Date post: | 22-Apr-2023 |
Category: |
Documents |
Upload: | independent |
View: | 1 times |
Download: | 0 times |
Mechanisms controlling bursting activity induced bydisinhibition in spinal cord networks
Pascal Darbon, Luke Scicluna, Anne Tscherter and JuÈrg StreitDepartment of Physiology, University of Bern, BuÈhlplatz 5, CH-3012 Bern, Switzerland
Keywords: network, multielectrode array, central pattern generator, locomotion, burst
Abstract
Disinhibition reliably induces regular synchronous bursting in networks of spinal interneurons in culture as well as in the intactspinal cord. We have combined extracellular multisite recording using multielectrode arrays with whole cell recordings to
investigate the mechanisms involved in bursting in organotypic and dissociated cultures from the spinal cords of embryonic rats.
Network bursts induced depolarization and spikes in single neurons, which were mediated by recurrent excitation throughglutamatergic synaptic transmission. When such transmission was blocked, bursting ceased. However, tonic spiking persisted in
some of the neurons. In such neurons intrinsic spiking was suppressed following the bursts and reappeared in the intervals after
several seconds. The suppression of intrinsic spiking could be reproduced when, in the absence of fast synaptic transmission,
bursts were mimicked by the injection of current pulses. Intrinsic spiking was also suppressed by a slight hyperpolarization. Anafterhyperpolarization following the bursts was found in roughly half of the neurons. These afterhyperpolarizations were combined
with a decrease in excitability. No evidence for the involvement of synaptic depletion or receptor desensitization in bursting was
found, because neither the rate nor the size of spontaneous excitatory postsynaptic currents were decreased following the bursts.Extracellular stimuli paced bursts at low frequencies, but failed to induce bursts when applied too soon after the last burst.
Altogether these results suggest that bursting in spinal cultures is mainly based on intrinsic spiking in some neurons, recurrent
excitation of the network and auto-regulation of neuronal excitability.
Introduction
Local neural networks in the spinal cord of vertebrates provide the
rhythmic output for locomotion (Grillner et al., 1998). Studies in
lower vertebrates like the lamprey and the tadpole suggest that such
central pattern generators are composed of local oscillator networks
of spinal interneurons, which are coupled by mutual inhibitory and
excitatory synaptic connections (Grillner & WalleÂn, 1985; Dale &
Kuenzi, 1997). This concept has been successfully tested in computer
models (WalleÂn et al., 1992) and is now believed to be also valid in
the mammalian spinal cord, although less experimental data are
available there (Cazalets et al., 1995; Kiehn & Kjaerulff, 1998; Beato
& Nistri, 1999).
The study of how local oscillator networks function and how they
are activated is compromised by the limited experimental access to
such networks in intact spinal cord preparations. Cultures of spinal
cord slices have therefore been developed to maintain and analyse
isolated oscillator networks in vitro (Braschler et al., 1989; Ballerini
& Galante, 1998). Indeed, rhythmic activity can be induced in spinal
cord slice cultures. The observed frequencies of the rhythmic activity
are comparable to those seen in the intact spinal cord during ®ctive
locomotion, although the range of possible induction protocols varies
in both preparations (Ballerini et al., 1999). By mapping the
spatiotemporal structure of rhythmic activity in the slice cultures,
we have previously shown that rhythm sources are usually present on
both sides of the slices close to the ventral ®ssure. However, the
activity always spreads from these sources to the whole slice, thus
leading to synchrony of the rhythms on both sides (Tscherter et al.,
2001). These ®ndings suggest the presence of local oscillator
networks on both sides of spinal cord slice cultures. However, the
proper coupling of these networks into functional pattern generators
with an alternating output for both sides is not formed.
Rhythmic bursting activity has also been described in randomized
cultures of dissociated neurons of the spinal cord (Ransom et al.,
1977; O'Brien & Fischbach, 1986; Gross et al., 1993a) and the brain
(Maeda et al., 1995; MuÈller & Swandulla, 1995). We have previously
found similar patterns of rhythmic activity in such cultures as in slice
cultures, although with differences in the details and in the induction
protocols (Streit et al., 2001; Tscherter et al., 2001). Nevertheless the
similarities bring up the following hypothesis: local oscillator
networks function through basic mechanisms which are not speci®c
to spinal networks but which are inherent properties of many neural
networks without the requirement of high levels of structure.
Among such inherent properties, synaptic depression (O'Donovan
& Rinzel, 1997; Staley et al., 1998) and activity dependent changes in
neuronal excitability (Maeda et al., 1995; Sanchez-Vives &
McCormick, 2000; Streit et al., 2001) have been proposed to be
critically involved in bursting. In the present work, we have
investigated the contributions of both mechanisms to bursting
induced by disinhibition in dissociated and organotypic spinal cord
cultures. The behaviour of the network was compared to that of
individual neurons by combining multielectrode arrays (MEAs)
recording and stimulation (Gross et al., 1982; Gross et al., 1993b;
Jimbo et al, 1993) with whole cell patch recordings (Hamill et al.,
1981; Vogt et al., 1995). Our ®ndings suggest that bursting is based
Correspondence: Dr JuÈrg Streit, as above.E-mail: [email protected]
Received 7 September 2001, revised 5 December 2001, accepted 3 January2002
European Journal of Neuroscience, Vol. 15, pp. 671±683, 2002 ã Federation of European Neuroscience Societies
on the intrinsic spiking of some neurons, recurrent excitation of the
network and an auto-regulation of neuronal excitability.
Materials and methods
Cultures
All cultures were made from the spinal cord of rats at embryonic age
14. The cultures were prepared in the same way as described
previously (Braschler et al., 1989; Spenger et al., 1991; Streit et al.,
2001; Tscherter et al., 2001). The embryos were delivered by
caesarian section from deeply anaesthetized rats (0.4 mL pentobarbi-
turate i.m.) and killed by decapitation. Following the delivery of the
embryos, the mother rat was killed by intracardiac injection of
pentobarbiturate. Animal care was in accordance with guidelines
approved by Swiss local authorities. The backs of the embryos were
isolated from their limbs and viscera and cut into 225 mm-thick
transverse slices with a tissue chopper. For the dissociated cultures,
slices of all regions of the spinal cord, without dorsal root ganglia,
were exposed to a 0.3% trypsin solution for 3 min at 37 °C. These
slices were then mechanically dissociated by forcing them through
®ne-tipped pipettes several times. The cells were plated on MEAs or
on glass coverslips at a density of 150 000 or 75 000/150 mL,
respectively. MEAs were produced as described previously
(Tscherter et al., 2001) and coated for one hour with diluted
(1 : 50) Matrigelâ (Falcon/Biocoat, Becton Dickinson AG,
Switzerland). The glass coverslips were coated with polylysine
(1 mg/mL overnight at 37 °C). The cells were restricted to an area
around the electrodes (» 50 mm2) using cloning glass cylinders
attached to the MEAs or coverslips by silicone sealant. They were
maintained in culture dishes containing 150 mL of nutrient medium
and incubated in a 5% CO2-containing atmosphere at 36.5 °C for up
to 12 weeks. Serum-free NeurobasalTM medium (Gibco BRL, Life
Technologies AG, Switzerland) supplemented with B27 and
Glutamax (both Gibco BRL) was used for the MEA cultures and
some of the cultures on the glass coverslips. The other cultures were
kept in a MEM Eagle's Medium supplemented with 10% fetal bovine
serum, 0.2% glucose, B27 and Glutamax. Half of the medium was
changed weekly.
For the organotypic cultures, the spinal cord slices with their
attached dorsal root ganglia were ®xed on MEAs using reconstituted
chicken plasma (Cocalico Biologicals, Pennsylvania, USA) coagu-
lated by thrombin (Sigma, Fluka Chemie AG, Switzerland). The
cultures were maintained in sterile plastic tubes containing 3.5 mL of
nutrient medium and incubated in roller drums rotating at 120 rph in
a 5% CO2-containing atmosphere at 36.5 °C. The medium had the
following composition: 79% Dulbeccos MEM with Glutamax, 10%
fetal bovine serum (Life Technologies), 10% H2O and 5 ng/mL 2.5 S
nerve growth factor (Alomone Laboratories Ltd, Jerusalem, Israel).
During the ®rst week of incubation a medium with 10 ng/mL nerve
growth factor was used. Half of the medium was removed and
replaced with fresh medium weekly.
Recordings
Recordings were made in a chamber mounted on an inverted
microscope (Nikon, Japan) from cultures of 1±3 weeks of in vitro age
(slice cultures) and 2±12 weeks (dissociated cultures). The medium
was replaced by an extracellular solution containing (in mM): NaCl,
145; KCl, 4; MgCl2, 1±2; CaCl2, 2; HEPES, 5; Na-pyruvate, 2;
glucose, 5 at pH 7.4. Recordings were either made in the presence of
continuous superfusion at 1 mL/min or with solution changes every
10±15 min. No differences in the patterns of activity were seen
between these two protocols. All recordings were made at room
temperature.
MEA recording and analysis
MEAs contained 68 electrodes, laid out either in the form of a
rectangle (dissociated cultures) or in the form of a hexagon (slice
cultures). Channels showing activity (usually 10±30 for the dissociated
cultures and 20±50 for the slice cultures) were selected by eye and their
recordings digitized, visualized and stored on hard disc using custom
made virtual instruments within Labviewâ (National Instruments,
Switzerland) as described previously (Tscherter et al., 2001).
FIG. 1. Combined recordings of disinhibition-induced bursting in the wholenetwork and in a single neuron of a dissociated culture. The recordingswere made in the presence of bicuculline (20 mM) and strychnine (1 mM).(A) Event raster plot of MEA recordings (see Materials and methods)showing the activity of 20 channels. (B) Network activity plot (seeMaterials and methods) derived from the recordings in A. To visualize theamount of activity in the whole network, the number of detected eventswere counted within a sliding time window of 10 ms. Note the increasinglevels of activity in the intervals. (C) Whole cell recording from a singleneuron in the network. Note that the activity ceases following the bursts dueto an afterhyperpolarization. (D) Superposition of the network activity(grey) and the whole cell activity (black) for one burst at an expanded time-scale. (E) Whole cell recording from the same neuron as shown in Cshowing persisting spiking in the presence of CNQX (10 mM) and APV(50 mM).
672 P. Darbon et al.
ã 2002 Federation of European Neuroscience Societies, European Journal of Neuroscience, 15, 671±683
Detection of the extracellularly recorded action potentials and further
analysis were carried out of¯ine in the software package IGOR
(WaveMetrics Inc., Oregon, USA) as described previously. The
electrical noise of individual channels was very stable. The electrodes
recorded fast voltage transients (< 4 ms), which correspond to single
action potentials in neurons or axons (single-unit activity). These fast
voltage transients often appeared in clusters (multiunit activity)
originating from closely timed action potentials of several neurons or
axons seen by one electrode. During such multiunit activity, individual
fast voltage transients could not be clearly linked to underlying action
potentials. Therefore no attempt was made to sort spikes seen by one
electrode. For the analysis of the spinal network activity a detector was
designed to ®nd fast voltage transients, and its output was de®ned as
event. During multiunit activity the fast voltage transients could not be
attached to individual action potentials, therefore the detection device
was designed in a way to represent multiunit activity by periods at a
de®ned level of activity. This activity detector reliably translated on
the one hand single-unit activity into one event and on the other
multiunit activity into trains of events at 333 Hz (see Tscherter et al.,
2001). The selectivity of event detection was controlled using
recordings obtained in the presence of tetrodotoxin (TTX, 1.5 mM)
as a zero reference. The processed data were visualized in the form of
event raster plots and network activity plots. Event raster plots show
the time-markers of detected activity of each selected channel (e.g.
Fig. 1A). Network activity plots (e.g. Fig. 1B) show the total activity
of all selected channels within a sliding time window of 10 ms, shifted
by 1 ms.
Stimulation by MEA electrodes
Biphasic monopolar voltage pulses with a duration of 0.5 ms and
amplitudes of 0.5±1.25 V were addressed to one of the MEA
electrodes. No recording was possible with the electrode connected to
the stimulator.
Whole cell patch clamp recording and analysis
Intracellular voltage and current measurements were obtained from
individual neurons using the whole cell patch technique (Hamill et al.,
1981) with either an Axopatch 1B or an Axoclamp 2B ampli®er (Axon
Instruments Inc., California, USA). The patch pipettes were ®lled with
a solution containing (in mM): K-gluconate, 100; KCl, 20; HEPES, 10;
Mg-ATP, 4; Na2-GTP, 0.3; Na2-phosphocreatine, 10; pH 7.3 (with
KOH); for voltage clamp recordings only: EGTA 10. The electrodes
had a resistance of 4±5 MW. No series resistance compensation
was used. Control resting membrane potentials were in the range of
±40 mV to ±70 mV (mean: ±48 6 1.1 mV, n = 29). Cells with a
potential less negative than ±40 mV were discarded. Synaptic currents
were recorded at a clamped membrane potential of ±60 mV. Voltage
and current recordings were digitized, visualized and stored on
computer using either pClamp (Axon Instruments inc.) software or
virtual instruments within Labview. They were analyzed of¯ine using
custom made programs in IGOR. The detection of synaptic currents
was based on a threshold set to the ®rst derivative of the original
current recording at three times the noise level. Bursts were de®ned as
the superposition of four or more subsequent currents separated by an
interval < 10±15 ms. The end of such burst was de®ned when the
interval between two subsequent currents exceeded a threshold of 15±
25 ms.
Statistics
Averages are expressed as mean 6 SEM. Statistics are based on
Student's t-test and ANOVA. A correlation was considered signi®cant
when its coef®cient (r) was signi®cantly different from 0 (t-statistics,
P < 0.05).
Chemicals
All agents were applied by exchanging the bath solution for a solution
containing the drugs. The following agents were used: bicuculline,
strychnine, APV, CNQX, TTX (all Sigma).
FIG. 2. Whole cell recording from a single neuron during block of fastsynaptic transmission by bicuculline (20 mM), strychnine (1 mM), CNQX(10 mM) and APV (50 mM) in a dissociated culture. Intrinsic spiking isstopped by changes in the membrane potential and spike rate due toinjected current pulses. (A) Pulses of depolarizing current injections withincreasing amplitudes induce spiking at increasing rates during the pulseswithout spike frequency adaptation. The grey bars visualize the duration(length of the bars) and the strength (thickness of the bars, range: 50±200pA) of the current pulses. Note that the intrinsic ®ring in the intervalsbetween the pulses is eventually suppressed. (B) Spike intervals fordifferent voltage steps induced by current injections as shown in A. Notethat except for the ®rst three spikes there is practically no spike frequencyadaptation. (C) Steady state response curve of the neuron shown in A andB. Spike rate is a linear function of the membrane potential with a thresholdat ±52 mV and a slope of 2.2 Hz/mV. (D) After the stop of currentinjection, intrinsic activity returns following a refractory period ofapproximately 5 s. Note that the intrinsic activity is ®rst phasic and thengradually returns to the initial tonic activity. (E) Hyperpolarization inducedby pulses of negative current injections (grey bars) stop intrinsic spiking.Note that hardly detectable changes in membrane potential are suf®cient tostop spiking (right pulse) without leaving an oscillating membrane potential.
Bursting activity induced by disinhibition 673
ã 2002 Federation of European Neuroscience Societies, European Journal of Neuroscience, 15, 671±683
Results
Network and single cell activity
The results are based on the investigation of 104 dissociated spinal
cultures and 22 organotypic slice cultures. From the dissociated
cultures 26 were investigated by MEAs (on average 21 channels per
culture), 66 by whole cell patch clamp and 12 by a combination of
both methods. It has been shown previously that disinhibition by a
combination of strychnine (1 mM) and bicuculline (20 mM) induces a
slow and more or less regular bursting of activity in the whole
network of dissociated cultures, in slice cultures and in the intact
spinal cord (Droge et al., 1986; Streit, 1993; Bracci et al, 1996). MEA
recordings in dissociated cultures showed that such bursting was
characterized by periods of high activity in the whole network,
interrupted by intervals of low network activity (Fig. 1A and B). The
network activity abruptly increased at the burst onset, then slowly
decreased during the bursts and ®nally rapidly decreased at the end of
the burst (Fig. 1D). During the onset of the bursts, a wavefront was
spreading from one electrode (source) to the whole network within
50±100 ms. During the intervals between the bursts, activity at low
rates (around 0.1±1 Hz) persisted in some of the channels. In general,
network activity was low immediately following the bursts and it
increased slowly during the intervals (Fig. 1B and C).
Whole cell recordings of spinal neurons showed that the network
bursts caused depolarization, always leading to one or several spikes
at the beginning of the bursts. These initial spikes were followed by
repetitive spiking at rates of 1±20 Hz in 75% of the cells (Fig. 3). In
25% of the neurons a strongly depolarized plateau potential (at
±19 6 1 mV, n = 10, compared to a depolarization to ±37 6 1.8 mV
in the other neurons, n = 29) was seen during the bursts (Fig. 1D).
These neurons had on average higher spike rates during control
activity (in the absence of disinhibition by strychnine and bicuculline)
than those with repetitive spiking (19.1 6 9.3 vs. 5.9 6 1.4 spikes
per s). This suggests that the plateau potentials are caused by strong
synaptic inputs leading to large scale depolarization with inactivation
of sodium channels. In line with this hypothesis, spikes often
reappeared at the end of the bursts when the membrane potential
repolarized (Fig. 1D). In approximately half of the neurons, bursts
were followed by long lasting afterhyperpolarizations. In the intervals
between the bursts, the spike rate was low. Some of the neurons,
however, started to repetitively ®re at increasing rates (1±10 Hz) in
the second half of the intervals (Fig. 1C), when the afterhyperpolar-
izations had disappeared. Such spontaneous ®ring was characterized
by gradually depolarizing membrane potentials between the spikes.
This behaviour of individual neurons explains the increasing network
activity during the intervals observed with MEA recordings.
It has been shown previously that the bursting is based on recurrent
excitation of the network through glutamatergic synapses (Streit,
1993). Blocking such synapses by a combination of APV and CNQX
stopped the bursting in most of the experiments, while asynchronous
low rate activity persisted at approximately half of the electrodes in
MEA recordings (Streit et al., 2001). As previously shown, such
intrinsic neuronal activity is preferentially found at sites where bursts
are frequently initiated (Streit et al., 2001). As shown in Fig. 1E, in
whole cell recordings of single cells, intrinsic spiking was found in 9
out of 22 neurons in the presence of strychnine, bicuculline, CNQX
FIG. 3. Whole cell recordings showing the alterations in neuronal excitability during disinhibition-induced bursting (bicuculline, 20 mM and strychnine, 1 mM)in a dissociated culture. For this analysis burst rate was slowed by 1 mM CNQX. (A) Example of a neuron, in which bursts were followed by anafterhyperpolarization with a decrease in excitability. Action potentials in the intervals were induced by injecting depolarizing current pulses of 50 msduration applied at 1 Hz. The amplitude of the pulses was adjusted to be just above threshold for most of the time. (B) Example of a neuron, in which burstsare followed by an afterdepolarization with an increase in excitability. The current pulses were adjusted to be just below threshold for most of the time. (C)The percentage of failures to induce action potentials for seven consecutive pulses at 1 Hz following a burst and six pulses preceding the next burst (n = lastpulse before the next burst). Each point represents the mean of six experiments (6 SEM) as illustrated in A (®lled points), and seven experiments asindicated in B (open points).
674 P. Darbon et al.
ã 2002 Federation of European Neuroscience Societies, European Journal of Neuroscience, 15, 671±683
and APV. The mean spike rate in these neurons was 3.7 6 1.2 Hz
(n = 6). The membrane potential between the spikes was gradually
depolarizing from the hyperpolarized level following the spikes, and
it was on average more depolarized than the resting membrane
potential in non±spiking neurons (±41.4 6 1.3 mV, n = 6 vs.
±48.8 6 3.3 mV, n = 4). In summary, combining extracellular
MEA recordings with intracellular recordings from single neurons,
we could con®rm our previous hypothesis: intrinsic ®ring of some
neurons can activate the whole network through recurrent excitation
and cause a burst (Streit et al., 2001). Intrinsic ®ring ceases following
the bursts and slowly recovers during the intervals (Fig. 1C).
Auto-regulation of intrinsic spiking
The modulation of intrinsic ®ring during bursting could be caused by
the synaptic activity during bursts because neurotransmitters and
neuromodulators are known to change neuronal excitability
(Legendre et al., 1989; White et al., 1996). Alternatively, in the
sense of an auto-regulation of neuronal spiking, it could be in¯uenced
by the rate of spiking itself. In order to distinguish between these two
possibilities we mimicked bursting in the absence of synaptic
transmission by current injections into single neurons in the presence
of blockers of synaptic transmission (bicuculline 20 mM, strychnine
1 mM, CNQX 10 mM and APV 50 mM). We found that intrinsic
spiking was highly sensitive to current injections, i.e. depolarizing
currents increased the ®ring rate. The larger the current pulse, the
stronger was the depolarization, the higher was the spike rate and the
smaller and slower were the spikes (Fig. 2A). A linear relationship
was found between the spike rate and the membrane potential
between spike threshold at ±56.3 6 1.6 mV (n = 6 experiments) and
±30 mV (Fig. 2C). The mean slope of this relationship was
2.0 6 0.2 Hz/mV (n = 6 experiments). In the course of strong
depolarization the spikes became smaller and slower and could ®nally
die out, thus leading to the plateau potentials comparable to those
shown in Fig. 1 for the synaptically induced bursts. However, similar
to the ®ndings in interneurons in slice cultures (Ballerini et al., 1999),
practically no increase in spike intervals (spike frequency adaptation)
FIG. 4. Whole cell recordings showing spontaneous excitatory postsynaptic currents (EPSCs) in a voltage clamped neuron under disinhibition (bicuculline,20 mM, strychnine, 1 mM) in a dissociated culture. The holding potential was ±60 mV (A) EPSCs are organized into bursts. The grey line shows the de®nedbursts (lower level) and intervals (upper level). Note the summation of EPSCs during the bursts, especially at the beginning. The small inset shows thesuperposition of two EPSCs occurring during the intervals. Note the small size and the large variability of these EPSCs. (B) Correlation between the durationof the bursts and the preceding interval; r = 0.78. (C) Correlation between the initial peak amplitude of the summated currents during the bursts and thepreceding interval; r = 0.78. (D) Peak amplitudes of EPSCs occurring during the intervals are plotted vs. interval time. Six consecutive intervals aresuperimposed. (E) Rates of EPSCs occurring in the intervals are plotted vs. interval time. Six consecutive intervals are superimposed.
Bursting activity induced by disinhibition 675
ã 2002 Federation of European Neuroscience Societies, European Journal of Neuroscience, 15, 671±683
was found in these cells (Fig. 2B). Only the interval between the ®rst
and second spike of a pulse was usually shorter than the following
intervals.
Pulses of strong current injections suppressed the spontaneous
®ring in the intervals between the pulses in all ®ve neurons (Fig. 2A).
When the current pulses were stopped, intrinsic ®ring re-appeared
after a latency of several seconds: ®rst, phasic activity came up, then
the activity gradually turned into the tonic activity present before the
current injections (Fig. 2D). Because the membrane potential was
usually close to the spike threshold potential, intrinsic spiking was
stopped during pulses of hyperpolarizing current injections, even
when the amount of hyperpolarization did not exceed a few millivolts
(Fig. 2E). Under these conditions no subthreshold oscillation in
membrane potential and no fast postsynaptic potentials were seen. All
these ®ndings are consistent with the hypothesis of an auto-regulation
of intrinsic spiking during bursting.
Changes in the threshold of excitation during bursting
The auto-regulation of spiking could be either a property of
intrinsically ®ring neurons or a general property of neurons in the
cultures independent of their intrinsic activity. In order to distinguish
between these two possibilities, we measured, by whole-cell record-
ing, the excitability of neurons during disinhibition-induced bursting
in dissociated cultures. The neurons selected for these experiments
(n = 13) were not intrinsically spiking in the intervals between the
bursts. The excitability was measured by injecting short current
pulses (50 ms long at 1 Hz) which induced a depolarization just at the
threshold for spike initiation. Both slightly subthreshold and supra-
threshold stimuli were tested to measure changes in excitability in
both directions. The percentage of failures to induce spikes for
constant stimuli was used to express excitability. Approximately half
of the cells (7/13) showed an increase in the failure rate following the
bursts indicating a decrease in excitability (Fig. 3), from which the
cells recovered within 5±10 s. This decrease in excitability went in
parallel with a long-lasting afterhyperpolarization. The other half of
the neurons (6/13) showed a decrease in the failure rate following the
bursts indicating an increase in excitability. In these cells excitability
reached a minimum in the middle of the interval between the bursts.
The low failure rate at the end of the bursts was accompanied by a
slight afterdepolarization. These ®ndings show that modulation of
excitability is not only occurring in intrinsically spiking neurons, and
that excitability can be modi®ed in both directions during bursting.
Excitatory synaptic currents underlying bursting
It has been proposed previously that bursts of activity may be
terminated by a depletion of transmitter causing synaptic depression
(Staley et al., 1998; Fedirchuk et al., 1999). In order to investigate
whether such a mechanism may contribute to disinhibition induced
bursting in dissociated spinal cultures, synaptic currents were
recorded from eight neurons clamped at ±60 mV. Following
disinhibition of the network, bursts consisting of superimposed
synaptic currents appeared (Fig. 4A). Usually, summation of indi-
vidual currents caused a large inward current at the beginning of the
bursts, which was followed by considerably smaller currents of which
the amplitudes gradually declined during the bursts. During the
intervals between the bursts, small currents occurred at rates too low
for a summation of currents. Similar to previous ®ndings (Streit et al.,
2001), the duration of the bursts was positively correlated to the
duration of the preceding intervals (r = 0.53 6 0.08, n = 10, signi®-
cant in 6/10 experiments, Fig. 4B). In these experiments, a stronger
correlation was found between the amplitude of the initial peak
currents of the bursts and the duration of the preceding intervals
FIG. 5. Electrical stimulation by MEA electrodes from a dissociated culture. (A) Original recordings from four MEA channels showing the stimulation artifact(S) and the induced spikes in the ®rst 30 ms following the stimulus. Note that the stimulus-evoked activity is concentrated on the ®rst 15 ms. The location ofthe stimulation (arrow) and recording electrodes (number) is shown in C. (B±D) Spatial patterns of activity evoked by stimulation with two differentintensities (B and C) and two different locations (C and D). Several axons are activated by stimulation at one electrode. The size of the black circlesrepresents the average intensity of the evoked activity in the ®rst 10 ms following the stimulus for each active channel (the channels which recorded activity).Stimulus-evoked activity spans a range from 0 (smallest dots) to 3 events/10 ms (largest circles in C. The upper level is the maximum set by the detectionalgorithm (see Materials and methods). The stimulation electrode (arrow) was not used for recording during stimulation.
676 P. Darbon et al.
ã 2002 Federation of European Neuroscience Societies, European Journal of Neuroscience, 15, 671±683
(r = 0.73 6 0.07, n = 10, signi®cant in 8/10 experiments, Fig. 4C).
Such a correlation suggests that one or several parameters, which
control peak synaptic currents as well as burst duration, are slowly
recovering during the intervals from the previous burst. Because the
peak currents represent a summation of the contribution from several
active presynaptic neurons, such parameters could include the
excitability of the presynaptic neurons, their synaptic ef®cacy, or
both. In order to investigate the changes in synaptic ef®cacy during
the intervals between the bursts, we measured the rate and peak
amplitudes of the synaptic currents occurring spontaneously during
these intervals. The range of peak amplitudes of these currents (10±
60 pA) was comparable to that previously reported for miniature
currents in spinal neurons (Ulrich & LuÈscher, 1993; Vogt et al.,
1995), suggesting that most of these currents were due to spontaneous
release of glutamate at single synapses. During the intervals, no
change in the peak amplitudes of these currents was seen (Fig. 4D).
The difference between the mean peak amplitude of the ®rst 10
currents following a burst and that of the last 10 currents before the
next burst was 0.84 6 0.66% (n = 9 experiments), which was not
signi®cantly different from 0 (P = 0.24). The rate of spontaneous
currents was high following a burst and decreased towards the next
burst within 5±10 s (Fig. 4E). A signi®cant difference (P = 0.0001)
was found between the mean rate of the ®rst 10 currents following a
burst and that of the last 10 currents before the next burst (±39 6 6%,
n = 9 experiments). These ®ndings exclude transmitter depletion and
the desensitization of postsynaptic receptors as parameters involved
in the recovery process in the intervals between the bursts.
Bursts induced by electrical stimulation
Our ®ndings so far suggest that the initiation of bursts requires
intrinsically ®ring cells and the readiness of the network to respond to
their activity. Both intrinsic ®ring and network activation threshold
are modi®ed during the bursts. In order to investigate the relative
contribution of the network activation threshold to the bursting
patterns, we stimulated dissociated cultures under disinhibition.
Single MEA electrodes were addressed as stimulation electrodes and
the response of the network to the stimulation was measured by the
remaining MEA electrodes. The activities evoked in the network by
stimulation were classi®ed on the basis of their latency either in early
activity (latencies < 10 ms), which was not blocked by CNQX and
APV, or in late activity (latencies > 10 ms), which was due to
synaptic transmission, because it disappeared when excitatory
synaptic transmission was blocked. Early activity, which could be
used as a control for the success of stimulation is shown in Fig. 5 for
different stimulus strengths and locations. Usually several electrodes,
which were distributed in a quite large area, responded to the stimuli
with 1±3 events. Only at the threshold of stimulation (0.5 V) could
the responding electrodes be restricted to a small area (Fig. 5) close
to the stimulating electrode. Changing the stimulating electrode
changed the patterns of the responding electrodes. The reason why
several electrodes responded to single stimuli is probably that several
axons cross the stimulating electrode.
In order to evoke a burst, usually stimuli, which were strong enough
to activate several neurons, were required (0.75±1.25 V). However,
even with such strong stimuli, the network was not able to follow the
stimuli when their frequency was too high. On the other hand, when the
frequency of stimulation was too low, spontaneous bursting was
interfering with the stimulation. Figure 6 shows an experiment in
which the rate of spontaneous bursts was very low (» 2 per min). When
this network was stimulated once every 10 s, it reliably followed the
stimuli with long-lasting bursts. When the frequency of stimulation
was increased to one stimulus every 5 s, the network was still able to
reliably follow stimulation, however, with shorter bursts. We found a
positive correlation between the stimulus interval and the burst
duration in dissociated as well as in slice cultures (Fig. 9B; nine
experiments in slice cultures). When the frequency of stimulation was
further increased to one stimulus every 2 s, the network could follow
every second stimulus only. In ®ve experiments in dissociated cultures,
the success rate of stimuli to induce bursting was 49.2 6 3.2% for
stimulus intervals of 2 s and 98.9 6 1.6% for 5 s intervals
(P < 0.001). As reported previously for cortical cultures (Maeda
et al., 1995), stimuli applied during or immediately following a burst
did not usually prolong the bursts or evoke a new burst, respectively. In
dissociated cultures, the frequency of approximately 0.3 Hz seemed to
represent the highest bursting frequency that could be reached by this
network. Even stimulation at 5 Hz or 10 Hz induced bursting at 0.3 Hz
(n = 8). The mean minimum burst period was 5.8 6 0.9 s (range: 3.2±
10.3 s). The period of spontaneous bursting in the absence of
stimulation in the same group of experiments was 9.2 6 4.5 s
(range: 4.7±17.1 s). In all of these experiments, the maximum
frequency of stimulated bursting was faster than spontaneous bursting.
Together, these ®ndings suggest that the readiness of the network to
respond to active neurons is low following the bursts and slowly
increases during the intervals within several seconds. This network
refractory period determines the maximum possible burst frequency,
whereas the actual spontaneous burst frequency is additionally
controlled by the recovery of the intrinsically ®ring neurons.
The low excitability of some neurons and the refractoriness of the
network at the end of the bursts both suggest that the excitability of
the neurons is on average decreasing during the bursts. In order to
investigate the average network excitability during standardized
bursts, high frequency trains of stimuli were applied by one electrode
in the absence of functional synaptic transmission (in the presence of
strychnine, bicuculline, CNQX and APV). As shown before, several
neurons were directly activated by stimulation via one electrode.
Therefore the total amount of the activity in the network within the
®rst 10 ms following a stimulus was taken as a measure of stimulus-
FIG. 6. Network activity plots of stimulus-locked bursting underdisinhibition in a dissociated culture. In the presence of bicuculline (20 mM)and strychnine (1 mM) the network reliably responds to single stimuli withbursts, if the frequency of stimulation is below approximately 0.3 Hz. Notethat the duration of the bursts is decreasing with increasing stimulationfrequency (compare also Fig. 9B). At a frequency of 0.5 Hz every secondstimulus failed to induce bursts, resulting in an evoked burst rate of0.25 Hz. At even higher frequencies (5 Hz) many stimuli failed to inducebursts resulting in an evoked burst frequency of approximately 0.3 Hz. Thespontaneous burst frequency in the absence of stimulation in thisexperiment was low (< 0.1 Hz). The stimulation strength was 1 V.
Bursting activity induced by disinhibition 677
ã 2002 Federation of European Neuroscience Societies, European Journal of Neuroscience, 15, 671±683
induced direct (non-synaptic) excitation of the network. Such
network excitation was usually not affected by stimulus frequencies
up to 5±10 Hz (Fig. 7A). At 20 Hz stimulation it decreased during
the trains (Fig. 7A and B). The decrease occurred within several
seconds and seemed to be also dependent on the history of the train,
i.e. the recovery from previous trains. The decrease in excitability
was not equal for all electrodes that measured activity following the
stimuli (Fig. 7B). This ®nding excludes the possibility that it was due
to high frequency-induced changes in the current ¯ow through the
electrodes. In 11 experiments such a decrease in network excitability
was seen during 20 Hz trains, in some experiments it occurred, to a
lesser extent, already in 10 Hz or even 5 Hz trains. Similar to
intracellular current injection (Fig. 2), trains of 20 Hz extracellular
stimulation suppressed the intrinsic activity (measured in the absence
of synaptic transmission by MEAs) at some of those electrodes,
where spiking was directly induced by stimulation (Fig. 7C, arrows).
In summary, these ®ndings con®rm that, at least in some of the
neurons, spiking at high rates decreases excitability, thus leading, at
the end of a burst, to a suppression of intrinsic spiking on the one
hand and of the responsiveness of the network on the other.
Bursting in slice cultures is ruled by similar mechanisms
The previous ®ndings suggest, that in dissociated spinal cultures,
disinhibition-induced bursting is mainly ruled by intrinsic ®ring and
by the modulation of excitability during bursting, whereas synaptic
depletion or receptor desensitization are not involved. Previously,
however, we have proposed that the fast oscillations seen within
bursts in slice cultures are mainly ruled by synaptic depression (Streit,
1993; Senn et al., 1996; Streit, 1996). To investigate, whether the
bursting in slice cultures is ruled by similar mechanisms as in the
dissociated cultures, we looked for evidence for intrinsic ®ring which
initiates the bursts, and for a modulation of excitability during
bursting in slice cultures. In the dissociated cultures, the suppression
of intrinsic ®ring by the bursts leads to the increase in the rate of
network activity in the intervals towards the next burst (Fig. 1A and
B). In the slice cultures, the rate of activity within the intervals was
lower than in the dissociated cultures (Tscherter et al., 2001).
Nevertheless, the bursts in the slice cultures were often preceded by
intrinsic activity in some channels (Fig. 8A). When the intrinsic
activity in the intervals was measured in time windows of 100 ms, the
windows preceding the bursts contained increasing amounts of
activity. Most of the activity was seen in the last 100 ms before the
bursts (Fig. 8B). Sites with high activity before the bursts were
preferentially situated in the ventral parts of the spinal slices around
the ventral ®ssure (Fig. 8C). This roughly corresponded to the
distribution of the burst sources in the slice (Fig. 8D and E and
Tscherter et al., 2001). When the slices were stimulated by one
electrode, bursts were reliably evoked only in a de®ned range of
FIG. 7. MEA recordings from a dissociated culture showing the response of a network to stimulation (1 V) in the absence of fast synaptic transmission(bicuculline 20 mM, strychnine 1 mM, CNQX 10 mM, APV 50 mM). Several axons are activated by stimulation at one electrode. (A) Response during period ofstimulation with various frequencies from 0.5 to 20 Hz. Note the developing decrease in response during consecutive periods of 20 Hz stimulation. (B)Spatial patterns of activity evoked during a period of 10 Hz stimulation and at the end of period of 20 Hz stimulation, when the network response isdecreased. The size of the black circles represents the average intensity of the evoked activity in the ®rst 10 ms following the stimulus for each activechannel. Stimulus-evoked activity spans a range from 0 (smallest dots) to 3 events/10 ms (largest circles, see also Fig. 5). The location of the stimulatingelectrode is shown by an arrow. Note that the activity at 20 Hz is not reduced at all channels and even increased at one channel. (C) Event raster plots forthree channels during three consecutive periods of 20 Hz stimulation. Note the suppression of intrinsic activity (arrows) following the stimulation periods(bars). Longer stimulus trains seem to provoke stronger suppression. Different experiment from A and B.
678 P. Darbon et al.
ã 2002 Federation of European Neuroscience Societies, European Journal of Neuroscience, 15, 671±683
frequencies. When the frequency was too low, the spontaneous bursts
interfered with the evoked bursts; when the frequency was too high,
stimulation was not reliable any more and some stimuli failed to
evoke bursts (Fig. 9A). Within the range of reliable stimulation
(intervals of 3±35 s, Fig. 9B±E), the duration of the evoked bursts,
the maximum activation of the network during the bursts and the total
activity within the ®rst 10 ms after the stimulus (excitability)
normally increased with increasing intervals between the stimuli,
whereas the time taken to reach maximal activation decreased
(Fig. 9B±E, for nine experiments). This shows that the earlier after
the last burst the network is stimulated, the less neurons can be
activated, the longer it takes to activate the whole network and the
faster the critical threshold for burst termination is reached. All these
®ndings are in line with those previously described in the dissociated
cultures. Therefore they suggest that the same mechanisms are
involved in the control of bursting in slice cultures as in dissociated
cultures.
Discussion
In the present study we investigated the mechanisms involved in
disinhibition-induced bursting in spinal cultures. These ®ndings
con®rm the hypothesis that bursting is due to intrinsic ®ring in a
subset of spinal neurons, which activate the whole network through
recurrent excitation. Such network activation (the burst) is terminated
after one to several seconds mainly due to a decrease in neuronal
excitability. It is followed by a network refractory period (the
interval), which is mainly determined by an auto±regulation of
neuronal excitability and not by synaptic parameters like depletion or
receptor desensitization.
Intrinsic spiking
We have previously shown that at many of the sites, where bursts are
preferentially initiated (burst sources), some low level of activity
persists when fast synaptic transmission is blocked (Streit et al.,
2001). We have suggested that such activity originates from
intrinsically ®ring neurons. Intracellular recordings from such
neurons now show that intrinsic spiking is due to slow depolarizations
between the spikes which can be switched off by a slight
hyperpolarization. Under such conditions, the membrane potential
is ¯at without spontaneous oscillations as seen in motoneurons and
interneurons in the presence of NMDA (Hochman et al., 1994;
Schmidt et al., 1998; Tresch & Kiehn, 2000). The neurons ®red at
rates of 0.3±7 Hz, similar to rates previously reported for mouse
cultures (Latham et al., 2000). However, as the ®ring rate was highly
dependent on the membrane potential, it was variable and could also
cease and reappear spontaneously. While endogenous bursting cells
have been reported in brainstem slices (Koshiya & Smith, 1999) and
their presence is still a matter of debate in mouse cultures (Legendre
et al., 1985; Latham et al., 2000), most of the intrinsically active cells
found in our cultures were tonically active. However, we found
occasional evidence for intrinsic bursters in extracellular MEA
recordings (Streit et al., 2001) and in one whole cell patch recording
(our unpublished observation). It has to be considered that the
intrinsic ®ring patterns may be modulated by the network activity
(Turrigiano et al., 1994).
Auto-regulation of excitability
While intrinsically spiking neurons certainly provided a source of
activity in the network, the patterns of intrinsic active neurons did not
determine its bursting patterns. Instead, bursting was a network
phenomenon based on the recurrent excitation of the network, which
was initiated by the intrinsically ®ring cells. Following the bursts,
intrinsic ®ring is nearly suppressed and recovers in the subsequent
FIG. 8. MEA recordings from an organotypic slice culture showing thatbursts are preceded by an increase in spike rate in the areas of burstinitiation. (A) Event raster plot from 16 channels at the beginning of oneburst. Note the unitary events preceding the burst onset in three channels.(B) Histogram showing the number of events per 100 ms in the networkduring the whole intervals and in the last ®ve bins of 100 ms of the intervalbefore the start of the burst (n = 20). (C) Sites in the slice where the spikerate increased before the bursts. The circles represent the percentage ofbursts with at least one event during the last 100 ms preceding the burst.(D) Sites of burst initiation. The size of the circles shows at whatpercentage bursts started at the location indicated. (E) Average propagationof the wave-front of the bursts in the slice. The lines (isochrones) show howlong it takes for the wavefront of the burst to reach the different electrodes(5.5 ms between 2 lines). The isochrones are extrapolated from measure-ments of the latencies at the electrodes.
Bursting activity induced by disinhibition 679
ã 2002 Federation of European Neuroscience Societies, European Journal of Neuroscience, 15, 671±683
interval. Such a behaviour has been predicted previously in models
based on spike frequency adaptation (Latham et al., 2000) or slow
synaptic depression (Tabak et al., 2000). Spike frequency adaptation
has been described for neurons in the lamprey spinal cord (Elmanira
et al., 1994) and for motoneurons in the cat spinal cord (Kernell,
1965). In rat organotypic cultures spike frequency adaptation was
found in motoneurons (Streit et al., 1991), but not however in spinal
interneurons (Ballerini et al., 1999). Furthermore, it is well known
that neuronal excitability can be in¯uenced by neurotransmitters for
example through Ca2+-dependent K+-conductances (White &
Neuman, 1980; WalleÂn et al., 1989; Knopfel et al., 1990) or by
growth factors like BDNF (Desai et al., 1999a). To investigate a
possible role of such mechanisms in the suppression of intrinsic
spiking following bursts, we simulated bursts in the absence of fast
synaptic transmission by current injection in single neurons and by
high frequency trains of stimulation in the network. These experi-
ments suggest that the suppression of intrinsic activity following
bursts is directly caused by the spiking activity during the bursts, in
the sense of an auto-regulation of excitability, and not by the synaptic
activity. Nevertheless, we found neither evidence for spike frequency
adaptation nor for an accumulation of afterhyperpolarization in these
neurons (Fig. 2B), at least not during current injection. Following the
current pulse, there was a brief afterhyperpolarization, which was
much shorter than the suppression of the intrinsic activity. Thus, the
mechanism that underlies the suppression of intrinsic activity
following bursts remains unclear. It may involve a direct effect of
activity on ionic conductances like Na+ and K+ channels as recently
shown for pyramidal neurons (Desai et al., 1999b), however, on a
much longer time-scale, or it may be related to regulation of
excitability by intracellular Ca2+ (Turrigiano et al, 1994). In line with
FIG. 9. Stimulus-induced bursting in slice cultures which were disinhibited by bicuculline (20 mM) and strychnine (1 mM). (A) Network activity plots showingthat bursts reliably follow external stimuli only in a de®ned range of frequencies. At too low a frequency a spontaneous burst intermingled with the stimulus-induced bursts (upper trace). At too high a frequency some stimuli failed to induce bursts (lower trace). Plots of the burst duration (B), the maximumactivation during the burst (C), the time to reach maximum activation (D), and the network excitability (E) vs. the stimulus interval in nine experiments. Thenetwork excitability was determined by counting the events induced by stimulation at all electrodes in the ®rst 10 ms following the stimulus. All data wereobtained in the range of reliable burst induction.
680 P. Darbon et al.
ã 2002 Federation of European Neuroscience Societies, European Journal of Neuroscience, 15, 671±683
the hypothesis of an auto-regulation of excitability, the recovery of
intrinsic spiking following the cessation of current injections
(Fig. 2D) usually ®rst led to phasic activity after 5±10 s, which
gradually turned into tonic activity (a damped oscillation of the spike
rate, which is characteristic for an unbalanced auto-regulation).
A decrease in excitability following the bursts was not only seen in
intrinsically ®ring neurons, but also in roughly half of the neurons
without intrinsic activity, as indicated by the decrease in response to
depolarizing current pulses (Fig. 3). It went in parallel with a slow
afterhyperpolarization following the bursts (Figs 1 and 3A). A similar
slow afterhyperpolarization following bursts in the hippocampus has
been suggested to be based on Ca2+ -induced K+ conductances (Alger
& Nicoll, 1980; Elmanira et al., 1994; Traub & Jefferys, 1994;
Empson & Jefferys, 2001). However, it may also be related to the
increased activity of the electrogenic Na/K pump following the bursts
(Ballerini et al., 1997). The other half of the cells without intrinsic
activity were depolarized and the response to depolarizing current
pulses was increased at the end of the bursts (Fig. 3B). These ®ndings
are dif®cult to interpret, because the behaviour of the neurons at the
end of the bursts is in¯uenced by their intrinsic properties, as well as
by ongoing activity in the network.
Burst initiation and termination
As previously mentioned, intrinsic spiking must activate the network
through recurrent excitation in order to initiate a burst. When single
or repetitive spikes were evoked in single neurons by current
injection, bursts could not be initiated. However, when several
neurons were activated together by extracellular electrical stimuli,
bursts could be initiated provided the stimuli were not delivered too
soon following the last burst. Besides the frequency of stimulation
(see below) the success rate of stimulation was dependent on the
strength of the stimulus (i.e. the number of stimulated neurons) and
the location of the stimulating electrode (affecting the number and
selection of the stimulated neurons). These ®ndings suggest that in
order to initiate bursts, either several neurons have to be simultan-
eously active or the right neurons (privileged by their connectivity)
must be active. Even when the stimulation site and strength were
optimized to activate the network, burst initiation was frequency
dependent. When the stimulation frequency was too high, some of the
stimuli failed to initiate bursts (Figs 6 and 9). Similar observations
have been made in the intact spinal cord of the neonatal rat for dorsal
root stimulation (Bracci et al., 1997) and in dissociated cortical
cultures (Robinson et al., 1993; Maeda et al., 1995). These ®ndings
suggest, that the bursts are followed by network refractory periods,
which determine the maximum possible burst rate (Maeda et al.,
1995). The duration of these refractory periods correspond well to the
decrease in excitability, suggesting that they are mainly based on the
modulation in excitability. A similar mechanism has been proposed to
underlie the bursting induced by high potassium in hippocampal
slices (Chamberlin et al., 1990; Traub & Dingledine, 1990) and the
slow oscillations in neocortical slices (Sanchez-Vives & McCormick,
2000).
During the bursts there was a slow steady reduction in network
activity (Fig. 1D), which was probably due to a general reduction in
excitability in the network. This ®nally turned into a rapid reduction
which terminated the burst. The turning point probably corresponds
to the time point when the number of activated neurons is not
suf®cient to maintain full network activation (network activation
threshold). In most of the experiments both the burst duration and the
peak activity of the bursts are controlled by the duration of the
preceding interval (Fig. 6). This is most clearly seen in experiments
in which bursting is reliably induced by stimulation at various
frequencies (Fig. 9B). This correlation between burst parameters and
the preceding interval can be explained with the relative network
refractory period. The earlier in this period a new burst is initiated,
the less neurons are initially activated (Fig. 9E) and the sooner the
network activation threshold is reached.
Synaptic depression
Frequency-dependent synaptic depression has been shown to be
prominent in the intact spinal cord of the neonatal rat (Pinco &
Levtov, 1993) as well as in slice cultures of embryonic rat spinal cord
(Streit et al., 1992). In the latter it has been proposed to underlie the
high frequency oscillations, which are often seen within the bursts
(Streit, 1993; Tscherter et al., 2001). In the spinal cord of the chick
embryo as well as in hippocampal slices, refractory periods following
bursts have also been attributed to slowly recovering synaptic
depression (O'Donovan & Rinzel, 1997; Staley et al., 1998; Tabak &
O'Donovan, 1998; Fedirchuk et al., 1999). In dissociated cultures of
mouse spinal cord, however, depression seems to be less prominent in
synapses connecting interneurons (Jia & Nelson, 1986). Synaptic
depression is often attributed to depletion of the readily releasable
pool of vesicles on the presynaptic side (Pieribone et al., 1995;
Dittman & Regehr, 1998; Wu & Borst, 1999), or to the desensitiza-
tion of the postsynaptic receptors (Trussell et al., 1993; Otis et al.,
1996). To investigate a contribution of a slow recovery from synaptic
depression to the network refractory period following the bursts, we
studied the peak amplitudes and the rate of the spontaneous synaptic
currents occurring during the intervals. Such spontaneous currents
have the same range of peak amplitudes as previously reported for
miniature currents, suggesting that most of them are due to
spontaneous release of transmitter in the absence of presynaptic
spikes (Ulrich & LuÈscher, 1993; Vogt et al., 1995). However, we
have shown before, that spikes occur during the intervals, especially
towards the end, suggesting that a small percentage of these currents
could be evoked by presynaptic spikes. In both cases, the rate and
mean amplitude of such currents should be decreased, when the
readily releasable pool of vesicles were depleted following the bursts
(Highstein & Bennett, 1975; Glavinovic, 1995; Zhou et al., 2000).
This was clearly not seen (Fig. 4E), suggesting that depletion was not
involved. In contrast, the rate of currents was increased following the
bursts and decayed during the intervals. This could be explained by
the high Ca2+ concentration in the presynaptic terminals following
the bursts, as it has been shown that high presynaptic Ca2+
concentration increases the rate of miniature potentials (Erulkar &
Rahamimoff, 1978; Llano et al., 2000). On the other hand, if the
postsynaptic receptors are desensitized following the bursts, the
peak amplitudes of the currents in the intervals should be
decreased. Again, this was not the case (Fig. 4D), excluding such
a mechanism. In summary, we found no evidence for a slow
recovery from synaptic depression during the intervals between
the burst.
In summary, our results support the hypothesis, that the generation
and termination of bursts induced by disinhibition in spinal cultures
(dissociated and organotypic) is mainly controlled by intrinsic spiking
in some neurons, recurrent excitation through glutamatergic synaptic
transmission and auto±regulation of neuronal excitability.
Acknowledgements
We would like to thank Ruth Rubli for carrying out the culture work, MarcHeuschkel for producing the arrays, Denis de Limoges, Hans Ruchti, JuÈrg
Bursting activity induced by disinhibition 681
ã 2002 Federation of European Neuroscience Societies, European Journal of Neuroscience, 15, 671±683
Burkhalter and Hans-Ueli Schweizer for the electronic and mechanicalequipment of the recording set-up, Nissim Buchs for software support and DrLucinda Davies and Dr David Thurbon for helpful comments on themanuscript.
This work was supported by the Swiss National Science Foundation (SNFgrants no. 31±59080.99 and 3152±053761.98).
Abbreviations
APV, (+/±)-2-amino-5-phosphonopentanoic acid; AMPA, (+/±)-a-amino-3-hydroxy-5-methylisoxazole-4-proprionic acid hydrobromide; CNQX, 6-cyano-7-nitroquinoxaline-2-3-dione; GABA, g-aminobutyric acid; HEPES,4-(2-hydroxyethyl)-piperazin-1-ethansulphonic acid; 5-HT, 5-hydroxytrypta-mine (serotonin); MEA, multielectrode array; NMDA: N-methyl-D-asparticacid; r, correlation coef®cient; TTX, tetrodotoxin.
References
Alger, B.E. & Nicoll, R.A. (1980) Epileptiform burst afterhyperolarization:calcium-dependent potassium potential in hippocampal CA1 pyramidalcells. Science, 210, 1122±1124.
Ballerini, L., Bracci, E.M. & Nistri, A. (1997) Pharmacological block ofelectrogenic sodium pump disrupts rhythmic bursting induced by strychnineand bicuculline in the neonatal rat spinal cord. J. Neurophysiol., 77, 17±23.
Ballerini, L. & Galante, M. (1998) Network bursting by organotypic spinalslice cultures in the presence of bicuculline and/or strychnine isdevelopmentally regulated. Eur. J. Neurosci., 10, 2871±2879.
Ballerini, L., Galante, M., Grandolfo, M. & Nistri, A. (1999) Generation ofrhythmic patterns of activity by ventral interneurones in rat organotypicspinal slice culture. J. Physiol. (Lond.), 517, 459±475.
Beato, M. & Nistri, A. (1999) Interaction between disinhibited bursting and®ctive locomotor patterns in the rat isolated spinal cord. J. Neurophysiol.,82, 2029±2038.
Bracci, E., Ballerini, L. & Nistri, A. (1996) Spontaneous rhythmic burstsinduced by pharmacological block of inhibition in lumbar motoneurons ofthe neonatal rat spinal cord. J. Neurophysiol., 75, 640±647.
Bracci, E., Beato, M. & Nistri, A. (1997) Afferent inputs modulate the activityof a rhythmic burst generator in the rat disinhibited spinal cord in vitro. J.Neurophysiol., 77, 3157±3167.
Braschler, U.F., Iannone, A., Spenger, C., Streit, J. & LuÈscher, H.-R. (1989) Amodi®ed roller tube technique for organotypic cocultures of embryonic ratspinal cord, sensory ganglia and skeletal muscle. J. Neurosci. Meth., 29,121±129.
Cazalets, J.R., Borde, M. & Clarac, F. (1995) Localization and organization ofthe central pattern generator for hindlimb locomotion in newborn rat. J.Neurosci., 15, 4943±4951.
Chamberlin, N.L., Traub, R.D. & Dingledine, R. (1990) Spontaneous EPSPsinitiate burst-®ring in rat hippocampal neurons bathed in high potassium. J.Neurophysiol., 64, 1000±1008.
Dale, N. & Kuenzi, F. (1997) Ionic currents, transmitters and models of motorpattern generators. Curr. Opin. Neurobiol., 7, 790±796.
Desai, N.S., Rutherford, L.C. & Turrigiano, G.G. (1999a) BDNF regulates theintrinsic excitability of cortical neurons. Learn Mem., 6, 284±291.
Desai, N.S., Rutherford, L.C. & Turrigiano, G.G. (1999b) Plasticity in theintrinsic excitability of cortical pyramidal neurons. Nature Neurosci., 2,515±520.
Dittman, J.S. & Regehr, W.G. (1998) Calcium dependence and recoverykinetics of presynaptic depression at the climbing ®ber to Purkinje cellsynapse. J. Neurosci., 18, 6147±6162.
Droge, M.H., Gross, G.W., Hightower, M.H. & Czisny, L.E. (1986)Multielectrode analysis of coordinated, multisite, rhythmic bursting incultured CNS monolayer networks. J. Neurosci., 6, 1583±1592.
Elmanira, A., Tegner, J. & Grillner, S. (1994) Calcium-dependent potassiumchannels play a critical role for burst termination in the locomotor networkin lamprey. J. Neurophysiol., 72, 1852±1861.
Empson, R.M. & Jefferys, J.G. (2001) Ca(2+) entry through L-type Ca(2+)channels helps terminate epileptiform activity by activation of a Ca(2+)dependent afterhyperpolarisation in hippocampal CA3. Neuroscience, 102,297±306.
Erulkar, S.D. & Rahamimoff, R. (1978) The role of calcium ions in tetanic andpost-tetanic increase of miniature end-plate potential frequency. J. Physiol.(Lond.), 278, 501±511.
Fedirchuk, B., Wenner, P., Whelan, P.J., Ho, S., Tabak, J. & O'Donovan, M.J.
(1999) Spontaneous network activity transiently depresses synaptictransmission in the embryonic chick spinal cord. J. Neurosci., 19, 2102±2112.
Glavinovic, M.I. (1995) Decrease of quantal size and quantal content duringtetanic stimulation detected by focal recording. Neuroscience, 69, 271±281.
Grillner, S., Parker, D. & el Manira, A. (1998) Vertebrate locomotion: alamprey perspective. Ann. N Y Acad. Sci., 860, 1±18.
Grillner, S. & WalleÂn, P. (1985) Central pattern generators for locomotion,with special reference to vertebrates. Annu. Rev. Neurosci., 8, 233±261.
Gross, G.W., Rhoades, B.K. & Kowalski, J.M. (1993a) Dynamics of burstpatterns generated by monolayer networks in culture. In Bothe, H.-W.,Samii, M. & Eckmiller, R. (eds), Neurobionics. Elsevier, Amsterdam, pp.89±121.
Gross, G.W., Rhoades, B.K., Reust, D.L. & Schwalm, F.U. (1993b)Stimulation of monolayer networks in culture through thin-®lm indium-tin oxide recording electrodes. J. Neurosci. Meth., 50, 131±143.
Gross, G.W., Williams, A.N. & Lucas, J.H. (1982) Recording of spontaneousactivity with photoetched microelectrode surfaces from mouse spinalneurons in culture. J. Neurosci. Meth., 5, 13±22.
Hamill, O.P., Marty, A., Neher, E., Sakmann, B. & Sigworth, F.J. (1981)Improved patch-clamp techniques for high-resolution current recordingfrom cells and cell-free membrane patches. P¯uÈgers Arch., 391, 85±100.
Highstein, S.M. & Bennett, M.V. (1975) Fatigue and recovery of transmissionat the Mauthner ®ber-giant ®ber synapse of the hatchet®sh. Brain Res., 98,229±242.
Hochman, S., Jordan, L.M. & Schmidt, B.J. (1994) TTX-resistant NMDAreceptor-mediated voltage oscillations in mammalian lumbar motoneurons.J. Neurophysiol., 72, 2559±2562.
Jia, M. & Nelson, P.G. (1986) Calcium currents and transmitter output incultured spinal cord and dorsal root ganglion neurons. J. Neurophysiol., 56,1257±1267.
Jimbo, Y., Robinson, H.P. & Kawana, A. (1993) Simultaneous measurementof intracellular calcium and electrical activity from patterned neuralnetworks in culture. IEEE Trans. Biomed. Eng., 40, 804±810.
Kernell, D. (1965) High-frequency repetitive ®ring of cat lumbosacralmotoneurons stimulated by long-lasting injected currents. Acta Physiol.Scand., 65, 74±86.
Kiehn, O. & Kjaerulff, O. (1998) Distribution of central pattern generators forrhythmic motor outputs in the spinal cord of limbed vertebrates. Ann. N YAcad. Sci., 860, 110±129.
Knopfel, T., Vranesic, I., Gahwiler, B.H. & Brown, D.A. (1990) Muscarinicand beta-adrenergic depression of the slow Ca2+-activated potassiumconductance in hippocampal CA3 pyramidal cells is not mediated by areduction of depolarization-induced cytosolic Ca2+ transients. Proc. Natl.Acad. Sci. USA, 87, 4083±4087.
Koshiya, N. & Smith, J.C. (1999) Neuronal pacemaker for breathingvisualized in vitro. Nature, 400, 360±363.
Latham, P.E., Richmond, B.J., Nirenberg, S. & Nelson, P.G. (2000) Intrinsicdynamics in neuronal networks. II. experiment. J. Neurophysiol., 83, 828±835.
Legendre, P., Guzman, A., Dupouy, B. & Vincent, J.D. (1989) Excitatoryeffect of serotonin on pacemaker neurons in spinal cord cell culture.Neuroscience, 28, 201±209.
Legendre, P., McKenzie, J.S., Dupouy, B. & Vincent, J.D. (1985) Evidence forbursting pacemaker neurones in cultured spinal cord cells. Neuroscience,16, 753±767.
Llano, I., Gonzalez, J., Caputo, C., Lai, F.A., Blayney, L.M., Tan, Y.P. &Marty, A. (2000) Presynaptic calcium stores underlie large-amplitudeminiature IPSCs and spontaneous calcium transients. Nature Neurosci., 3,1256±1265.
Maeda, E., Robinson, H.P.C. & Kawana, A. (1995) The mechanisms ofgeneration and propagation of synchronized bursting in developingnetworks of cortical neurons. J. Neurosci., 15, 6834±6845.
MuÈller, W. & Swandulla, D. (1995) Synaptic feedback excitation hashypothalamic neural networks generate quasirhythmic burst activity. J.Neurophysiol., 73, 855±861.
O'Brien, R.J. & Fischbach, G.D. (1986) Excitatory synaptic transmissionbetween interneurons and motoneurons in chick spinal cord cell cultures. J.Neurosci., 6, 3284±3289.
O'Donovan, M.J. & Rinzel, J. (1997) Synaptic depression: a dynamicregulator of synaptic communication with varied functional roles. TrendsNeurosci., 20, 431±433.
Otis, T., Zhang, S. & Trussell, L.O. (1996) Direct measurement of AMPAreceptor desensitization induced by glutamatergic synaptic transmission. J.Neurosci., 16, 7496±7504.
682 P. Darbon et al.
ã 2002 Federation of European Neuroscience Societies, European Journal of Neuroscience, 15, 671±683
Pieribone, V.A., Shupliakov, O., Brodin, L., Hil®ker-Rothen¯uh, S., Czernik,A.J. & Greengard, P. (1995) Distinct pools of synaptic vesicles inneurotransmitter release. Nature, 375, 493±497.
Pinco, M. & Levtov, A. (1993) Modulation of monosynaptic excitation in theneonatal rat spinal cord. J. Neurophysiol., 70, 1151±1158.
Ransom, B.R., Christian, C.N., Bullock, P.N. & Nelson, P.G. (1977) Mousespinal cord in cell culture. II. Synaptic activity and circuit behavior. J.Neurophysiol., 40, 1151±1161.
Robinson, H.P., Torimitsu, K., Jimbo, Y., Kuroda, Y. & Kawana, A. (1993)Periodic bursting of cultured cortical neurons in low magnesium: cellularand network mechanisms. Jpn. J. Physiol., 43, S125±S130.
Sanchez-Vives, M.V. & McCormick, D.A. (2000) Cellular and networkmechanisms of rhythmic recurrent activity in neocortex. Nature Neurosci.,3, 1027±1034.
Schmidt, B.J., Hochman, S. & MacLean, J.N. (1998) NMDA receptor-mediated oscillatory properties: potential role in rhythm generation in themammalian spinal cord. Ann. N Y Acad. Sci., 860, 189±202.
Senn, W., Wyler, K., Streit, J., Larkum, M., LuÈscher, H.-R., Mey, H., MuÈller,L., Stainhauser, D., Vogt, K. & Wannier, T. (1996) Dynamics of a randomneural network with synaptic depression. Neural Networks, 9, 575±588.
Spenger, C., Braschler, U.F., Streit, J. & LuÈscher, H.-R. (1991) Anorganotypic spinal cord-dorsal root ganglion-skeletal muscle coculture ofembryonic rat. I. The morphological correlates of the spinal re¯ex arc. Eur.J. Neurosci., 3, 1037±1053.
Staley, K.J., Longacher, M., Bains, J.S. & Yee, A. (1998) Presynapticmodulation of CA3 network activity. Nature Neurosci., 1, 201±209.
Streit, J. (1993) Regular oscillations of synaptic activity in spinal networksin vitro. J. Neurophysiol., 70, 871±878.
Streit, J. (1996) Mechanisms of pattern generation in cocultures of embryonicspinal cord and skeletal muscle. Int. J. Dev. Neurosci., 14, 137±148.
Streit, J., LuÈscher, C. & LuÈscher, H.-R. (1992) Depression of postsynapticpotentials by high frequency stimulation in embryonic motoneurons grownin spinal cord slice cultures. J. Neurophysiol., 68, 1793±1803.
Streit, J., Spenger, C. & LuÈscher, H.-R. (1991) An organotypic spinal cord-dorsal root ganglia-skeletal muscle coculture of embryonic rat. II.Functional evidence for the formation of spinal re¯ex arcs in vitro. Eur.J. Neurosci., 3, 1054±1068.
Streit, J., Tscherter, A., Heuschkel, M.O. & Renaud, P. (2001) The generationof rhythmic activity in dissociated cultures of rat spinal cord. Eur. J.Neurosci., 14, 191±202.
Tabak, J. & O'Donovan, M.J. (1998) Statistical analysis and intersegmentaldelays reveal possible roles of network depression in the generation ofspontaneous activity in the chick embryo spinal cord. Ann. N Y Acad. Sci.,860, 428±431.
Tabak, J., Senn, W., O'Donovan, M.J. & Rinzel, J. (2000) Modeling of
spontaneous activity in developing spinal cord using activity-dependentdepression in an excitatory network. J. Neurosci., 20, 3041±3056.
Traub, R.D. & Dingledine, R. (1990) Model of synchronized epileptiformbursts induced by high potassium in CA3 region of rat hippocampalslilce. Role of spontaneous EPSPs in initiation. J. Neurophysiol., 64,1009±1018.
Traub, R.D. & Jefferys, J.G. (1994) Are there unifying principles underlyingthe generation of epileptic afterdischarges in vitro? Prog. Brain Res., 102,383±394.
Tresch, M.C. & Kiehn, O. (2000) Motor coordination without action potentialsin the mammalian spinal cord. Nature Neurosci., 3, 593±599.
Trussell, L.O., Zhang, S. & Raman, I.M. (1993) Desensitization of AMPAreceptors upon multiquantal neurotransmitter release. Neuron, 10, 1185±1196.
Tscherter, A., Heuschkel, M.O., Renaud, P. & Streit, J. (2001) Spatiotemporalcharacterization of rhythmic activity in spinal cord slice cultures. Eur. J.Neurosci., 14, 179±190.
Turrigiano, G., Abbott, L.F. & Marder, E. (1994) Activity-dependent changesin the intrinsic properties of cultured neurons. Science, 264, 974±977.
Ulrich, D. & LuÈscher, H.-R. (1993) Miniature excitatory synaptic currentscorrected for dendritic cable properties reveal quantal size and variance. J.Neurophysiol., 69, 1769±1773.
Vogt, K., LuÈscher, H.-R. & Streit, J. (1995) Analysis of synaptic transmissionat single identi®ed boutons on rat spinal neurons in culture. P¯uÈgers Arch.,430, 1022±1028.
WalleÂn, P., Christenson, J., Brodin, L., Hill, R., Lansner, A. & Grillner, S.(1989) Mechanisms underlying the serotonergic modulation of the spinalcircuitry for locomotion in lamprey. Prog. Brain Res., 80, 321±327.
WalleÂn, P., Ekeberg, O., Lansner, A., Brodin, L., TraÊveÂn, H. & Grillner, S.(1992) A computer-based model for realistic simulations of neuralnetworks. II. The segmental network generating locomotor rhythmicity inthe lamprey. J. Neurophysiol., 68, 1939±1950.
White, S.R., Fung, S.J., Jackson, D.A. & Imel, K.M. (1996) Serotonin,norepinephrine and associated neuropeptides: effects on somaticmotoneuron excitability. Prog. Brain Res., 107, 183±199.
White, S.R. & Neuman, R.S. (1980) Facilitation of spinal motoneuroneexcitability by 5-hydroxytryptamine and noradrenaline. Brain Res., 188,119±127.
Wu, L.G. & Borst, J.G. (1999) The reduced release probability of releasablevesicles during recovery from short-term synaptic depression. Neuron, 23,821±832.
Zhou, Q., Petersen, C.C. & Nicoll, R.A. (2000) Effects of reduced vesicular®lling on synaptic transmission in rat hippocampal neurones. J. Physiol.(Lond.), 525, 195±206.
Bursting activity induced by disinhibition 683
ã 2002 Federation of European Neuroscience Societies, European Journal of Neuroscience, 15, 671±683