Output-based comparison of alternative kinetic schemes for the NMDA receptor within a glutamate spillover model.

Recent experimental and theoretical work continues to explore the mechanisms and implications of neurotransmitter spillover. Here we examine N-methyl-D-aspartate receptor (NMDA-R) kinetics to determine their implication(s) in glutamate spillover by comparing two mechanistically different NMDA-R models, the 5-state Lester and Jahr (LJ) model and the 8-state Banke and Traynelis (BT) model, within the context of a glutamate spillover model. We employ a search-survey-and-summarize strategy to analyze the relationships within model behavior (model relational analysis) and form a model output landscape. Our results indicate that model relational analysis can reveal differences in models whose outputs would be considered the same. The analysis reveals that the BT model, with its more complex kinetics, is less reliant on diffusion compared to the LJ version, resulting in differences in the relationships between open probability and glutamate concentration despite the fact that both model versions were able to produce the same target output values. Additionally, model relational analysis is able to distinguish between the BT and LJ NMDA-R model versions even though factor analysis indicates that the overall model output space dimensions are the same for both NMDA-R models. Furthermore, the work presented here suggests that model relational analysis may be broadly applicable as a means to examine the complex interactions hidden within overall model behavior.