Neural Connectivity in Parkinson's Disease: The Interplay of Network Structure, Synchrony, and Neuroplasticity

Parkinson’s disease (PD) is a chronic neurological disorder characterized by the loss of dopamine producing neurons in the basal ganglia and a number of debilitating motor symptoms including bradykinesia, tremor, and freezing of gait. The level of synchronous neural oscillations in the beta frequency range (13 – 30 Hz) is correlated with the severity of bradykinesia and rigidity. The reciprocally connected subthalamic nucleus (STN) and globus pallidus externus (GPe) circuit within the basal ganglia (BG) has been implicated in the generation of pathological beta activity in PD. While it is clear from a dynamical systems perspective that the structure of a network influences the synchronisation of the elements within that network, the role of network structure in the generation of synchronous beta activity has been overlooked in PD. In addition, changes in network structure at the level of individual synaptic connections have been observed in rodent models of PD over the past decade and a half. However, it is not clear if a mechanistic link between these changes and the presence of beta oscillations in the STN-GPe circuit exists. To address these questions, the primary aim of this thesis was to explore how network structure influences and is influenced by neural activity in the STN-GPe circuit. The approach taken utilized computational modelling in conjunction with network science to establish a network perspective on the generation of beta oscillations in the parkinsonian BG. The aim of the first study was to incorporate techniques used in the study of complex networks to better understand how network structure can influence the level of synchrony in the STN-GPe circuit. This was achieved using several standard network topologies to simulate different underlying connectivity structures in a conductance-based spiking neuron model of the STN-GPe circuit. A synaptic scaling rule was developed to isolate the contributions of network structure to the level of synchrony from other contributory factors. It was found that the number of connections within a network was partially indicative of the level of synchrony, however networks with an intermediate number of connections could display a wide range of synchronous activity. A correlative relationship was observed between an analytically derived measure of network synchronisability known as algebraic connectivity, and the level of synchrony across different network topologies. These results demonstrate how changes in network structure can contribute to the level of synchrony present in the STN-GPe circuit, furthering the understanding of PD pathology, and providing guidance on the development of spiking neural network models. The second study presents network-based methods for desynchronizing pathological oscillations in the STN-GPe circuit. An algorithm was developed which uses the mathematical definition of algebraic connectivity to identify the connections that supported synchronous oscillations. This algorithm was used to explore the effects of synapse removal, neuron removal, and targeted DBS. The removal of synaptic connections identified using the proposed algorithm outperformed the removal of random connections. These effects were more apparent when the synaptic weights were rescaled using the synaptic scaling rule from the previous chapter. As with synapse removal, removing the neurons identified by the algorithm resulted in greater beta reduction than random removal, and this effect was enhanced when synaptic weights were rescaled. Simulated DBS performed similarly to node removal and outperformed the stimulation of random neurons. The techniques developed in this study provide a means of reducing beta activity in a minimally invasive way, by targeting the connections supporting synchronisation. Applying these techniques to the next generation of neurostimulation devices has the potential to further improve motor symptoms while minimizing stimulation-induced