Brain cells can be enticed into forming uniform functioning patterns using a nano-engineering trick.

The technique could allow the development of sophisticated biological sensors that use functioning brain cells, the researchers say. This type of device would identify a compound - a deadly nerve agent or poison, for example - by measuring its effect on a functioning network of neurons.

A team led by Yael Hanein of Tel Aviv University in Israel used 100-micrometre-wide bundles of nanotubes to coax rat neurons into forming regular patterns on a sheet of quartz.

The neurons cannot stick to the quartz surface but do bind to the nanotube dots, in clusters of about between 20 and 100. Once attached, these neuron bundles are just the right distance from one another to stretch out projections called axons and dendrites to make links with other clusters nearby.

Electrical activity

Axons and dendrites carry electrical signals between neurons. The electrical activity of the neural network can easily be measured because carbon nanotubes conduct electricity and so can function as electrodes.

Existing methods for growing networks of neurons cannot produce such neat patterns and clean links between cells. This is because neurons are normally deposited on surfaces that do not prevent them from growing out of ordered clusters onto projections, which makes for a messier network. This is not a problem for Hanein's group. "There is no chance of the cells migrating outside of there," she says.

Uniform networks

The process makes it possible to create more uniform neural networks, Hanein says. In experiments they last longer than other artificial networks, surviving for up to 11 weeks. This could be crucial for building biosensors using the cells, she claims.

"It is clear they grow very nicely and cleanly," says Leslie Smith, a neural computing researcher at the University of Stirling in the UK.

Smith says finding ways to connect to individual neurons in similar arrangements would be even more useful. "That's the holy grail," he told New Scientist. "The best labs can only really put an electrode near a neuron of interest. Finding a way to connect directly and non-invasively is much harder."

Journal reference: Journal of Neural Engineering (vol3, p95)