Nathan Gray Hedrick, PhD - US grants

During learning, neurons in the cerebral cortex form new synaptic connections that convey novel information about the new experience. In order for these new connections to have a meaningful impact on neuronal function and animal behavior, they must somehow provide strong enough input to drive action potential firing, the fundamental output of neurons, in the target cell. Achieving this efficiently (i.e. without massive rewiring of the brain) is likely critical for the fast, flexible, and effective learning seen in mammals. Some lines of evidence suggest that an efficient solution to this problem lies in how the inputs are spatially arranged: by clustering the learning-related inputs onto dendrites of the target neuron, the inputs can have a greater impact on action potential firing. While a large body of work has demonstrated that such supra-linear synaptic integration is possible when synapses are close to one another, whether such an arrangement actually arises as a result of learning has been controversial. Further, whether such an arrangement actually drives action potential firing in a way that is relevant to learning is unclear. The research proposed here will use cutting edge imaging techniques to measure the activity of synapses on neurons of the cortex in awake animals while they learn, allowing the determination of whether such clustered activity exists, and how it evolves over learning to impact neuronal firing as well as the execution of a learned behavior. Further, this research will make use of specific genetic tools to gain clues about how such clustering arises during learning. This work will thus help to characterize a basic strategy employed by the brain to efficiently encode information during learning, as well as the mechanisms that allow such a phenomenon to arise.

Animal learning is thought to involve the precise re-wiring of neuronal circuitry so as to encode novel information, facilitating everything from the learning of a sensory stimulus to the development of a motor skill. While considerable effort has been placed into identifying the specifics of the circuit rearrangements underlying animal learning, the complexity of neural tissue makes the scope of this task a daunting one. This challenge is compounded by the simultaneous need for functional information about how the activity of specific connections relates to the learned behavior. Ultimately, the ideal platform to overcome this sizeable obstacle would involve the study of a minimal ?unit? of neuronal computation, where the collective of connectivity changes occurring over learning can be understood as contributing to the function of this unit. Because of their unique role as computationally semi-isolated integrative compartments, neuronal dendrites are well suited to this demand. Further, many lines of evidence now support small groups of synapses on individual dendrites ? so-called synaptic ?clusters? ? as disproportionate contributors to the function of dendrites and, therefore, to their parent neurons. Thus, the study of learning-related synaptic clusters is an ideal starting point for extracting meaningful information about impactful connectivity changes occurring over learning. The current research seeks to use a novel combination of imaging and molecular approaches to reconstruct the circuit details of learning-related synaptic clusters. Specifically, this work will capture both the functional activity of synapses ? especially those in clusters ? across the length of dendrites as they relate to behavior and identify the brain regions of origin of the inputs impinging on these synapses. These efforts will thus permit a reconstruction of the partial ?connectome? of individual dendrites, with simultaneous information about how the different connections interact with each other (i.e. the coherence of their activity) in the integrative compartment of the dendrite. Such results will have strong implications for how different information streams are integrated at the level of dendrites, and further, will reveal how these integration schemes are sculpted by learning. By using a well characterized model of learning in the form of a simple motor learning task, this work will additionally allow a quantitative description of how specific learning-related connectivity changes are associated with a learned behavior. This work will thus provide an unprecedented level of detail about specific patterns of connectivity changes that occur over learning and will give insights into how such patterns inform learned neuronal activity and, in turn, a learned behavior. These efforts will provide a critical tool set to the neuroscience community and will establish a detailed framework in which diseases affecting neuronal connectivity can be understood.