Microdevice Studier sfo f so Cultured Neural Networks Jerome Pine Caltech 256-48, 1200 E. California Blvd., Pasadena, CA 91125 Abstract cultureA . d networadvantagee th s kha s tha networe th t two-dimensionas ki l and easily observed, that the biochemical environment can be controlled, and that conventional electrodes as well as extracellular electrodes incorporated into the cultured substrate can be used to selectively stimulate and record from individual neurons in the network. It is possible to study small numbers of connected neurons, from a few to hundreds. This talk will describe two techniques, the multielectrode array and the silicon neurochip theid an ,r applicatio long-tero t n m communication wit networa h k by mean simultaneouf so s recordin r stimulatiogo manf no y neurons. INTRODUCTION In 1978 I stopped doing research in High Energy Physics and spent a year inten- sively studying neurobiology. As a physicist, the idea of studying intact nervous systems, with hundreds of millions or even many billions of interconnected cells was very intimidating, so I was intrigued when I learned about cultured neural networks indisha culturen I . , network growe b n nsca which hav thousando t e w frofe ma s of neurons two-dimensionaa n i , l geometry, easily visible wit phase hth e contrast microscope, and easily accessible to physiological measurements and biochemical manipulations decideI . studo t learo dt w y nho culture s electrophysiologicallyd ,an I had an idea. Instead of just one or two recording electrodes, there could be an arra manyf yo , embedde bottoe culture th th n di f m o e dish, sensing nearby neurons by extracellular recording, and also able to selectively stimulate the culture. Microfabrication using photolithograph d becomha y a commoe n electrica- en l gineering technique, so I set out to fabricate arrays of small electrodes on glass substrates that would become the bottom of culture dishes. And I learned to grow neurons in culture, with the help of expert neurobiologists. Figure 1 shows part of wha I thet n calle "electrin da c petri dish - whic" woulw hno callee db dmula - tielectrode array or MEA. The phase contrast photo shows part of a mass culture of sympathetic neurons, a few thousand in the dish, growing on my array of 16 electrodes. The neurons had been dissociated from the superior cervical ganglion of a newborn rat about eight days before the picture was made. They adhered to CP501, Stochastic Dynamics and Pattern Formation in Biological and Complex Systems, edited by S. Kirn, K. J. Lee, and W. Sung © 2000 American Institut Physicf eo s 1-56396-914-9/007$ 17.00 203 Downloaded 02 Oct 2007 to 131.215.225.176. Redistribution subject to AIP license or copyright, see http://proceedings.aip.org/proceedings/cpcr.jsp FIGURE 1. Cultured rat superior cervical ganglion neurons growing on a multielectrode array. polylysina e treated surfac regred an e w their axon dendrited san foro st mdensa e synaptically connected network, visible in the photograph. e electrodeTh figure th blace n ar esi k dots nea ende th rf gol so d leads whico hg up and down in the picture to terminals at the edge of the glass. The leads are covered with an insulating layer of silicon dioxide, and a horizontal groove which is visible in the figure near the ends of the leads was etched through to expose small 10 micron square gold areas. These were plated with platinum blac provido kt e impedancw lo a e electrical connectio e culturth o nt e medium. Figur showa 2 e s a scanning electron micrograph of a platinized electrode, and the spongy deposit effectively increases the electrode surface area by one to two orders of magnitude. Figur illustrateb e2 recordinsa g made fro electrode mth e nea cela r l labeledB wit arron ha s penetratee celFigurn wi wa Th l . e1 d wit hshara p electroded an , short positive stimulus currents, of successively larger amplitude, were passed into 2 msec V 100 JJV FIGURE 2. At left, a platinized electrode. At right, intracellular and extracellular recordings from a stimulated neuron. 204 Downloaded 02 Oct 2007 to 131.215.225.176. Redistribution subject to AIP license or copyright, see http://proceedings.aip.org/proceedings/cpcr.jsp FIGUR biophysice Th . E3 extracellulaf so r recording p traceto e e scell th Th sho. w recordings from this electrode, which shows five successive cele stimulth l responsed an i . Whe stimulue nth s exceede thresholdda , e celth l fire action a d n potential abou millivolt0 7 t millisecond3 r o s2 higd han s widee extracellulaTh . r electrode recordin e bottos showgi th n no m tracee Th . stimuli produce almost no response, but the action potentials are accompanied by sharp negative spikes. The signal to noise ratio is excellent, although the signals are only about 50 microvolts high. The noise from these electrodes and an inexpensive jfet preamp is negligible. Figur illustratee3 biophysice sth extracellulaf so r recording schematia t lef s A i .t c cell wit axo n manha d nan y dendrites. Whe t fireni action a s n potentia larga l e current flow througn i s e sodiuhth m channels e intracellulaTh . r potential rises s showa Figurn d currenni an , e2 t flows radially outward throug e dendriteshth , leakin t througgou membranese hth returnind an , currenn gi t loops e showth y nb dashed arrowfiguree th f thern extracellulan I i sa . s ei r electrode nea e cells th r a , figuree indicate th potentia e n i th , x y dt b thaa l t poin loweres i t d frome thath t a t outer bordee dendrite th e ohmif o th r y cb s voltag ee retur droth f po n currenn i t the culture medium. The remote boundary of the dendritic tree is essentially at the "ground" reference potential of the medium, and a negative going signal with respect to ground will be seen at x. The signal is generated by current into the cell, and if the cell looks approximately like a "point sink" then the simple analysis diagrammed at the right in Figure 3 holds true, where I is the total current, p is the resistivit culture th f yo e medium, roughl distance th ohm-cm0 s i y10 r e d froan , m the center of the cell. Note that the large intracellular voltage changes generated by the stimulus currents are barely visible. It is the current into the cell that generates the signal, not capacitative current through the cell membrane. Inasmuch as this current into the cell is dominated by charging of the cell membrane capacitance, the current, and the extracellular signal, is approximately the derivative of the intracellular voltage. Durin summee gth 1979f o r , writins whewa nI experiment p gu juse son t like eth discussed [1], I searched the literature and learned that others before me had seen 205 Downloaded 02 Oct 2007 to 131.215.225.176. Redistribution subject to AIP license or copyright, see http://proceedings.aip.org/proceedings/cpcr.jsp the possible benefit of an MEA in culture. In 1972, Thomas and his collaborators built an array for studying neurons but only succeeded in recording from heart cells [2]1977In . ,colleague Groshis and sals [3] so buil arrayan t onlbut , y usefor dit recording fro invertebratn celle a m th f so e ganglion recordinge th o S . sI describe d wer e firseth t from dissociated neurons. Since then, Gros mand an s y others have used array recoro st d from neuron cultures. MEA APPLICATIONS pase Inth t several years numbea , experimenterf o r s have studied mass cultures of mammalian brain neurons with MEAs. They have also laid brain slices ovee th r array recoro t s d from them. Kanaws coworkerhi d an a s have done exceptionally interesting experiments with cultures [4,5] and brain slices [6], and he will speak about that work at this conference. I will therefore describe other MEA applica- tions, with whic have hw e been involved whicd an , hI hop e will illustrat ebroaa d rang possibilitiesf eo describe A ME e dTh . above evolved int omora e well-developed 1980se th n i . b Figurla devicr illustrateeou 4 n ei MEAe sth buile sw t then, which wer experimente useth r dfo swilI l soon describe arrae Th .diagrammes yi Fign di - ure 4a, along with a schematic cross section, and consists of 61 electrodes spaced 70 microns apart. The spacing was chosen for experiments we wanted to do on small cultures of a few neurons, so that there would always be an electrode close enougneuroy an o recoro ht nt d r stimulatfroo conductore t mi Th . eit transe sar - parent indium tin oxide, on glass, and the exposed electrode areas are platinized by electroplating. The insulation was originally polyamide plastic, but has evolved to be silicon nitride or silicon dioxide, which are more durable. Figure 4b shows the complete MEA assembly. The glass substrate is approximately 2x4 cm, and the the leads at the edge are connected to a printed circuit board with elastomer "zebra" connectors e substratTh .thick m f m o s onl facilitato ei t 4 , e y0. us e th e FIGURE 4. At left, the electrode structure, and at right the complete multielectrode array. 206 Downloaded 02 Oct 2007 to 131.215.225.176. Redistribution subject to AIP license or copyright, see http://proceedings.aip.org/proceedings/cpcr.jsp FIGURE 5.