASD. However, whether altered striatal function is a shared pathophysiology across genetically diverse ASD mouse models has yet to be comprehensively assessed. For this dissertation project, I investigated this question in two unique ASD mouse models. First, I developmentally deleted the ASD risk gene Tsc1 selectively from the two primary types of striatal projection neurons (SPNs). I found that loss of Tsc1 selectively from SPNs of the direct pathway (dSPNs), but not the indirect pathway (iSPNs), increases cortical drive of these neurons, likely through increased glutamate release from cortical inputs onto these cells. This increased corticostriatal drive of the direct pathway increases motor learning, measured through enhanced performance in the accelerating rotarod assay. Altered cortical input to the striatum has arisen as a potential convergent change across a number of ASD mouse models, an idea that is reviewed and expanded upon in the second chapter of this dissertation. To investigate whether striatal function and striatum-associated behaviors are altered in a genetically distinct mouse model, I utilized mice with brainwide loss of the ASD risk gene Cntnap2. I found that mice lacking Cntnap2 also exhibited increased cortical drive, of both dSPNs and iSPNs. In this model however, cortical synaptic input onto SPNs was unchanged. Instead, the intrinsic excitability of SPNs in Cntnap2-/- mice was significantly increased, in particular in dSPNs, which likely underlies the increased cortical drive of these cells. Behaviorally, Cntnap2-/- mice also exhibit increased performance in the accelerating rotarod task, as well as increased spontaneous repetitive behaviors, and cognitive inflexibility in a reversal learning task. Together, these findings support a role for striatal dysfunction in the manifestation of stereotyped, inflexible behaviors across ASD mouse models with varying genetic causes. In particular, this data supports an emerging theory that corticostriatal alterations, in particular enhanced cortical activation of the movement-initiating direct pathway, may occur commonly in the case of ASD-associated repetitive behaviors.

Tuberous Sclerosis (TSC) is a multisystem developmental disorder, which is associated with early-onset epilepsy, varying degrees of intellectual disability and high risk for psychiatric conditions including autism. It is defined by the presence of “cortical tubers”, which are focal cortical malformations composed of dysplastic neurons and astrocytes that can become seizure foci. TSC is caused by mutations in either TSC1 or TSC2, which encode for proteins forming a complex that is a negative regulator of mTORC1, a key cellular signaling node controlling cell growth and metabolism. Animal models have greatly increased our understanding of TSC; however they fail to recapitulate some of the key neuropathological features, namely the cortical tubers. This may be due to the contracted timeframe over which rodent cortical development occurs (days) compared to human cortical development (months) as well as the presence of human-specific progenitor cells. My research goal was to overcome these limitations by creating a human neuronal model of TSC that develops on a human timescale and recapitulates key features of early human cortical development.

I used CRISPR/Cas9 to create an isogenic panel of human embryonic stem cell lines harboring heterozygous or homozygous loss-of-function mutations in TSC1 or TSC2. I differentiated these cells into 3D cortical spheroids over five months in culture to model mid-gestational human fetal development. I found that while early forebrain development was largely normal, TSC1-/- or TSC2-/- cells in five-month-old spheroids were highly abnormal. Specifically, loss of function of TSC1/2 led to the formation of highly enlarged, dysplastic cells that strongly resembled the cortical tuber cells seen in patients. Furthermore, I revealed a significant differentiation bias of TSC1/2 mutant cells towards glial cell fates and away from neuronal fates, again reflecting what is seen in patient cortical tubers.

Cortical tubers are thought to arise due to a “second-hit” somatic mutation during progenitor cell proliferation that results in loss of heterozygosity. I further refined our human TSC model to mimic a somatic second-hit event, by generating a TSC2c/- stem cell line, in which all cells were heterozygous for TSC2 but would undergo biallelic inactivation in the presence of Cre recombinase. I also engineered a fluorescent Cre- reporter into these cells to enable lineage tracing of TSC2-/- cells over time. I found that cells with a second-hit mutation in TSC2 were large, dysplastic and primarily expressed glial markers, while the adjacent TSC2+/- cells showed normal morphology and development, mimicking a cortical tuber. Chronic treatment with rapamycin, a potent mTOR inhibitor, starting early in development rescued tuber cell morphology and caused marked shifts of differentiation towards a neuronal fate in both TSC2+/- and TSC2-/- cells, demonstrating mTORC1’s bidirectional control over human neuronal differentiation.

To elucidate the developmental programs of normal human corticogenesis and how these are altered by loss of TSC2, I have performed an in-depth single cell RNA- sequencing analysis across development in TSC2c/- spheroids. My analysis has confirmed the differentiation bias of TSC2-/- cells away from neuronal cell fates and revealed that this becomes pronounced primarily after two months of differentiation, before which TSC2+/- and TSC2-/- cells share similar identities. Furthermore, I discovered novel genetic markers of TSC2-/- cells: CLU, PTGDS, APOE, B2M that may contribute to their aberrant differentiation. Together, my thesis research has established novel, validated human brain spheroid models for TSC, which have generated new insights into the early molecular, cellular, and developmental mechanisms contributing to both normal and abnormal human brain development.