Planar electrode arrays and network dynamics

Vincent A Mazzarella vamg6792 at uxa.cso.uiuc.edu
Sat May 18 12:45:46 EST 1991

The following is a one page synopsis of a recent one-hour seminar.
The impressions are my own and do not reflect accuracy of the facts
contained in the presentation. Corrections and discussion are welcomed.

     Mouse spinal cord explant (organotypic) cultures were most
often used for these experiments, but mouse olfactory bulb
cultures were also used; such cultures had average lifetimes of
greater than 3 mo. (compared to 3-4 wks. for cerebellum and hcx
cultures). Cultures of approximately 300 neurons were grown on an
insulation layer covering a glass plate containing indium/tin
oxide conductors with gold tips. Laser blasts were used to make
holes in the insulation layer, allowing neurons to contact the
gold tips of the electrodes as they grew in culture. 100 kHz, 3V
signals from the elctrodes are being tried to attract the neurons
to grow toward the electrodes during culture. The culture is
grown for 3-4 wks., at the end of which, the 64 electrodes of the
plate are connected to amplifiers, and a perfusion system
connected to the chamber. (Perfusion flows are precalibrated with
dyes to characterize flows over areas of the chamber from each
perfusion pump.) Hoffman-modulation optics are used to view
neuronal process growth over the electrode surfaces. Multiple
neuron processes usually cross any one elctrode.
     Only supra-threshold spikes were measured, and RC
integration used to store data. Thus bursts were not measured as
many individual spikes but as waveforms each representing an
RC-integrated envelope of a burst. RC-integration waveforms were
categorized and a library of shapes stored in a computer for
later pattern recognition. Phase-space diagrams were used to
compare bursting patterns of separate electrodes, as the raw data
is difficult to discern patterns. Information then was gathered
about the pattern of bursts, the phase delays between bursts, the
timing of bursts, and the cross-correlations between bursts seen
at separate electrodes. Six levels of activity were recognized:
1) resting 2) lo frequency spiking 3) spiking with lo frequency
random bursts 4) patterned bursting 5) periodic bursting and 6)
hi frequency spiking and burst fusion, progressing sometimes to
cell damage. Replacement of CaCl2 by MgCl2 in the culture medium
blocks all activity recorded at the electrodes; this shows
synaptic activity is recorded, since Ca2+ is required for
synaptic activity.
     Network ignition can be initiated by a cell (or group of
cells or processes that one elctrode represents) reaching a
certain firing frequency or firing intensity (i.e. burst
frequency or burst waveform intensity, since RC integrated
waveforms are the units of measure). At this threshold frequency
or intensity of firing, other neurons (or groups of neurons or
processes represented by other electrodes) are recruited to fire.
These other neurons (electrode groups) may fire bursts
synchronously with a "leader" neuron (electrode group), which may
or may not be the igniting neuron (electrode group). The
followers may exhibit delays in synchronization from the leader,
a short delay (few msec) indicating a direct connection with the
leader and a long delay (few hundred msec) indicating some
processing intervening between the bursts of the leader and
follower. "Coarse grain synchronization" is the term that
indicates that synchronization may differ between separate
follower neurons (electrode groups), such that phase differences
of each from the leader of each may differ. Furthermore, the
leader may change with time. By examining delays between each
leader and its various followers, some network mapping can be
achieved for any one culture.
     Differences between ventral and dorsal spinal cord
hemisection cultures can be seen: dorsal shows a wide range of
bursting behaviors whereas ventral shows primarily short bursts
of variable frequency. Using ventral horm hemisection- derived
cultures, non-oscillatory bursts of activity are elicited by
stimulation with ACh. This activity differs from the oscillatory,
long-lasting bursts of activity seen by GABA disinhibition by
bicucilline or by glycine disinhibition by strychnine. Washout of
these disinhibitors will return firing neurons (electrode groups)
to their pre-disinhibition basal firing patterns. GABA or glycine
themselves will inhibit burst activity altogether, in a dose
dependent manner (the ED50 differing from culture to culture.)
Some LTP can be seen using an NMDA agonist: a neuron (electrode
group) with a pre-stimulus bursting pattern can be thrown into a
periodic pattern with bicucilline, as above. Washout normally
returns the neuron to its pre-stimulation pattern, but addition
of NMDA causes the periodic pattern to remain for some time, even
after bicucilline washout. 
     In addition to such pharmacologic manipulations of the
culture network, a He-Ne laser can be used to transect neuronal
processes with a + 2 u resolution.

More information about the Neur-sci mailing list

Send comments to us at biosci-help [At] net.bio.net