Jyotika Bahuguna (Research Centre Jülich) | Functionally classifying an ensemble of healthy and pathological basal ganglia network models

Abstract

Our understanding of the circuitry of the basal ganglia has been refined in recent years due to the discovery of sub-populations within nuclei previously assumed to be homogeneous or/and additional novel projections between the existing nuclei. Although, computational models can be used to understand the implications of these structural developments, the lack of data about effective connectivities pose difficulty to their use. Faced with this high degree of under-specification, one can either choose to make simplifying assumptions on the unknown parameters or strive for an unique or locally optimal solution by performing an extensive parameter fit. We propose a yet another alternative, where we perform an extensive parameter search but instead of striving for an unique solution, we embrace the uncertainty and generate a large ensemble of network configurations for both healthy and PD conditions. In order to deal with the high dimensionality of the solution space, we then re-project these solutions on a reduced space of sensible functional features.

We use the recent discovery of two GPe subpopulations: arkypallidal and prototypical neurons [1] as an example of a novel structural development. The phase and firing rate relationships in [1] were used to generate many valid network models for PD and healthy conditions.

These ensembles of healthy and pathological network models were then classified on the basis of dynamical properties, namely the ability to suppress activity in GPi (GS) and susceptibility to oscillations (SO). These dynamical features were used since they are known to be network analogs of PD, i.e. insufficient suppression of GPi is associated with stymied movement [2] and akinetic symptoms of PD patients have been shown to grow worse with an imposed low frequency stimulation of 10-20Hz in STN [3]. Such a functional classification reveals a distinct separation between the PD and healthy ensembles, with majority of PD networks showing insufficient suppression of GPi activity and high susceptibility to oscillations and healthy networks showing high suppression of GPi and low susceptibility to oscillations. There are however some PD and healthy networks that overlap in this functional space.

We propose that such an approach where multiple network models are generated and then projected to a lower-dimensional functional space gives a better chance at understanding a complex pathology like PD which involves deficits in multiple pathways in the basal ganglia.

Literature

 1.Abdi, a., Mallet, N., Mohamed, F. Y., Sharott, a., Dodson, P. D., Nakamura, K. C., et al. (2015). Prototypic and Arkypallidal Neurons in the Dopamine-Intact External Globus Pallidus. Journal of Neuroscience 35, 6667–6688. doi:10.1523/JNEUROSCI.4662-14.2015

2. Boraud, T., Bezard, E., Bioulac, B., and Gross, C. E. (2000). Ratio of inhibited-to-activated pallidal neurons decreases dramatically during passive limb movement in the MPTP-treated monkey. Journal of neurophysiology 83, 1760–1763

3. Timmermann, L., Wojtecki, L., Gross, J., Lehrke, R., Voges, J., Maarouf, M., et al. (2004). Ten-hertz stimulation of subthalamic nucleus deteriorates motor symptoms in Parkinson’s disease. Movement Disorders 19, 1328–1333. doi:10.1002/mds.20198