Membrane-Based Artificial Neurons for Brain-Like Signal Processing at the Edge of Biology

Bottom-up synthetic biology has transformed the design and modification of custom biomolecules and living cells. It also provides paths for developing functional biomolecular materials for use in applications such as targeted drug delivery, tissue regeneration and even smart, wearable bioelectronics. In contrast to traditional electronic components that are heavy, non-biocompatible, and power hungry, synthetic biomolecular materials are soft, biocompatible, and leverage similar structures and transport properties of biological cells. These are needed to improve sensing and signal processing functions at the biotic/abiotic interface: the <i>edge of biology</i>. Specifically, synthetic biomolecular materials would benefit from the adaptive properties of memristors and neuromorphic devices that enable brain-like signal processing, in-memory computing capabilities, and activity-dependent plasticity. Recent research directions in this field intend to bridge the gap between modern day computing and biological tissues by carefully engineering new neuromorphic devices inspired by how the brain processes and stores information.<br/> <br/>A new type of two-terminal, neuromorphic device—a <i>biomolecular memristor</i>—has been introduced by Sarles <i>et al</i>. Biomolecular memristors are assembled at the interfacial bilayer of lipid-coated aqueous droplets dispersed in oil using the droplet interface bilayer (DIB) technique. DIB membrane-based devices are highly modular and scalable, can be functionalized to various types of stimuli, and offer tunable memristance and activity-dependent plasticity. Herein, we examine new functionality of biomolecular memristors doped with voltage-activated ion channels: spontaneous voltage spiking (self-regenerative pulses akin to action potentials in nerve cells) upon injection of a DC current. We hypothesize that spontaneous spiking voltages can be generated in response to injected DC currents due to the combined effects of membrane capacitance and voltage-dependent ion channel gating that show negative differential resistance (NDR). Preliminary results show that NDR can be achieved in DIBs using alamethicin peptides and protamine introduced on one side of the membrane. Spiking behaviors are examined by injecting DC currents and measuring the resulting changes in membrane voltage. The addition of NDR, a property of “active” devices, to DIB memristive elements can improve the storage density and reduce power consumption through local amplification in neuromorphic circuits. Our goal is to build a scalable material system of biomolecular <i>neuristors </i>that emulate the “leaky integrate and fire” and nonlinear dynamic functions of neurons. This brings us a step closer to enabling spiking neural networks (SNNs) functionality for real-time spatio-temporal data processing at the edge of biology using soft, compatible bioelectronic materials.