Treffer: Computer simulations of a dynamic sodium pump-mediated hyperpolarization and short-term motor memory in the spinal locomotor network of Xenopus frog tadpoles.

Simple four-neuron computational models comprising bilateral pairs of excitatory dIN and inhibitory cIN neurons were used to test several hypotheses concerning the role of electrogenic sodium pumps in shaping swimming CPG output in Xenopus tadpoles. The initial model had no sodium pumps and generated continuous swim-like rhythmic activity. In real tadpoles, activity-dependent "dynamic" sodium pumps are proposed to mediate the postswim ultraslow afterhyperpolarization (usAHP) apparent in most cINs that reduces subsequent swim episode durations, producing a form of short-term motor memory (STMM). Dynamic pumps were therefore incorporated into model cINs, which then generated a usAHP causing swimming episodes to self-terminate, and when interswim intervals were varied, the model also replicated STMM. In real tadpoles, no usAHP is normally apparent in dINs, but one can be revealed by pharmacologically blocking the hyperpolarization-activated current, I<subscript>h</subscript> , which is exclusively expressed in dINs. Dynamic pumps and HCN channels mediating I<subscript>h</subscript> were therefore added to the model dINs. If HCN conductance was locked at its resting level, the dINs now showed a substantial pump-generated usAHP, but this was almost completely cancelled when HCN conductance was allowed to respond normally. Complete cancellation could be achieved by including a speculative cAMP-mediated modulation of the HCN activation kinetics. The models thus confirm the plausibility of published hypotheses regarding the generation of the usAHP in cINs, its apparent absence in dINs due to masking by I<subscript>h</subscript> , and its role in mediating STMM. They also suggest the involvement of the usAHP in swim termination and possible regulation by cyclic nucleotides. NEW & NOTEWORTHY The tadpole locomotor network, an important model system in motor control, has been extensively modeled. Dynamic sodium pumps can generate a slow afterhyperpolarization (usAHP) that contributes to swimming. Here, we present novel computer models that incorporate these pumps and replicate both the usAHP and spinal motor memory. We also show that the usAHP can be masked by HCN channels, validating the conclusions of physiological experiments and suggesting new mechanisms of network function.