21
1 A History of MEA Development Jerome Pine Introduction For this volume about the current state of the art in MEA electrophysiology, it is valuable to set the stage with an overview of what has come before, extending up to the present, including some interesting antecedents of the work described here and also some descriptions of work that supplements these chapters. The time span is over thirty years, and most readers will not be familiar with all this past work. We believe there is value in knowing the tradition on which we are building. 1.1 Beginnings and Basics More than 30 years ago, in 1972, Thomas et al. published the first paper de- scribing a planar multielectrode array for use in recording from cultured cells (Thomas et al., 1972). Their introductory paragraph is prescient: Currently available and rapidly improving techniques permit the in vitro culture of an increasing variety of bio-electrically active tissues and single cells. Perhaps the most inter- esting questions to be asked of such cultures are those dealing with the development and plasticity of electrical interactions among the cultured elements (tissues or single cells). Exploration of these questions would be greatly facilitated by a convenient non-destructive method for maintaining electrical contact with an individual culture, at a large number of points, over periods of days or weeks. This report describes one approach to the development of such a method. The multielectrode array that was developed for tests had two rows of 15 elec- trodes each, spaced 100 µm apart, and was intended for experiments with cultured dorsal root ganglion neurons. The main features of multielectrode arrays to this day are shown in Figure 1.1 from their paper. The array was on glass, with gold electrodes and leads over an adhesion layer, insulated with photoresist. The electrodes were plated with platinum black to 3

Advances in Networks Electrophysiology

Embed Size (px)

DESCRIPTION

Electrophysiology advances in molecular biology netowrks.

Citation preview

Page 1: Advances in Networks Electrophysiology

1A History of MEA Development

Jerome Pine

Introduction

For this volume about the current state of the art in MEA electrophysiology, itis valuable to set the stage with an overview of what has come before, extendingup to the present, including some interesting antecedents of the work describedhere and also some descriptions of work that supplements these chapters. Thetime span is over thirty years, and most readers will not be familiar with all thispast work. We believe there is value in knowing the tradition on which we arebuilding.

1.1 Beginnings and Basics

More than 30 years ago, in 1972, Thomas et al. published the first paper de-scribing a planar multielectrode array for use in recording from cultured cells(Thomas et al., 1972). Their introductory paragraph is prescient:

Currently available and rapidly improving techniques permit the in vitro culture of anincreasing variety of bio-electrically active tissues and single cells. Perhaps the most inter-esting questions to be asked of such cultures are those dealing with the development andplasticity of electrical interactions among the cultured elements (tissues or single cells).Exploration of these questions would be greatly facilitated by a convenient non-destructivemethod for maintaining electrical contact with an individual culture, at a large number ofpoints, over periods of days or weeks. This report describes one approach to the developmentof such a method.

The multielectrode array that was developed for tests had two rows of 15 elec-trodes each, spaced 100 µm apart, and was intended for experiments with cultureddorsal root ganglion neurons. The main features of multielectrode arrays to thisday are shown in Figure 1.1 from their paper.

The array was on glass, with gold electrodes and leads over an adhesion layer,insulated with photoresist. The electrodes were plated with platinum black to

3

Page 2: Advances in Networks Electrophysiology

4 Jerome Pine

Figure 1.1. Schematic diagram of the MEA structure. (From Thomas et al, 1972.)

reduce the impedance of their connection to the culture medium (Gesteland et al.,1959; Robinson, 1968) and were 7 µm square. Standard photolithography methodswere used. Initial experiments to try to record from dissociated chick dorsal rootganglion neurons were unsuccessful, and this was ascribed to a confluent glial layeron which the neurons were grown, which insulated them from the electrodes. Turn-ing to dissociated chick myocytes, they found it possible to record robust signals20 to 1000 microvolts high, after the myocytes had formed a confluent contractinglayer over the electrodes. Isolated cells or clumps did not yield measurable signals.It seems likely that the current inflow to initiate myocyte contraction, which canproduce extracellular voltages, is over a large area of the cell; and thus for a singlecell or small group of cells it does not produce a large enough localized change inpotential near an electrode.

Five years later, in 1977, with a very similar introduction to that of Thomas et al.,Guenter Gross and his collaborators proposed the idea of a multielectrode array,without knowledge of the previous work (Gross et al. 1977). Their electrodes weregold, insulated with a thermosetting polymer, and were about 10 µm in diameter,deinsulated with UV laser pulses. There were 36 electrodes spaced 100 or 200 µmapart. The initial test reported in 1977 showed recordings from an isolated snailganglion laid over the electrodes, with single-action potentials having amplitudesup to 3 mv, depending upon the cell size. More specifics for the same preparationwere provided in a later paper (Gross, 1979).

The first successful recordings from single dissociated neurons were reported byme in 1980 from a multielectrode array with two parallel lines of 16 gold electrodes,platinized, and insulated with silicon dioxide (Pine 1980). The electrodes wereabout 10 µm square and 250 µm apart. Though replicating many of the featuresof the earlier work, I did not know of it while making the array and doing theexperiments. Figure 1.2 below shows a scanning micrograph of a typical “fluffy”platinum black deposit on one of the electrodes, which has up to 100 times thesurface contact area with the medium as would smooth gold.

The rat superior cervical ganglion neurons used in these experiments had beengrowing for one to three weeks in culture, and had formed rich interconnectednetworks. They were grown on a fibrous collagen substrate 3 to 5 µm thick. Thecells were about 20 µm in diameter, and recordings were made from 19 cells in

Page 3: Advances in Networks Electrophysiology

1. A History of MEA Development 5

Figure 1.2. A platinum black coated electrode. (From Pine, 1980.)

9 cultures, not in contact with electrodes but typically about 25 µm away. Signalswere on average 50 µv, with signal-to-noise ratios of 5 to 15:1. It was found thatthe electrodes could be used for stimulation with a voltage pulse of 0.5 volts andduration of 1 millisecond.

It was important to ask if the results were compatible with what might beexpected from known neuron properties. It is easy to show that for a point currentsink, or outside a spherical sink, a voltage is generated equal to Iρ/4π r, where Iis the current, ρ is the resistivity of the medium, and r is the distance from thecenter of the current sink. It is believed that typically a neuron action potential isactivated by current flow into the axon hillock, which would represent a point sink(Angelides et al., 1988; Stuart and Sakmann, 1994; Claverol-Tinture and Pine,2002). At a distance of 25 µm, a current of 16 nanoamperes at the peak of theaction potential, which is reasonable, would produce a signal of 50 µv like thoseobserved. These signals are also similar to those observed in vivo, yet somewhatsmaller because the culture medium has lower resistivity than brain.

Figure 1.3 shows oscilloscope traces recorded simultaneously from anintracellular stimulating pipette and an extracellular MEA electrode about 25 µmaway. The upper intracellular traces show five successively larger short stimuliwhich elicit action potentials for the two largest. The lower traces show thesimultaneous extracellular recordings, from three nonexcitations and two actionpotentials, so similar they overlap. The difference of a factor of about 1000 in the

Page 4: Advances in Networks Electrophysiology

6 Jerome Pine

Figure 1.3. Five superimposed intracellular oscilloscope traces are at the top. Action po-tential signals are seen for the largest two. Below are extracellular recordings made simul-taneously with the intracellular ones. The two below baseline are from the action potentials,and are too similar to resolve. (From Pine, 1980.)

size of the signals points up the challenge of making good recordings with MEAs,and the shape of the recorded signal shows very well that it comes at the peak ofthe inward current at the start of the action potential. A great deal of detail aboutthe theory and design issues for MEAs is provided in a book chapter by Kovacs(1994).

1.2 The 1980s

Following up on his earlier work, Gross used his arrays to record from dissociatedspinal cord cultures in 1982. Good signals were obtained from spontaneous activity,which was shown to be very temperature dependent below about 30◦C, decreasingrapidly to a small value at room temperature. Periodic and aperiodic bursts ofactivity were seen, which were studied in more detail in followup experiments(Droge et al., 1986.) Figure 1.4 shows some of these culturewide bursts, recordedfrom three electrodes separated by up to 1 mm. A great variety of patterns wasseen, including some very precisely periodic ones.

In 1981, Jobling et al. reported pioneering work with a nine-electrode MEA inwhich the electrodes were the gates of FET transistors on a silicon chip (Joblinget al., 1981). They demonstrated its effectiveness in recordings from hippocampalslices with good signal-to-noise ratio, while stimulating the slice with a conven-tional stimulating electrode in a fiber tract. Both slice recording and FET-basedMEAs were developed much further by others, but this group did not report anyfurther work.

Page 5: Advances in Networks Electrophysiology

1. A History of MEA Development 7

Figure 1.4. Simultaneous recordings on three widely spaced electrodes of a monolayercortical culture, about four weeks in vitro. (From Gross et al., 1982.)

Wheeler and Novak built passive multielectrode arrays tailored to the need ofanalyzing hippocampal slice activity by performing current source density analysesof field potentials. (Wheeler and Novak, 1986; Novak and Wheeler, 1988). Theybuilt an 8 × 4 array of 32 electrodes, 20 µm in diameter and 200 µm on centers,using conventional fabrication techniques, polyimide insulation, and platinization.Figure 1.5 shows in the center a view of a hippocampal slice and the electrode

Figure 1.5. Responses of a hippocampal slice to a stimulus of the Shaffer collaterals.The center of the figure shows the slice, the stimulating electrode, and the 32 recordingelectrodes. At left are the normal responses and at right the result when picrotoxin wasadded to the bath to reduce inhibitory transmission. (From Novak and Wheeler, 1988.)

Page 6: Advances in Networks Electrophysiology

8 Jerome Pine

array. At left are recordings generated by a stimulus to the Shaffer collateralswhich synapse on the CA3 neurons. At right, the slice has been treated withpicrotoxin, which reduces inhibition. The array can be seen to sample the regionsof the dendrites and cell bodies of these neurons. The size and timings of the fieldpotentials can be used to infer the inward or outward currents in the underlyingneurons.

MEA data was used for calculating a current source distribution in detail(Wheeler and Novak, 1986). By adding picrotoxin to the slice bath they couldcreate epileptic seizures and use their analysis to infer the current sources andsinks. Figure 1.6 shows the epileptic MEA recordings at left and the correspond-ing current sources (positive) and sinks (negative) at right. The traces extend fora time of 30 milliseconds. The slant lines in the current source density analysisshow a propagation of the activity along the slice with a speed of about 250 µmper millisecond.

Many large invertebrate neurons are “identifiable” by their size and location inganglia, can be dissected out, and can be used with other identified neurons toform simple networks in culture that replicate some or all of their connections invivo. The MEA can provide a means for long-term noninvasive communicationwith such networks for stimulation and recording, much superior to conventional

Figure 1.6. Epileptiform activity in the hippocampal slice, positioned as in Figure 1.5. Atleft are the recordings. The time span is 30 msec and the vertical scale spans +2 to −2 mv.At right are the results of a current source density analysis for the locations of the electrodes.The time span is 30 msec and the vertical scale is arbitrary units oriented so that currentsources are positive.

Page 7: Advances in Networks Electrophysiology

1. A History of MEA Development 9

Figure 1.7. The 61-electrode array built in the Pine lab. At left, the electrode structure, andat right a completed MEA. (From Meister et al., 1994 and Regehr et al., 1989.)

electrodes. In 1989, Regehr et al. studied invertebrate neurons using a 61-electrodearray fabricated originally by Gilbert and Pine (Regehr et al., 1989). This “Pinearray” was used in a variety of experiments in the Pine lab and elsewhere. Figure 1.7shows the layout of the electrodes and their construction at left. They are in a close-packed hexagonal pattern, 70 µm apart. The array was designed to be able to recordfrom any neuron of a low-density culture which is on its area, based on the resultsof Pine’s 1980 experiments.

The array was fabricated on a thin glass coverslip to facilitate observation ofcells at high magnification with a short working distance lens and an inverted mi-croscope. The conductors were transparent indium tin oxide, originally introducedby Gross, so as not to interfere with microscopy. The 61 leads end on pads at theedge of the cover slip from where they are connected to external printed circuit-board traces with silicone rubber “zebra connectors”. A complete array is shownat the right in Figure 1.7.

The experimenters wanted to explore these possibilities by recording from avariety of invertebrate neurons, from snails, aplysia, and leeches. Not surprisingly,the activity of these large neurons was easy to record with large signals. Sometimesthe neurons were right over an electrode and could form a seal over it, and at othertimes the electrode was in a more extracellular location near a cell body or aprocess. These varied geometries led to a variety of MEA signals and a few areillustrated in Figure 1.8.

The two traces at the left in Figure 1.8 show recordings from “B19”-identifiedsnail neurons, the top one from an electrode covered over by a cell and the bottomone from an electrode near a process from the cell body. At right are recordingsfrom an aplysia Retzius cell stimulated by an intracellular electrode. At the top isthe intracellular recording, and below are, first an MEA electrode near a process,and second an electrode covered over by the cell body. It seems clear that these bigcells can sometimes seal over an electrode well enough that capacitative couplingcan produce a replica of the intracellular action potential. However, if an MEAelectrode is just adjacent, or makes a poor seal, then the recording replicates

Page 8: Advances in Networks Electrophysiology

10 Jerome Pine

Figure 1.8. At left, recordings from B19 identified snail neurons, at top from an electrodecovered by a cell, and at bottom from one near a cell. At right, simultaneous recordings fromtwo electrodes, from a stimulated aplysia cell. At top, the intracellular signal; immediatelybelow, a signal from an electrode near the cell; and at bottom the signal from an electrodecovered by the cell. (From Regehr et al., 1989.)

the inward current during the action potential. In the paper a detailed model isconstructed that can predict the variety of recorded signals.

In 1989, Meister et al. used a Pine lab MEA to study activity from an explantedsalamander retina positioned over the array (Meister et al., 1989; Meister et al.,1993). The ganglion cell layer of the retina was in contact with the electrode arrayand the photoreceptors were illuminated from above by light patterns generatedon a CRT screen. Very clean extracellular signals were seen, made larger by therelatively low conductivity of the overlying retinal tissue. The retina remainedhealthy and responsive for many hours. Figure 1.9 shows at left a schematic view ofthe array, from which simultaneous recordings from 50 neurons near the electrodeslabeled in black were obtained in one experiment. Spike sorting separated multiple

Figure 1.9. Multi-neuron signals from a salamander retina. At left is a schematic view ofthe locations on the array where signals from 50 different neurons were recorded. Twelvetypically varied poststimulus time histograms are shown, from a two-second flash indicatedby the dark line. (From Meister et al., 1994.)

Page 9: Advances in Networks Electrophysiology

1. A History of MEA Development 11

Figure 1.10. A spontaneous wave of activity propagating across a five-day postnatal ferretretina. The size of the black circles is proportional to the 0.5 sec firing rate. (From Meisteret al., 1991.)

neurons on some electrodes, and in some cases axons produced a trail of recordingsextending from a cell body.

The illumination was a two-second flash, uniform over the retina, and the post-stimulus time histograms for a few cells in the upper right region are shown at theright in Figure 1.9. They were derived from 100 successive stimuli at 10-secondintervals. The light flash is indicated by the heavy line below the histograms. Thevariety of responses to such a simple stimulus is striking. Some neurons fire veryprecisely timed signals, whereas others produce responses that decay with varyingtimes or even remain steady. Cells are seen that respond to the turning off of theflash as well as when it turns on. In later experiments, the retina was illuminated bya multicolored checkerboard pattern changing randomly every 15 milliseconds. Bylooking backward in time, the “spike-triggered average” stimulus for each neuronwas obtained, indicating its receptive field and color sensitivity, simultaneouslyfor all the recorded cells.

The experimental setup used for the salamander experiments was later usedwith retinas from newborn ferrets and cats, where development of the retinal con-nections is still taking place (Meister et al., 1991). The animals are blind at birthand for some time afterwards. Very good recordings were again obtained, with nolight stimulus, showing spontaneous bursting. When the time and space depen-dence was analyzed, it was seen that waves of activity passed across the retina,with varying originating points and directions. Figure 1.10 shows the propagationof such a wave across the retina of a five-day postnatal ferret. The size of theblack circles is scaled to the firing rate over 0.5-second intervals. The propaga-tion of a wave from lower right to upper left during a 3-second period is clearlyevident.

1.3 The 1990s

The MEA extracellular electrode is not adapted to the detection and measurementof subthreshold synaptic potentials. For a synaptic potential of 10 millivolts, theexpected recording near a cell body will be much less than one tenth that for an ac-tion potential, because it will be generated by a diffuse outward capacitative currentfrom the cell body and neighboring dendrites in contrast to the large current into

Page 10: Advances in Networks Electrophysiology

12 Jerome Pine

Figure 1.11. At left, dye recordings from a stimulated neuron and one that receives synapticinput from it. At right, the same signals, averaged over six trials. (From Chien and Pine,1991.)

the axon hillock associated with an action potential. Thus, as in vivo, subthresholdsynaptically generated potential changes are not expected to be observable. Forstudies of network development and plasticity this is a serious shortcoming, andmotivated a combination of an MEA for stimulation and voltage sensitive dyes forrecording. In 1991, Chien and Pine investigated the use of Pine lab MEAs combinedwith voltage-sensitive dye recording (Chien and Pine, 1991). Because the dyes aredirectly sensitive to membrane potential, subthreshold signals are only reducedlinearly in comparison with the action potential. Nonetheless, their measurementprovides a challenge, inasmuch as the dye-recorded signal-to-noise ratio is oftensmall.

Figure 1.11 illustrates the result of one experiment. A neuron in a culture of ratsympathetic neurons was stimulated with an MEA electrode and the postsynapticpotential in a neuron driven by that cell was recorded. At left in the figure isan optical recording of the action potential of the stimulated neuron and ofa postsynaptic potential about one tenth as large, while at right the averagefor 6 trials is shown, and clearly shows the postsynaptic potential. For opticalrecording from single cultured neurons the dominant noise is shot noise fromthe photon flux, so that signal averaging improves the signal-to-noise ratio bya factor of the square root of the number of trials. The data show that evenwhen the dye signal is only 1% of the total fluorescence for an action poten-tial, the stimulus-locked synaptic potentials can be cleanly measured. In manyinstances the dye signal is several times larger and the technique will be much lesschallenging.

Page 11: Advances in Networks Electrophysiology

1. A History of MEA Development 13

Figure 1.12. FET recordings. At left, a simultaneous intracellular and extracellular record-ing of a spontaneous action potential. At top right, spontaneous activity from a cell mak-ing a good seal and below it from a cell making a poor seal. (From Fromherz et al.,1991.)

In 1991, Fromherz and his collaborators investigated the use of a field ef-fect transistor to record action potentials from large aplysia Retzius cells, about50 µm in diameter (Fromherz et al., 1991). The insulated gate of the FET,about 6 by 10 µm, was completely covered by the cell. Figure 1.12 showsat left the very clean FET recording and the intracellular voltage on a stimu-lating electrode. At right are two examples of spontaneous activity, one withvery good contact between the cell and the FET and one that is like the ex-tracellular voltage which is the derivative of the intracellular potential change.A wide range of signals was seen, and it was hypothesized that they resultedfrom variations of the contact between the cell and the gate. This began a se-ries of investigations in the Fromherz lab aimed at understanding the FET-neuroninterface.

In 1995, Welsh et al. published an account of MEA experiments using the Pinelab MEA which capitalized on the possibility of using it to record from a neuralnetwork for long periods of time (Welsh et al., 1995). The suprachiasmatic nucleusin the mammalian brain generates the diurnal circadian rhythm. Suprachiasmaticneurons were dissociated and cultured, and their spontaneous activity monitoredfor days with the MEA. A surprising result, shown in Figure 1.13, was that thecultured network did not synchronize, but that each neuron independently exhib-ited oscillation of its activity with an approximately 24-hour period. The figureshows the firing rates of four neurons as a function of time over a three-day pe-riod. They are not synchronous, nor do they have exactly identical periods. Thus,each neuron must have its own self-contained circadian rhythm generator. Theintracellular machinery for doing this was uncovered eventually by molecularbiologists.

Thiebaud et al. reported in 1997 on work designed to improve recording fromslices with MEAs (Thiebaud et al., 1997). They constructed perforated siliconsubstrates that supported an array of electrodes which were either platinum bumps

Page 12: Advances in Networks Electrophysiology

14 Jerome Pine

Figure 1.13. Spontaneous activity of suprachiasmatic neurons in culture. The diurnal vari-ation of firing rate is large and unsynchronized. (From Welsh et al., 1995.)

or silicon pyramids 45-µm high. The open area of the substrate was 27%. Theperforated support was intended to preserve the viability of the slice, and thethree-dimensional electrodes to make better contact with the cells of the slice. Areservoir of cell culture medium was connected through the perforated substrateto the slice. They used cultured slices grown for up to 15 days in culture, and theycould record for many days. They could also use acute slices for eight hours. Adiagram of their system is shown in Figure 1.14.

Page 13: Advances in Networks Electrophysiology

1. A History of MEA Development 15

Figure 1.14. Schematic view of a slice measurement setup with a perforated silicon sub-strate. (From Thiebaud et al., 1997.)

The development and plasticity of small neural networks is a long term interestof the Pine lab. It was hoped that network connectivity could be determined at agiven time by stimulating each neuron of a network and observing the responsesof the others. However, this was made problematic for a conventional MEA by thelack of single neuron specificity for both stimulation and recording. The networkof processes over the electrodes made it very difficult to limit the electrode inter-actions to specific neurons. To address this issue, a silicon-based “neurochip” wasfabricated by Maher et al. (1999). The initial version was a 4 × 4 array of wellsspaced 100 µm apart on a silicon chip. A thinned area allowed for the creation ofwells 15 µm deep designed to contain a single neuron, with an electrode in thebottom of each. A scanning micrograph of one well seen in section after breakinga chip is shown in Figure 1.15. The structure is based on a truncated pyramidalcavity created with anisotropic etching. A gold electrode at the bottom of the wellis about 6 µm square.

Initially, the idea was to place neurons into the well through a top hole andhave them grow out through small holes at the corners of an overlying siliconnitride cover. For hippocampal neurons, which move along an axon to reach theirtargets during development, this did not work. The cell body escaped throughthe axon’s corner hole. Therefore long thin tunnels were created for outgrowth ofaxons and dendrites, too small for the cell body to escape through. In the figure,a thin silicon nitride “umbrella” extends outward from the well, and five raisedareas can be seen radiating outward. Under them are tunnels 0.5 µm high and10 µm wide, through which axons and dendrites grew. Recording and stimula-tion with these “neurochip” electrodes was easier than for flat arrays. However,

Page 14: Advances in Networks Electrophysiology

16 Jerome Pine

Figure 1.15. A neurochip well, broken open to show the structure. The scale bar is20 microns long. There are radial tunnels under a top silicon nitride layer for out-growth of axons and dendrites, and a gold electrode in the bottom. (From Maher et al.,1999.)

a serious problem arose from the growth of glial cells (necessary for survival ofthe cultured neurons) because they grew over the wells and could be reached byneurons inside and used as a path to escape. The neurons seemed in fact to beattracted to the glial cells. Further work on the project was stopped, pending aredesign.

During the late 1990s, groups in Japan at the Matsushita and NTT laboratories,led by Taketani and Kawana, fabricated 64-electrode MEAs for use in slice ex-periments and with cultures of dissociated cortical neurons (Maeda et al., 1995;Oka et al. 1999). For slices, a rocking device was developed for keeping cultured“organotypic” slice cultures alive over many weeks, so that their development couldbe observed over time (Kamioka et al., 1997). For cortical cultures, experimentsprobed plasticity of connections as a result of tetanic stimulation (Jimbo et al.,1999). The striking result was obtained that stimulation of one electrode affectedall the neurons driven by it to either enhance or reduce their response. But, fordifferent stimulus electrodes, the response of individual neurons could be eitherenhanced or reduced. Figure 1.16, originally in color but here in monochrome,illustrates this effect in the graphs across the top or at the right side of thefigure.

The top graphs of the figure show how the typical responses to test pulsesapplied to each of two stimulating electrodes changed after tetanic stimulation bythe electrode. All the neurons on a “pathway” from that electrode showed eitherenhancement or reduction of their firing probability. In contrast, the graph at right,for a typical neuron, shows that its response to stimuli from the 64 electrodes canbe either enhanced or reduced by the tetanic stimuli. These unexpected results were

Page 15: Advances in Networks Electrophysiology

1. A History of MEA Development 17

Figure 1.16. Response of a dense cortical culture to stimuli at 64 different electrodes, alongrows. The firing rates after stimulation for 71 neurons are displayed in columns. Graphs attop and bottom show firing rate changes of responses to stimuli for two rows and of onecolumn after tetanizations of all rows. (From Jimbo et al., 1999.)

part of an ongoing study in the Kawana group of the dynamics of dense corticalcultures grown on MEAs (Maeda et al., 1995).

1.4 The 21st Century

Several of those whose early work has been described above have continued inthe period from 2000 to the present. Some of the latest developments are summa-rized here, without duplicating the work described in the successive chapters ofthis volume. One significant issue for those trying to study plasticity of corticalnetworks has been the culturewide bursts that occur at intervals of seconds, inwhich typically every electrode of the MEA records. Jimbo et al. have studiedthe dynamics of high-density cortical cultures in detail, and have found that asingle stimulus can produce such a burst, even when delivered to a single neuron

Page 16: Advances in Networks Electrophysiology

18 Jerome Pine

Figure 1.17. Dishwide bursting in a dense cortical culture. (From Jimbo et al., 2000.)

(Jimbo et al., 2000). Figure 1.17 shows such culturewide activity, which typicallylasts over 100 milliseconds.

The Potter group has studied these dishwide bursts, and is engaged in workaimed at understanding and controlling them (Wagenaar et al., 2001; Madhavenet al., 2003). It is hypothesized that their goal of achieving controlled induction ofplasticity in the network connections will be thwarted by these bursts.

Not unlike the Potter group, Shahaf and Maron have been trying to detect learn-ing in dense cortical cultures (Shahaf and Maron, 2001). Their paradigm in oneexperiment was to stimulate one electrode and identify a neuron that respondedwith a low probability during a time 50+/−10 msec afterward. Then, with contin-uing stimulation, they stopped when two out of ten responses occurred and paused.Then the stimulation was resumed and again stopped when a desired response oftwo or more out of ten successive stimuli occurred. Figure 1.18 shows an exampleof “learning” with this paradigm, with an increased response after training duringthe targeted time window from 40 to 60 msec.

The Fromherz group, continuing work with FET electrode arrays, has developeda structure that embodies a picket fence around each electrode to hold a largeinvertebrate neuron in place, and has used these structures to study small connectednetworks (Zeck and Fromherz, 2001). In addition, they have built a closely spacedline of 96-FET electrodes 20 µm apart, and used it for a current source densityanalysis on a hippocampal slice (Besl and Fromherz, 2002).

Meister and his collaborators have continued studies of retinal signal processing.Studies of motion-sensitive neurons have revealed that the stimulus motion in their

Page 17: Advances in Networks Electrophysiology

1. A History of MEA Development 19

Before training After training

time (msec) relativeto stimulus

time (msec) relative to stimulus

500500

Figure 1.18. Learning in a cortical culture. Ten responses at one electrode after stimulationof another electrode are shown at the left. This pair is chosen for poor response in the targetperiod from 40 to 60 msec after the stimulus. At right is shown a “learned” response duringthis time period for ten trials after training. (From Marom and Shahaf, 2002.)

receptive field must be different from the background motion for them to respond.Figure 1.19 shows raster plots for jittering motion, which is normal for the eye,that dramatically show this effect. The object jittering motion is the same in cases(a) and (b), and the background jitter is in phase in (b) and not in (a) (Olvevczkyet al., 2003).

Another study from Meister’s lab found that groups of up to seven ganglioncells fire synchronously and that such groups may account for more than 50%of all the spikes recorded from the retina (Schnitzer and Meister, 2003). Thesepatterns convey messages about the visual stimulus far different from what hasbeen inferred from studies of single ganglion cells.

Finally, the Pine lab has returned to the development of the neurochip, utilizinga new fabrication strategy in which “cages” are created on a flat substrate, initiallysilicon, but possibly glass as well. The technique involves creating cage structures

Page 18: Advances in Networks Electrophysiology

20 Jerome Pine

Figure 1.19. Raster plots of the response of a motion-sensitive neuron is modified by thebackground motion of its surround. In traces a,c, and d the motions are not synchronous,whereas in b they are, and there is no response. (From Olveczky et al., 2003.)

of parylene plastic, which allows a great deal of latitude in the shapes of structures,beyond what would be practical with the usual silicon micromachining technology.(Tooker et al., 2004). Figure 1.20 is a schematic view of a cage structures nowbeing investigated. The idea of tunnels for outgrowth, to keep the cell in the cage,is continued from the earlier work. The height of the cage is about 8 µm, but itcan be made higher if necessary. The top of the cage is overhanging on the inside,to prevent neurons from climbing out. An additional outward-projecting overhangcan be added to keep glial cells from growing over or in. Initial tests have shownnormal neuron outgrowth and connected networks developing for cells in a 4 × 4array of cages.

Figure 1.20. Schematic view of a parylene neuron cage. (From Tooker et al., 2004.)

Page 19: Advances in Networks Electrophysiology

1. A History of MEA Development 21

Summarizing, during thirty years there have been a great variety of explorationsof the possibilities provided by MEAs. Some have led to new understanding, andothers have built technologies that promise new future knowledge. It is an excitingtime for the MEA community.

References

Angelides, K.J., Elmer, L.W., Loftus, D., and Elson, E. (1988). Distribution and lateralmobility of voltage-dependent sodium channels in neurons. J. Cell Biol. 106: 1911–1925.

Besl, B. and Fromherz, P. (2002). Transistor array with an organotypic brain slice: fieldpotential records and synaptic currents. Eur. J. Neurosci. 15: 995–1005.

Chien, C.B. and Pine, J. (1991). Voltage-sensitive dye recording of action potentials andsynaptic potentials from sympathetic microcultures. Biophys. J. 60: 697–711.

Claverol-Tinture, E. and Pine, J. (2002). Extracellular potentials in low-density dissociatedneuronal cultures. J. Neurosci. Meth. 117: 13–21.

Droge, M.H., Gross, G.W., Hightower, M.H., and Czisny, L.E. (1986). Multielectrode anal-ysis of coordinated, multisite, rhythmic bursting in cultured CNS monolayer networks.J. Neurosci. 6: 1583–1592.

Fromherz, P., Offenhausser, A., Vetter, T., and Weis, J. (1991). A neuron-silicon junction:A Retzius cell of the leech on an insulated-gate field effect transistor. Science 252:1290–1293.

Gesteland, R.C., Howland, B., Lettvin, J.Y., and Pitts, W.H. (1959). Comments on mico-electrodes. Proc. IRE 47: 1856–1862.

Gross, G.W. (1979). Simultaneous single unit recording in vitro with a photoetched laserdeinsulated gold multi-microelectrode surface. IEEE Trans. Biomed. Eng. 26: 273–279.

Gross, G.W., Reiske, E., Kreutzberg, G.W., and Mayer, A. (1977). A new fixed-array mul-timicroelectrode system designed for long-term recording of extracellular single unitactivity in vitro. Neurosci. Lett. 6: 101–105.

Jimbo, Y., Kawana, A., Parodi, P., and Torre, V. (2000). The dynamics of a neuronal cultureof dissociated cortical neurons of neonatal rats. Biol. Cybern. 83: 1–20.

Jimbo, Y., Robinson, H.P.C., and Kawana, A.(1993). Simultaneous measurement of intra-cellular calcium and electrical activity from patterned neural networks in culture. IEEETrans. Biomed. Eng. 40: 804–810.

Jimbo, Y., Tateno, T., and Robinson, H.P.C. (1999). Simultaneous induction of pathway-specific potentiation and depression in networks of cortical neurons. Biophys. J. 76:670–678.

Jobling, D.T., Smith, J. G., and Wheal, H.V. (1981). Active microelectrode array to recordfrom the mammalian central nervous system in vitro. Med. Biol. Eng. Comp. 19: 553–560.

Kamioka, H., Jimbo, Y, Charlety, P.J., and Kawana, A. (1997). Planar electrode arrays forlong-term measurement of neuronal firing in cultured cortical slices. Cellular Eng. 2:148–153.

Kovacs, G.T.A, (1994). Introduction to the theory, design, and modeling of thin-filmmicroelectrodes for neural interfaces. In: Stenger, D.A. and McKenna, T.M., eds.,Enabling Techniques for Cultured Neural Networks. Academic Press, San Diego,pp. 121–165.

Page 20: Advances in Networks Electrophysiology

22 Jerome Pine

Maeda, E., Robinson, H.P.C., and Kawana, A. (1995). The mechanisms of generation andpropagation of synchronized bursting in developing networks of cortical neurons. J.Neurosci. 15: 6834–6845.

Mahavan, R., Wagenaar, D.A., and Potter, S.M. (2003). Multisite stimulation quiets burstsand enhances plasticity in cultured networks. Society for Neuroscience Annual Meeting,Abstract 808.14.

Maher, M.P., Pine, J., Wright, J., and Tai, Y.-C. (1999). The neurochip: A new multielectrodedevice for stimulating and recording from cultured neurons. J. Neurosci. Meth. 87: 45–56.

Meister, M., Pine, J., and Baylor, D.A. (1989). Multielectrode recording from the vertebrateretina. Invest. Ophthalmol. Vis. 30(Suppl.): 68.

Meister, M., Pine, J., and Baylor, D.A. (1994). Multi-neuronal signals from the retina –acquisition and analysis.Neurosci. Meth. 51: 95–106.

Meister, M., Wong, R.O.L., Baylor, D.A., and Schatz, C.J. (1991). Synchronous bursts ofaction potentials in ganglion cells of the developing mammalian retina. Science 252:939–943.

Novak, J.L. and Wheeler, B.C. (1988). Multisite hippocampal slice recording and stimula-tion using a 32 element microelectrode array. J. Neurosci. Meth. 23: 149–159.

Oka, H., Shimono, K., Ogawa, R., Sugihara, H., and Taketani, M. (1999). A new planarmultielectrode array for extracellular recording: Application to hippocampal acute slice.J. Neurosci. Meth. 93: 61–67.

Olveczky, B.P. , Baccus, S.A., and Meister, M. (2003). Segregation of object and backgroundmotion in the retina. Nature 423: 401–408.

Pine, J. (1980). Recording action potentials from cultured neurons with extracellular mi-crocircuit electrodes. J. Neurosci. Meth. 2: 19–31.

Regehr, W.G., Pine, J., Cohan, C.S., Mischke, M.D., and Tank, D.W. (1989). Sealing culturedneurons to embedded dish electrodes facilitates long-term stimulation and recording. J.Neurosci. Meth. 30: 91–106.

Robinson, D. A. (1968). The electrical properties of metal electrodes. Proc. IEEE 56: 1065–1071.

Schnitzer, M.J. and Meister, M. (2003). Multineural firing patterns in the signal from eyeto brain. Neuron. 37: 499–511.

Shahaf, G. and Maron, S. (2001). Learning in networks of cortical neurons. J. Neurosci. 21:8782–8788.

Stuart, G.J. and Sakmann, B. (1994). Active propagation of somatic action potentials intoneocortical pyramidal cell dendrites. Nature 367: 69–72.

Thiebaud, P., de Rooij, N.F., Koudelka-Hep, M., and Stoppini, L. (1997). Microelectrodearrays for electrophysiological monitoring of hippocampal organotypic slice cultures.IEEE Trans. Biomed. Eng. 44: 1159–1163.

Thomas, C.A., Springer, P.A., Loeb, G.E., Berwald-Netter, Y., and Okun, L. M. (1972). Aminiature microelectrode array to monitor the bioelectric activity of cultured cells. Exp.Cell Res. 74: 61–66.

Tooker, A., Meng, E., Erickson, J, Tai, Y.-C., and Pine, J. (2004). Development of biocom-patible parylene neurocages. IEEE EMBS Meeting, 2004.

Wagenaar, D.A., DeMarse, T.B., Potter, S.M., and Pine, J. (2001). Development of complexactivity patterns in cortical networks cultured on microelectrode arrays. Ann. Mtg. Soc.Neurosci. Abstr. 922.3.

Welsh, D.K., Logothetis, D.E., Meister, M., and Reppert, S.M. (1995). Individual neuronsdissociated from rat suprachiasmatic nucleus express independently phased circadianfiring rhythms. Neuron. 14: 697–706.

Page 21: Advances in Networks Electrophysiology

1. A History of MEA Development 23

Wheeler, B.C., and Novak, J.L. (1986). Current source density estimation using microelec-trode array data from the hippocampal slice preparation. IEEE Trans. Biomed. Eng. 33:1204–1212.

Zeck, G. and Fromherz, P. (2001) Noninvasive neuroelectric interfacing with synapticallyconnected snail neurons immobilized on a semiconductor chip. Proc. Nat. Acad. Sci. 98:10457–10462.