Evolution of Adaptive Neural Networks: The Role of Voltage-Dependent K + channels

Abstract

The vestibular pathway of the mollusk Hermissenda crassicornis mediates a reflexive, unconditioned response to disorientation, clinging, that has been conserved during evolution even to the emergence of our own species. This response becomes associated with a visual stimulus (mediated by a precisely ordered visual-vestibular synaptic network) according to principles of Pavlovian conditioning that are also followed in human learning. It is not entirely surprising therefore that molecular and biophysical cascades responsible for this associative learning appear to function in both mollusks and mammals. In brief, combinational elevation of [Ca²+]_i, diacylglycerol, and arachidonic acid activates protein kinase C to phosphorylate the Ca²⁺ and guanosine triphosphate-binding protein, cp20 (now called calexcitin [Nelson T, et al. Proc Natl Acad Sci USA 1996;93:13808–13]), which potently inactivates postsynaptic voltage-dependent K⁺ currents and thereby increases synaptic weight. Longer term changes included rearrangement of synaptic terminals and modified protein synthesis. This cascade has also been implicated in other associative-learning paradigms (e.g., spatial maze, olfactory discrimination) and as a pathophysiologic target in early Alzheimer's disease. Recent molecular biologic experiments also demonstrate the dependence of associative memory (but not long-term potentiation) on voltage-dependent K⁺ currents. Theoretic learning models based on these findings focus on dendritic spine clusters and yield computer implementations with powerful pattern-recognition capabilities.