Recurrent circuits are a hallmark of mammalian sensory cortex. How they impact dynamics of sensory representation is not understood. Because recurrent circuits provide a majority of the synaptic excitation to cortical neurons in response to sensory stimulation, the intrinsic dynamics of these cortical recurrent circuits are expected to be a critical determinant of the timing of the sensory response in cortex. Previous methods could not isolate dynamics of these intra-cortical recurrent circuits from those of thalamic afferents during sensory processing. I now accomplish this by developing an approach to optogenetically silence thalamus in a model system: the mouse visual pathway. Silencing thalamus revealed the time course over which visually evoked activity in visual cortex was maintained by the intra-cortical recurrent circuits themselves, in isolation from thalamic input. I found that, at all time points during the cortical sensory-evoked response, optogenetically silencing thalamus led to a fast decay of sensory-evoked activity in cortical recurrent circuits. This activity decay time course was fit by a 10 ms network time constant, similar to a neuron’s integration time window. This decay time course was invariant across all tested visual stimulation conditions and behavioral states but depended on cortical inhibition. In awake mice, the dynamics of this time course predicted the time-locking of cortical activity to thalamic input at frequencies <15 Hz and the attenuation of the cortical response to higher frequencies. Under anesthesia, however, dynamics of depression at thalamocortical synapses disrupted the fidelity of sensory transmission. Thus, I determine sensory-evoked dynamics intrinsic to the intra-cortical recurrent circuits in isolation from thalamus and show how these dynamics transform afferent input in time.

Ever since Hodgkin and Huxley described the sodium dependence in action potential nearly 70 years ago, neuroscientists have discovered an entire family of voltage-gated sodium channels (NaVs) in neuronal cells. Uncovering the differential distributions of these NaV isoforms between and within each type of neuron helped us understand the distinct roles they play in cellular functions. So crucial to normal neuronal physiology, genetic variants that result in protein structure changes can cause pronounced alteration in brain function and lead to disease. NaV1.2, encoded by SCN2A, is one NaV isoform whose genetic variants have been linked to neurodevelopmental disorders such as autism spectrum disorder (ASD). Studies have uncovered the importance of NaV1.2 in neuronal intrinsic properties and synaptic physiology in the forebrain. However, how these channels maintain normal cellular functions in the cerebellum, in which some of the strongest expression of NaV1.2 are detected, is unclear. The cerebellum is heavily implicated in ASD due to its functions in motor learning, fine movement control, and more recently discovered, cognition. Therefore, understanding how SCN2A dysfunctions alter cerebellar circuit has become critical to study disease manifestation in channelopathies. In my thesis work, I found that Scn2a haploinsufficiency in mice decreased cerebellar granule cell excitability, impaired high-frequency action potential propagation along their axons, and ultimately compromised the downstream synaptic plasticity that is critical to cerebellar learning. I also used cerebellum-dependent vestibulo-ocular reflex (VOR) to uncover that mice with Scn2a heterozygosity exhibited saturated VOR gain and deficit in VOR adaptation, which are correlated with abnormal Purkinje cell firing pattern during behavior. In addition, children with LoF SCN2A variants showed a similarly elevated baseline VOR gain. Lastly, I revealed that the VOR adaptation deficit in mice can be rescued by upregulating Scn2a in the entire mouse brain using a novel CRISPR activation technique. I hope this work will contribute to our understanding of sodium channel function in the cerebellum and ASD.

Prefrontal cortex (PFC) is an associative center of the brain that integrates local and long-range inputs from cortical and subcortical structures to support working memory and complex decision-making behaviors. Layer 5 pyramidal cells in PFC are themselves associative centers, as they have dendrites that span all cortical layers and can integrate inputs from different input basal and apical streams. In mouse PFC, layer 5 pyramidal cells can be classified based on their dopamine receptor expression pattern. Recent work has identified a novel class of D3 dopamine receptor (D3R)-expressing pyramidal cell that is distinct from the D1 and D2 dopamine receptor-expressing pyramidal cell classes. D3R-expressing pyramidal neurons have been characterized in terms of their morphologies, projection patterns, and intrinsic properties, but this opens up more questions: (1) how do D3R-binding ligands, such as second generation antipsychotic drugs, change the activity of these cells, (2) how are inputs to PFC layer 5 processed across D1, D2, and D3R-expressing neurons, and (3) how do these cells contribute to decision-making behavior?

Here, I use whole-cell patch-clamp recordings, two-photon laser-scanning microscopy, and calcium imaging to show that some second generation antipsychotic drugs, thought of antagonists for G-protein signaling at D3R, can also act as agonists for arrestin signaling at D3R. Arrestin recruitment to D3R can, acutely, result in modulation of calcium at the axon initial segment and, chronically, lead to D3R internalization and degradation. I go on to show that D3R-expressing pyramidal cells, more so than their D1R- and D2R-expressing counterparts, in their apical dendrites exhibit supralinear calcium transients evoked by a burst of backpropagating action potentials. A lack of hyperpolarization-activated cyclic nucleotide-gated channels appears to contribute to this difference across cell classes. These data suggest that there are several features of D3R-expressing pyramidal neurons that differentiate them from other PFC layer 5 pyramidal cells and suggest that information flow through PFC layer 5 is not equivalently processed. Finally, while not a direct test of D3R function, I use a complex decision-making behavior to show that animals with Scn2a haploinsufficiency, which have reduced dendritic excitability in PFC layer 5 pyramidal cells, exhibit no behavioral differences across several behavioral metrics. Taken together, these data further our understanding of dopamine receptors, second generation antipsychotic drug action, and ion channels in layer 5 pyramidal neuron subtypes and in prefrontal cortical processing.