| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Swammerdam Institute for Life Sciences, Section Neurobiology, University of Amsterdam, 1098 SM Amsterdam, Netherlands
Submitted 30 September 2002; accepted in final form 8 April 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
150% of the first response in the train. In contrast, the
plateau levels of Alveus inputs to interneurons were not different from the
first responses for frequencies 
40 Hz. Paired-pulse facilitation of
Schaffer input was stronger than for Alveus input. Cells in stratum oriens
with horizontal dendritic trees appeared to be a special group of interneurons
because Alveus input to these cells showed strong facilitation with plateau
levels of 200% of the first responses. Schaffer input to CA1 basket and
bistratified cells showed similar synaptic dynamics compared with Schaffer
input to pyramidal cells for frequencies 80 Hz. The synaptic dynamics of
Schaffer and Alveus input depended only weakly on the stimulus intensity. The
difference between the dynamics of Alveus and Schaffer input to CA1
interneurons implies that the relative contribution of feedforward and -back
inhibition to network activity depends on the frequency of the input signal at
the afferent fibers, adding a level of complexity to transient responses. |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
;
Dobrunz and Stevens 1999). The
signal transfer depends on the properties of the involved neurons, on the
connectivity within the network, and on the strengths of the synapses between
neurons. The latter are dynamic parameters that depend on previous synaptic
activity (Dobrunz and Stevens
1999; Markram et al.
1998; Varela et al.
1997). The CA1 network of the hippocampus contains principal
cells, which are responsible for the output signals to other brain areas, and
interneurons with many local projections. Interneurons receive inputs from
many principal cells and project to a large fraction of the principal cells
(Buhl et al. 1994a). They exert
strong control over the signal transfer of the network
(Cobb et al. 1995;
Dvorak-Carbone and Schuman
1999; Yanovsky et al.
1997). To understand the signal transfer during in vivo network
activation, knowledge about the dynamic properties of the involved synapses is
essential.
The connectivity of an interneuron in the network determines when and how
it can participate in population activity
(Karnup and Stelzer 1999;
Wierenga and Wadman 2003).
After Schaffer collateral activation, interneurons in the hippocampal CA1 area
can receive monosynaptic input from the Schaffer collaterals or disynaptic
input from the CA1 pyramidal cells, which are first activated by the Schaffer
collaterals (Andersen et al.
1963,
1964;
Buzsáki and Eidelberg
1982; Kandel et al.
1961; Lacaille
1991; Sah et al.
1990). Disynaptic input to interneurons seems to reflect the
activity of a large fraction of the pyramidal cells. Interneurons receiving
this type of input are therefore well suited to normalize or scale the mean
pyramidal cell output (Wierenga and Wadman
2003). Interneurons receiving monosynaptic input from the Schaffer
collaterals play a direct role in information processing by the CA1 pyramidal
cells (Pouille and Scanziani
2001).
Use-dependent changes in synaptic efficacy (often referred to as short-term
plasticity or synaptic dynamics) occur on the time scale of tens to thousands
of milliseconds. The dynamics of excitatory and inhibitory synapses are an
important factor in shaping network activity. Depression of excitatory
synapses between neocortical pyramidal cells can function as a gain control
mechanism, enabling the postsynaptic neuron to detect a change in the
presynaptic firing rate of an input irrespective of absolute presynaptic
activity levels (Abbott et al.
1997; Markram et al.
1998). The dynamic properties of both inhibitory and excitatory
synapses in the hippocampus increase the ability of pyramidal cells to detect
specific intervals in presynaptic spike trains. This may be important in
transforming temporal information into a spatial code segregated over
different populations of neurons
(Buonomano 2000;
Buonomano et al. 1997).
Furthermore, differences in synaptic dynamics of inhibitory and excitatory
connections may affect the stability of neuronal networks by adding dynamic
properties to the balance between inhibition and excitation
(Galarreta and Hestrin 1999;
Varela et al. 1999).
In the present study, we investigated the synaptic dynamics of excitatory inputs to CA1 interneurons and pyramidal cells. We examined the three excitatory inputs that are activated in the CA1 network after Schaffer collateral stimulation: direct Schaffer input to CA1 interneurons and to the pyramidal cells and the subsequent input from the pyramidal cells to interneurons. The use-dependent synaptic dynamics were revealed in the synaptic responses to trains of extracellular stimuli at the Schaffer collaterals or the Alveus fibers. We characterize the differences in the dynamics of the synaptic inputs and discuss the consequences for the frequency-dependent signal transfer in the CA1 network.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
Wistar rats of 14–28 days old were used for this study (mean age was
21 days; rats were obtained from Harlan B.V., Zeist, the Netherlands). After
decapitation the brain was rapidly removed and the hippocampus was dissected
in ice-cold artificial cerebrospinal fluid (ACSF). The ACSF contained (in mM)
125 NaCl, 2.4 KCl, 1 MgCl2, 2 CaCl2, 1.1
NaH2PO4, 26 NaHCO3, and 25
D-glucose and was continuously gassed with 95% O2-5%
CO2 to set pH at 7.3. All chemicals were obtained from Sigma (St.
Louis, MO). Slices (300 µm thick) were cut with a McIllwain tissue
chopper under an angle of 20° with respect to the direction of the
Alvear fibers. This direction was chosen to keep as many Schaffer collaterals
intact in the slice as possible (Ishizuka
et al. 1990). The hippocampal slices were incubated at 32°C in
ACSF for 1 h and then kept at room temperature in ACSF until they were
transported to the recording chamber.
Extracellular recordings and stimulation
All recordings were done with ACSF as the perfusion medium at a temperature
of 30–32°C. Stimulations were applied as biphasic current pulses of
200 µs through bipolar stimulation electrodes (made in house from
straight-cut, isolated stainless steel 60-µm-thick wire; 50- to 100-µm
tip separation). One stimulation electrode was placed over the
Schaffer-Commissural afferent fibers in stratum radiatum and a second
stimulation electrode was placed at the Alveus, both at the CA1–CA3
border (Fig. 1C). The
stimulation current for maximal response varied from slice to slice between
250 and 700 µA. The extracellular recording electrode (borosilicate glass;
2–4 M
; filled with ACSF) was placed in the CA1 stratum
pyramidale. Signals were amplified (1,000 times) with a custom-built amplifier
and sampled at 4 kHz.
|
Schaffer collaterals project to CA1 s. radiatum but also to s. oriens
(Ishizuka et al. 1990;
Li et al. 1994). Alveus
stimulation incidentally activated fibers in the Alveus other than axons of
the pyramidal cells. This showed up in the extracellular recording as a small
field potential after the antidromic population spike. Careful placement of
the Alveus stimulus electrode and the use of only half-maximal intensity for
most stimulations minimized this activation. Its occasional presence did not
affect the conclusions reached in this paper. Synapses onto interneurons do
not contribute to the recorded field potential because interneuron morphology
is usually not polarized and there are relatively few interneurons. Activation
of fibers in the Alveus other than axons of pyramidal cells and that only
contact interneurons cannot be excluded. Cholinergic input did not play a role
because addition of 1 µM atropine to the bath did not affect the amplitude
or synaptic dynamics of Alveus or Schaffer input (n = 6; data not
shown).
Patch-clamp recordings
An upright microscope (Nikon Optiphot; Nikon, Tokyo) with a x40
water-immersion objective and a CCD camera with a high-pass 700 nm filter were
used to visually select CA1 interneurons and pyramidal cells. Patch-clamp
recordings were made using an Axopatch 200B amplifier (Axon Instruments, Union
City CA). The patch pipettes (borosilicate glass, 3–5 M) were
filled with intracellular solution containing (in mM) 130 K-gluconate, 10
HEPES, 1 EGTA, 10 KCl, 0.5 CaCl2, and 2 MgATP (pH adjusted to 7.3;
280–290 mOsm). In most experiments, the quartenary lidocaine derivative
QX314 (3–10 mM) was added to block sodium action potentials and
postsynaptic GABAB receptors. Recorded interneurons were located in
s. oriens, s. pyramidale, and s. radiatum and had cell bodies with a different
shape or orientation than the majority of cells in the pyramidal layer.
Interneurons showed higher firing rates, narrower action potentials and higher
input resistances than pyramidal cells
(Table 1).
|
The tip of the patch electrode was filled with the intracellular solution
without QX314. Immediately after the establishment of the whole cell
configuration, the amplifier was switched to current-clamp mode to record the
firing pattern of the cell during injection of 200-pA depolarizing current. We
verified that at this time the evoked action potentials were overshooting with
constant amplitudes, indicating that possible washout of intracellular factors
and QX314 diffusion into the cells did not yet play a role. Current clamp
recordings were sampled at 10 kHz. Subsequently, whole cell voltage clamp
recordings were made at a membrane potential of –70 mV. We checked the
stability of the series resistance (4–10 M) and whole cell
capacitance (15–30 pF) by applying small voltage steps during the
experiments. Voltage-clamp recordings were low-pass filtered at 2 kHz with an
8-pole Bessel filter (in house built) and sampled at 4 kHz. Recordings were
done with an ATARI TT030 computer, using custom-made interface and
software.
During the recordings, cells were loaded with biocytin (0.5–1%) via
the patch pipette. After the recordings the slice was fixed overnight in
phosphate buffered saline with 4% paraformaldehyde. For reconstruction of the
cells using a drawing tube, biocytin was visualized by using an avidin-HRP
reaction (ABC elite peroxidase kit; Vector laboratories, Burlingame, CA)
according to the instructions of the manufacturer. Slices were not resectioned
for this procedure. Some interneurons were lost before the end of the
recordings (probably as a result of the high-frequency stimulations) and could
not be reconstructed (n = 17 of 42). The distinction between
interneurons that received monosynaptic input (directly from the Schaffer
collaterals) and interneurons receiving disynaptic feedback input (from CA1
pyramidal cells, stimulated by Schaffer stimulation) was based on the timing
of the synaptic current with respect to the simultaneously recorded population
spike (Maccaferri and McBain
1995; Wierenga and Wadman
2003). Monosynaptic Schaffer input started before or during the
population spike, while disynaptic input occurred >1 ms after the peak of
the population spike. Furthermore, only monosynaptic Schaffer input could be
observed at low stimulus intensities in the absence of a population spike,
while the occurrence of disynaptic input was causally related to the
occurrence and amplitude of the population spike
(Wierenga and Wadman 2003). In
the rare cases (<5%; n = 2) when there was doubt to which
experimental group the recorded cell belonged, the cell was not further
analyzed.
Synaptic dynamics
The synaptic dynamics of the inputs were determined using trains of 10 stimuli at the Schaffer collaterals or at the Alveus fibers at various frequencies (40 s between consecutive trains). Stimulus intensity was chosen to evoke a half–maximal response to the first stimulus, unless stated otherwise. Recordings were accepted only when the change in input resistance during the recordings did not exceed 20% and when the responses to the stimulus trains were reliably repeated. Responses were then averaged over two to five trials. The amplitude of the synaptic responses to each stimulus was estimated by fitting the decay phase of the preceding response with an exponential function (Fig. 1B) using the least-squares method (custom-made software). Comparison of the synaptic responses of each cell was done after normalization to the mean amplitude of the first synaptic response of all trains (average over 20–50 samples; in the following referred to as reference amplitude).
The synaptic dynamics of Alveus (n = 7) and Schaffer (n = 3) input to interneurons in slices of which the CA3 region was cut off were not different from the dynamics in "intact" slices. This suggests that the contribution of recurrent activation of CA3 pyramidal cells to the synaptic dynamics was negligible in our recordings.
Ten stimuli were sufficient to reach a plateau level in the synaptic response amplitude for most frequencies. The plateau level was determined from the mean of the last three synaptic responses in the train. Recordings of longer stimulus trains (15–30 stimuli) at frequencies <50 Hz confirmed that this level was accurate within 10%. Longer trains, especially at high frequencies, were avoided because they often induced persistent changes in the synaptic response that could last for minutes. During the experiments we monitored the stability of the amplitude of the first response (Fig. 1A). The ratio between the amplitudes of the second and first response in a train will be indicated in the following as paired-pulse ratio. For each frequency, we compared the plateau levels and paired-pulse ratios of all cells between two groups with the Student's t-test. P < 0.05 was used to indicate a significant difference.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
Interneurons were selected based on their appearance under the microscope
and the location of their soma. During the recordings, cells were filled with
biocytin, allowing posthoc inspection of their morphology. The interneurons
were classified according to their dendritic and axonal morphology following
earlier descriptions (Ali et al.
1998; Buhl et al.
1994b,
1996;
Halasy et al. 1996). Putative
bistratified cells had axons that ramified mainly in the proximal half of s.
radiatum and throughout s. oriens. Their dendrites rarely extended
significantly into s. lacunosum-moleculare. Putative basket cells had axons
that cover the entire depth of the pyramidal layer and only very proximal
regions of the adjacent strata and their dendrites usually extended into s.
lacunosum-moleculare. Of the 25 interneurons that were reconstructed, 14 were
identified as putative basket cells and 10 as putative bistratified cells. The
remaining interneuron was located in s. radiatum and projected mainly to s.
lacunosum-moleculare. The axon of one putative basket cell ramified only in
half of s. pyramidale and in the most proximal part of s. oriens and therefore
may have been an axo-axonic cell. Our data gave no indication that the
synaptic inputs or their dynamics were different in basket cells and
bistratified cells in agreement with recent findings
(Losonczy et al. 2002). We
therefore will treat these interneurons as one group in the further report.
The axons of three reconstructed pyramidal cells ran through s. oriens and
followed the fibers in the Alveus. In addition, we recorded from five
interneurons in s. oriens with horizontal dendrites confined to that layer.
These cells will be considered separately.
The firing properties of the recorded cells were probed by a 200-pA current
injection (corresponding to a 17- to 29-mV depolarization) when the cells
were held in current clamp at –70 mV. Although the firing rate measured
in this way strongly correlated with input resistance, probing interneuron
firing in this very incomplete way illustrated the heterogeneity in the firing
properties of the interneurons. Accommodation (usually moderate) of the firing
rate was observed in 26 (of 43) interneurons, of which 7 showed clear burst
firing. The other 17 interneurons were classical non-accommodating fast
spiking interneurons. Spontaneous activity was observed in six interneurons.
In agreement with previous reports (Ali et
al. 1998; Gupta et al.
2000; Losonczy et al.
2002; Parra et al.
1998), firing properties and other electrophysiological parameters
did not correlate with morphological cell types. General cell characteristics
are given in Table 1.
Characterization of the responses to the stimulus trains
To investigate the synaptic dynamics of input to CA1 interneurons, we applied trains of 10 stimuli through the Schaffer or Alveus stimulation electrode at several frequencies. We defined the reference amplitude of each cell as the mean amplitude of the first responses in all trains (Fig. 1A). Mean amplitudes, rise times, and decay time constants of the synaptic currents are given in Table 1. The amplitude of the synaptic current to each stimulus of the train was determined by assuming linear addition of the synaptic responses. For this purpose, we fitted the decay of the preceding responses with an exponential function (Fig. 1B).
At 0.5 Hz, the amplitude of the synaptic current was not different for successive stimuli in all cells (mean amplitude was 102 ± 1% of reference amplitude). For frequencies between 5 and 80 Hz, the synaptic responses systematically showed an initial increase in amplitude during the first two to five stimuli and then usually declined to reach a plateau level. We expressed all amplitudes (including the plateau level and the paired-pulse ratio) relative to the reference amplitude. In the extracellular recordings, the second and last fiber volleys or antidromic population spikes in a train were not different from the first for all frequencies (Fig. 1D). This indicates that the observed changes in synaptic currents were not due to changes in stimulus efficacy.
Schaffer and Alveus input to interneurons
The aim of this study was to compare the synaptic dynamics of the input
from the Schaffer collaterals to CA1 interneurons (feedforward activation)
with the dynamics of the input from the pyramidal cells to the interneurons
(part of the inhibitory feedback loop in the network). These inputs were
evoked monosynaptically by stimulating the Schaffer collaterals or the Alveus.
Schaffer collateral stimulation evoked activity in the pyramidal cell
population, which could, via the pyramidal axons, result in disynaptic input
to the interneurons. We therefore distinguished between interneurons in which
Schaffer stimulation evoked monosynaptic input and interneurons that received
disynaptic input. This distinction was based on the timing of the synaptic
input with respect to the simultaneously recorded population spike after
Schaffer stimulation (see METHODS). Examples of these two types of
input to interneurons are given in Fig. 2,
A1 and B1. The mean time differences between the
peak of the population spike and the onset of the synaptic current are given
in Table 1 (
t Schaffer
input). Schaffer collateral stimulation evoked predominantly monosynaptic
input in 28 interneurons (17 from s. radiatum, 8 from s. oriens, and 3 from s.
pyramidale) and disynaptic input in 14 interneurons (4 from radiatum, 9 from
oriens, and 1 from pyramidale).
|
In interneurons with disynaptic Schaffer input, we used Alveus stimulation to evoke this input monosynaptically (in the following referred to as Alveus input). Alveus stimulation elicited a synaptic response in all of these interneurons. Stimulation of the fibers in the Alveus always activated axons of the CA1 pyramidal cells as it evoked an antidromic population spike in the field potential recordings (latency of the peak <3 ms; Fig. 2, A2 and B2, top). The latency after the stimulus of monosynaptically evoked Alveus input was comparable to the latency of monosynaptically evoked Schaffer input [4.8 ± 0.2 ms (n = 23) vs. 4.3 ± 0.1 ms (n = 21)] as was expected because the two stimulation electrodes were placed at similar distances from the recorded interneurons.
We used Alveus stimulation also in the interneurons that received
monosynaptic input from the Schaffer collaterals to estimate the contribution
of Alveus input in these cells. In 13 of these cells (of 20 tested), Alveus
stimulation also evoked synaptic input. For a comparison of the amplitudes of
the synaptic currents evoked by Schaffer or Alveus stimulation, both stimuli
should ideally recruit the same pyramidal cells, which was impossible to
realize in our experiments. The amplitude of the maximal population spike is
an approximate measure of the total number of pyramidal cells that can be
activated by the stimulus (Varona et al.
2000). We therefore compared synaptic responses in interneurons
only in those recordings in which Alveus and Schaffer stimulations could evoke
similar sized population spikes (i.e., when the amplitudes of the maximal
antidromic and orthodromic population spike did not differ >30%). For
interneurons that received disynaptic Schaffer input, the maximal synaptic
amplitude after Alveus stimulation was 76 ± 16% (n = 11) of
the maximal amplitude after Schaffer stimulation. For interneurons that
predominantly received monosynaptic Schaffer input, this value was only 29
± 6% (n = 20; P < 0.01). This difference confirmed
our distinction between the two types of input after Schaffer stimulation.
Recordings of the synaptic dynamics were done at half-maximal stimulus
intensity. Based on stimulus-response curves that were measured (data not
shown) (see Wierenga and Wadman
2003), we estimated that a stimulus that evoked half–maximal
monosynaptic Schaffer input, evoked disynaptic Schaffer input with only 12
± 9% of the maximal amplitude. The contribution of disynaptic input in
our recordings of the synaptic dynamics of monosynaptic Schaffer input was
therefore expected to be <10%.
Differences between Schaffer and Alveus input dynamics
Repetitive Schaffer stimulation evoked synaptic currents in the
interneurons that showed paired-pulse ratios >1 in most cells (96%) for
frequencies >5 Hz. After the strong initial facilitation, synaptic
responses to later stimuli in the train decreased compared with the second,
but synaptic responses at the end of the train were still larger (150%)
than the first (Fig.
3A). At 5 Hz, the pattern was slightly different, 22% of
the cells showed paired-pulse ratios <1 and synaptic responses usually
increased throughout the train until the plateau level was reached
(Fig. 3A).
|
Alveus input also showed the initial facilitation for frequencies >5 Hz. In 83% of the cells paired-pulse ratios were >1, but values were smaller than for Schaffer input. In contrast to Schaffer input, plateau levels of Alveus input were close to 1 for most frequencies (Fig. 3, B and C). At 5 Hz, 52% of the cells showed paired-pulse ratios <1 and the synaptic responses during the train were usually all similar to the first response.
The mean plateau level and the mean paired–pulse ratio at different frequencies are given in Fig. 4, A and B, for Schaffer and Alveus input to CA1 interneurons. Paired-pulse ratios of both inputs increased with frequency and reached a maximum at 30 Hz, but values for Alveus input were smaller than for Schaffer input. At frequencies >10 Hz, the paired-pulse ratio of Alveus input was significantly smaller than that of the Schaffer input. The plateau level of Alveus input stayed close to 1 until 40 Hz and then decreased. The plateau levels of Schaffer input increased until 20 Hz and then decreased at frequencies >40 Hz. The largest difference in plateau levels between Schaffer input and Alveus input was observed for frequencies between 5 and 30 Hz (Fig. 4).
|
The difference in the dynamic properties between Schaffer and Alveus input
described above was evident despite a large variability between cells. At 20
Hz for instance, the plateau level of Alveus input varied between values of
0.4 and 1.8 (mean was 1.0) and for Schaffer input between 0.8 and 2.8 (mean
was 1.6). This large cell-to-cell variability was also recently reported by
Losonczy et al. (2002). In
nine of the interneurons described in the preceding text we could measure both
Alveus and Schaffer input. These recordings showed that within the same cell
Schaffer input always showed more facilitation than Alveus input, which
confirmed the difference observed in Fig.
4. The differences in plateau levels between individual
interneurons were not related to cell-specific parameters, such as input
resistance, membrane time constant, firing rate, or the age of the rat
(Spearman's rank test).
In the preceding text we argued that, if a disynaptic input component was present after Schaffer stimulation, it would be small compared with the direct Schaffer input. In addition, the difference in Alveus and Schaffer input dynamics (Fig. 4) further decreased the probability that such component will distort the recordings of the dynamics of the Schaffer input. Furthermore, there was no correlation between plateau levels (or paired–pulse ratios) of the monosynaptic Schaffer input and the relative contribution of the disynaptic input as determined from Alveus stimulation in the individual interneurons.
Schaffer input to pyramidal cells
To elucidate the role of the interneurons in the CA1 network, we compared
the dynamic properties of Schaffer input to interneurons and to pyramidal
cells. Pyramidal cells in the CA1 s. pyramidale (n = 10) were
recorded during stimulation of the Schaffer collaterals, and their input was
analyzed in the same way as for the interneurons. The onset of the synaptic
current in the pyramidal cells with respect to the peak of the population
spike was later than monosynaptic input to interneurons but earlier than
disynaptic input to interneurons (Table
1; t Schaffer input). The amplitude of the
synaptic current elicited in pyramidal cells was 62% larger than in
interneurons (Table 1 and
Fig. 2). Interestingly, the
synaptic dynamics of Schaffer input to pyramidal cells were quantitatively
similar to Schaffer input to CA1 interneurons, both showed similar
facilitation in the same frequency range
(Fig. 4, A and
B). This implies that during repetitive activation of the
CA1 network, interneurons and pyramidal cells always receive similar input
from the Schaffer collaterals.
Horizontal interneurons
Schaffer and Alveus inputs converging on the same interneuron showed
different dynamics, while Schaffer input to interneurons and pyramidal cells
showed comparable dynamics. This suggests that the presynaptic fibers and not
the postsynaptic cells determine the synaptic dynamics. However, analogous to
reports from other brain areas (Markram et
al. 1998; Rozov et al.
2001), synaptic properties of some connections in the CA1 network
depended on the identity of the postsynaptic cell. A special group of
interneurons in s. oriens was easily distinguished because they have
horizontally oriented dendritic trees. Three (3/5) of these cells were
reconstructed, and we could follow the axon branching into s.
lacunosum-moleculare (Fig.
5B). Horizontal interneurons did not receive monosynaptic
input from Schaffer collaterals, but synaptic inputs were elicited by Alveus
stimulation (Blasco-Ibáñez
and Freund 1995). In contrast to CA1 interneurons described in the
previous paragraphs, Alveus input to horizontal interneurons showed strong
facilitation for frequencies between 5 and 50 Hz
(Fig. 5). These findings were
in agreement with earlier reports (Ali and
Thomson 1998; Losonczy et al.
2002). Plateau levels of >200% of the reference response were
reached in these cells. This facilitation was stronger than for Schaffer input
to other CA1 interneurons (compare the y axes of
Fig. 5, C and
D, with Fig.
4).
|
Relation to stimulus intensity
In addition to the frequency dependence of the synaptic dynamics, we investigated how these dynamics depended on stimulus intensity. The dependence of the amplitude of the first synaptic response of the train on the stimulus intensity could be described by a sigmoidal curve in all cells (Fig. 6A). The synaptic dynamics of the Alveus input were hardly affected by the stimulus intensity as is illustrated in Fig. 6B, where we normalized the amplitudes to the first response (last panel). There was a weak correlation between stimulus intensity and plateau levels or paired-pulse ratios in only 33% of the interneurons (Spearman's ranked test). Pearson's correlation coefficients were –0.4 ± 0.1 and –0.3 ± 0.1 for the plateau levels and paired-pulse ratios, respectively (n = 6). This is illustrated in Fig. 6C, which shows the small decrease in paired-pulse ratio and plateau level with increasing stimulus intensity.
|
Increasing the stimulus intensity of Schaffer stimulation increases the number of recruited synapses that make contact with CA1 interneurons. In addition, the activation of pyramidal cells will recruit (feedback) inhibitory loops in the network. These may affect the intensity dependence of the plateau levels or paired-pulse ratios of Schaffer input. However, the same analysis as described for Alveus input resulted for Schaffer input in similar values for the mean Pearson's correlation coefficient for plateau levels (–0.4 ± 0.1) and paired-pulse ratios (–0.2 ± 0.2; n = 5). This is illustrated in Fig. 6D. The values mentioned in the preceding text were obtained with trains of 20 Hz, but similar results were obtained with 40-Hz trains. For 5-Hz trains, correlation coefficients were small but positive (0.3 ± 0.2 for plateau levels of Schaffer input; n = 9; P < 0.05). These results indicate that the dynamics of the Alveus and Schaffer input to CA1 interneurons did not critically depend on the stimulus intensity.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
40 Hz. Schaffer input to CA1 interneurons
and to pyramidal cells showed facilitation, with paired-pulse ratios reaching
180% and plateau levels reaching 150% of the first amplitude. The largest
differences between Schaffer and Alveus input occurred at frequencies between
5 and 30 Hz. Cells in s. oriens with horizontal dendritic trees formed a
special group of interneurons in which Alveus input showed strong
facilitation. Comparison with paired recordings
Previously, recordings from pairs of connected neurons revealed some of the
dynamic properties of individual synaptic connections in the hippocampal CA1
area (Ali and Thomson 1998;
Ali et al. 1998; Debanne et al.
1995,
1996). In this study, we did
not record from pairs because the probability to find a connected pair of CA3
pyramidal cell and CA1 interneuron (Schaffer input) is very low
(Debanne et al. 1995). In
contrast to paired recordings, where very few neurons of the local population
are activated by the presynaptic neuron, extracellular stimulation activates a
substantial fraction of the local network, including pyramidal cells and
interneurons that project to each other and to the recorded cell. Recruitment
of the synapses and cells underlying the observed responses is controlled by
stimulus intensity. The weak dependence on stimulus intensity
(Fig. 6) suggests that the
influence of the local network on the observed synaptic dynamics is only
small. It also indicates that although the properties of individual synapses
may vary substantially, there exists a certain uniformity of the contributing
synapses recruited over the range of stimulus intensities. We examined a large
range of physiological frequencies in a systematic way for the first time. Our
data generally agreed with data from paired recordings, but differences were
also encountered. Recordings from pairs of pyramidal neurons and interneurons
always showed paired-pulse depression (Ali
et al. 1998), whereas we often observed paired-pulse facilitation
of the Alveus input. It is not clear to what extent this is the result from
sampling bias in the paired recordings or the result from additional network
interactions in our experiments. The amount of facilitation in horizontal
cells observed here is smaller than previously reported by Ali et al.
(1998). Voltage-dependent
Na+ conductances are proposed to contribute to the large excitatory
postsynaptic current (EPSC) facilitation in these cells
(Martina et al. 2000). In our
recordings, these channels were blocked by intracellular QX314. In addition,
presynaptic inhibition (by other interneurons in the network activated by the
stimulus) might have decreased paired-pulse facilitation in our
recordings.
We chose to present our experimental data without the bias of a model.
Markram and Tsodyks have supplied a very useful phenomenological description
of the dynamics of single synaptic connections (paired recordings)
(Markram et al. 1998;
Tsodyks and Markram 1997). The
model does not include more than one presynaptic cell and possible effects of
the local network (e.g., presynaptic inhibition), which are very likely to
affect synaptic dynamics in our recordings. Nonetheless, this description
fitted our data reasonably well for most cells. The model description tended
to underestimate the observed paired-pulse facilitation, but it captured the
difference between the Schaffer and Alveus input to interneurons. The latter
showed a significantly larger utilization of synaptic efficacy (U; P
< 0.01) and a significantly shorter facilitation time constant
(
F; P < 0.05). The results of these fits were: for
Alveus input to interneurons (n = 12): U = 0.19 ±
0.03, rec = 226 ± 62 ms, F = 99
± 16 ms; for Schaffer input to interneurons (n = 20):
U = 0.11 ± 0.01, rec = 307 ± 52 ms,
F = 195 ± 30 ms; and for Schaffer input to pyramidal
cells (n = 8): U = 0.14 ± 0.03, rec =
186 ± 25 ms, F = 129 ± 24 ms (U,
utilization parameter of synaptic efficacy; rec, time constant
of recovery from depression; F, facilitation time constant;
values are means ± SE over all cells). For reasons outlined in the
preceding text, we considered it not valid to analyze our data in greater
detail with this formalism.
Inhibition
Stimulation of the Schaffer collaterals activated a large part of the CA1
network, presumably including feedforward and feedback inhibitory loops, while
Alveus stimulation most likely only activated feedback inhibition in the
network. Inhibitory synaptic currents do not show up in our recordings because
of their reversal close to holding potential. However, inhibitory synapses are
often strategically located at or near the soma
(Gulyás et al. 1999;
Megías et al. 2001),
and their activity may shunt excitatory inputs arriving at the dendrites
(Borg-Graham et al. 1998;
Staley and Mody 1992). The
difference in evoked inhibition during Alveus and Schaffer stimulation could
theoretically influence the measured synaptic dynamics of the excitatory
input. Our experimental conditions did not allow measuring inhibitory inputs
to the interneurons during the stimulus trains. We therefore cannot exclude
that differences in the strength or dynamics of the inhibitory inputs to the
recorded interneurons contributed to the observed differences in synaptic
dynamics after Schaffer and Alveus stimulation. In vivo, when synaptic input
changes the membrane potential of the neurons, the influence of inhibitory
synapses will not be negligible (Bracci et
al. 2001). However, under voltage-clamp conditions, shunting
inhibition only slightly affects the amplitude of dendritic synaptic current
measured in the soma (<10%) (unpublished observations). The observed
differences in dynamics between Schaffer and Alveus input were much larger
than the estimated contribution of inhibition, and we therefore think that our
recordings mainly reflect the synaptic dynamics of the excitatory
synapses.
Underlying mechanisms
The relation between paired-pulse ratios and plateau levels was frequency
dependent. For instance, Schaffer input showed plateau levels that were larger
than paired-pulse ratios at 5 Hz, whereas at higher frequencies paired-pulse
ratios were usually larger. These complex dynamics reflect the interplay
between (multiple) synaptic facilitation and depression processes with
different time constants. Dynamic properties of the synapses at short time
scales (tens to thousands of milliseconds) are often considered to have mainly
a presynaptic origin. Facilitation probably results from a build-up of
intracellular calcium in the presynaptic terminal
(Fisher et al. 1997;
Zucker, 1999), whereas
depletion of the fast releasable pool of presynaptic vesicles may be important
in short-term depression (Hagler and Goda
2001; Markram et al.
1998; Tsodyks and Markram
1997). However, postsynaptic processes (such as desensitization of
postsynaptic receptors) may also be involved
(Jones and Westbrook 1996).
Under our conditions, presynaptic modulation of transmitter release by other
activated neurons can also not be excluded. Cholinergic input did not play a
role in these experiments (see METHODS).
It was recently shown that postsynaptic cells can express receptor subunits
in an input-specific manner (Gardner et
al. 2001; Nyíri et al.
2001; Tóth and McBain
1998). Targeting of specific AMPA channels to mossy fiber and
recurrent synapses was reported in CA3 interneurons
(Tóth and McBain 1998).
Calcium-permeable AMPA channels show strong facilitation during repetitive
activation by a relief from polyamine block
(Rozov et al. 1998). Analogous
to CA3 interneurons, CA1 interneurons may specifically target these channels
at synapses made by the Schaffer collaterals. Calcium-impermeable AMPA
receptors, which are not sensitive to polyamines and show less facilitation,
may be targeted to synapses made by CA1 pyramidal cells. This could be an
intriguing postsynaptic mechanism by which CA1 interneurons could
differentially regulate the dynamics of specific inputs.
Hippocampal rhythms
In vivo, oscillations occur in the hippocampus, which are proposed to be
relevant to memory and other cognitive processes
(Buzsáki and Chrobak
1995; Hasselmo et al.
2001; Lisman
1999). Models of this activity are based on the generation of
rhythmic, synchronous synaptic potentials
(Hasselmo et al. 2001;
Traub et al. 1996;
Wang and Buzsáki 1996).
So far, short-term synaptic dynamics have not been included in these models.
Our findings, as well as several other reports (e.g.,
Bracci et al. 2001;
Losonczy et al. 2002), clearly
show that the involved synapses are dynamic synapses. Their dynamics and
especially the observed difference between Alveus and Schaffer input in the
relevant frequency range of 5–30 Hz could have important consequences
for understanding the generation and maintenance of in vivo hippocampal
rhythms (Klausberger et al.
2003).
The special status of the horizontal cells in the CA1 network might be
particularly interesting in this respect. The majority of excitatory afferents
to horizontal interneurons (>75% of all excitatory boutons) originates from
CA1 pyramidal cells
(Blasco-Ibáñez and Freund
1995). The axons of the three reconstructed horizontal cells in
this study ramified into s. lacunosum–moleculare, the region of
entorhinal input to CA1 pyramidal cells, whereas most other CA1 interneurons
projected to s. oriens, s. pyramidale, and s. radiatum. Activation of
horizontal cells in s. oriens decreases synaptic transmission in s. radiatum
and s. lacunosum-moleculare (Yanovsky et
al. 1997). This suggests that horizontal cells are specialized to
control and/or modulate entorhinal input to the pyramidal cells as a function
of population activity (i.e., in a feedback manner). The strong facilitation
of Alveus input to horizontal interneurons compared with other CA1
interneurons may reflect a special function for horizontal interneurons and
may lead to oscillations in enthorhinal input to the CA1 area during network
oscillations (Hasselmo et al.
2001; Klausberger et al.
2003). Other feedback interneurons in the CA1 network may have a
more general function in controlling the population activity (see following
text).
Functional implications
A strong convergence of pyramidal axons onto single interneurons exists in
the CA1 area (Blasco-Ibáñez
and Freund 1995; Buhl et al.
1994a; Gulyás et al.
1999). The Alveus input to a single interneuron therefore reflects
the activity of the pyramidal cell population. The plateau level of Alveus
input to CA1 interneurons was comparable to reference values for frequencies
40 Hz (Fig. 4A).
This implies that the gain of the inhibitory feedback loop will be relatively
independent of the frequency at which the pyramidal cells fire. This may be an
important contribution to the stability of the network.
Interneurons receiving monosynaptic Schaffer input are activated before or
simultaneously with the pyramidal cell population
(Karnup and Stelzer 1999).
They may inhibit specific pyramidal cells or modulate specific inputs to
pyramidal cells, analogous to the role of feedforward inhibition in shaping
receptive fields of cortical neurons
(Dykes 1997;
Dykes et al. 1984). Somatic
feedforward inhibition shortens the spike integration window for synaptic
summation in CA1 pyramidal cells, improving their detection of coincident
inputs (Pouille and Scanziani
2001). These feedforward inhibitory actions modulate information
processing by the CA1 pyramidal cells. The similar dynamics of synapses made
by the Schaffer collaterals onto pyramidal cells and onto interneurons may
serve to keep input modulation by feedforward inhibition relatively
independent of strength and frequency of the Schaffer input to the pyramidal
cells.
Activity in the CA1 network in vivo operates over a wide frequency range
(Buzsáki et al. 1983).
At input frequencies between 5 and 30 Hz, Schaffer input to interneurons shows
stronger facilitation than Alveus input. As a consequence of this difference,
the relative contribution of feedforward and feedback inhibition has become a
dynamical property of the circuit. This means that the temporal properties of
the input signal strongly contribute to the functional connectivity in the CA1
network.
|
DISCLOSURES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
|
FOOTNOTES |
|---|
Address for reprint requests: W. J. Wadman, Swammerdam Institute for Life Sciences, Section Neurobiology, University of Amsterdam, Kruislaan 320, 1098 SM Amsterdam, Netherlands (E-mail: wadman{at}science.uva.nl).
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
Ali AB,
Deuchars J, Pawelzik H, and Thomson AM. CA1 pyramidal to basket and
bistratified cell EPSPs: dual intracellular recordings in rat hippocampal
slices. J Physiol 507:
201–217, 1998.
Ali AB and
Thomson AM. Facilitating pyramid to horizontal oriens-alveus interneurone
inputs: dual intracellular recordings in slices of rat hippocampus.
J Physiol 507:
185–199, 1998.
Andersen P, Eccles JC, and Løyning Y. Recurrent inhibition in the hippocampus with identification of the inhibitory cell and its synapses. Nature 198: 540–542, 1963.[Medline]
Andersen P,
Eccles JC, and Løyning Y. Pathway of postsynaptic inhibition in the
hippocampus. J Neurophysiol 27:
608–619, 1964.
Blasco-Ibáñez JM and Freund TF. Synaptic input of horizontal interneurons in stratum oriens of the hippocampal CA1 subfield: structural basis of feed-back activation. Eur J Neurosci 7: 2170–2180, 1995.[ISI][Medline]
Borg-Graham LJ, Monier C, and Freqnac Y. Visual input evokes transient and strong shunting inhibition in visual cortical neurons. Nature 393: 369–373, 1998.[Medline]
Bracci E,
Vreugdenhil M, Hack SP, and Jefferyes JGR. Dynamic modulation of
excitation and inhibition during stimulation at gamma and beta frequencies in
the CA1 hippocampal region. J Neurophysiol
85: 2412–2422,
2001.
Buhl EH, Halasy K, and Somogyi P. Diverse sources of hippocampal unitary inhibitory postsynaptic potentials and the number of synaptic release sites. Nature 368: 823–828, 1994a.[Medline]
Buhl EH, Han Z,
Lörinczi Z, Stezhka VV, Karnup SV, and Somogyi P. Physiological
properties of anatomically identified axo-axonic cells in the rat hippocampus.
J Neurophysiol 71:
1289–1307, 1994b.
Buhl EH, Szilágyi T, Halasy K, and Somogyi P. Physiological properties of anatomically identified basket and bistratified cells in the CA1 area of the rat hippocampus in vitro. Hippocampus 6: 294–305, 1996.[ISI][Medline]
Buonomano DV. Decoding temporal information: a model based
on short-term synaptic plasticity. J Neurosci
20: 1129–1141,
2000.
Buonomano DV,
Hickmott PW, and Merzenich MM. Context-sensitive synaptic plasticity and
temporal-to-spatial transformations in hippocampal slices. Proc
Natl Acad Sci USA 94:
10403–10408, 1997.
Buzsáki G and Chrobak JJ. Temporal structure in spatially organized neuronal ensembles: a role for interneuronal networks. Curr Opin Neurobiol 5: 504–510, 1995.[ISI][Medline]
Buzsáki G and Eidelberg E. Direct afferent excitation and
long-tem potentiation of hippocampal interneurons. J
Neurophysiol 48:
597–607, 1982.
Buzsáki G, Leung LW, and Vanderwolf CH. Cellular bases of hippocampal EEG in the behaving rat. Brain Res 287: 139–171, 1983.[Medline]
Cobb SR, Buhl EH, Halasy K, Paulsen O, and Somogyi P. Synchronization of neuronal activity in hippocampus by individual GABAergic interneurons. Nature 378: 75–78, 1995.[Medline]
Debanne D,
Guérineau NC, Gähwiler BH, and Thompson SM. Physiology and
pharmacology of unitary synaptic connections between pairs of cells in areas
CA3 and CA1 of rat hippocampal slice cultures. J
Neurophysiol 73:
1282–1294, 1995.
Debanne D, Guérineau NC, Gähwiler BH, and Thompson SM. Paired-pulse facilitation and depression at unitary synapses in rat hippocampus: quantal fluctuation affects subsequent release. J Physiol 491: 163–176, 1996.[ISI][Medline]
Dobrunz LE and Stevens CF. Response of hippocampal synapses to natural stimulation patterns. Neuron 22: 157–166, 1999.[ISI][Medline]
Dvorak-Carbone H and Schuman EM. Patterned activity in stratum
lacunosum moleculare inhibits CA1 pyramidal neuron firing. J
Neurophysiol 82:
3213–3222, 1999.
Dykes RW. Mechanisms controlling neuronal plasticity in somatosensory cortex. Can J Physiol Pharmacol 75: 535–545, 1997.[ISI][Medline]
Dykes RW,
Landry P, Metherate R, and Hicks TP. Functional role of GABA in cat
primary somatosensory cortex: shaping receptive fields of cortical neurons.
J Neurophysiol 52:
1066–1093, 1984.
Fisher SA, Fischer TM, and Carew TJ. Multiple overlapping processes underlying short-term synaptic enhancement. Trends Neurosci 20: 170–177, 1997.[ISI][Medline]
Galarreta M and Hestrin S. A network of fast-spiking cells in the neocortex connected by electrical synapses. Nature 402: 72–75, 1999.[Medline]
Gardner SM,
Trusell LO, and Oertel D. Correlation of AMPA receptor subunit composition
with synaptic input in the mammalian cochlear nuclei. J
Neurosci 21:
7428–7437, 2001.
Gulyás AI, Megías M, Emri Z, and Freund TF. Total
number and ratio of excitatory and inhibitory synapses converging onto single
interneurons of different types in the CA1 area of the rat hippocampus.
J Neurosci 19:
10082–10097, 1999.
Gupta A, Wang
Y, and Markram H. Organizing principles for a diversity of GABAergic
interneurons and synapses in the neocortex. Science
287: 273–278,
2000.
Hagler DJ Jr
and Goda Y. Properties of synchronous and asynchronous release during
pulse train depression in cultured hippocampal neurons. J
Neurophysiol 85:
2324–2334, 2001.
Halasy K, Buhl EH, Lörinczi Z, Tamás G, and Somogyi P. Synaptic target selectivity and input of GABAergic basket and bistratified interneurons in the CA1 area of the rat hippocampus. Hippocampus 6: 306–329, 1996.[ISI][Medline]
Hasselmo ME, Bodelón C, and Wyble BP. A proposed function for hippocampal theta rhythm: separate phases of encoding and retrieval enhance reversal of prior learning. Neural Comput 14: 793–817, 2001.
Ishizuka N, Weber J, and Amaral DG. Organization of intrahippocampal projections originating from CA3 pyramidal cells in the rat. J Comp Neurol 295: 580–623, 1990.[ISI][Medline]
Jones MV and Westbrook GL. The impact of receptor desensitization on fast synaptic transmission. Trends Neurosci 19: 96–101, 1996.[ISI][Medline]
Kandel ER,
Spencer WA, and Brinley FJ Jr. Electrophysiology of hippocampal neurons.
I. sequential invasion and synaptic organization. J
Neurophysiol 24:
225–242, 1961.
Karnup S and
Stelzer A. Temporal overlap of excitatory and inhibitory afferent input in
guinea pig CA1 pyramidal cells. J Physiol
516: 485–504,
1999.
Klausberger T, Magill PJ, Márton LF, Roberts JDB, Cobden PM, Buzsáki G, and Somogyi P. Brain-state- and cell-type-specific firing of hippocampal interneurons in vivo. Nature 421: 844–848, 2003.[Medline]
Lacaille J-C. Postsynaptic potentials mediated by
excitatory and inhibitory amino acids in interneurons of stratum pyramidale of
the CA1 region of rat hippocampal slices in vitro. J
Neurophysiol 66:
1441–1454, 1991.
Li X-G, Somogyi P, Ylinen A, and Buzsáki G. The hippocampal CA3 network: an in vivo intracellular labeling study. J Comp Neurol 339: 181–208, 1994.[ISI][Medline]
Lisman JE. Relating hippocampal circuitry to function: recall of memory sequences by reciprocal dentate-CA3 interactions. Neuron 22: 233–242, 1999.[ISI][Medline]
Losonczy A,
Zhang L, Shigemoto R, Somogyi P, and Nusser Z. Cell type dependence and
variability in the short-term plasticity of EPSCs in identified mouse
hippocampal interneurones. J Physiol
542: 193–210,
2002.
Maccaferri G and McBain CJ. Passive propagation of LTD to stratum oriens-alveus inhibitory neurons modulates the temporoammonic input to the hippocampal CA1 region. Neuron 15: 137–145, 1995.[ISI][Medline]
Markram H, Wang
Y, and Tsodyks M. Differential signaling via the same axon of neocortical
pyramidal neurons. Proc Natl Acad Sci USA
95: 5323–5328,
1998.
Martina M, Vida
I, and Jonas P. Distal initiation and active propagation of action
potentials in interneuron dendrites. Science
287: 295–300,
2000.
Megías M, Emri Z, Freund TF, and Gulyás AI. Total number and distribution of inhibitory and excitatory synapses on hippocampal CA1 pyramidal cells. Neuroscience 102: 527–540, 2001.[ISI][Medline]
Nyíri G,
Freund T, and Somogyi P. Input-dependent synaptic targeting of
2-subunit-containing GABAA</
; n = 21) and
Alveus stimulation (
; n = 23) for stimulus trains at 5, 20, and
40 Hz. Amplitudes are normalized to the 1st amplitude of the trains.
Significant differences are indicated (* P < 0.05; ** P
< 0.01).
) as a function of frequency. B: same for paired-pulse ratios.
Schaffer input to interneurons and pyramidal cells showed more facilitation in
plateau levels and paired–pulse ratios than Alveus input to
interneurons. Significant differences between Schaffer and Alveus input to CA1
interneurons are indicated (* P < 0.05; ** P < 0.01).
Schaffer input to interneurons and pyramidal cells had indistinguishable
parameters.
