Harvard researchers have created a chip that can record the activity of hundreds of synapses simultaneously, giving a unique insight into brain function.
The chip consists of an array of vertically-aligned nanometer-scale electrodes on the surface, which are operated by an underlying integrated circuit. Each nanoelectrode is coated with platinum powder, which enhances its ability to pass signals. Neurons are cultured directly on the chip. The integrated circuit sends a current to each coupled neuron through the nanoelectrode to open tiny holes in its membrane, creating intracellular access. At the same time, the integrated circuit also amplifies the voltage signals from the neuron picked up by the nanoelectrode through the holes.
“In this way we combined the high sensitivity of intracellular recording and the parallelism of the modern electronic chip,” said first author Jeffrey Abbott, a postdoctoral fellow in Harvard’s Department of Chemistry & Chemical Biology and the John A Paulson School of Engineering and Applied Sciences (SEAS).
In experiments, the array recorded more than 1,700 rat neurons. Just 20 minutes of recording provided what is claimed to be a never-before-seen picture of the neuronal network, allowing the Harvard team to map more than 300 synaptic connections. The work is published in Nature Biomedical Engineering.
“We also used this high-throughput, high-precision chip to measure the effects of drugs on synaptic connections across the rat neuronal network,” said Abbot, “and now we are developing a wafer-scale system for high-throughput drug screening for neurological disorders such as schizophrenia, Parkinson’s disease, autism, Alzheimer’s disease, and addiction.”
As well as the biological and medical implications of the breakthrough, the synapse-mapping technology also has ramifications for AI and machine learning
“The mapping of the biological synaptic network enabled by this long-sought-after parallelisation of intracellular recording also can provide a new strategy for machine intelligence to build next-generation artificial neural network and neuromorphic processors,” said co-senior author Donhee Ham, Professor of Applied Physics and Electrical Engineering at SEAS.