This work concerns the use of engineered genetic tools to build maps of the mammalian nervous system. Within the practice of circuit neuroscience, one of the most effective tools to emerge in recent years are the neurotropic viruses. Among these are modified strains of rabies virus which are made safe for laboratory use. We introduce here a novel form of engineered rabies virus with substantially improved utility for exploring the structure and function of neural circuits. Additionally, using this new tool, an investigation of an important motor circuit, the cortico-striatal circuit, is presented.

The mammalian serotonin system modulates a wide variety of emotional behaviors and states, including reward processing, anxiety, and social interaction. To reveal the underlying patterns of neural activity, we visualized serotonergic neurons in the dorsal raphe nucleus (DRN5-HT) of mice using miniaturized microscopy and calcium imaging during diverse emotional behaviors. In total, we imaged the activity of over 2,000 genetically-identified serotonin neurons. We discovered discrete ensembles of cells with highly correlated activity and found that DRN5-HT neurons are preferentially recruited by emotionally salient stimuli as opposed to neutral stimuli. Individual DRN5-HT neurons responded to diverse combinations of salient stimuli, with some preference for valence and sensory modality. Anatomically-defined subpopulations projecting to either a reward-related structure, the ventral tegmental area, or an anxiety-related structure, the bed nucleus of the stria terminalis, contained all response types, but were enriched in reward- and anxiety-responsive cells, respectively. Our results suggest that the DRN serotonin system responds to emotional salience using correlated ensembles with mixed selectivity and biases in downstream connectivity.

The mammalian brain is a complex organ, the result of a very specific and regulated differentiation process. Although there are many different cell types in the mammalian brain, neurons make up the bulk of the tissue. Neurons come from the divisions of radial glial progenitors (RGPs), which are columnar stem cells in the developing brain. These cells undergo two types of division: symmetric or asymmetric. Symmetric divisions expand the stem cell population, resulting in two new RGPs. Symmetric divisions are critical for ensuring the stem cell population is not depleted too quickly in development. Asymmetric divisions are neurogenic, producing one RGP and one cell that will either differentiate into one neuron, or an intermediate progenitor (IP) that will divide again and produce two to four neurons (Shitamukai, Konno, and Matsuzaki 2011). Several factors have been linked to this determination, including mitotic spindle orientation, centrosomal inheritance, and exposure to proliferative factors, like sonic hedgehog and Notch (Chenn and McConnell 1995; Gaiano and Fishell 2002; Han 2016). This work will focus on spindle orientation, which has been linked to cell fate in many contexts (Lancaster and Knoblich 2012; Williams and Fuchs 2013; Chenn and McConnell 1995). Spindle orientationmust be tightly controlled in order to expand the RGP cell population in early development, then, with more randomized spindles, to shift to producing neural precursors during cortical expansion (Götz and Huttner 2005). While the exact mechanism is still unknown, the orientation of the mitotic spindle relative to the ventricular surface at the time of division affects what type of division occurs (Lancaster and Knoblich 2012). A related process in RGP neural production is interkinetic nuclear migration (INM), in which the RGP nucleus travels apically and basally in a cell-cycle dependent manner (Noctor et al. 2001; Sauer 1935; Hu et al. 2013). The RGP only divides when the nucleus reaches the apical surface; why this occurs is still not known. INM ensures that only a small population of RGPs is dividing in a controlled manner, allowing for cells to interpret polarity cues and orientthe spindle while dividing. One protein that is important to multiple processes in neuronal development is Kif1A. Kif1A is a kinesin motor that has been shown to be critical for INM, in particular for transporting the nucleus basally after division. When Kif1A expression is reduced using shRNA, RGPs fail to migrate away from the ventricular surface, but continue to go through the cell cycle at a normal rate (Carabalona, Hu, and Vallee 2016). Additionally, RGPs that lack Kif1A also exhibit more horizontal and symmetric divisions. This indicates that Kif1a is involved in asymmetric, oblique divisions that produce neurons. Thus, without Kif1a, RGPs produce fewer neurons, instead expanding the RGP cell population. Another kinesin that may be involved in spindle orientation is Kif13B. Kif13B is in the same kinesin-3 subfamily as Kif1A. While structurally very similar to Kif1A, it does have distinct features. It contains a CAP-gly domain, used for binding to the plus end of microtubules. This domain is absent from other kinesin-3 family members, including the most closely related,Kif13A. Kif13B has been shown to be critical for spindle orientation in polarized Drosophila S2 cells, as well as in neuroblasts (Carabalona, Hu, and Vallee 2016; Siegrist and Doe 2005). Kif13B functions to anchor the mitotic spindle to other factors at the cell cortex during mitosis. This occurs through direct interaction with Discs large (Dlg1), which then connects to other factors at the cell membrane, including G?i, LGN, and NuMA. This is a critical process to ensure daughter cells are properly specified. Many of these factors, including LGN and NuMA have been identified as important spindle regulators in RGP divisions as well. Kif13B binds to Dlg1 and to 14-3-3 ?, which is bound to 14-3-3 ?, bound to NudE and Dynein, connecting the Kif13B to Dynein (Lu and Prehoda 2013). Kif13B, as a kinesin, moves along microtubules towards the plus end. Dynein moves in the opposite direction, towards the minus end. The connection of two opposing motors moving in opposite directions may serve to put tension on the spindle and prevent it from freely moving within the cell. When Kif13B is knocked down or removed in cells, the spindle orients randomly in the cell, not in line with LGN or NuMA at the cell cortex (Siegrist and Doe 2005; Lu and Prehoda 2013). This indicates that in mammalian systems, it likely is important for maintaining orientation, and its loss in RGPs would result in random orientation as well. This would result in more neurogenic divisions in RGPs, which is the opposite of the effect seen with Kif1a shRNA. By using in utero electroporation of embryonic rat brains as well as a mouse model ofKif13b knockout in RGPs, I have shown that Kif13B and Kif1A have opposing roles in neurogenesis. This difference can be traced to an alteration of IP production, which Kif1A shRNA decreases, and Kif13b shRNA increases. This can be further traced to the opposing effects on spindle orientation of dividing RGPs. Kif1a shRNA results in more horizontal spindle angles while Kif13b shRNA or deletion results in more random spindle angles. While the kinesin-3 family members are very similar in structure, there are key differences between them. Kif1A has a cargo binding domain at its C terminus, the pleckstrin homology (PH) domain. Kif13B contains a CAP-gly domain. This difference in tail domains would presumably allow Kif13B to bind to microtubule plus ends, while Kif1A would dissociate from the spindle. This difference, therefore, could explain why these two very similar kinesins appear to be performing the opposite roles in spindle orientation. This work provides evidence for a novel mechanism of regulation of neuron production in the mammalian cortex.

Somatosensory neurons densely innervate skin, our largest sensory organ. Adult skin continually remodels throughout the lifespan to maintain a protective barrier for our bodies. How sensory neurons maintain their peripheral endings in the face of continual turnover of their target tissue is not well understood. To address this gap in knowledge, I analyzed the temporal dynamics and mechanisms of structural plasticity of touch receptors in healthy adult skin. My studies focused on the terminals of Merkel-cell afferents in mouse touch domes. These two-part touch receptors comprise epithelial Merkel cells innervated by branching axons of fast-conducting sensory neurons. I show that Merkel cells and their afferents are structurally plastic over the course of hair growth in adults. These two components simplify during active hair growth, with fewer terminal neurites and fewer Merkel cells per touch dome at this stage compared with other phases of hair growth. Merkel-cell removal was observed with multiple molecular markers. Additionally, mice showed diminished touch-evoked behavior during hair growth compared with follicle quiescence. Next, I showed that Sarm1, a key effector of Wallerian degeneration, is not required for structural plasticity of Merkel cell-neurite complexes in young adulthood. Finally, I developed a technique to perform time-lapse in vivo imaging of identified Merkel cells and afferent terminals over the course of a month. These structures were highly plastic, with afferent terminals undergoing frequent growth and regression, as well as both Merkel cells and terminal branches being added or removed. Together, these studies reveal that peripheral nerve terminals undergo a previously unsuspected amount of structural plasticity in healthy tissue.

In the mammalian brain, information is believed to be encoded at the cellular level through alterations in synaptic weights. Furthermore, changes in synaptic strength are correlated with structural changes at dendritic spines, such as growth and shrinkage, which may serve to shape inputs into functional domains and increase the computational power of neurons. Neuroanatomical alterations in dendritic spines have been described in humans with intellectual disability, further supporting the relationship between neuronal structure and function. Fragile X Syndrome (FXS) is the most common single-gene neurodevelopmental disorder, and a hallmark feature of this disorder is the increased density of long spines in several brain regions including the hippocampus. Identification of FXS spines as filopodia-like has led to the theory that these spines are immature, and that altered spine development underlies the cognitive dysfunction in this disorder. However, the functional capacity of the long spines observed in FXS is not well understood. For my thesis work, I used two photon imaging, glutamate uncaging and electrophysiology to perform a high-resolution characterization of dendritic spine structure, function, and plasticity in the hippocampus of the FXS mouse model in order to determine what gives rise to these alterations and how this contributes to the observed neuronal dysfunction in this disorder. From my dissertation research, I find that while Fmr1 KO neurons have region-specific alterations in both dendrite and spine morphology, the functional responses of single synapses in FXS mutant neurons are grossly normal. FXS spines respond proportionally to increased levels of glutamate release, and the linear relationship between structure and function is preserved at these synapses. In addition, structural plasticity, both growth and shrinkage, at single inputs is similar in magnitude to control neurons following synaptic potentiation and depression, respectively. However, upon more detailed examination of structural plasticity, either at single or multiple inputs, I find several deficits. First, following structural plasticity, I observe aberrant heterosynaptic plasticity in Fmr1 KO neurons, where unstimulated mutant spines located in close proximity to activated spines become significantly larger compared to neighboring spines in control neurons, which showed no significant change in size. Next, competition for mGluR-LTD does not occur in Fmr1 KO neurons, leading to an increase in spines that undergo spine shrinkage. I conclude from this work that while spine morphology is altered in FXS, spines develop with functional synapses that have the capacity to express bidirectional forms of structural plasticity. However, these spines undergo abnormal structural plasticity across stimulated inputs, leading to the expression of aberrant heterosynaptic structural plasticity. As activity is integrated across a dendritic branch, such excess plasticity observed in Fmr1 KO neurons could contribute to the altered spine morphology as well as cognitive dysfunction observed in FXS.

Comprised of six distinct layers, the neocortex is a key brain structure for many of our advanced cognitive abilities, ranging from sensation to decision making to movement. Each layer contains distinct cell types differing in their genes, biophysical properties, and connectivity with other parts of the brain. Yet how these diverse cortical layers and cell types contribute to any given behavior remains unresolved. Because sensory cortical areas have stereotyped anatomies and the six cortical layer organization is highly conserved across all mammals, understanding computations in one cortical area, such as the mouse barrel cortex within the primary somatosensory cortex, may inform us of computations being performed by similar microcircuits across the neocortex. This thesis is an investigation of cortical circuit function as it pertains to (1) distinct functional role of cortical layers in sensory discrimination, (2) increased cortical connectivity enhancing sensation, (3) a cautionary tale of selecting appropriate transgenic mouse lines for in vivo manipulations, (4) and the role of proprioception in the establishment of long-term visuospatial memory. Investigating layer-specific function first requires a cortex-dependent task. Yet, despite our extensive understanding of cortical anatomy and physiology, the contributions of different cortical layers to behaviors remain unknown. We developed a two-alternative forced choice paradigm in which head-fixed mice use a single whisker to either discriminate textures of parametrically varied roughness or detect the same textured surfaces. Lesioning barrel cortex revealed that texture discrimination, but not detection, was cortex-dependent. Paralyzing the whisker pad demonstrated that passive can rival active perception and cortical dependence is not movement-related. Transgenic Cre lines were used to target inhibitory opsins to excitatory cortical neurons of specific layers for selective perturbations. Discrimination required all layers, but deep layers (layers 5/6) were critical for accumulation of sensory evidence whereas superficial layers (layers 2-4) appeared to provide top-down motor input. This thesis shows that superficial layers contextually interpret sensory evidence to modify the deep layer output in behaviorally appropriate ways. Having identified distinct functional roles of deep and superficial layers through perturbation experiments, we next sought to enhance texture task performance by selectively activating texture-encoding neurons. However, given that all layers are involved in the task and the technical difficulties of targeting stimulus-selective cells, we turned to humanized mouse model (SRGAP2C) that exhibits increased local and long-range cortico-cortical connections and increased response selectivity to whisker stimulations in layer 2/3 pyramidal neurons in the barrel cortex. This thesis demonstrates that the increased cortico-cortical connectivity not only improved sensory coding accuracy in SRGAP2C mice, but the humanized animals trained on the texture discrimination task displayed increased learning rate and were more likely to learn the task compared to control. Next, we provide a cautionary tale of selecting appropriate mouse lines for in vivo experiments. Advances in optogenetics and transgenic Cre mouse lines enable us to probe the function of genetically defined neuronal populations, but transgene expression can adversely affect cell health and cause neural and behavioral abnormalities. We discovered learning impairments specific to cortex-dependent sensory discrimination behaviors in Emx1-Cre animals that express inhibitory opsins in excitatory cortical neurons. We suggest Nex1-Cre line as a more reliable and robust alternative to Emx1-Cre animals. The thesis highlights the importance of characterizing and selecting appropriate transgenic lines for in vivo optogenetic experiments.  In addition to touch, the primary somatosensory cortex processes other tactile information including temperature, pain, and proprioception. Creating a spatially accurate representation of the visual world requires transforming spatially inaccurate visual information coming from a constantly moving retina into a representation that can be used for accurate perception and action. This thesis shows that the dysgranular zone, the proprioceptive region of the primary somatosensory cortex, is required to establish long-term visuospatial memory.

This thesis documents three main projects performed during my PhD. Chapter 3 describes a published project in which detailed behavioral analysis based on machine learning approaches for pose-estimation were used to characterize a novel sensorimotor transformation in which mice use whisker information to rapidly modify their gait in order to rapidly avoid an obstacle in their path (Warren et al., 2021). I contributed to designing experiments, data collection and analysis related to this project spanning roughly from Aug. 2018 to Aug. 2019. Appendix 1 describes a follow-up study in which I performed multi-site silicon probe recordings and anatomical reconstruction of recording sites across the deep cerebellar nuclei in head-fixed mice performing the same obstacle avoidance behavior mentioned above. Data collection for this project spanned roughly from May 2019 to Jan. 2021. This data was initially analyzed in collaboration with Richard Warren and is currently being analyzed in collaboration with Ramin Kajeh in Dr. Larry Abbott’s group. Finally, Chapter 2 reports on the major independent work undertaken as part of my thesis, spanning from Sept. 2021 to present. As such, the Introduction relates solely to Chapter 2. The goal of this ongoing project is to extend the Sawtell laboratory studies of the mechanisms for sensory prediction and cancellation in the cerebellum-like circuitry of the electrosensory lobe (ELL) of electric fish to a cerebellum-like circuit in mammals, the dorsal cochlear nucleus (DCN) in the auditory brainstem. In particular, my work provides initial insights into the function of the cartwheel cell (CWC), a previously enigmatic cell type that occupies a similar place in the circuitry of the dorsal cochlear nucleus as the Purkinje cell of the cerebellum and the medium ganglion (MG) cell of the ELL. We have demonstrated that CWCs convey tonotopically-specific signals that are well-suited for canceling self-generated auditory responses in fusiform cells (FCs), the principal output cells in the DCN. Additionally, our findings reveal that the two characteristic types of spikes observed in CWCs—the axonal simple spikes (comparable to simple spikes in Purkinje cells and narrow spikes in MG cells) and dendritic complex spikes (similar to complex spikes in Purkinje cells and broad spikes in MG cells)—are distinctly modulated by both self-generated behavior and external acoustic stimuli, suggesting that these two types of spikes serve separate functional roles in the processing of the cancellation signal, as well as auditory information, within the DCN circuitry. This finding is consistent with the reported distinct functions of narrow and broad spikes in MG cells within the circuitry of the ELL, suggesting an evolutionarily conserved role of Purkinje-like cells in cerebellum-like circuits.