The intricate patterning and complex circuitry of the nervous system beg the question of how such precise and ordered connections are formed during development. Genetically encoded information certainly instructs coarse projections of neuronal processes, and it has long been appreciated that sensory experience can fine tune connections. However, many functional neuronal circuits are present prior to birth and therefore must be organized in the absence of sensory experience. What instructs the formation of these circuits?

Spontaneous activity is prevalent in the developing nervous system and may play a role in organizing functional circuits. Spontaneous activity has been most extensively studied in the developing retina, where bursts of action potentials propagate among neighboring ganglion cells in a wave like fashion. These `retinal waves' exist for an extended period of time prior to vision during which connections between retinal projection neurons and target regions of the brain are being refined. It is thought that information relevant for the refinement of these connections is provided in the spatial and temporal patterns of retinal waves.

In my thesis work, I set out to determine how the developing retina generates the distinct spatial and temporal patterns of retinal waves. A population of interneurons called starburst amacrine cells generates retinal waves during the first week after birth in mice. These starburst cells release acetylcholine onto neighboring starburst cells and retinal ganglion cells. How does this excitatory network generate precise patterns of activity?

Using a combination of calcium imaging, electrophysiology, and a cell-based sensor of acetylcholine, I characterized the features of the developing starburst cell network that give rise to the initiation, propagation, and termination of retinal waves. I incorporated these features of the starburst network into a computational model to explain how the spatial and temporal features of waves rely upon the properties of the underlying network. Finally, I described the molecular mechanisms of a key feature in starburst cells, a minute long slow afterhyperpolarization, that sets the frequency and spatial coverage of waves. These findings provide a detailed understanding of how spatial and temporal patterns of spontaneous activity are generated in the retina, and provide a basis for further experiments to determine the relevant information within these patterns that instruct the wiring of visual circuits.

Last, I provide evidence that the slow afterhyperpolarizartions in starburst amacrine cells are mediated by two-pore potassium channels, which are traditionally thought to be leak channels. Our findings suggest that slow variations of second messengers, including cAMP and IP3, are read out by two-pore channels to alter neuronal excitability. These findings would have relevance to similar conductances found throughout the nervous system.

Across the developing nervous system, immature networks generate spontaneous activity that is highly correlated amongst neighboring cells, which is required for the correct establishment of adult neural circuits. Remarkably, correlated activity persists following disruption of the underlying circuits that mediate it, indicating that plasticity mechanisms exist to ensure correlated activity is maintained. Here, we examine this phenomenon in the developing mouse retina, where correlated activity is mediated by cholinergic transmission and propagates across the retina as a wave. The absence of cholinergic signaling leads to the generation of "recovered" waves that propagate through a distinct, gap junction mediated circuit. Our findings show that stimulation of melanopsin-expressing intrinsically photosensitive retinal ganglion cells (ipRGCs) facilitates recovery of correlated activity in the absence of cholinergic waves, which results in the emergence of a light-sensitive network. We tested whether pharmacological blockade of cholinergic waves altered retinal light-response properties. We observed an increase in the duration of light-evoked activity and number of light-responsive cells, which arose from signaling via gap junctions. These observations suggest that electrical coupling of ipRGCs increases in the absence of cholinergic input, allowing melanopsin-driven signals to propagate to other neurons. Furthermore, we show that light-sensitive waves are strongly modulated by dopamine signaling--a potent neuromodulator of gap junction coupling. We determine that this light-sensitive wave circuit is present but latent in wild type retina, where it is usually suppressed by a combination of cholinergic and dopaminergic signaling. Our observations indicate that dopamine signaling acts as a switch for network reconfiguration, where high dopamine silences the light-sensitive, gap junction coupled network under cholinergic waves and reduced dopamine activates it in the absence of cholinergic waves. We conclude that the wiring diagram of the developing retina includes several overconnected circuits, in which some circuits are closed and others activated depending on the internal state of the system.

Throughout the developing nervous system, spontaneous oscillatory patterns of activity have been observed. The developing murine retina is no exception, where spontaneous activity manifests as spatially correlated waves of depolarizations. These retinal waves propagate between neighboring neurons within retinal layers during the two postnatal weeks just prior to eye-opening and development of the light response. Waves are necessary for the normal patterning of connections of the retinal projections to their primary targets in the brain, indicating they contribute to a general self-organizing principal of neural development.

There are three distinct circuits that mediate these waves at different developmental stages: gap junction mediated waves that occur around birth, acetylcholine-receptor mediated that occur during the first postnatal week, and glutamate-receptor mediated that occur between postnatal day 10 (P10) and P13. At this age, retinal circuitry appears adult-like such that excitatory synapses connect neurons across layers but inhibitory synapses connect neurons within layers. Thus it remains a mystery as to how depolarization propagates between neighboring within-layer neurons during waves.

This dissertation addresses the cellular and synaptic basis underlying the initiation and propagation of glutamatergic retinal waves. Using a combination of multielectrode array recording and two-photon calcium imaging, I identified a role for electrical coupling, diffuse release of glutamate, and inhibitory circuits in the retinal wave generation. In addition, I studied the role of endogenous copper in modulating retinal waves. This research represents a near-complete description of the intraretinal circuits that underlie the spatial and temporal properties of glutamatergic retinal waves.

Spontaneous activity is a hallmark of developing neural systems. In the developing visual system, patterned spontaneous activity begins in the retina, where waves of depolarizations, called retinal waves propagate across a network of neurons. Retinal waves propagate laterally across the retina, but also downstream in the visual circuit through the dorsal lateral geniculate nucleus, the superior colliculus (SC), and the primary visual cortex. Throughout the development of the mouse visual system, retinal waves are observed in three stages starting from embryonic day 16 (E16) to eye opening, corresponding to postnatal day 13 (P13). The circuits that mediate each stage of retinal waves change throughout development. Stage 2 waves are observed between P1-P10 and are initiated by the release of acetylcholine (ACh) from starburst amacrine cells (SACs), leading to the activation of nicotinic acetylcholine receptors (nAChRs). They propagate through a network of SACs that themselves express nAChRs during development. Stage 3 waves are observed between P10-P14 and are initiated by the release of glutamate from bipolar cells, leading to the activation of ionotropic glutamate receptors. Stage 3 waves propagate through a network of bipolar and amacrine cells via glutamate transmission and gap junction coupling. Stage 1 waves are observed between E16-P0 and the initiation and propagation mechanisms for Stage 1 waves have been thought to mainly rely on gap junction coupling between retinal ganglion cells (RGCs), unlike the other two stages the details of their initiation and propagation mechanisms are poorly understood.

The focus of this dissertation is to describe the spatiotemporal properties and circuit mechanisms of Stage 1 waves, identify how distinct these waves are from Stage 2 waves, and then begin to probe their function in the development of the retina. In Chapter 1, I summarize the current understanding of the mechanisms shaping the spatiotemporal properties of all stages of retinal waves. In Chapter 2, I used a custom-built macroscope and a two-photon microscope to perform population calcium imaging in embryonic retinas. I first characterized the spatiotemporal properties of Stage 1 waves. Then, using genetic and pharmacological manipulations, I determined the relative role of gap junctions and nAChRs in mediating Stage 1 waves. Stage 1 waves initiate at several locations across the retina and propagate across finite regions of varying areas. Additionally, I found that waves that spanned a large retinal surface area were more sensitive to nAChRs and gap junction antagonists, as they were either abolished or significantly reduced in frequency in the presence of those pharmacological agents. I also found that waves in mice lacking the 2 subunit of nAChRs (β2-nAChR-KO) exhibited reduced propagation areas. The application of a general nAChR antagonist in the β2-nAChR-KO had no effect on wave properties. However, the application of a gap junction antagonist greatly reduced both the area and frequency of waves, suggesting a compensatory role for gap junctions in the β2-nAChR-KO.

The disruption of Stage 1 and 2 waves in the β2-nAChR-KO allowed for further investigation into the role of retinal waves in regulating the maturation of RGCs. We specifically investigated a subpopulation of RGCs, called intrinsically photosensitive RGCs (ipRGCs). Using immunohistochemistry, we observed a significant reduction in ipRGC density between P1 and P7 retinas in both WT and β2-nAChR-KO mice, suggesting that ipRGC cell death observed in the first postnatal week of development occurs independently of normal retinal waves. Moreover, there was no significant change in density or spatial distribution between WT and β2-nAChR-KO mice at P1 and P7, also indicating this developmental step occurs independently of normal waves. In Chapter 3, I discuss the implications of these results on our current understanding of retinal waves. Specifically, I highlight newly found key similarities and differences between Stage 1 and 2 waves, and future experiments necessary to fully understand the distinction between them.

The generation of new neurons in the hippocampus in adult mammals, or adult hippocampal neurogenesis, is a potent source of plasticity in the brain. Adult neurogenesis is highly regulated by the organism’s environment, and stress is a powerful regulator of neurogenesis. Understanding the molecular basis of this stress-induced regulation helps us better appreciate the mechanisms and functional relevance of this fascinating process. Neurogenesis depends on the development of neural stem cells, which are tightly controlled by their surrounding cellular microenvironment, especially by a cell type of the brain that has shown to be highly important in regulating brain function, the astrocyte. The following studies describe how astrocyte-based signaling imparts key molecular regulatory mechanisms that are modulated by stress in the adult rat. Chapter 2 describes how an acute stressor enhances neural stem cell proliferation via astrocytic FGF2 signaling. Chapter 3 describes the suppression of stem cell proliferation by chronic stress via a late-onset astrocytic increase of BMP signaling. Taken together, these studies further advance our understanding of stress modulation of brain plasticity, and take into consideration an understudied cell type in stress biology, the astrocyte.

The mouse brain contains over 70 million neurons that form around a trillion connections, or synapses, with one another. Different synaptically connected networks of neurons are organized with varying circuit motifs to receive information and transform that information into a relevant output for the animal. The visual system, and in particular the retina, has served as a classic model for elucidating how neuronal circuits perform these computations. The focus of this dissertation was to understand how the inter-neuronal and sub-neuronal aspects of the direction selective circuit in the retina develop and operate to extract the directions of object motion from the milieu of information in the visual scene. The key neuron in this circuit is the starburst amacrine cell. The dendrites of starburst cells exhibit specific responses to motion in the visual scene that depend upon the dendrite’s orientation within the retina. These signals are then propagated through a stereotyped asymmetric arrangement of synaptic connections from multiple starburst cells onto direction selective ganglion cells, which transmit this directional information to the brain. I begin by summarizing the current understanding of how the direction selective circuit develops and operates in Chapter I. Then, experimental evidence is provided that seeks to answer outstanding questions regarding the development of direction selective circuit components and their roles within the mature circuit. First, in Chapter II, I investigate how the asymmetric synaptic connections between starburst cells and direction selective ganglion cells develop. I find that the mature direction selective circuit wiring diagram results from a subcellularly specified synaptogenesis program, rather than synaptic pruning or strengthening. Next, in Chapter III I investigate how the direction selective responses within the starburst cell dendrites are generated. I find that a skewed distribution of excitatory inputs onto the dendrite is sufficient to produce these responses. In Chapter IV, I explore how starburst cell dendritic morphology influences direction selective circuit computations. I find that starburst cell morphology aligns the subcellular starburst cell computation with the appropriate post-synaptic ganglion cell and that the dense plexus of starburst dendrites is required to produce directionally discriminate responses in the ganglion cell. In addition, I find that the subcellular directional computation in starburst cells is performed at the end of the bipolar cell input distribution along the dendrite identified in Chapter III. In Chapter V I seek to identify the molecular determinants of the asymmetric wiring program identified in Chapter II. These preliminary results will provide a basis for further exploration of direction selective circuit development. Together, these results reveal how the morphology, connectivity, and biophysical properties of neurons are arranged during development to generate the amazing computational abilities of neuronal circuits.