A scientist from Eindhoven University of Technology TU/e, together with colleagues from the Italian Institute of Technology and Stanford University, has developed and tested an artificial synapse with living neurons.
The researcher, Yoeri van de Burgt, had previously made working samples of the artificial synapses, the structures of neurons that are responsible for signal transmission between neurons. The first version was already created in 2017, and two years later, in 2019, a small network with nine synapses was built. The researchers publish in Nature Materials how their synapses interact with living tissue and can thus be used to read signals from neurons.
The artificial synapses are unique because, unlike existing methods of detecting the activity of neurons, they function not electrically, but electrochemically. This makes their functioning much more similar to that of real synapses, where neurotransmitters bridge the distance between two neurons. In addition, the artificial synapses have the ability to learn, which could emulate the particularly energy-efficient process by which neurons process information.
Photo: Yoeri van de Burgt
The synapse consists of two electrodes of soft polymers that are biocompatible. Between the electrodes is a trench filled with electrolyte. When a living neuron is placed on one of the electrodes and releases neurotransmitters, ions are formed at that electrode which in turn travel through the conductive trench and cause a change in its conductivity at the second electrode. In this way the potential of that electrode changes and the signal is transferred.
Because not all of the charge disappears at the receiving electrode, the conductivity remains permanently affected, so a weaker signal would be enough to retrigger the electrode or synapse. This learning capacity of the artificial synapses was tested with neurons from rats using dopamine as a neurotransmitter.
Although the scientists indicate that the research is still at a very early stage, the artificial synapses could eventually be used to allow electronics, including prosthetics, to communicate with neurons. Thanks to the self-learning ability, algorithms that are still needed for signal processing could become redundant with the new synapses. The signals could be automatically adjusted by the learning neural network on the basis of biofeedback.