The normal aging process is accompanied by mild declines in visual function in both humans and non-human primates, independent of ocular deficiencies, indicating an involvement of structures in the central visual pathway. Studies examining loss of cortical neurons as a potential explanation have concluded that neuron numbers are largely preserved with age both in visual cortex and as well as other cortices. In contrast to the stability of neuron numbers with age, a significant increase in the total number of glia was found in the infragranular layers of the primary visual cortex, in rhesus monkeys (Giannaris and Rosene, 2012). Unpublished data indicates that increase in glial density is correlated with decreased visual function assessed by the behavioral performance metric. In order to understand the basis of glial increase with age in the rhesus monkey, we used immunohistochemistry to parcellate the total number of glia in primary visual cortex into three subtypes: microglia with Iba1, astrocytes with GFAP and oligodendrocytes with Olig2. These were then quantified using unbiased stereology in a subset of 12 animals whose ages ranged from young to old (6 male and 6 female), from the original study of 26 monkeys. Adding together all three subtypes in the current subset of animals showed a modest but non-significant trend toward the increase observed in the larger sample of 26 animals. In this study, examining the three subtypes showed no significant increase and the total number of glial cells was found to be unchanged with age. A definitive answer to how the different subtypes contribute to the overall increase in glia will require analyzing the remainder of the full data set.

The olfactory epithelium (OE) possesses the rare capacity among neuronal tissues to regenerate throughout life. As a result, progenitor cells continuously proliferate and differentiate into olfactory sensory neurons (OSNs) that project their axons to the olfactory bulb (OB) where they establish connections to the central nervous system. The olfactory epithelium is therefore an attractive model for the study of axonal growth and synapse formation. The present set of studies attempts to provide insights into synapse formation and axonal development of olfactory sensory neurons. First, I sought to understand the regulation of expression of pre-synaptic molecules in the olfactory epithelium. I established by in situ hybridization that as OSNs mature, they express sequentially groups of pre-synaptic genes. Genes encoding for proteins that play a structural role at the active zone showed an early onset of expression, whereas genes encoding for proteins associated with synaptic vesicles showed a later onset of expression. In particular, the signature molecule for glutamatergic neurons VGLUT2 shows the latest onset of expression. The sequential onset of expression suggests the existence of discrete steps in pre-synaptic development. In addition, contact with the targets in the olfactory bulb is not controlling pre-synaptic protein gene expression, suggesting that olfactory sensory neurons follow an intrinsic program of development. Second, in order to visualize simultaneously OSN axonal arborizations and their pre-synaptic specializations in vivo, I developed a method based on post-natal electroporation of the mouse nasal cavity. This technique allowed me to perform a temporal study where I followed the elaboration of axons and synapses in olfactory sensory neurons at different post-natal ages. The results show that olfactory sensory axons develop with exuberant growth and synapse formation. Exuberant branches and synapses are eliminated to achieve the mature pattern of connectivity in a process likely to be regulated by neural activity.

Drosophila larval and adult stage forms are very different. Drosophila larvae move using undulatory body muscle contractions, while adult flies walk and fly using legs and wings. Leg motor neurons control multi-jointed leg movement by coordinately regulating leg muscle contractions. With the aim of understanding how Drosophila leg motor neurons are specified, in Chapter One, I give a general introduction of the mechanisms that regulate motor neuron generation and specification. In Chapter Two, I show that adult Drosophila leg motor neurons are mostly generated de novo during larval stages in a lineage dependent manner. Although leg motor neurons are born from 11 lineages, nearly two thirds of leg motor neurons are born from two major lineages: Lin A and Lin B. I describe the individual leg motor neuron birth orders, axonal and dendritic morphologies by using single cell labeling methods. Each motor neuron that has unique axonal targeting and dendritic architecture is born in a stereotypic birth order from a specific lineage. Leg motor neurons targeting similar muscles share dendritic territory in the CNS and subsequently form a dendritic myotopic map in the CNS. These findings provide critical information about how individual leg motor neurons are generated, and how individual leg motor neuron axons and dendrites look like. In Chapter Three, I describe the results of a candidate gene approach. In vertebrate systems, Hox genes and Hox cofactors regulate spinal cord motor neuron identity. Although much work has been done addressing the function of Hox genes and Hox cofactors in vertebrate motor neuron development, the function of Hox genes and Hox cofactors in motor neuron dendritic arborization has not been clearly addressed. With this aim in mind, I describe the function of Hox genes and Hox cofactors in Drosophila leg motor neuron development by removing Hox genes and Hox cofactors in both entire lineages and individual motor neurons. I show that Hox genes and Hox cofactors are required for motor neuron survival, and proper axonal and dendritic targeting. In Chapter Four, I discuss about how segmental and temporal identities of leg motor neurons are specified and how the axonal targeting of leg motor neurons at the early stage is achieved. Finally, in the Appendix, I show my attempts to find leg motor neuron specific molecular markers.

This thesis is the combination of two separate lines of work linked by one common goal: understanding the nervous system as an information-processing machine. David Marr (1982) put forth the idea that in order to fully understand an information-processing machine one must understand it at three separate levels. The computational goal of the system must be understood separately from the algorithm by which it is computed and the hardware in which it is computed. During my time as a graduate student I have been fortunate enough to work on two different levels in two very different systems. Chapter 1 focuses on the hardware of neural circuitry, specifically on how inhibitory interneurons connect to excitatory neurons. Chapter 2 focuses on the algorithmic problem of how flies could use gyroscopic sensors to calculate angular velocity.

For a wide variety of organisms on the planet, the sense of smell is of critical importance for survival. The mouse olfactory system mediates both learned and innate odor-driven behaviors, including activities as diverse as the localization of food sources, the avoidance of predators, and the selection of mates. How a chemical stimulus in the environment ultimately leads to the generation of an appropriate behavioral response, however, remains poorly understood. All of these behaviors begin with the binding of an odorant in the external environment to receptors on sensory neurons in the olfactory epithelium. These sensory neurons transmit this odor information to neurons in the olfactory bulb via spatially stereotyped axonal projections, and a subset of these bulbar neurons, mitral and tufted cells, in turn transmit this information to a number of higher brain regions implicated in both learned and innate odor-driven behaviors, including the piriform cortex and amygdala. Previous work has revealed that odorants drive activity in unique, sparse ensembles of neurons distributed across the piriform cortex without apparent spatial preference. The patterns of neural activity observed, however, do not reveal whether mitral and tufted cell projections from a given glomerulus to piriform are segregated or distributed, or whether they are random or determined. Distinguishing between these possibilities is important for understanding the function of piriform cortex: a random representation of odor identity in the piriform could accommodate learned olfactory behaviors, but cannot specify innate odor-driven responses. In addition, behavioral studies in which the function of the amygdala has been compromised have found that innate odor-driven behaviors are disrupted by these manipulations while learned odor-driven behaviors are left intact, strongly suggesting a role for the amygdala in innate olfactory responses. How odor information is represented in the amygdala, as well as the amygdala's exact role in the generation of olfactory responses, however, remain poorly understood. We therefore developed a strategy to trace the projections from identified glomeruli in the olfactory bulb to these higher olfactory centers. Electroporation of TMR dextran into single glomeruli has permitted us to define the neural circuits that convey olfactory information from specific glomeruli in the olfactory bulb to the piriform cortex and amygdala. We find that mitral and tufted cells from every glomerulus elaborate similar axonal arbors in the piriform. These projections densely fan out across the cortical surface in a homogeneous manner, and quantitative analyses fail to identify features that distinguish the projection patterns from different glomeruli. In contrast, the cortical amygdala receives spatially stereotyped projections from individual glomeruli. The stereotyped projections from each glomerulus target a subregion of the posterolateral cortical nucleus, but may overlap extensively with projections from other glomeruli. The apparently random pattern of projections to the piriform and the determined pattern of projections to the amygdala are likely to provide the anatomic substrates for distinct odor-driven behaviors mediated by these two brain regions. The dispersed mitral and tufted cell projections to the piriform provide the basis for the generation of previously observed patterns of neural activity and suggest a role for the piriform cortex in learned olfactory behaviors, while the pattern of mitral and tufted cell projections to the posterolateral amygdala implicate this structure in the generation of innate odor-driven behaviors. We have also developed high-throughput methods for imaging odor-evoked activity in targeted populations of neurons in multiple areas of the olfactory system to investigate how odor information is represented and transformed by the mouse brain. We have used a modified rabies virus that drives expression of GCaMP3, a calcium-sensitive indicator of neural activity, to image odor-evoked responses from mitral and tufted cells, as well as a modified adenoassociated virus that drives expression of GCaMP3 to image odor-evoked responses from neurons in piriform cortex. These imaging methods have permitted us to examine odor-evoked responses in a transgenic mouse where 95% of sensory neurons express a single kind of olfactory receptor (M71). In these mice, there is a 1,000-fold increase in sensory neurons expressing the M71 receptor ligand acetophenone, and a 20-fold reduction in neurons expressing olfactory receptors from the endogenous repertoire. These M71 transgenic mice provide a useful tool for examining the role that the normally stereotyped pattern of sensory neuron input to the bulb plays in olfactory processing, as well as how odor information is transformed as is moves from the sensory periphery to the cortex. In control mice, odors evoke activity in unique ensembles of spatially distributed, narrowly tuned mitral and tufted cells, and the number of cells responding to odor increases linearly with stimulus concentration. Surprisingly, despite the fact that there is a significant decrease in sensory neuron activity in response to odors other than acetophenone in M71 transgenics, a wide variety of odorants are able to evoke mitral and tufted cell activity in these mice. Furthermore, the number of cells responding to these odors as well as the magnitude of these odor-evoked responses are higher in M71 transgenics compared to controls. However, despite a massive increase in acetophenone-evoked sensory neuron input to the bulb in M71 transgenics, mitral and tufted cell responses to acetophenone are similar in M71 transgenics and controls. Our results provide evidence for excitatory mechanisms that amplify weak sensory neuron input as well as inhibitory mechanisms that suppress strong, pervasive odor-evoked input, suggesting that a major role of the olfactory bulb is to aid in the comprehensive detection and refinement of olfactory signals from the environment. Despite the fact that the representation of odor in the olfactory bulb of M71 transgenic mice differs from that observed in controls, we find the representations of odor in the piriform cortex of M71 transgenic mice and controls is quantitatively indistinguishable. Our results suggest that circuits intrinsic to the piriform significantly transform the representation of odor information as it moves from the olfactory bulb to the piriform cortex. Moreover, in comparison to the olfactory bulb, the piriform encodes odor in a more sparse, distributed manner within a much narrower dynamic range. The nature of the representation of odor we observe in piriform cortex further supports a role for this area in mediating odor discrimination and associative odor-driven behaviors. The work described in this thesis has provided insight into the way odor is represented in several areas of the mouse olfactory system, clues about how odor information is transformed as it passes through the brain, and the role that different areas of the olfactory system play in odor-driven perception and behavior. In the future, the novel techniques and methods described in this thesis can be applied to the study of many different areas of the mammalian brain, giving our work the potential to have a significant impact on our understanding of how patterns of neural activity may ultimately underlie the generation of perceptions, emotions, and behaviors.

While much progress has been made in understanding the development, differentiation, and organization of the spinal motor system, the complex circuitry that is integrated to determine a motor behavior has yet to be fully understood. The activity of motor neurons is influenced by sensory feedback, excitatory and inhibitory interneurons, and supraspinal control from higher brain regions in the CNS. Descending pathways from the cortex and midbrain are involved in the control of voluntary motor output. This is made possible by their projections onto spinal interneurons and, to a degree that varies between species, directly onto motor neurons. However, the somatotopic organization and molecular diversity of supraspinal projection neurons, and the circuitry that underlies their contribution to motor output, remain incompletely understood. The evolutionary emergence of direct descending projections onto motor neurons has been considered to reflect a specialized level of organization for precise control of individual forelimb muscles. Unlike their polysynaptic counterparts, monosynaptic connections represent direct, unfiltered access to the motor neuron circuit. The direct circuit is thought to represent a neural specialization for the increase in fractionated digit movements exhibited by primates and humans. The progressive realization that rodents have a greater degree of manual dexterity than was previously thought has evoked renewed interest in the role of direct supraspinal projections in other mammalian species. Lesion studies in the rodent indicated that, of the two major supraspinal pathways involved in the control of voluntary movement, the rubrospinal tract had a greater role in control of distal forelimb musculature. However, the degree to which this reflected direct projections onto motor neurons was not clear. Earlier anatomical tracing studies in the rat indicated that there are close appositions between labeled rubrospinal axons and motor neurons projecting to intermediate and distal forelimb muscles. To confirm that these contacts correspond to synapses, I developed a viral tracing strategy to visualize projections from the midbrain. Using an established technique of high-magnification confocal imaging combined with co-localization of the rubrospinal synaptic terminal marker, vglut2, I established the existence of monosynaptic connections from the ventral midbrain at the level of the red nucleus onto a restricted population of forelimb motor neurons at a single spinal level (C7-C8) in the rodent. To determine whether the motor neurons that receive synaptic input correspond to specific motor pool(s), I first established a positional map of forelimb muscle motor pools in the cervical enlargement of the mouse spinal cord. A single motor pool, that which innervates the extensor digitorum muscle, appeared to be situated in the dense dorsolateral termination zone of rubrospinal ventral fibers. The extensor digitorum muscle plays a key role in digit extension and arpeggio movements during skilled reaching. Anterograde labeling of rubrospinal descending fibers combined with retrograde labeling of extensor digitorum motor neurons revealed a direct circuit from the red nucleus onto this population of motor neurons. Surprisingly, neighboring motor pools innervating digit flexor muscles did not receive rubrospinal inputs. Moreover, other modulatory inputs onto motor neurons, including corticospinal, proprioceptive, and cholinergic interneuron afferents did not distinguish between extensor and flexor digitorum motor neurons. My data therefore reveal a previously unrecognized level of motor pool specificity in the direct rubrospinal circuit. The identification of a small number of rubrospinal fibers that project onto extensor digitorum motor neurons suggested a considerable degree of heterogeneity between rubrospinal neurons. I therefore investigated the anatomical and molecular organization of subpopulations of rubrospinal neurons using retrograde labeling to identify subpopulations of rubrospinal neurons projecting, respectively, to cervical and lumbar levels of the spinal cord. Two rubrospinal populations could be identified within the red nucleus: a rostral population of intermingled cervical and lumbar projection neurons which express the Pou transcription factor Brn3a, and a caudal population containing segregated cervical and lumbar domains, which co-express Brn3a and a novel member of the C1q/TNF protein family, C1qL2. Following laser capture microdissection and genetic profiling of these three populations, I identified and validated molecular correlates of the topographic domains within the rodent red nucleus. The transcription factors tshz3 and mafB are expressed in the caudal cervical domain, whereas the chemokine fam19a4 is restricted to the caudal lumbar domain. KitL is an axon guidance molecule that is expressed in both the rostral population and the caudal cervical population. Finally, I identified two genes, cxcl13 and gpr88, that characterize subpopulations within these topographic divisions. Although the functional role of these genes in the establishment of the rubrospinal circuit remains to be determined, the data reveal a high level of molecular heterogeneity within the red nucleus. I hypothesize that this diversity allows rubrospinal neurons to form circuits in a precise and specific manner during development. Overall, my data provide evidence for a novel organization within the rodent motor system in which direct projections from the rubrospinal tract onto motor neurons appear to control a very specific aspect of skilled movement: the stereotypic extension and separation of the digits in preparation for a task requiring digit manipulation. Identifying molecular correlates of the direct rubrospinal population is the logical next step in further understanding the specific circuitry that encodes descending motor commands. My results will provide a basis for the dissection of the rubro-motoneuronal circuit, enabling the establishment of a direct link between neural connectivity and individual muscle control during a skilled movement.

One of the essential requirements for accurate functioning of the nervous system is that synaptic connections are formed and neural circuits are assembled with precision. Two major contributors to the establishment of selective synapse formation are thought to be the positional and molecular identities of neurons. In many instances, the fine-grained precision of synaptic connectivity is thought to occur through a process of molecular recognition that depends on the interaction of complementary recognition molecules expressed on pre- and post-synaptic partners. However, the lack of experimental observations suggests that this is perhaps not the predominant mechanism used in assembling neural networks. In addition to molecular recognition mechanisms, the range of alternative postsynaptic targets can be reduced by organized patterns of neuronal position and axonal growth and termination to deliver the terminals of appropriate pre- and postsynaptic partners to restricted volumes of the developing nervous system. Thus, the positional identities of neurons carry significance in establishing neural networks. The selectivity with which sensory axons form connections with spinal motor neurons drives coordinated motor behavior. The precise profile of monosynaptic sensory-motor connectivity has been suggested to have its origins in the recognition of motor neuron subtypes by group Ia sensory afferents. Here I present an analysis of sensory-motor connectivity patterns in mice in which the normal clustering and positioning of motor neurons has been scrambled through genetic manipulations to conditionally knock out the transcription factor FoxP1. FoxP1, together with an intricate network of Hox genes, drives molecular differentiation programs that give rise to the molecular diversity observed in limb level motor neurons. Conditional ablation of FoxP1 in motor neurons causes scrambling of the motor neurons as well as normalization of molecular identity among all limb level motor neurons. My findings in the conditional FoxP1 mutant mice indicate that critical steps in the patterning of sensory-motor connectivity are governed more by the dorsoventral position of motor neurons than by their identity. My findings imply that sensory-motor specificity in monosynaptic reflex arcs depends on the ability of group Ia sensory afferents to target discrete dorsoventral domains of the spinal cord in a manner that is independent of motor neuron subtype identities, and even of motor neurons themselves. Motor pool clustering and positioning may therefore have evolved to ensure that the motor neurons that innervate a specific limb muscle are able to receive synaptic input from the group Ia sensory afferents supplying the same muscle.

Memory storage and memory-related synaptic plasticity rely on precise spatiotemporal regulation of gene expression. To explore the role of small RNAs in memory-related synaptic plasticity we carried out massive parallel sequencing to profile the small RNAs of Aplysia. We identified 170 distinct 21-23 nt sized miRNAs, 13 of which were novel and specific to Aplysia. Nine miRNAs were brain-enriched, and several of these were rapidly down-regulated by transient exposure to serotonin, a modulatory neurotransmitter released during learning. Two abundant, and conserved brain-specific miRNAs, miR-124 and miR-22 were exclusively present pre-synaptically in a sensory-motor synapse where they constrain synaptic facilitation through regulation of the transcriptional factor CREB1 and translation factor CPEB respectively. We therefore provide the first evidence that a modulatory neurotransmitter important for learning can regulate the levels of small RNAs and present a novel role for miR-124 and miR-22 in long-term plasticity of synapses in the mature nervous system. While mining the small RNA libraries for miRNAs, we discovered an unexpected and abundant expression in brain of a 28-nt sized class of piRNAs, which had been thought to be germ-line specific. These piRNAs have unique biogenesis patterns and predominant nuclear localization. Moreover, we find that whereas miRNAs are down-regulated by exposure to serotonin, piRNAs are up-regulated. Importantly, we find that the piwi/piRNA complex facilitates serotonin-dependent methylation of a conserved CpG island in the promoter of CREB2, the major inhibitory constraint of memory in Aplysia, leading to the persistence of long-term synaptic facilitation. Taken together, these findings provide a new serotonin-dependent, bidirectional, small-RNA mediated gene regulatory mechanism during plasticity where miRNAs provide translational control and piRNAs provide long-lasting transcriptional control for the persistence of memory.

We live in a complicated visual world where stimuli are constantly clamoring for our limited attentional resources. We use our eyes to explore the world and our brain must make moment-to-moment decisions about which points of space contain the most information or which points are associated with rewarding outcomes. In our neural representation of the visual world, stimuli are locked in a constant battle for spatial priority and a single winner must emerge each time an eye movement is to be made, though the mechanisms by which this winner emerges are unclear. In this thesis we explore how competition between neural representations of visual stimuli in the parietal cortex may be implemented by changes in the activity and reliability of neural signals. The macaque lateral intraparietal area (LIP) is part of an oculomotor attentional network and its activity represents the relative priority of spatial locations. We demonstrate how neurons in LIP use surround suppressive mechanisms to resolve conflict between spatial locations and explore the role of shared variability in the priority map network. We manipulate the cognitive state of the monkey by changing his expected reward and show that the activity, reliability, and noise correlation are affected by the context of the monkeys' choice. Finally, we demonstrate how behavioral variables such as the monkeys' performance and saccade latency are modulated during competitive choice.

One cardinal promise of stem cell research is that many intractable, common, and poorly understood diseases may be studied in an entirely new way: in vitro in the specific human cell types affected in vivo. Embryonic stem (ES) cells have the pluripotency to generate all somatic cells types, and the invention of somatic cell reprogramming techniques has allowed the creation of cell lines with both ES-cell grade pluripotency--induced pluripotent stem (iPS) cells--and the genetic determinants of diseases. If iPS cells derived from patients with genetic disease are to enable studying the affected human cell types in vitro then it is necessary to: first, precisely define the appropriate cellular phenotypes in vivo; second, selectively generate those cell types in vitro; and third, demonstrate that iPS cells retain similarly predictable and tractable cellular potential as ES cells. In the motor neuron degenerative disease Amyotrophic Lateral Sclerosis (ALS) spinal motor neurons innervating different types of muscles and individual muscle groups show selective vulnerability or resistance to disease. We therefore set out to define the subtypes of human motor neurons in vivo and to generate these in vitro. Here we report that human motor neurons in vivo share with mouse the molecular markers of motor neuron column, division, and pool organization, as well as positional expression of HOX proteins which regulate this diversity in chick and mouse. We then used combinations of these markers to classify motor neuron subtypes derived from human ES cells in vitro under standard differentiation conditions. These human ES cell-derived motor neurons expressed marker combinations appropriate to each motor column, but were strongly biased to cervical phenotypes. In order to access a greater diversity of motor neuron subtypes, including some with differential responses to ALS in vivo, we defined a developmental strategy to generate more caudal ES-cell derived motor neurons. We show that FGF treatment, in a patterning window we defined, generated human ES-cell derived motor neurons with more caudal (brachial, thoracic, and lumbar) phenotypes. We then participated in a long term collaboration to generate iPS cell lines from donors with ALS-genotypes (familial ALS), and no clinical motor dysfunction (controls). We first showed that ALS and control iPS cells from patients of advanced age could generate motor neurons in vitro. To address questions about the variability of iPS cells, and their comparability to ES cells for making defined neuronal subtypes, we generated a panel of iPS lines from donors of varying demography, thoroughly characterized these cells by standard assays for pluripotent cells, and assessed their ability to generate functional motor neurons in comparison to a panel of ES cell lines. We showed that iPS cells were equivalent to ES cells, and that human genetic diversity may influence the efficiency of motor neuron generation. Next, we used these lines to show that iPS cells could generate the same diversity of motor neurons in vitro, and that the rostrocaudal output of this diversity was rationally manipulable. Finally, since ALS is an adult onset disease, we anticipated that if ES and iPS cell-derived motor neurons could reach significant landmarks of functional maturation in vitro, then the chances of manifesting disease phenotypes would be increased. Therefore we developed methods for long term cultures in which ES and iPS cell-derived motor neurons showed progressive molecular, morphological, and electrophysiological maturation. Together these results enable future studies to ask if ALS-patient iPS cell-derived motor neurons will show pan-motor neuron or subtype-specific ALS phenotypes in vitro. In turn these which may help elucidate mechanisms of disease resistance and vulnerability and identify novel therapeutic targets.