1Department of Neurological Surgery and 2Department of Physiology and Biophysics, University of Washington, Seattle, Washington 98195
![]() |
ABSTRACT |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Hochman, Daryl W. and
Philip A. Schwartzkroin.
Chloride-Cotransport Blockade Desynchronizes Neuronal Discharge
in the "Epileptic" Hippocampal Slice.
J. Neurophysiol. 83: 406-417, 2000.
Antagonism of the
chloride-cotransport system in hippocampal slices has been shown to
block spontaneous epileptiform (i.e., hypersynchronized) discharges
without diminishing excitatory synaptic transmission. Here we test the
hypotheses that chloride-cotransport blockade, with furosemide or
low-chloride (low-[Cl]o) medium,
desynchronizes the firing activity of neuronal populations and that
this desynchronization is mediated through nonsynaptic mechanisms.
Spontaneous epileptiform discharges were recorded from the CA1 and CA3
cell body layers of hippocampal slices. Treatment with
low-[Cl
]o medium led to cessation of
spontaneous synchronized bursting in CA1
5-10 min before its
disappearance from CA3. During the time that CA3 continued to burst
spontaneously but CA1 was silent, electrical stimulation of the
Schaffer collaterals showed that hyperexcited CA1 synaptic responses
were maintained. Paired intracellular recordings from CA1 pyramidal
cells showed that during low-[Cl
]o
treatment, the timing of action potential discharges became desynchronized; desynchronization was identified with phase lags in
firing times of action potentials between pairs of neurons as well as a
with a broadening and diminution of the CA1 field amplitude. Continued
exposure to low-[Cl
]o medium increased the
degree of the firing-time phase shifts between pairs of CA1 pyramidal
cells until the epileptiform CA1 field potential was abolished
completely. Intracellular recordings during 4-aminopyridine (4-AP)
treatment showed that prolonged low-[Cl
]o
exposure did not diminish the frequency or amplitude of spontaneous postsynaptic potentials. CA3 antidromic responses to Schaffer collateral stimulation were not significantly affected by prolonged low-[Cl
]o exposure. In contrast to CA1,
paired intracellular recordings from CA3 pyramidal cells showed that
chloride-cotransport blockade did not cause a significant
desynchronization of action potential firing times in the CA3 subregion
at the time that CA1 synchronous discharge was blocked but did reduce
the number of action potentials associated with CA3 burst discharges.
These data support our hypothesis that the anti-epileptic effects of
chloride-cotransport antagonism in CA1 are mediated through the
desynchronization of population activity. We hypothesize that
interference with Na+,K+,2Cl
cotransport results in an increase in extracellular potassium ([K+]o) that reduces the number of action
potentials that are able to invade axonal arborizations and
varicosities in all hippocampal subregions. This reduced efficacy of
presynaptic action potential propagation ultimately leads to a
reduction of synaptic drive and a desynchronization of the firing of
CA1 pyramidal cells.
![]() |
INTRODUCTION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Synchronization of spontaneous neuronal firing
activity is thought to be an important feature of a number of normal
and pathological processes in the CNS. Examples include synchronized
oscillations of population activity such as gamma rhythms in neocortex,
which are thought to be involved in cognition (Singer and Gray
1995), and theta rhythm in hippocampus, which is thought to
play roles in spatial memory and in the induction of synaptic
plasticity (Huerta and Lisman 1995
, 1996
;
O'Keefe 1993
). Epileptiform activity is identified with
spontaneously occurring hypersynchronized discharges of neuronal
populations (Traub and Jefferys 1994
). Most research on
the processes underlying the generation and maintenance of spontaneous
synchronized activity has focused on synaptic mechanisms (Abeles
et al. 1994
; Bland and Colom 1993
;
McNamara 1994
; Stanford et al.
1998
; Traub and Miles 1991
;
Williams and Kauer 1997
). However, much data have
supported the notion that nonsynaptic mechanisms also could play
important roles in the modulation of synchronization in normal and
pathological activities in the CNS (Andrew 1991
; Dudek and Traub 1989
; Faber and Korn
1989
; Jeffreys 1995
).
In previous reports, we have shown that inhibition of cation chloride
cotransport, either by a reduction of extracellular chloride
([Cl]o) or by treatment with transporter
antagonists such as loop-diuretics (furosemide and bumetanide), blocks
spontaneous epileptiform activity (Hochman et al. 1995
,
1999
; Schwartzkroin et al. 1998
). Further, because furosemide and low-[Cl
]o
effectively blocked spontaneous bursting elicited by a variety of
treatments that work through a spectrum of different synaptic mechanisms, it appears that chloride-cotransport antagonism affects general processes critical for maintenance of spontaneous synchronized neuronal activity. Several observations from our previous studies on
hippocampal slices suggest that the anti-epileptic effects of
furosemide and low-[Cl
]o are mediated by
nonsynaptic mechanisms: excitatory synaptic transmission (as measured
by field responses of area CA1 to Schaffer collateral stimulation) is
not diminished; spontaneous field shifts induced by
low-Ca2+ treatment
a type of epileptiform activity thought
to be mediated through nonsynaptic mechanisms (Jefferys and Haas
1982
; Taylor and Dudek 1982
)
can be blocked;
and activity-evoked intrinsic optical changes of the tissue are blocked
concomitant with the blockade of spontaneous epileptiform activity,
suggesting that changes in the volume fraction of the extracellular
space (ECS) or ionic fluxes associated with cell volume regulation
might be related to the anti-epileptic effects of furosemide
(Holthoff and Witte 1996
; MacVicar and Hochman
1991
).
Our previous study (Hochman et al. 1999) suggested that
it was the blockade of the
Na+,K+,2Cl
cotransporter that was
critical to the anti-epileptic effects of chloride-cotransport
antagonism. Further, prolonged exposure of the tissue to
low-[Cl
]o medium or furosemide resulted in
blockade of spontaneous synchronized population discharges, even though
individual cells continued to discharge in bursts of action potentials.
Two hypotheses followed from these observations: the anti-epileptic
effects of furosemide and low-[Cl
]o
treatments are a result of the desynchronization of the activity of the
neuronal population, and this desynchronization is independent of
effects on excitatory synaptic interactions. In the present study, we
test these hypotheses by characterizing the effects of
low-[Cl
]o and furosemide on the
synchronization of spontaneous and stimulation-evoked action potential
discharges and spontaneous synaptic activity in the CA1 and CA3 regions
of hippocampal slices.
![]() |
METHODS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Slices were prepared from Sprague-Dawley adult rats as in
previous studies (Hochman et al. 1995, 1999
). Transverse
hippocampal slices 400-µm thick were cut with a vibrating cutter.
Slices typically contained the entire hippocampus and subiculum. After
cutting, slices were stored in an oxygenated holding chamber at room
temperature for
1 h before recording. All recordings were acquired in
an interface chamber with oxygenated (95% O2-5%
CO2) artificial cerebrospinal fluid (ACSF) at
34-35°C. Normal ACSF contained (in mmol/l) 124 NaCl, 3 KCl, 1.25 NaH2PO4, 1.2 MgSO4, 26 NaHCO3, 2 CaCl2, and 10 dextrose.
Sharp-electrodes for intracellular recordings from CA1 and CA3 pyramidal cells were filled with 4 M potassium acetate. Field recordings from the CA1 and CA3 cell body layers were acquired with low-resistance glass electrodes filled with 2 M NaCl. For stimulation of the Shaffer collateral or hilar pathways, a small monopolar tungsten electrode was placed on the surface of the slice. Spontaneous and stimulation-evoked activities from field and intracellular recordings were digitized (Neurocorder, Neurodata Instruments, New York, NY) and stored on videotape. AxoScope software (Axon Instruments) on a personal computer was used for off-line analysis of data.
In some experiments, normal or
low-[Cl]o medium was
used containing bicuculline (20 µM), 4-amino pyridine (4-AP; 100 µM), or high-K+ (7.5 or 12 mM). In all of our
experiments, low-[Cl
]o
solutions (7 and 21 mM) were prepared by equimolar replacement of NaCl
with Na+-gluconate (Sigma) (see Hochman et
al. 1999
for rationale). All solutions were prepared so that
they had a pH of ~7.4 and an osmolarity of 290-300 mosmol at 35°C
at equilibrium from carboxygenation with 95%
O2-5% CO2.
After placement in the interface chamber, slices were superfused at ~1 ml/min. At this flow-rate, it took 8-10 min for changes in the perfusion media to be completed. All of the times reported here have taken this delay into account and have an error of approximately ±2 min. This accuracy was sufficient for the purposes of this study.
![]() |
RESULTS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Timing of cessation of spontaneous epileptiform bursting in areas CA1 and CA3
The relative contributions of the factors that modulate
synchronized activity vary between areas CA1 and CA3. These factors include differences in the local circuitry (Schwartzkroin
1993) and in region-specific differences in cell packing and
volume fractions of the extracellular spaces (McBain et al.
1990
). If the anti-epileptic effects of chloride-cotransport
antagonism are due to a desynchronization in the timing of neuronal
discharge, chloride-cotransport blockade might be expected to
differentially affect areas CA1 and CA3. To test this hypothesis, we
performed a set of experiments to characterize differences in the
timing of the blockade of spontaneous epileptiform activity in areas CA1 and CA3 (Fig. 1).
|
Field activity was recorded simultaneously in areas CA1 and CA3b
(approximately midway between the most proximal and distal extent the
CA3 region), and spontaneous bursting was induced by treatment with
high-[K+]o (12 mM;
n = 12), bicuculline (20 µM; n = 12),
or 4-AP (100 µM; n = 5). Single electrical stimuli
were delivered to the Schaffer collaterals, midway between areas CA1
and CA3, every 30 s so that the field responses in areas CA1 and
CA3 could be monitored throughout the duration of each experiment. In
all experiments, 20 min of continuous spontaneous epileptiform
bursting was observed before switching to
low-[Cl
]o (21 mM) or
furosemide-containing (2.5 mM) medium.
In all cases, after 30-40 min exposure to furosemide or low-chloride medium, spontaneous bursting ceased in area CA1 before the bursting ceased in area CA3 (cf. Fig. 1, A and B). The temporal sequence of events typically observed included an initial increase in burst frequency and amplitude of the spontaneous field events, then a reduction in the amplitude of the burst discharges that was more rapid in CA1 than in CA3. After CA1 became silent, CA3 continued to discharge for 5-10 min, until it too no longer exhibited spontaneous epileptiform events (Fig. 1C).
This temporal pattern of burst cessation was observed with all
epileptiform-inducing treatments tested, regardless of whether the
agent used for blockade of spontaneous bursting was furosemide or
low-[Cl]o medium.
Throughout all stages of these experiments, stimulation of the Schaffer
collaterals evoked hyperexcited field responses in both the CA1 and CA3
cell body layers. Immediately after spontaneous bursting was blocked in
both areas CA1 and CA3, hyperexcited population spikes still could be
evoked (Fig. 1C) as was reported previously (Hochman
et al. 1999
).
We considered the possibility that the observed cessation of bursting in CA1 before CA3 was an artifact of the organization of synaptic contacts between these areas relative to our choice of recording sites. If the cessation of bursting occurs in the different subregions of CA3 at different times, the results of the aforementioned set of experiments might arise not as a difference between CA1 and CA3 but rather as a function of variability in bursting activity across CA3 subregions. Therefore immediately after the spontaneous bursting ceased in CA1, we surveyed the CA3 field with a recording electrode. Recordings from several different CA3 locations (from the most proximal to the most distal portions of CA3), showed that all subregions of area CA3 were spontaneously bursting during this time that CA1 was silent (not shown).
Effects of reduced extracellular chloride on the synchronization of CA1 and CA3 field population discharges
The observations from the previous set of experiments suggested a
temporal relationship between the exposure-time to
low-[Cl]o or
furosemide-containing medium and the characteristics of the spontaneous
burst activity. Further, this relationship was different between areas
CA1 and CA3. To better characterize these temporal relationships, we
compared the occurrences of CA1 action potentials and the population
spike events in the field responses of CA1 and CA3 subfields during
spontaneous and stimulation-evoked burst discharges. Figure
2 shows CA1 intracellular activity in relationship to spontaneous burst discharges (fields) during
low-[Cl
]o blockade of
bicuculline-induced (20 µM) activity.
|
Intracellular recordings were obtained from CA1 pyramidal cells, with
the intracellular electrode placed close (<100 µm) to the CA1 field
electrode. The slice was stimulated every 20 s with single stimuli
delivered to the Schaffer collaterals. After continuous spontaneous
bursting was established for 20 min, the bathing medium was switched
to bicuculline-containing low-chloride (21 mM) medium. After ~20 min,
the burst frequency and amplitude was at its greatest. Simultaneous
field and intracellular recordings during this time (Fig.
2A) showed that the CA1 field and intracellular recordings
were synchronized closely with the CA3 field discharges. During each
spontaneous discharge, the CA3 field response preceded the CA1
discharge by several milliseconds (Fig. 2A, left). During stimulation-evoked events, action potential discharges of the CA1
pyramidal cell were closely synchronized to both CA3 and CA1 field
discharges (Fig. 2A, right).
With continued exposure to
low-[Cl]o medium, the
latency between the spontaneous discharges of areas CA1 and CA3
increased, with a maximum latency of 30-40 ms occurring after 30- to
40-min exposure to the bicuculline-containing low-chloride medium (Fig. 2B, left). During this time, the amplitude of both the CA1
and CA3 spontaneous field discharges decreased. Stimulation-evoked discharges during this time (Fig. 2B, right) closely
mimicked the spontaneously occurring discharges in morphology and
relative latency. However, the initial stimulus-evoked depolarization
of the neuron [presumably, the monosynaptic excitatory postsynaptic potential (EPSP)] began without any significant increase in latency (Fig. 2B, bottom right). The time interval during
which these data were acquired corresponds to the time interval shown
at the beginning of the traces in Fig. 1B immediately before
the cessation of spontaneous bursting in CA1.
After 40- to 50-min perfusion with low-[Cl-] medium, the spontaneous bursts were nearly abolished in CA1 but were unaffected in CA3. Schaffer collateral stimulation during this time showed that monosynaptically triggered responses of CA1 pyramidal cells occurred without any significant increase in latency but that stimulation-evoked field responses were almost abolished (Fig. 2C). The time interval during which these data were acquired corresponds to the moments immediately before the cessation of spontaneous bursting in CA3, shown at the beginning of the traces in Fig. 1C.
After prolonged exposure to low-[Cl-] medium, large increases (>30
ms) developed in the latency between Schaffer collateral stimulation
and the consequent CA3 field discharge (Fig.
3, left). Eventually, no field
responses could be evoked by Schaffer collater stimulation in either
areas CA1 or CA3 (Fig. 3, right). However, action potential
discharge from CA1 pyramidal cells, in response to Schaffer collateral
stimulation, could be evoked with little change in response latency.
Indeed, for the entire duration of our experiments (>2 h), action
potential discharges from CA1 pyramidal cells could be evoked at short
latency by Schaffer collateral stimulation. Further, although
stimulation-evoked hyperexcited discharges of CA3 eventually were
blocked after prolonged exposure to
low-[Cl]o medium, the
antidromic response in CA3 appeared to be preserved (Fig. 3,
top).
|
Effects of chloride-cotransport antagonism on the synchronization of burst discharges in CA1 pyramidal cells
These foregoing data suggest that the disappearance of the field responses may be due to a desynchronization of the occurrence of action potentials among neurons. That is, although synaptically driven excitation of CA1 pyramidal cells was preserved, action potential synchrony among the CA1 neuronal population was not sufficient to summate into a measurable DC field response. To test this hypothesis, paired intracellular recordings of CA1 pyramidal cells were acquired simultaneously with CA1 field responses. In these experiments, both of the intracellular electrodes and the field recording electrodes were placed within 200 µm of one another.
During the period of maximum spontaneous activity induced by
bicuculline-containing
low-[Cl]o medium,
recordings showed that action potentials between pairs of CA1 neurons
and the CA1 field discharges were synchronized tightly both during
spontaneous and stimulation-evoked discharges (Fig.
4A). After continued exposure
to low-chloride medium, when the amplitude of the CA1 field discharge
began to broaden and diminish in amplitude, both spontaneous and
stimulation-evoked discharges showed a desynchronization in the timing
of the occurrences of action potentials between pairs of CA1 neurons,
and between the action potentials and the field responses (Fig.
4B). This desynchronization was coincident with the
suppression of CA1 field amplitude. By the time that spontaneous
bursting in CA1 ceased, a significant increase in latency had developed
between Schaffer collateral stimulation and CA1 field discharge (Fig.
4C). At this time, paired intracellular recordings showed a
dramatic desyncronization in the timing of action potential discharges
between pairs of neurons and between the occurrence of action
potentials and the field discharges evoked by Schaffer collateral
stimulation.
|
It is possible that the observed desynchronization of CA1 action potential discharge is due to the randomization of mechanisms necessary for synaptically driven action potential generation, such as a disruption in the timing of synaptic release or random conduction failures at neuronal processes. If this was the case, then one would expect that the occurrence of action potentials between a given pair of neurons would vary randomly with respect to one another from stimulation to stimulation. We tested this hypothesis by comparing the patterns of action potential discharge of pairs of neurons between multiple consecutive stimuli of the Schaffer collaterals (data not shown). During each stimulation event, the action potentials occurred at nearly identical times with respect to one another and showed an almost identical burst morphology from stimulation to stimulation. We also checked to see whether the occurrence of action potentials between a given pair of neurons, during spontaneous field discharges, was fixed in time. The patterns of action potential discharges from a given pair of CA1 neurons was compared between consecutive spontaneous field bursts during the time when the occurrence of action potentials was clearly desynchronized (e.g., Fig. 4B). Just as in the case of stimulation-evoked action potential discharge described earlier, the action potentials generated during a spontaneous population discharge occurred at nearly identical times with respect to one another and showed a nearly identical burst morphology from one spontaneous discharge to the next.
Synchronization of burst discharges in CA3 pyramidal cells
The results of the previous experiments support the hypothesis
that spontaneous bursting ceases in CA1 because of a
low-[Cl]o-induced
desynchronization of the action potentials of the CA1 pyramidal cell
population. To test whether a desynchronization of action potential
discharge also underlies the cessation of bursting in CA3, we carried
out a series of similar experiments with paired intracellular and field
recordings acquired in area CA3 and a stimulating electrode placed on
the hilus for activation of the mossy fiber afferent input pathway
(Fig. 5). In addition to testing our
hypothesis regarding the desynchronization of neuronal activity, we
were interested in whether chloride-cotransport antagonism affected a
different hippocampal circuit (in this case, the hilus-to-CA3 circuit)
in a manner similar to its effects on the CA3-to-CA1 circuit.
|
Slices (n = 6) were perfused with
high-K+ medium (8 or 12 mM) until continuous
spontaneous bursting had been elicited in CA3 for 20 min and
thereafter bathed with high-[K+]o
medium containing furosemide (2.5 mM). As previously reported (Hochman et al. 1999
), an initial period of increased
burst amplitude was observed in CA3 after exposure to furosemide.
Simultaneous field and paired intracellular recordings during this
period of increased excitability showed a tight synchronization in the
occurrence of action potentials between a given pair of neurons, and
between the occurrence of action potentials and the morphology of the CA3 field burst (Fig. 5A). Both spontaneously occurring
discharges (Fig. 5A, left) and stimulation-evoked discharges
(Fig. 5A, right) showed similar synchronization. After 40 min of furosemide exposure, immediately before the complete blockade of
spontaneous bursting, no changes in the synchronization of neuronal
activity was observed during either spontaneously occurring (Fig.
5B, left) or stimulation-evoked (Fig. 5B, right)
discharges. Note the nearly identical morphology of the pattern of
action potential firing patterns between the spontaneous and
stimulation-evoked discharges. Most importantly, note that the number
of action potentials (1-3) associated with each discharge after
prolonged furosemide exposure is significantly less than the number of
action potentials (6-10) associated with discharges during the peak of
spontaneous burst activity. Although we did not have an electrode in
the CA1 region during these experiments, it is reasonable to assume
(based on previous experiments) that the CA3 responses illustrated in
Fig. 5B were acquired during a time when spontaneous
bursting in CA1 had ceased.
As in the case of the previous experiments that focused on the CA1 subregion, we were interested in whether the suppression of activity in CA3 was due a randomization of processes involved in the generation of action potentials. In all our experiments, we noted for comparison the patterns of action potential discharge between a given pair of neurons during a consecutive series of discharges (data not shown). Consecutive spontaneous discharges, immediately before (<2 min) the complete furosemide-blockade of spontaneous bursting, showed a nearly identical pattern of firing activity during each burst. The firing patterns during consecutive trials of stimulation-evoked discharges were also nearly identical from discharge to discharge and had a similar morphology to the spontaneously occurring discharges.
Axon conduction in Schaffer collaterals during blockade of spontaneous epileptiform burst discharge
We considered the possibility that the desyncronization of
CA1 action potential generation during
low-[Cl]o treatment was
due to changes in the ability of the Schaffer collaterals to conduct
action potentials. Such changes might involve alterations in axonal
conduction velocities; this could explain the increasing latencies
between CA3 and CA1 field discharge during low-[Cl
]o exposure. To
test this hypothesis, CA3 antidromic field responses were recorded
during prolonged exposure to low-chloride (7 mM) medium (Fig.
6). To eliminate any artifact due to
changes in synaptic excitability, excitatory and inhibitory synaptic
responses were blocked [bicuculline, 20 µM;
2-amino-5-phosphonovaleric acid (APV), 100 µM;
6-cyano-7-nitroquinoxalene-2,3-dione (CNQX), 50 µM]. The field
electrode was adjusted to optimize the amplitude of the antidromic
response in the CA3b region.
|
Figure 6A shows the orthodromic CA1 response
(top) and the antidromic CA3 response (bottom) to
Schaffer stimulation in normal medium. After blockade of synaptic
transmission (Fig. 6B) and prolonged exposure to
low-chloride medium, no significant changes were observed in the
latency of the antidromic responses in CA3. A small increase in the
peak latency (<1 ms) of the CA3 antidromic response can be observed
after further low-[Cl]o
exposure (Fig. 6C). However, this latency increase is not
sufficient to explain the large latency changes (>30 ms) between CA3
discharge and CA1 bursts observed in our previous sets of experiments.
Note also that the amplitudes of the CA3 antidromic responses in
control conditions and after prolonged
low-[Cl
]o exposure were
not significantly different. These observations suggest that the
synchronized discharges of action potentials in CA3, evoked
antidromically by Schaffer stimulation, was not significantly affected
by low-[Cl
]o exposure.
Effects of low-chloride treatment on spontaneous synaptic activity
It is possible that the anti-epileptic effects associated with
chloride-cotransport antagonism are mediated by some action on
transmitter release. Blockade of chloride cotransport could alter the
amount or timing of transmitter released from terminals, thus affecting
neuronal synchronization. To test whether
low-[Cl]o exposure
affected mechanisms associated with transmitter release, intracellular
CA1 responses were recorded simultaneously with CA1 and CA3 field
responses during a treatment that dramatically increases spontaneous
synaptic release of transmitter from presynaptic terminals (Fig.
7). Increased spontaneous release of
transmitter was induced by treatment with 4-AP (100 µM). After 40-min
exposure to 4-AP-containing medium, spontaneous synchronized burst
discharges were recorded in areas CA1 and CA3. Switching to
4-AP-containing low-[Cl
]o medium led
initially, as was shown previously, to enhanced spontaneous bursting
(20 min after exposure to
low-[Cl
]o medium; Fig.
7A). High-gain intracellular recordings showed that
high-amplitude spontaneous synaptic activity was elicited by 4-AP
treatment (Fig. 7A, bottom). Further exposure to
low-chloride medium blocked spontaneous burst discharge in CA1,
although CA3 continued to discharge spontaneously (Fig. 7B,
top). At this time, CA1 intracellular recordings showed that
spontaneous synaptic noise was increased further and remained so for
prolonged exposure times to 4-AP-containing low-chloride medium (Fig.
7C). These data suggest that mechanisms responsible for
synaptic release from terminals are not adversely affected by
low-chloride exposure in a manner that could explain the blockade of
4-AP-induced spontaneous bursting in CA1. These results also eliminate
the possibility that the effects of
low-[Cl
]o exposure are
due to alterations in CA1 dendritic properties that would compromise
their efficiency in conducting PSPs to the soma.
|
Finally, in six experiments, we observed the affects of furosemide treatment on synaptic activity measured with paired intracellular recordings of CA3 pyramidal cells in slices in which spontaneous bursting activity was elicited by exposure to high-K+-medium (8 or 12 mM; data not shown). In these experiments, hyperpolarizing current was injected into the cells to prevent action potential discharge between bursts so that the EPSPs could be distinguished more easily. During the period of spontaneous bursting, large EPSPs that were synchronized between pairs of neurons, and synchronized with small deflections in the field recording, were recorded routinely. The amplitude, frequency, and degree of EPSP synchrony were maximal just after a spontaneous burst and diminished just before the subsequent burst. After prolonged exposure to furosemide, which blocked all spontaneous bursting (>40 min), the frequency and amplitude of the EPSPs were significantly reduced, although infrequent synchronized EPSPs still could be distinguished.
![]() |
DISCUSSION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
The goal of this study was to test the hypotheses that blockade of
chloride cotransport results in a desynchronization of neuronal
activity and that the anti-epileptic effects of chloride-cotransport blockade are independent of direct affects on excitatory synaptic transmission. Presumably extracellular field recordings from the cell
body layers of hippocampal slices represent some statistical average of
the activity of the population of neurons near the recording electrode.
We suppose that a necessary degree of temporal synchronization of
action potential firing is required in a geometrically organized
neuronal population so that dipoles summate into a detectable field
response (Andersen 1975; Faber and Korn
1989
). Given this interpretation, several observations from our
previous studies (Hochman et al. 1995
, 1999
) motivated
the first hypothesis
that chloride-cotransport blockade leads to a
desynchronization of population neuronal activity. First, tissue
exposure to furosemide or
low-[Cl
]o medium
blocked spontaneous epileptiform (i.e., synchronized) burst discharges
in the cell body layers of areas CA1 and CA3. Second, during prolonged
exposure to low-[Cl
]o
medium, action potential discharge capabilities of CA1 pyramidal cells
remained intact even though the field recording was silent. Our second
hypothesis, that the antiepileptic effects of chloride-cotransport blockade are independent of effects on transmitter release at the
synapse, was motivated by the observations that furosemide and
low-[Cl
]o blocked
spontaneous field discharges but that hyperexcited CA1 field response
still could be evoked by Schaffer collateral stimulation. These
hyperexcited field responses to stimulation (indicated by multiple
population spikes), seen during furosemide or
low-[Cl
]o treatment,
likely reflect diminished GABAA inhibitory currents induced
by the blockade of cation-chloride cotransport (Hochman et al.
1999
; Misgeld et al. 1986
; Thompson and
Gähwiler 1989
; Thompson et al. 1988
).
Loss of coherence in action potential firing times
Several important aspects of the effects chloride-cotransport
blockade on the firing times of CA1 neurons should be noted: 1) the loss in the precise timing of action potential firing
between pairs of neurons was coincident with the blockade of
spontaneous epileptiform activity, suggesting that this change in the
timing of action potentials is involved in the anti-epileptic effects of chloride-cotransport blockade; 2) although a significant
offset in the times of occurrences of action potentials between pairs of CA1 neurons developed with
low-[Cl]o exposure, the
offset for a given pair of neurons remained constant with respect to
one another during a fixed interval of the experiment. Thus rather than
describing the consequence of
low-[Cl
]o exposure as a
loss of "synchronization," it is more accurate to describe it as an
induction of "fixed phase lags" in the timing of action potential
firing between neurons in the CA1 population or as a transition from a
state in which the firing patterns of the population of CA1 pyramidal
cells are "coherent" to a state where they are "incoherent"
(Block 1997
). Before
low-[Cl
]o exposure, the
CA1 population of pyramidal cells fired action potentials in phase with
one another and with the field, whereas after prolonged exposure, the
firing of action potentials among the population became sufficiently
incoherent so that the field response eventually was diminished below
the threshold of detectability. We propose that the anti-epileptic
effects of chloride-cotransport blockade in CA1 are a result of its
action on the coherence of action potential firing.
Changes in excitatory synaptic transmission
Our observations are consistent with the hypothesis that a loss of
coherence of CA1 action potential discharges, induced by low-[Cl]o exposure, is
independent of changes in excitatory transmission. First, even after
the CA1 field response to Schaffer collateral stimulation had been
blocked completely by prolonged exposure to
low-[Cl
]o medium,
Schaffer collateral stimulation still faithfully evoked monosynaptic
EPSPs (and action potentials) with short latency. This observation
suggests that the excitatory synaptic mechanisms associated with the
stimulation of presynaptic fibers were still functional. Second, the
magnitude and frequency of 4-AP-induced spontaneous transmitter release
from presynaptic terminals did not appear to be decreased by
low-[Cl
]o exposure.
Indeed, the magnitude of the 4-AP-induced PSPs actually increased with
prolonged exposure to
low-[Cl
]o medium. This
result has two important implications for the interpretation of the
effects of low-[Cl
]o
exposure on the CA3-CA1 circuit: the mechanisms associated with the
presynaptic release of transmitter are not affected in a way that could
explain the blockade of spontaneous bursting in CA1, and transmission
of PSPs from the dendritic regions to the somata of CA1 cells are not
affected. Finally, the characteristics of the
low-[Cl
]o-induced phase
lags in the timing of action potentials (described in the preceding
text) are consistent with the interpretation that this loss of
coherence in action potential timing is independent of affects on
excitatory synaptic transmission. That is, if the observed phase
shifting effects were due to changes associated with synaptic
mechanisms [such as changes in the timing of transmitter release or
changes in the kinetics of transmitter reuptake by synaptic vesicles
(Diamond and Jahr 1997
; Nicholls and Attwell 1990
; Otmakhov et al. 1993
; Rusakov
and Kullmann 1998
; Tong and Jahr 1994
;
Wolosker et al. 1996
; Zador 1998
)], it
would be expected that the phase lags between pairs of neurons would
vary randomly with respect to one another, from burst to burst; it
would be unlikely that a given synapse would fail in exactly the same
way from one stimulation trial to the next. However, the pattern of action potential discharges between pairs of neurons in our experiments remained nearly invariant between consecutive discharges, suggesting that a randomization of PSP mechanisms was not responsible for the
generation of the observed phase lags.
During the time interval when CA1 field potentials were blocked by
low-[Cl]o exposure,
while CA3 continued to generate spontaneous discharges, intracellular
recordings from CA1 pyramidal cells showed that no PSPs were generated
by the spontaneous presynaptic bursts even though electrical
stimulation of the presynaptic fibers continued to evoke a discharge.
We can interpret this result in a manner consistent with the preceding
discussion: the disappearance of the spontaneously occurring CA1 field
response, although coincident with the incoherence of action potential
discharge, is due to a reduction in the ability of action potentials in
CA3 neurons to invade the synaptic terminals. This conclusion is most
clearly supported by the results of our 4-AP experiment (Fig. 7).
During the time that spontaneous bursting ceased in CA1 (but continued in CA3), it was clear that the EPSPs evoked by the synchronized discharges of CA3 were not propagated to CA1 even though
large-amplitude PSPs due to 4-AP-induced spontaneous transmitter
release were clearly present. This observation suggests that there was
a blockade of action potential propagation somewhere between the
presynaptic CA3 cell and the synaptic terminal
but that the terminal
itself was still capable of releasing transmitter. Further, this result suggests that this blockade affected only a fraction of the synapses because electrical stimulation of the presynaptic fibers was capable of
evoking a maximal postsynaptic response. We propose that this presynaptic effect, related to a blockade of action potential invasion
into synaptic terminals, is responsible for the changes in the timing
of postsynaptic action potential generation.
In contrast to spontaneous activity, electrical stimulation of
presynaptic axons was always sufficient to evoke synaptically driven
discharges in CA1 (Fig. 3). Electrical stimulation of the Schaffer
collaterals likely results in the simultaneous depolarization of all
the axons within the vicinity of the stimulating electrode, thus
initiating a highly synchronized discharge of action potentials in
Schaffer collateral (and other) axons. In contrast, because low-[Cl]o exposure
caused a diminution of spontaneous CA3 discharges (discussed in the
following text), spontaneous CA3 bursts undoubtedly represent a less
powerful synaptic drive, involving a smaller number of axons. The
ability of action potentials in presynaptic neurons to evoke a
postsynaptic response during chloride-cotransport antagonism is likely
dependent on a sufficiently large number of presynaptic neurons firing
action potentials in phase with one another.
Axonal conduction
The results discussed so far suggest that the anti-epileptic
effects of chloride-cotransport blockade in area CA1 are not due to
postsynaptic actions because action potential generation and the
propagation of PSPs from the dendrites to the soma in postsynaptic
cells are not affected. Further, impairment of mechanisms associated
with the release of synaptic transmitter does not seem to be involved.
Because the ability of presynaptic CA3 pyramidal cells to fire action
potentials is not impaired, it is likely that
low-[Cl]o exposure
mediates its effects by affecting propagation of action potentials at
some point between the soma and the terminals of CA3 pyramidal cells.
To test whether this effect was mediated by decreasing the ability of
CA3 axons to support action potentials, we examined the consequences of
low-[Cl
]o exposure on
the antidromic response of CA3 pyramidal cells (Fig. 6). Prolonged
exposure to low-[Cl
]o
medium did not significantly alter the antidromic response of CA3 to
Schaffer collateral stimulation. This result rules out gross changes in
action potential propagation by the major axonal branches of CA3
pyramidal cells. However, it may be that action potential propagation
by smaller axonal arborizations, or the ability of action potentials to
invade the axonal varicosities immediately before the synaptic
terminals, is affected (Andersen 1975
; Shepherd
and Harris 1998
; Westrum and Blackstad 1962
).
We observed that during chloride-cotransport antagonism, CA3 continued to burst spontaneously for a brief period of time after bursting ceased in CA1. We did not observed any noticeable loss of coherence in CA3 action potential firing. Both of these observations imply that area CA3 is less vulnerable than CA1 to treatments that affect chloride cotransport. This difference between the responses of areas CA1 and CA3 may be due to the existence of local excitatory synaptic contacts between CA3 pyramidal cells; this could provide a more robust means for maintaining coherence in the firing of action potentials. It should be noted that after prolonged exposure to furosemide, CA3 neurons fired fewer action potentials during burst discharges. This finding suggests that the axons comprising the recurrent excitatory synaptic connections between CA3 pyramidal cells are affected by chloride-cotransport antagonism.
It was noted that the latency between spontaneous discharges of areas
CA1 and CA3 significantly increased with exposure to low-[Cl]o medium. As
well, there was a significant increase in the latency between
electrical stimulation of the Schaffer collaterals and the consequent
CA1 epileptiform burst discharge. These latencies increased from a few
milliseconds before
low-[Cl
]o exposure, to
as much as 30-40 ms after prolonged
low-[Cl
]o exposure,
just before the cessation of bursting in area CA1. It is unlikely that
the long latencies between the activation of the presynaptic fibers and
CA1 epileptiform discharge were due to changes in synaptic
transmission, because even after the CA1 field response was blocked
completely by prolonged exposure to
low-[Cl
]o medium, CA1
pyramidal cells were found that would discharge action potentials at
short latency to Schaffer stimulation. However, prolonged exposure to
low-[Cl
]o medium did
induce an increase in the latency the between Schaffer stimulation and
part of the antidromically evoked hyperexcitable burst discharge in
area CA3. In this case, there were two distinct components to the CA3
response
an antidromic population spike that showed no significant
increase in latency and a burst discharge the latency of which
increased with exposure time to
low-[Cl
]o medium until
it was blocked completely (Fig. 3). Because the burst discharge that
followed the CA3 population spike was likely due to local recurrent
excitatory interactions in CA3 (Wong and Traub 1983
),
this finding suggests that the effects of
low-[Cl
]o exposure on
the generation of epileptiform discharges in CA3 (involving local
recurrent excitatory connections) were similar to those observed in
CA1
although a significantly longer time of exposure to
low-[Cl
]o medium was
required to affect the CA3 response.
Chloride-cotransporter blockade and the fidelity of axonal conduction
We propose, as a possible explanation of our observations
described in the preceding text, that blockade of chloride cotransport results in the failure in the ability of action potentials to invade a
subpopulation of axonal arborizations (or varicosities) of both the
Schaffer collateral pathway and the local recurrent connections in area
CA3. The total number of axonal branches affected, or the degree to
which a specific subpopulation is affected, are likely dependent on the
time of exposure to treatments that antagonize chloride cotransport.
This mechanism would result in a time-dependent reduction in the
fidelity of transmission of spontaneous burst discharges from CA3 to
CA1 as well as a reduction in the strength of local recurrent
excitatory activity in CA3. This explanation is consistent with our
conclusion that the major effects of
low-[Cl]o treatment on
area CA1 neuronal activity are not dependent on direct actions at
excitatory synapses and cannot be attributed to changes in the
electrophysiological properties of either presynaptic or postsynaptic
neurons. Assuming that electrical stimulation of the Schaffer
collateral pathway activates primary axonal branches necessary for the
generation of the antidromic CA3 population spike, our observations
suggest that the mechanisms affected by low-[Cl
]o exposure in
the CA3-to-CA1 circuit are likely to be presynaptic with respect to
area CA1 but must be subsequent to the primary axonal branches of CA3
pyramidal cells. An effect on a specific, nonvarying subset of axonal
structures, distal to the primary branches, could explain the
consistent phase lags in action potential firing times observed in CA1.
If action potentials were prevented from invading some fraction of the
total number of synaptic terminals, the total amount of current
delivered to the postsynaptic neurons would be diminished. Further, a
reduction in the strength of recurrent excitatory activity in CA3 would
result in a smaller number of CA3 neurons firing at a given moment. The
consequence of these effects would be a weaker excitatory drive by CA3
onto CA1. We observed, just before the total blockade of the
stimulation-evoked CA1 field response, long latencies (>30 ms) between
Schaffer collateral stimulation and the CA1 population discharge. It
seems unlikely that these significant latencies can be explained on the
basis of changes in the monosynaptic drive of CA3 onto CA1. Stimulation of the Schaffer collaterals causes some fraction of the CA1 population to depolarize at short (monosynaptic) latency, but the effect appears
to be sufficiently desynchronized so that no field response is detected.
Accumulation of [K+]o has
been implicated in numerous systems as an important factor controlling
action potential propagation in axons (Adelman and Palti
1969; Grossman et al. 1979
; Luscher et
al. 1994a
,b
; Malenka et al. 1981
; Nicoll
and Alger 1979
; Parnas 1972
; Parnas et
al. 1976
; Rang and Ritchie 1968
; Smith
1980
). In response to a short train of high-frequency stimuli
delivered to the parallel fibers in rat cerebellar cortex slices
(Malenka et al. 1981
), it was found that the latency
increased and the amplitude decreased in the postsynaptic field
response
an effect paralleling our observations here. The magnitude of
these changes in the field responses was found to be correlated tightly
to the degree of extracellular potassium accumulation. A study on
dorsal root ganglion cells suggested that, under conditions such as
repetitive stimulation, failure of action potentials to invade the cell
soma occurred at sites of impedance mismatch such as branch points (Luscher et al. 1994b
). In cases of action potential
failure, the electrotonic responses at the soma to failed action
potentials had discrete amplitude levels, suggesting that failures of
action potential invasion always occurred at the same sites along the axon. This phenomenon of "deterministic failure" is a possible parallel to our observations that the phase lags between pairs of
neurons always are fixed with respect to one another.
A number of other factors undoubtedly are affected by the blockade of
chloride-dependent cotransportalthough it is unclear to what extent
they play a role in the consequent desynchronizing effects. For
example, blockade of activity-induced changes in the ECS could lead to
a reduction of ephaptic field interactions (Faber and Korn
1989
; McBain et al. 1990
). Previous optical
imaging studies of slices suggested that concomitant with the
furosemide-blockade of spontaneous bursting was a blockade of
activity-evoked volume changes of the ECS (Hochman et al.
1995
), indicating that activity-evoked shrinkage of the ECS is
blocked by chloride-cotransport antagonism. This action would
discourage ephaptic interactions that occur as ECS resistivity
increases. Because field interactions may play an important role in
neuronal synchronization (Dudek and Traub 1989
), the
blockade of such ephaptic effects might contribute to the antiepileptic
effects of chloride-cotransport antagonism. Another possible factor
affected by chloride-cotransport antagonism includes the changes in pH
dynamics of intra- and extracellular spaces, affecting putative
pH-dependent signaling mechanisms (Chesler 1990
;
Gottfried and Chesler 1994
; Pappas and Ransom
1994
; Ransom 1992
). It might be that ionic and
pH changes induced by the blockade of chloride cotransport reduce the
degree of electrotonic coupling between neurons through gap junctions
(MacVicar and Dudek 1981
); such coupling is known to be
labile and affected by changes in the ionic environment and pH
(Church and Baimbridge 1991
; Neyton and Trautmann
1985
). Recently a novel form of activity-dependent action
potential gating in CA3 cells has been proposed where the action
potential block requires a hyperpolarizing prepulse to activate an
A-type K+ channel (Debanne et al.
1997
; Kopysova and Debanne 1998
). This type of
action potential blockade also might play a role in our study because
it has been reported that treatments that cause a reduction of the
transmembrane chloride gradient in nonmyelinated mammalian axons can
lead to dramatic increases in the magnitude and duration of
afterpotential hyperpolarizations (Rang and Ritchie 1968
). Both furosemide and
low-[Cl
]o treatments
result in a diminished transmembrane chloride gradient in hippocampal
and cortical pyramidal cells (Misgeld et al. 1986
; Thompson and Gähwiler 1989
; Thompson et al.
1988
).
Taken together with the findings of our previous study (Hochman
et al. 1999), the present results suggest that ionic fluxes mediated by the glial
Na+K+2Cl
cotransporter directly affect the statistical coherence of action potential firing times in neuronal populations. Our data suggest that
antagonism of chloride cotransport can induce phase lags between the
action potential firing times of CA1 neurons sufficiently large so that
both spontaneous and stimulation-evoked population discharges are
blocked even though the individual postsynaptic neurons continue to
generate monosynaptic EPSPs (and action potentials) in response to
Schaffer stimulation. This incoherence in action potential firing
associated with burst discharges might underlie the potent
antiepileptic effects of chloride-cotransport antagonism. This effect
is "nonsynaptic" because it is not mediated by changes in synaptic
transmission per se but rather by changes in the "gating" of action
potentials allowed to invade the synaptic terminals of the presynaptic
neurons. Given the potency that modulation of chloride cotransport has
on population discharge under pathophysiological conditions, it may be
that this mechanism also plays a role in the modulation of synchronized
activity necessary for normal function. An understanding of how
chloride cotransport modulates the coherence of action potential firing
might shed light on the contributions of nonsynaptic mechanisms to the
generation and maintenance of synchronized activity as well as suggest
new strategies for the development of antiepileptic treatments.
![]() |
ACKNOWLEDGMENTS |
---|
This work was supported by National Institute of Neurological Disorders and Stroke Grant NS-35548 to P. A. Schwartzkroin and NS-07144 to D. W. Hochman.
![]() |
FOOTNOTES |
---|
Address for reprint requests: P. A. Schwartzkroin, Dept. of Neurological Surgery, Box 356470, University of Washington, Seattle, WA 98195.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Received 14 April 1999; accepted in final form 14 September 1999.
![]() |
REFERENCES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|