DESCRIPTION (provided by applicant): This application addresses a core question in cerebellar function: How do the deep cerebellar nuclei (DCN) integrate the inhibitory signal from the cerebellar cortex with excitatory input from the rest of the brain to generate the ultimate output of the cerebellum? An interdisciplinary approach of computer modeling and multisite recordings in awake rodents is taken to determine the coding properties of the DCN with respect to the coordination of the rhythmic behaviors of breathing and whisking. These behaviors are represented through modulation of activity in the vermis of the cerebellar cortex and in the medial cerebellar nuclei. Anatomical tract tracing studies are part of the proposed work and will elucidate the exact spatial layout of cerebellar cortical connections of the area in the vermis representing these rhythmic movements to the DCN. Simultaneous recordings from both the vermis and a connected area in the DCN will show how the cerebellar cortical activity is reflected in the output from the DCN. Other influences that will be considered in shaping DCN output are excitatory input from other brain areas and the intrinsic dynamical properties of DCN neurons themselves. Biologically realistic computer simulations of DCN neurons present the ideal tool to integrate the experimental results in a working model of cerebellar output generation. These models will simulate the full morphological structure, ion channel composition, and synaptic inputs of DCN neurons, and will open the door for realistic network simulations of the cerebellum. The intellectual merit of this project lies in the innovative combination of in vivo techniques and modeling to address the question of synaptic integration at the level of a single neuron in a behaving animal. We know that neurons in vivo receive thousands of inputs per second, but what transfer function extracts useful information from this barrage to generate a single output spike train remains poorly understood. The proposed close interaction of multisite recordings to generate data specifically to educate and constrain a model is unique. A further important intellectual component of the problem to be addressed is how neural computation can use inhibition as main signal. This is clearly the case in the signal transfer from the cerebellar cortex to the DCN. The results could be paradigmatic for other connections in the brain mediated by inhibition, e.g. the striato-pallidal, and pallido-subthalamic connections in the basal ganglia. The broader impact of this work is several fold: 1) The computer model of DCN neurons will be disseminated through publicly accessible databases, most notably NeuronDB located at Yale University. This will enable cerebellar researchers and modelers worldwide to incorporate realistic DCN neurons into their models of cerebellar function. This model is expected to become the de facto standard of simulating DCN neuron dynamics. 2) The computer model will be used in training courses at Emory and internationally to teach biologically realistic synaptic integration to a diverse audience. The P.I. is routinely involved in such international training courses, for example the Latin American Course in Computational Neuroscience (LASCON, Brazil), and the Okinawa Course in Computational Neuroscience (OCNC, Japan). This model would be the first one that is specifically calibrated to replicate synaptic integration in behaving animals. 3) The proposed work introduces the coordination of rhythmic behaviors as a new model system to studying the function of the deep cerebellar nuclei. This is a very promising new paradigm and could lead to a new area of cerebellar investigation similar to the explosion of work using the eye blink reflex following the pioneering experiments of Richard Thompson.