Electrotonic potentials couple short-term and long-term potentiation in lipid bilayer memristors and memcapacitors

Electrotonic potentials are associated with the passive conduction of ionic charges inside sensory receptor neurons and other excitable cells that do not involve voltage-dependent changes to membrane conductance. Also known as graded potentials, they do not propagate like voltage-dependent action potentials but attenuate rapidly, rendering them unsuitable for long-distance signaling. However, while action potentials can be transmitted over long distances, graded potentials conduct faster, enabling a higher bandwidth and information capacity. This allows for early signal processing optimization before encoding information in action potentials for distant transmission. In physiology, graded potentials involve activity-dependent modification of gap junctions, known as electrotonic couplings, which can be potentiated, just like chemically mediated postsynaptic action potentials, using the same electrical stimulation protocols that produce long-term synaptic potentiation (LTP) of long-term memory formation in the hippocampus. It has been shown that the effect that LTP has on electrotonic coupling in neurons is the result of increases in gap junction conductance.

Using the droplet interface bilayer (DIB) platform, we have developed both memristors (memory resistors) and memcapacitors (memory capacitors) as artificial synapses, consisting of diphytanoylphosphatidylcholine (DPhPC) lipid bilayers doped with antimicrobial peptide ion channels (alamethicin and gramicidin), and have demonstrated both short-term (msec to sec, STP) and long-term (minutes to hours, LTP) potentiation in DIB membranes. Here, we describe how the electrical training protocols that induce LTP in pure lipid bilayer memcapacitors can couple to STP in memristors consisting of lipid bilayers that contain peptides, through long-term modulation of the ion channel conductances. We go on to discuss the implications these coupled dynamical membrane models have for increased understanding of information processing in the brain, and for discovering novel functionality in soft matter-based memelements for neuromorphic computational devices.