|Jerry Pine (2008), Scholarpedia, 3(10):7766.
|revision #137146 [link to/cite this article]
The Neurochip is a caged neuron multielectrode array. It allows stimulation and recording of small neuron networks grown in a controlled environment.
About 50 years ago, neuroscientists first began studying neurons growing in culture when they attached to the bottom of a Petri dish. If neurons are dissected from intact brains and then dissociated using enzymes that free them from their tissue, they will regrow axons and dendrites and form networks in culture. In this environment it is easy to study the large scale response to drugs and the chemistry of their environment and also the electrical behavior of single cells (see intracellular recording). Figure 1 is a phase contrast micrograph of a culture of rat hippocampal neurons about eight days after they were placed in a culture dish. They are about 20 micrometers across and have rapidly regenerated a complex network of axons and dendrites. See Banker and Goslin (1998) for full details on neuron culture.
The planar electrode array
Extracellular electrical stimulation and recording
Beginning in 1972, researchers have explored the use of planar electrode arrays to record from neurons growing over them in culture and to stimulate them as well. The planar multielectrode array has evolved since then and been used for a variety of studies (Pine, 2005). Figure 2 illustrates extracellular recording with an array electrode. The top trace was recorded from an intracellular electrode in a neuron about 30 micrometers away from an extracellular electrode. The figure illustrates the responses to five short successively larger intracellular stimuli, the two largest generating action potentials in response.
The extracellular signal is generated by the current flowing into the neuron from the culture medium, which is largest during the rapid rise of the action potential. It is short and small, about 1,000 times smaller than the intracellular one, so the electronics must be sensitive and free of noise. Current flowing out of the neuron (rectifying K+ current) also generates a component of the extracellular signal. It is typically much smaller than the main Na+ stroke.
Extracellular stimulation can also be done with an electrode near a neuron. Passing a current into the electrode from the medium will lower the potential in its neighborhood, making it possible to depolarize a nearby neuron by more than about 15 mV and cause it to fire an action potential.
Most multielectrode arrays now in use have approximately 60 electrodes, and Figure 3 is an example that illustrates the main features. This array was made in the Pine lab in the 1980s. There are 61 electrodes, 10 micrometers in diameter, spaced about 70 micrometers apart. The array is on a glass coverslip at the bottom of a culture dish, and the electrodes are connected by conductors to the periphery where connections can be made to external electronics. The conductors are transparent indium tin oxide, with porous platinum plating at the electrode sites to enhance the connection between electrodes and the medium.
The Neuron Cage
The big idea
Experiments with planar arrays depend on chance for a neuron to be close enough to an electrode for recording or stimulation. Furthermore, there is often movement of neurons over time, so such a connection may not be stable. For stimulation, driving an electrode can stimulate nearby axons and antidromically drive other cells unintentionally. Many experiments with dense cultures have been done with planar arrays in which the collective behavior rather than the signals from specific neurons has been studied. This obviates these difficulties. However, in order to create specific two-way communication links with single neurons, the neuron cage idea has been pursued in the Pine Lab (Maher et al., 1999; Erickson et al., 2008). Figure 4 illustrates it.
A neuron placed in a cage can grow axons and dendrites out through tunnels at its base. It is trapped in close proximity to an electrode for recording or stimulation. When it grows out, it can synapse on neighboring neurons or receive inputs from them on its dendrites. An array of cages forms a caged neuron multielectrode array, and neurons in the cages form a cultured neural network which can be studied in detail as never before.
The cage is made of parylene plastic, which is biocompatible and not attacked by common solvents. A neuron can be loaded into it from the top and when it grows out through the tunnels, which are 10 micrometers wide and only one micrometer high, it is trapped. This cage is on a silicon substrate, for convenience in fabrication, but it can be on glass as well. The electrode and leads are gold, insulated with silicon nitride, and the electrode is off-center to provide more space for neuron attachment. Anchors penetrate the silicon and hold the cage firmly in place.
The fabrication uses what are called sacrificial layers to produce a template on which the parylene is deposited, shown schematically in Figure 6, not to scale. The top picture shows the silicon substrate in green, a gold electrode, a blue aluminum layer, and pink layers of photoresist. The picture at the bottom shows the parylene in yellow, conformally deposited from a gas phase, with a hole etched in the top. The photoresist has been dissolved, leaving the cage, and the aluminum is etched away, leaving the tunnels striped in blue.
A trial 16 cage array
In order to illustrate how the caged neuron array can be used to study neural networks 4 x 4 arrays of cages have been constructed, shown in a scanning electron micrograph in Figure 7.
Leads from the electrodes go to the periphery of a 1 x 2 cm. silicon chip, the neurochip, and are wire-bonded to a carrier and socket for connection to external amplifiers, as shown in Figure 8. The array, 300 micrometers square, is in the center of the right half of the chip. The area on the left is for depositing cells to be carried to the cages.
The electronics uses low noise preamplifiers and a computer interface for digitizing the recordings and for generating controlled stimuli. An example of a recording from a cage is shown in Figure 9. The dotted line symbolizes the underlying intracellular signal, and the signal to noise ratio of the recording is very high.
Growth and development
Neurons placed in the cages are cultured in a standard way, incubated at 37°C. in an atmosphere with 5% CO2 with standard culture media. Full details are in the Erickson et al. 2008 paper. They grow and develop in the same way as cultures in standard Petri dishes. Figure 10 shows a Nomarski picture of a culture of ten neurons after eight days in culture. There are a great many more fine processes not visible in the picture. The lifetime of the cultures is three to four weeks, as in standard low density cultures.
The caged neuron array makes possible unique observations of the network connectivity, and how it develops over time. Each neuron can be stimulated and the responses of all the others observed to map the connections.
The diagrams in Figure 11 show the way connections developed rapidly during a short time, beginning at about two weeks in vitro. Synapses are known to begin to develop earlier, but many synaptic inputs are required to cause a target cell to generate an action potential that reveals the connection. The arrows show the direction of the connections, with the red paths believed to be monosynaptic and the green ones disynaptic.
A complementary view of connectivity which more clearly shows the connections at one date, and also their time delays, is shown in Figure 12 for the same culture at 22 div (days in vitro). For each cage, if it contained a neuron that was stimulated it is shown by a black square in one of the small 16-square arrays. The neurons it stimulated are shown by colored boxes in the 16-square array. The colors show the delays, and for disynaptic connections the box is segmented to show the two delays. These are inferred from summed sequential delay times that match the response delay.
The Culture State display ( Figure 12) shows that there is no strong bias for connections to adjacent cages, nor especially short delays in that case. It is consistent with a culture-wide “neuropil” for these small networks, with connections made over the entire area.
Even with this small array it will be possible to explore some important issues: How does imposed activity sculpt the connectivity? Do neurons from different parts of the hippocampus connect in a way that mirrors their behavior in vitro? How will neurons derived from embryonic stem cells connect to hippocampal neurons in a network? Much more detailed data can be envisioned for a larger array. Figure 13 shows a trial 60 cage array on thin glass, making it possible to use optical tweezers for placing neurons in the cages in a controlled and automated way (Pine and Chow, 2008). The Pine lab will help others who want to create neurochip systems.
- Banker, G. and Goslin, K. (1998) Culturing Nerve Cells. MIT Press, Cambridge, MA.
- Erickson, J., Tooker, A., Tai, Y-C., and Pine, J. (2008) Caged neuron MEA: A system for long term investigation of cultured neural network connectivity. J. Neurosci. Meth. 175:1, 1-16.
- Maher, M. P., Pine, J., Wright, J., and Tai, Y.-C. (1999) The neurochip: A new multielectrode device for stimulating and recording from cultured neurons. J. Neurosci. Meth. 87:45-56.
- Meister, M., Pine, J., and Baylor, D.A. (1994) Multineural signals from the retina: acquisition and analysis. J. Neurosci. Meth. 51: 95-106.
- Pine, J. and Chow, G. (2008) Moving live dissociated neurons with an optical tweezers, IEEE Trans. Biomed. Eng., in press.
- Pine, J. (2005). A History of MEA Development. In Advances in Network Electrophysiology Using Multielectrode arrays. M. Taketani and M. Baudry, Eds., Kluwer, NY.
- Tooker, A., Meng, J., Erickson, J., Tai, Y-C., and Pine, J. (2005) Biocompatible parylene neurocages. IEEE Engineering in Medicine and Biology 24:30-33.
- Peter Jonas and Gyorgy Buzsaki (2007) Neural inhibition. Scholarpedia, 2(9):3286.
- Rodolfo Llinas (2008) Neuron. Scholarpedia, 3(8):1490.
- Don H. Johnson (2006) Signal-to-noise ratio. Scholarpedia, 1(12):2088.
- Philip Holmes and Eric T. Shea-Brown (2006) Stability. Scholarpedia, 1(10):1838.
- Advances in Network Electrophysiology Using Multielectrode arrays. (2005) M. Taketani and M. Baudry, Eds., Kluwer, NY.
- Caltech Micromachining laboratory (Yu-Chong Tai)