A computational model of GPe prototypic and arkypallidal neurons with automated parameter fitting.

Parkinson’s disease is characterized by pathological oscillations in the basal ganglia. To gain insight on the origin of these oscillations, we developed a computational model of the globus pallidus (GPe). Our model consists of interconnected prototypic (GPeP) and arkypallidal (GPeA) neurons [1, 5]. We modeled GPeP and GPeA neurons as single-compartment neurons using Hodgkin-Huxley formalism. The GPeA and GPeP neurons have similar ionic currents (I_NaP, I_NaF, I_HCN, I_SK, I_Kv3, I_Ca2+, I_leak) but differ from their conductance values. We tuned the parameters automatically with a multi-objective optimization approach, a variant of the differential evolution [4, 6]. From extensive simulations performed with the SiReNe software (Neural networks simulator, in french: Simulateur de Reseaux de Neurones [3]), we show that our model of GPeP and GPeA neurons are in good agreement with the physiological results of [1], i.e. F-I curves (see Fig. 1A), Voltage-Clamp and I-V relation (see Fig. 1B,C), shape of Action Potentials. Moreover, we show that our GPeP/A neurons interconnected with GABAergic synapses exhibit activity patterns similar to those observed in vivo [2]. This work aims at better understanding the influence of these two different types of neurons in Parkinson’s disease.