Andreas Burkhalter, Ph.D. - US grants

Mammalian cerebral cortex and rat primary visual cortex (area 17) specifically, contain both extrinsic and intrinsic neurons. Axons of extrinsic cells terminate within, but also outside area 17. Cells with intrinsic axons, on the other hand, only connect to targets within the area in which they reside; this cell type has frequently been associated with inhibitory actions which are thought to be crucial for normal cortical functions. The aim of the proposed research are to characterize the morphological and chemical types of intrinsic neurons and to assess the role of identified cells in cortical circuitry. To achieve these goals, anatomical tracing and immunocytochemical techniques will be used in combination with a new and powerful retrograde labeling method for visualizing local projection neurons in vitro. These combined approaches will permit intracellulr electro-physiological recordings and dye injections to be performed on identified, intrinsic cortical neurons in tissue slices. In the initial experiments, particular emphasis will be placed on the characterization of several intrinsic cell types which we previously identified in the geniculate input zone in deep layer 6. Two of these types stain for GABAergic markers, although they probably differ in their colocalization for somatostatin and, thus, may permit the distinction of two intrinsic, putative inhibitory systems. Analysis of these two systems will focus on the spatial distribution of their cell bodies, the vertical and horizontal laminar axonal projection patterns, the spatial relationships to afferent and efferent projection systems and their contribution to a GABAergic lattice in upper layers. Lower layer 6 receives thalamic and cortical afferents and, it is possible that at least one of these intrinsic systems provides monosynaptic input. Intracellular recordings from identified cells in deep layer 6 will, be performed, in order to determine the precise relationships. To examine the postsynaptic effects of the different intrinsic pathways originating in deep layer 6, we will record the responses in different cortical layers to stimulation of layer 6. In addition, we will investigate the possibility that GABA acts as neurotransmitter in some of the intrinsic neurons. Two-electrode experiments, combined with intracellular dye injections and immunocytochemistry, will attempt to identify types of inhibitory neurons and to determine how they are integrated into the intrinsic circuitry. These anatomical and immunocytochemical studies should contribute to the understanding of the organization of intrinsic, cortical system(s), where and how they act to process incoming excitatory activity and, therefore, how they provide for neuronal response properties. The physiological experiments should, in addition, permit determination and characterization of the synaptic responses of these inhibitory neurons and their postsynaptic targets.

Neurons are the cells of the brain that are the units for information processing. Understanding the properties of their cell membranes, including excitability and neurotransmission between neurons, is crucial to an understanding of brain function. In complex mammalian brains it has been technically difficult to study membrane properties of identifiable cells, particularly in the cerebral cortex. Conversely, the isolation of single cells in a dish ("in vitro") for membrane recordings makes it difficult to identify the type of neurons these cells were in the normal brain. This project will provide a new approach to try to establish reliable methods to culture and identify cell types or classes isolated from the cortex, and characterize the physiological and pharmacological responses of those neurons. Cells that project across the brain hemispheres, called callosal neurons, can be distinctly labelled by their uptake of non-toxic microscopic fluorescent beads. These cells will then be studied in vitro, either as cultured dissociated cells after removal from newborn rats, or in a thin-slice preparation taken from whole brain. Microelectrodes will be used to test the response of these cells to electrical stimuli across their membranes, to examine the nature of the currents of charged ionic particles that drive the excitable properties of the cells. These studies will compare changes in the cell properties with time since isolation or with time since birth, and will look for possible functional differ- ences among cells that appear morphologically similar. This approach to classify cortical cell types on the basis of membrane properties is technically very demanding, but success will be a breakthrough offering a powerful new way for electrophysiologists to study cortical processing.

This application is in response to the program announcement PA-92-07 for research grants on Neural Systems and Mental, Neurological and Aging Disorders. The goal of the proposed research is to determine how functional circuits develop in the human visual cortex and how these circuits change with advancing age. These questions will be approached by tracing connections in the aldehyde fixed, postmortem human brain with the fluorescent dye, DiI. This is a technology which we have first introduced for studying circuits in the human brain (Burkhalter and Bernardo, 1989). The study will be focused on connections in the visual cortex, not because understanding vision or visual disorders are the main interest here, but because this is the only system in the CNS whose functional architecture is understood well enough that meaningful structure/function relationships can be derived, and small alterations in the normal connectivity pattern can be detected. If successful, it is anticipated that the study will provide information on a number of important issues. It will show that l) microcircuits in the human brain can be studied by an experimental method, 2) neuronal circuits develop in a preprogrammed sequence which determines the maturation of specific functions, 3) cortical connections change throughout life, and 4) connectivity changes are restricted to systems that develop late in life and whose development may be strongly influenced by experience. The proposal is based on preliminary studies in the human primary visual cortex (VI) which revealed three principal findings: 1) local connections within cortical columns develop prenatally, 2) horizontal connections between columns develop postnatally in a manner that circuits related to the motion processing pathway develop before circuits the are more closely associated with form and color analysis, and 3) the laminar organization of feedback projections to striate cortex is selectively altered in aged brains. The experiments proposed in this application are focused on horizontal connections within V1, the second visual area V2 and feedback projections between V2 and V1. Specifically, we will determine: 1) how local connections form within V2, 2) whether feedback connections emerge before forward projections, 3) whether feedback projections are altered in aged brains and, 4) whether local connections within V1 change over the span of life. These connections have in common that they allow the intracortical comparison of distant points of the visual field. This property is essential for the segmentation of a figure from its surround. Psychophysical tests indicate that visual tasks that require integration over large portions of the visual field do not develop until 1 year of age and deteriorate with advanced age. Thus, it is possible that these changes are due to alterations of long-range connections within V1, V2 or modified feedback input from V2 to V1. Similar connections may be important in other cortical areas where they may underlie sensory-motor integration or provide for integration across functionally different systems.

The goal of the proposed research is to study the development of binocular connections that mediate sensitivity for the direction of stimulus motion, and to correlate the emergency of specific connections with the onset of stereopsis and symmetric motion processing. the project is based on the well established notion that stereoscopic vision is dependent on the normal development of direction selective neurons that are tuned for binocular disparity. The central hypothesis is derived from observations that pursuit eye movements in infant monkeys are biased for temporal-to-nasal directions of motion and that objects moving in nasal-to-temporal directions are ignored. Our hypothesis predicts that connections dominated by inputs from the nasal retina develop in advance of those that are dominated by inputs from the temporal retina. To test this prediction infant macaque monkeys will first be examined behaviorally for visual acuity, stereopsis and pursuit eye movements. Subsequently, animals will be used in anatomical tracing experiments for determining: 1) the segregation of lateral geniculate afferents into ocular dominance columns within layer 4Ca of the striate cortex, 2) the development of local connections from left and right eye columns within layer 4B of striate cortex, and 3) the development of striate cortical projections from layer 4B of left and right eye columns to thick stripes of V2 and MT. This experimental design will allow us to evaluate the development of relevant connections at several different levels of the motion processing pathway. Documenting the timing of the emergence of these connections in normal animals is important not only for understanding the neuroanatomic co-development of stereopsis and motion sensitivity, but also because it will provide a framework for understanding the cause of deficits in strabismic monkey and human.

A prominent feature of mammalian visual cortex is that it is composed of many different areas which are linked through reciprocal connections. Forward connections mediate the flow of information from peripheral to more central areas, whereas feedback projections convey information from higher to lower areas. The feedback projections that are of interest here are those that connect higher cortical areas with primary visual cortex. Such projections may have access to both intracortical and subcortical projection systems, on which they may exert different influences. The role of feedback input to corticocortically projecting neurons may be to modify receptive field properties of striate cortical neurons based on information extracted in functionally different cortical areas, and to provide for a coherent representation of a visual stimulus at an early stage in the cortical hierarchy. By contrast, the role of feedback input to subcortical projection systems, may be to select specific afferent information or to direct attention to a particular point of the visual field. The underlying hypothesis is that these different functions are implemented by different circuits. Because it is likely that each of the efferent projections systems, including local interneurons, receive feedback input the difference in the organization may lie in the strength, and the subcellular distribution of inputs. This organization may in part determine the timecourse and amplitude of monosynaptic potentials that activate different cell types. In addition, differences in postsynaptic responses may also arise from the activation of different transmitter receptors that are known to be preferentially distributed in different layers, and possibly also in neurons that give rise to different projections. Thus, the goal of the proposed project is to determine the strength and subcellular distribution of feedback input from higher cortical areas to interneurons and different types of projection neurons in primary visual cortex, and to determine the mechanism(s) that underlay feedback activation of these cell types. To achieve this goal, we will determine, using anterograde neuronal tracing and a combination of light- (LM) and electron microscopic (EM) analyses, the laminar distribution and relative strength of feedback projections from secondary to primary visual cortex in rat (aim 1). To determine feedback input to GABAergic interneurons of neuronal tracing and immunocytochemistry will be used to identify synaptic contacts under the LM and EM (aim 2). Anterograde and retrograde tracing and correlative LM and EM analyses will be employed to determine feedback input to striate cortical cells that project to secondary visual cortex (aim 3). A similar approach will be used to determine feedback input to colliculus projecting neurons (aim 4). Using intracellular recordings in in vitro slices we will examine whether activation of feedback input elicits different types of monosynaptic EPSPs in different cell types (aim 5). To examine whether the postsynaptic responses in different cells are mediated by different receptors we will use intracellular recordings and selective antagonists of different excitatory amino acid receptors (aim 6).

The overall goal of this proposal is to determine the area map of mouse extrastriate visual cortex, how the visual field is represented in these areas, whether different areas are connected to dorsal and ventral pathways and whether they represent functionally specialized processing streams. Preliminary studies, in which we have traced the outputs from three different points of V1 in the same animal, have revealed ten topographic maps in visual cortex. Receptive field mapping by multiunit recording has shown that each area contains an orderly map of the visual field. Anterograde tracing of outputs from each area have further revealed a network of connections, which suggests that areas are hierarchically organized and are linked by intertwined ventral and dorsal streams. Preliminary single unit recordings in ventral- and dorsal-stream areas show that the incidence of direction selective neurons is higher in the dorsal stream. The results suggest that ventral and dorsal pathways represent distinct visual processing streams, which may correspond to 'what' and 'where'streams in primates. This is the first demonstration that mouse visual cortex shares several principles of cortical organization with primates and suggests that the mouse is a good model of the human brain. Recent molecular genetic studies have identified dozens of gene mutations that affect cerebral cortex. In addition, genetic predispositions have been demonstrated for agnosias that affect the functioning of dorsal and ventral stream areas. To study the underlying molecular and synaptic mechanisms of these disorders it is essential to understand the structure and function of visual cortex in the mouse model. Given the impact of the Felleman and Van Essen (1991) area diagram for visual neuroscience, this must include first and foremost the identification of areas, the characterization of the areal hierarchy and the description of processing streams. Although area maps are available for rat and mouse, they differ enormously within and between species. The disagreements are due to pooling across animals and poor registration of partial maps using inadequate anatomical landmarks. We propose to resolve these conflicts with a novel approach that directly combines anatomical pathway tracing (Aim#1) and physiological receptive field mapping (Aim#2) to define areas. Interareal streams have previously not been studied in rodents. Thus, the proposed tracing studies of dorsal- and ventral-stream connections (Aim#3) combined with single unit recording of direction selective responses in dorsal and ventral stream areas (Aim#4) are conceptually novel. The significance of the proposed investigation is to complement functional studies in primate cortex and provide a mouse model for future studies of the molecular and synaptic basis of the human visual system.

DESCRIPTION (provided by applicant): The matrix of cerebral cortex consists of iterated circuits whose local, intra-areal connections account for 90% of excitatory synapses. When activated, these circuits generate positive feedback that is kept in check by local inhibition and plays a role in the selection of inputs and the restoration of signals from the outside world. Within mouse primary visual cortex (V1), non-overlapping groups of pyramidal cells with distinct projections to the higher visual areas LM and AL are connected within each population through horizontal networks. These networks provide influences from topographically distant points and are responsible for contour integration and image segmentation. This form of contextual processing is influenced by attention, expectation and perceptual task, suggesting that feedback from the ventral stream area, LM, and the dorsal stream area, AL, preferentially interacts with LM- and AL-projecting horizontal networks within V1. The top-down influences from LM and AL may be functionally specialized for object categories and temporal context, respectively, and disruption of interactions with specific horizontal V1 networks may lead to behavioral disorders. Direct evidence for interactions between context-processing local networks and top-down pathways, however, is lacking. Here, we propose to study whether: (1) LM- and AL-projecting neurons form separate V1 subnetworks, (2) feedback from the dorsal stream area, AL, and the ventral stream area, LM, preferentially interacts with the subnetwork from which it receives feedforward input, and (3) the LM-projecting ventral stream subnetwork is less strongly inhibited by feedback input from LM, than by feedback from the functionally different dorsal stream area, AL. We propose to study these questions by performing whole-cell patch clamp recordings form pairs of identified pyramidal neurons and interneurons in acute slices of mouse visual cortex, and to use subcellular channelrhodopsin-assisted circuit mapping to characterize the specificity of feedback inputs from dorsal and ventral streams to different excitatory and inhibitory V1 subnetworks. PUBLIC HEALTH RELEVANCE: Sensory processing in primary visual cortex is subject to powerful top-down influences by attention, expectation and perceptual task. Although it is widely accepted that these adaptive processes are determined by interactions between cortical areas and the modulation of intrinsic V1 circuits by feedback connections from higher cortical areas, these interactions have not been directly demonstrated. The proposed studies will address this problem in mouse visual cortex in which synaptic connections between different neuron types, areas, and functional streams can be examined more readily than in primates. If successful, the project will show that top-down influences are preferentially targeted to neurons that belong to the same V1 subnetwork and that excitatory top-down influences across functionally different interareal circuits are more strongly opposed by inhibition. The discovered synaptic circuits may enable switching between networks that enable redirecting attention.

DESCRIPTION (provided by applicant): The visual system is used for object recognition and visually guided actions. These diverse functions are generated by extracting spatial contrast, motion and chromatic cues from the visual scene and sending them to specialized circuits in the midbrain and thalamus. In the cortex the parallel input channels from the lateral geniculate nucleus converge, intermix through specific connections that create new response properties, which are sorted into distinct modules that send outputs into segregated streams of interconnected areas in dorsal and ventral extrastriate cortex. Most of this knowledge derives from groundbreaking studies in primates. At present, however, it is challenging to see how it will be experimentally possible in primates to link specific molecules, neurons and synaptic circuits to particular behaviors. Such advances appear to be critical for finding treatments for visuospatial disorders and agnosias. Many experimental tools for manipulating specific circuits are available to make progress in the mouse. These new opportunities have intensified the need to develop the mouse as model for research of networks underlying object recognition and visually guided actions. The present proposal originated from the rapidly spreading recognition that the mouse visual cortex shares basic similarities with the columnar, modular, areal and hierarchical organization of processing streams in primates. The application challenges the view that mouse primary visual cortex lacks columns and systematic maps of functionally distinct modules embodied in the daisy architecture of horizontal connections. We have arrived at this perspective by first finding that mouse visual cortex contains almost a dozen distinct areas which were recently shown to be functionally specialized. Our findings suggest that these diverse cortical properties emerge from parallel geniculocortical channels, specialized for the processing of spatial detail and rapidly changing stimuli. Further, we have found that within the cortex these inputs are distributed into dorsal and ventral streams of interconnected areas. Most recently we have discovered that mouse primary visual cortex contains a systematic array of type 2 muscarinic acetylcholine receptor expressing modules, which organize geniculocortical inputs, local horizontal connections within V1 and feedback pathways from higher visual areas. Our preliminary results demonstrate that these modules are functionally specialized and are differentially connected to dorsal and ventral streams. Thus, we hypothesize that modules play a role in mixing, sorting and sending distinct information into temporal circuits for object recognition and posterior parietal networks for visually guided actions. To test this general hypothesis we propose to determine by: 1) anatomical tracing and weight analysis of the inputs and outputs of V1 modules, 2) channelrhodopsin-assisted circuit mapping whether anatomical connections represent synaptic connections of a specific strength, and 3) single unit recordings whether modules have distinct receptive field properties.

DESCRIPTION (provided by applicant): The visual system is used for object recognition and visually guided actions. These diverse functions are generated by extracting spatial contrast, motion and chromatic cues from the visual scene and sending them to specialized circuits in the midbrain and thalamus. In the cortex the parallel input channels from the lateral geniculate nucleus converge, intermix through specific connections that create new response properties, which are sorted into distinct modules that send outputs into segregated streams of interconnected areas in dorsal and ventral extrastriate cortex. Most of this knowledge derives from groundbreaking studies in primates. At present, however, it is challenging to see how it will be experimentally possible in primates to link specific molecules, neurons and synaptic circuits to particular behaviors. Such advances appear to be critical for finding treatments for visuospatial disorders and agnosias. Many experimental tools for manipulating specific circuits are available to make progress in the mouse. These new opportunities have intensified the need to develop the mouse as model for research of networks underlying object recognition and visually guided actions. The present proposal originated from the rapidly spreading recognition that the mouse visual cortex shares basic similarities with the columnar, modular, areal and hierarchical organization of processing streams in primates. The application challenges the view that mouse primary visual cortex lacks columns and systematic maps of functionally distinct modules embodied in the daisy architecture of horizontal connections. We have arrived at this perspective by first finding that mouse visual cortex contains almost a dozen distinct areas which were recently shown to be functionally specialized. Our findings suggest that these diverse cortical properties emerge from parallel geniculocortical channels, specialized for the processing of spatial detail and rapidly changing stimuli. Further, we have found that within the cortex these inputs are distributed into dorsal and ventral streams of interconnected areas. Most recently we have discovered that mouse primary visual cortex contains a systematic array of type 2 muscarinic acetylcholine receptor expressing modules, which organize geniculocortical inputs, local horizontal connections within V1 and feedback pathways from higher visual areas. Our preliminary results demonstrate that these modules are functionally specialized and are differentially connected to dorsal and ventral streams. Thus, we hypothesize that modules play a role in mixing, sorting and sending distinct information into temporal circuits for object recognition and posterior parietal networks for visually guided actions. To test this general hypothesis we propose to determine by: 1) anatomical tracing and weight analysis of the inputs and outputs of V1 modules, 2) channelrhodopsin-assisted circuit mapping whether anatomical connections represent synaptic connections of a specific strength, and 3) single unit recordings whether modules have distinct receptive field properties.

Fear- and anxiety-driven behaviors result from interactions in a distributed network of functionally specialized areas of the cortex in which the amygdala plays an essential role by connecting sensory stimuli with their emotional meaning. To avoid detection, mice freeze in place when a looming visual stimulus mimics an approaching object. Alternatively, when detected by a predator the best option for survival may be to escape guided by a cognitive spatial map. Spatial maps are derived from landmark- and path integration-signals accumulated in the cortex from object features and optic flow patterns while navigating through the environment. The cortical network by which this is achieved involves inputs from the visual `where' and `what' streams to the postrhinal area (POR)66 of the parahippocampal cortex, whose output flows through the medial and lateral entorhinal cortex to the hippocampus. Our findings in mice show that POR is the only visual area with strong reciprocal connections with the amygdala, suggesting that POR processes information about emotionally salient objects and guides stimulus-appropriate actions. In direct support for this notion, we have found that POR carries shape, object-motion and self-motion signals. Our results further show that POR contains type 2 muscarinic acetylcholine receptor (M2)-positive and M2-negative modules of which only the M2-negative interpatches receive input from the amygdala. This suggests that interpatches, which in V1 are packed with motion selective cells33, are preferentially tuned by affective inputs from the amygdala. To test this hypothesis we propose to determine the connectional network between POR and the amygdala. Further, we propose to use single unit recordings in POR in awake head-fixed mice, and optogenetic manipulations of inputs from the amygdala, to examine whether the gain of responses to dynamically changing cues and static object features in M2-positive and M2-negative modules is differentially modulated. We expect to find that the amygdala selectively influences responses to object-motion in M2-negative interpatches rather than responses to shape-features represented in M2-postivite patches. If the expectations are confirmed, we will conclude that the amygdala selectively increases the saliency of coherently moving object elements, improves their recognition and optimizes the route13 of escape from of a predator pouncing from a hideout.

ABSTRACT In this proposal we are directing attention to L1, the target of higher thalamocortical connections, cortical feedback and site of input to apical dendrites of pyramidal cells (Pyr). Inputs to L1 were considered to be diffuse and non-specific until we have shown that inputs to L1 of mouse V1 are clustered55. The clustering challenged the notion that rodent visual cortex is non-columnar93 and revealed a mesoscale architecture which is preserved in primate cortex55. We have recently found a similar spatial clustering of neuropil and somata of GABAergic neurons (INT). The modularity involves PV- (parvalbumin), SOM- (somatostatin) and VIP- (vasoactive intestinal peptide) expressing cells. The clustering of INT contrasts with the uniformity envisioned by the canonical circuit model25, and suggests module-specific motifs for counterbalancing excitation with inhibition. The periodicity of INT-rich and INT-poor clusters interdigitates with thalamocortical and cortical long- range connections to L1, where they synapse with INT in source-specific fashion. The clustering of INT suggests a non-uninform distribution of inhibition across V1, which differs from previous proposals143. INT clusters resemble patches of Pyr cells with different spatiotemporal senitivities55. Preliminary results suggest that neurons with high temporal acuity are preferentially localized in INT-rich interpatches, whereas neurons with high spatial acuity reside in INT-poor patches. The overlap of high temporal acuity with INT-rich modules raises several important questions, whether: inhibition in these modules is stronger, feedforward inhibition mediated by PVs is responsible for it, the more strongly inhibited Pyr cells project to specific targets, and whether PVs play a role in the diverse spatiotemporal visual preferences in patches and interpatches. INT-rich and INT-poor modules do not exist in isolation. Only INT-poor patches receive input from the lateral geniculate nucleus and feedback from the higher visual areas, LM, AL and RL, whereas input from the lateral posterior thalamus (LP) and top-down projections to INT-rich interpatches originate from dorsal stream areas136, PM and AM. Thus, INT-rich interpatches are preferentially connected to dorsal stream areas and the LP, from where they receive attention and locomotion-related inputs used for spatial navigation and detection of unexpected motion incongruent with the running speed104,109,124. Previous studies2,70 have shown that suppressive top-down signals from the stimulus surround are mediated through SOM neurons, whereas top-down signals are mediated via VIP-cell-mediated disinhibition34,147. This suggests that top-down information for object segmentation and visually guided actions may differentially involve INT-rich and INT-poor modules. To test these hypotheses we propose to determine whether: 1) the distribution of PV, SOM and VIP neurons in V1 is modular, 2) the strength of PV-, SOM- and VIP-mediated inhibition in INT-rich and INT-poor modules is different, 3) INT-rich and INT-poor modules have different inputs and outputs, and (4) the visual sensitivities of INT-rich and INT-poor modules are differentially affected by inhibition from PV-, SOM- and VIP neurons.