Neurons in the mammalian central nervous system must be maintained throughout the lifetime of an organism because they are unable to regenerate. Retinal ganglion cells (RGCs) are the projection neurons of the retina that project directly to and form connections with the brain. Injury and disease can damage RGC axons, ultimately leading to RGC death and resulting in blindness. Learning how to protect these cells from dying and how to regenerate them after loss remain critical goals of vision neuroscientists; these endeavors aim to prevent blindness in those at risk and to restore vision in individuals already blind or visually impaired. Transcription factors regulate the development, maintenance, and survival of neurons. In this dissertation, I explore the roles of two transcription factors, Tbr2 and Isl2, in these important processes. My lab has previously demonstrated that Tbr2 is required during development for the formation and survival of intrinsically photosensitive RGCs (ipRGCs), a unique class of RGCs that are resilient to injury. The persistent expression of Tbr2 in a subset of adult RGCs, including all ipRGCs, led to the hypothesis that it is required for their maintenance and survival. In this dissertation, I establish that Tbr2 is indeed necessary for maintaining the defining aspect of ipRGC identity–the expression of the photopigment melanopsin, which enables them to directly respond to light. However, Tbr2 is not essential for the survival of adult ipRGCs. Additionally, I show that after optic nerve injury, ipRGCs lacking Tbr2 continue to survive better than non-ipRGCs, albeit to a lesser extent than wildtype ipRGCs. The function of Isl2 in RGC development has not previously been elucidated because Isl2 knockout mice die shortly after birth. Therefore, I designed an Isl2-flox mouse and subsequently generated Isl2 conditional knockout mice in which Isl2 is specifically removed from the retina. I demonstrate that Isl2 is required for the survival of Isl2-expressing RGC subtypes. My work uncovers important roles for Tbr2 and Isl2 in RGC development, maintenance, and survival. Additionally, the generation of a mouse with an Isl2 conditional knockout allele provides a new tool that can be used to study the development of other cell types in which Isl2 is expressed.

The retina detects the visual scene and coverts the information into separate channels of information. These channels then connect to diverse areas in the brain that initiate appropriate visually responsive behaviors. Retinal ganglion cells (RGCs) are the output neurons of the retina and come in ~30 different types, each with distinct receptive eld properties generated by circuits within the retina. Each RGC type processes a distinct feature of the visual scene making up a unique channel of visual information and tranferring it to the appropriate area of the brain. Intriguingly, all the RGC types arise from one common RGC progenitor cell. One of the questions in vision neuroscience is how does this diversity arise during development? Further, once the different RGC types have been speci ed, how do they maintain their identity? Answers to these questions will shed light on the mechanisms that set up the parallel channels of visual information and how visual circuitry is established.

Studies on the gene regulatory network that specifies RGCs have so far yielded several transcription factors that are involved. However, many other transcription factors could be involved in specifying, and then maintaining, the diversity of RGC types. A transcription factor called Tbr2 has been demontrated to be required for specifying a subset of RGCs. In my thesis work I have tested the hypothesis that

Tbr2 is also required for maintaining the identity of that RGC subset. Through loss of function and ectopic expression experiments in adult mice I demonstrate that Tbr2 has an epistatic relationship to melanopsin and is required to maintain the melanopsin- expression aspect of RGCs.

Another aspect of RGC identity is the response to axonal damage. Some RGC types are able to withstand axonal injury and resist death, while other types activate pro-apoptotic pathways within a very short time of the injury and die. The mechanisms responsible for this variation in response to injury is an important factor in understanding the pathogenesis of glaucoma, which is a disease that affects the retinal ganglion cells. There is evidence that the OFF RGCs are more susceptible to apoptosis following optic nerve injury, but still more work needs to be done in classifying RGC types and their response after damage. Melanopsin- expressing RGCs have also been found to survive better than other types following optic nerve injury. Considering that the transcription factor Tbr2 is expressed in all melanopsin RGCs, and that Tbr2 is required for RGC survival during development, I tested the hypothesis that Tbr2-expressing class of RGCs survive better following optic nerve injury as compared to other RGC types. My results show that Tbr2-expressing RGCs do survive better than other RGC types, however not due to Tbr2 expression. This implies the neuroprotective aspect of Tbr2- expressing RGCs is regulated by other factors, and future RNA-seq experiments will help reveal the genes involved.

The retina extracts relevant features from the visual scene and transmits these features to the brain through separate pathways that will eventually result in the perception of sight. The retinal ganglion cells (RGCs) are the only retinal cell type to send an axonal projection to the brain. This indicates that the signals generated by the RGCs are the end result of retinal processing, and the features detected by the RGCs are all that will be transmitted to the brain about the visual environment. Each RGC type represents a unique pathway that detects specific visual features. RGCs of the same type tile the retina so that the entire visual field is sampled. Different types of RGCs overlap so that each pixel in the visual field gets sampled by each pathway. To understand the different retinal pathways and what gets sent to the brain, it is necessary to know what types of RGCs are in the retina. Current classifications have used morphologies and physiology to describe the different RGC types. It is estimated that 20 different types of RGCs are present in the mammalian retina. However, very little is known about how these morphological features affect functional properties due to the inability to perturb the morphology and observe changes in physiology in many mammals. In order to overcome this limitation, I utilized a large-scale multielectrode array (MEA) approach to record and characterize responses from hundreds of RGCs in the mouse retina. The mouse model is advantageous due to the ease of genetic manipulation, allowing me to disrupt morphology and relate the changes to function. I first characterized and classified RGCs in the wild-type retina to develop a reference for mutant comparisons. I was able to classify up to 8 types of RGCs along with functional properties of the classes, such as tiling arrangements of the RGC receptive fields (RFs). Using a mutant mouse that has defects in dendritic and cell body spacing, I show that the dendritic structure is important for RF tiling and direction selective responses thus showing how morphology affects function. Finally, I used a transgenic mouse, in which RGCs expressing a particular gene was ablated, to show that these eliminated RGCs were a distinct functional subset that responded to light offset. The work that I performed will contribute to a complete classification of RGCs by linking the morphological types to the functional types, as well as the genetic programs that establish their properties. This will be necessary to determine what features are detected by the retina and how they ultimately lead to behavior.

AbstractNeuron Activation in Response to Auditory Stimulation Gabriela Llarena Gaining genetic access to neurons allows us to study the structure and function of the mammalian auditory system. Audition captures sound stimuli detected from the environment and sends this information to the brain through the auditory pathway. The information gathered by the auditory system is integrated with information gathered from the other sensory modalities: vision, olfaction, touch, and gustation; this allows us to modify our behavior accordingly and is crucial for survival. Methods to identify neuron populations based on their response properties are limited, as most genetic manipulation experiments identify neurons based on their anatomy or genetic composition. Neuroscientists have developed a novel tool that genetically targets different neuron populations based on their functional criteria. The fos/TRAP2 method, or “Targeted Recombination of Active Populations” (TRAP), genetically targets populations of auditory-responsive neurons when exposed to an auditory stimulus (Guenthner, Casey J, et al., 2013). Using a transgenic mouse line of Fos2A-iCreERT2, or TRAP2, knock-in mice, I designed an auditory stimulation experiment that utilizes a cre-inducible system in cells that express fos in the presence of 4-Hydroxytamoxifen to identify and locate the auditory-responsive neuron populations. I found that mice exposed to an auditory stimulus expressed greater numbers of active cells in areas receiving auditory inputs, such as the primary auditory cortex (A1), superior colliculus (SC), and inferior colliculus (IC), than in mice that had their ears plugged; in the primary somatosensory cortex (S1), both experimental conditions showed similar numbers of active neurons. These results demonstrate how the fos/TRAP2 method could serve as a tool to label neurons that respond to auditory stimuli; this method uses the effector gene R26tdTomato to fluorescently label cells, which can be analyzed using immunohistochemistry. The fos/TRAP2 method and analyses can provide insight into different auditory-responsive neuron populations; it can be used to determine which neuronal markers the cells express to identify their type, such as whether they are excitatory or inhibitory neurons, and the cells’ defining characteristics. Additionally, this method can be used to sort cells for RNA sequencing (RNA Seq) analysis or to induce expression of optogenetic tools used to assay auditory circuitry. Identifying the function of different auditory-responsive neuron populations can be used to further investigate the auditory pathway and sensory integration to better understand how mammals detect and process sound.

Abstract

James A. Shanks

Cortical-Thalamic Axons are Required for Retinal Ganglion Cell Targeting to the Mouse Dorsal Lateral Geniculate Nucleus

The human brain contains over 85 billion neurons, which make trillions of synapses in a very ordered and stereotypical manner (Williams and Herrup, 1988). The human cerebral cortex, which contains over 20 percent of the total number of neurons within the brain, is responsible for critical functions such as memory, attention, perception, awareness, language, thought, and consciousness, and it dedicates over 30 percent of the total area to the process of vision, 8 percent for touch, and 3 percent for hearing (Grady et al., 1993). One major question in the field of neuroscience is: how do neurons navigate the extremely complex environment of the developing brain that contains an endless amount of possible synaptic partners and continually, precisely, and stereotypically find their appropriate synaptic targets?

Many disorders that affect the human population can be attributed to the loss of the precise wiring within the brain, including mental retardation, schizophrenia, and autism (Tye and Bolton, 2013; Kana et al., 2011; Huttenlocker, 1991; Buchmann et al., 2014). Recent studies that utilized fMRI to investigate brain activity in people who suffer from schizophrenia, a disorder that renders the inflicted an inability to perceive what is real, show a distinct loss of cortical activity in the visual thalamus resulting from a loss of thalamo-cortical circuitry. Elucidating the mechanisms responsible for the proper wiring of the brain is of paramount importance, and could lead to possible remedies for afflictions such as schizophrenia.

The mouse visual system has become an increasingly popular model to study the development of neural circuits given the vast genetic tools and the ease with which the output neurons of the retina, retinal ganglion cells (RGCs), can be labeled. Multiple distinct targets of RGCs have been well characterized and many behaviors are associated with these nuclei that can be easily assayed to detect for dysfunction (Sweeney et al., 2014; Osterhout et al., 2015). It has been demonstrated that activity is necessary for the refinement of RGCs in the dLGN during development and the proper placement of RGCs within eye specific domains is reliant on specific receptor tyrosine kinases (Rossi et al., 1988; Pfeifenberger et al., 2005). Mounting evidence suggests that specific transmembrane adhesion molecules are necessary for the proper targeting of RGCs to specific visual nuclei responsible for multiple non-image forming behaviors, including circadian rhythms, the pupillary light reflex, and image stabilization (Osterhout et al., 2011; Osterhout et al., 2015; Sun et al., 2015; Su et al., 2011). Molecules necessary for the proper targeting of image forming nuclei including the dLGN and the superior colliculus (SC), a midbrain target necessary for reflexive head and eye movements, remain elusive.

While the retina sends projections to specific subcortical nuclei, many of these same nuclei also receive cortical input from either layer 5 or layer 6 of the visual cortex (V1). The timing and patterns of innervation of cortical layer 6 neurons in the dLGN has recently been elucidated (Seabrook et al., 2013). These data indicate that the removal of RGC axons from the dLGN during development allows cortical neurons to innervate the dLGN at earlier time points, suggesting RGC axons control the timing of innervation. Whether cortical neurons affect the innervation patterns of RGCs is not well understood.

In order to better understand the role of V1 input during the development of the mouse visual system, I utilized a transgenic mouse harboring a floxed conditional allele for a splicing factor that when removed with a cre-recombinase driver line, only expressed in cortical tissue, results in the almost complete loss of the mouse neo-cortex. Surprisingly, the animal is viable and thus allowed me to investigate the necessity of V1 innervation during the development of the mouse visual system including innervation patterns of RGCs within multiple subcortical nuclei. I was also able to investigate well known non-image forming behaviors, including the pupillary light reflex, to test for the necessity of cortical layer 5 input, and circadian rhythms to better understand if the cortex modulates sleep wake cycles. Also, I tested whether mice lacking cortical input and almost complete loss of RGC inputs to the dLGN can perceive light and whether this perception can be used to locate a cue in a complex environment.