| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Departments of Neurology and Physiology, The David Geffen School of Medicine at UCLA, Los Angeles, California 90095-1769
Submitted 7 February 2003; accepted in final form 24 March 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
; Diamond
2001; Isaacson
1999; Kullmann et al.
1999; Momiyama
2000; Rumbaugh and Vicini
1999; Sattler et al.
2000; Tovar and Westbrook
1999) and at certain presynaptic terminals
(Breukel et al. 1998;
Milner and Drake 2001;
Paquet and Smith 2000;
Woodhall et al. 2001). This
anatomical arrangement is also reflected by NMDA receptor subtype
heterogeneity. Thus the five NMDA-receptor subunits, NR1 (8 isomers, A-H) and
the four NR2 subunits (A-D), co-assemble into functional NMDA receptors,
presumably in the stoichiometry of two NR1 glycine-binding and two NR2
glutamate-binding subunits (Laube et al.
1998). In cell culture, synaptic NR2A-containing receptors appose
presynaptic release sites, whereas NR2B-containing receptors are located
predominantly outside the postsynaptic density
(Tovar and Westbrook 1999).
The latter subunits also appear to be present on presynaptic terminals of
entorhinal cortex (EC) neurons in acutely prepared brain slices
(Woodhall et al. 2001). There
is evidence from in vitro studies of heterotrimeric NMDA receptors such as
NR1/NR2A/NR2B subunit combinations
(Brimecombe et al. 1997;
Chen et al. 1997;
Vicini et al. 1998), but if
such receptors are expressed in vivo
(Pina-Crespo and Gibb 2002),
their subcellular localization is unknown.
NR2A and NR2B receptor subunits are coupled through different C-terminal
domains to distinct second-messenger pathways
(Chan and Sucher 2001;
Krupp et al. 2002;
Lieberman and Mody 1994;
Sattler et al. 2000), which in
addition to their separate localizations serve to diverge NMDA receptor
signaling. For example, extrasynaptic NR2B, but not synaptic NR2A receptors,
induce long-term genomic translational effects through activation of the
cAMP-responsive binding protein
(Hardingham et al. 2002).
Whenever NMDA receptors in the same intact synaptic structure can be expressed
on the presynaptic terminal and both in- and outside of the postsynaptic
density, it complicates the experimental design aimed at distinguishing
between the roles of the different receptors. Due to the poor signal-to-noise
ratio of spontaneous NMDA receptor-mediated events, stimulus-evoked events are
the usual choice for characterizing the activation of postsynaptic NMDA
receptors. A major difference between stimulus-evoked and spontaneous events
is that spontaneous events are monosynaptic in origin, causing a relatively
limited spillover of transmitter to extrasynaptic receptors or neighboring
postsynaptic sites. In contrast, events evoked by stimulation of afferent
fibers can be polysynaptic and may be due to the simultaneous activation of
neighboring release sites, causing considerable spillover activating
extrasynaptic receptors or neighboring synapses
(Barbour and Hausser 1997;
Diamond 2001;
Kullmann et al. 1999;
Lozovaya et al. 1999). To
further our understanding of this anatomical and functional receptor
heterogeneity in an intact system of synaptical structures, it is necessary to
establish physiological methods to characterize the mode of activation of NMDA
receptors at synapses. The purpose of the present study was twofold: first we
set out to distinguish the properties of spontaneous NMDA receptor-mediated
events in dentate gyrus granule cells including the underlying single-channel
conductance. Second, we compared spontaneous and stimulus-evoked events
recorded in the same cell to distinguish between the activation of NMDA
receptor subtypes by evoked and spontaneously released glutamate.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
Rat brains were obtained from 8 to 10 wk old Sprague Dawley rats (Harlan).
Briefly, rats were decapitated after pentobarbital (75 mg/kg ip) anesthesia,
and the head was quickly chilled. The brain was then rapidly dissected out in
artificial cerebrospinal fluid (ACSF) containing (in mM): 126 NaCl, 2.5 KCl, 2
CaCl2, 2 MgCl2, 26 NaHCO3, 1.25
NaH2PO4, 10 D-glucose, 0.3 ascorbate, and 1.0
pyruvic acid. The brain was glued with cyanocryalate to the platform of a
Leica 1000 VS microtome and cut into 350-µm-thick horizontal slices.
Osmolarity of ACSF was 300 ± 10 mosM, pH 7.4 ± 0.1, and
kynurenic acid (3 mM) was added to the ACSF used in the dissection procedures
only. Slices were hemi-sected and stored in a chamber with continuously
carbogenated ACSF for 1–4 h at 32°C before being individually
transferred to a recording chamber. NMDA responses were pharmacologically
isolated in a modified ACSF (ACSFNMDA) containing as described
above but with only 5 µM MgCl2 and with the following compounds
added (in µM) 10 DNQX, 30 picrotoxin, and 10 D-serine. For field
and whole cell recordings, we used slices that were located approximately
between 3.5 and 6.5 mm dorsal to the interaural line (IAL, also see
Fig. 8A). In a
separate set of experiments we also recorded bursts in whole cell and field
configuration in slices positioned approximately between 2 and 3 mm above the
interaural line (see Fig.
8B) (Paxinos and
Watson 1998). During the field recordings of these ventral slices,
a cut was made in the slice by a manipulator-controlled micro-scalpel to sever
the entorhinal cortex from the subiculum and dentate gyrus, thus transecting
the perforant path.
|
Field potential recordings
During recordings slices were held in place on lens tissue paper by small
platinum weights and continuously perfused with carbogenated
ACSFNMDA at 34–35°C at a flow rate of 1.5–2 ml/min
in a chamber with an internal carbogenated and humidified atmosphere. A
bipolar tungsten stimulating electrode was placed in the perforant
path/angular bundle immediately before crossing the hippocampal fissure and
set to deliver a 40-µs stimulus every 30 s. A recording glass electrode
filled with ACSFNMDA prepared as described below for whole cell
recordings was placed in the outer molecular layer of the dentate gyrus.
Current intensities for stimulation were set to evoke field potentials of
0.4 mV peak amplitude, i.e., 
60% of the stimulus intensity required
to evoke maximal peak responses.
Whole cell voltage-clamp recordings
Recording electrodes were made of borosilicate glass capillaries with an
inner filament (1.10 mm ID, 1.5 mm OD, Garner glass) and pulled on a two-stage
Narishige PP-83 puller or Zeitz horizontal puller (resistances: 5–8
M
). Intracellular solution (ICS) contained (in mM) 135 CsCl, 4 NaCl, 2
MgCl2, 10
N-2-hydroxyethylpiperazine-N'-2-ethanesulphonic acid
(HEPES), 0.05 EGTA, 5 QX-314, 5 tetraethylammoniumchloride (TEA), 2 Mg-ATP,
0.5 Na2-GTP. The pH of the ICS was adjusted to 7.2 with CsOH and
its osmolarity was 285–290 mosM. Type of chamber and ACSFNMDA
were similar for field and whole cell recordings. A stimulating electrode was
placed in the angular bundle at the border of entorhinal cortex and the
subiculum and set to deliver a stimulus every 20 s. Dentate gyrus granule
cells were recorded using blind-patching, i.e., by lowering patch electrodes
down through the granule cell layer while monitoring responses to 5-mV test
pulses (Staley et al. 1991). Seals (>7 G) were formed by applying
gentle suction to patch pipettes. Whole cell currents were amplified and
low-pass filtered (2 kHz, 8-pole bessel) using an Axopatch 200A amplifier
(Axon Instruments, Burlingame, CA) and 75% compensated for series resistance.
Cells were held for a period of 25 min during which series resistance and
capacitance were monitored every 3–4 min. After 10-min recording of
baseline spontaneous and evoked events, ifenprodil (10 µM) was added to the
perfusate and an additional 10 min were recorded after its addition. Cells in
which series resistance or capacitance deviated by >50% from initial values
were excluded from analysis. Also, cells with series resistances >18
M at any time during the recording were excluded from analysis.
Establishing the time course of evoked NMDA EPSCs
The true synaptic conductance was determined by the method of charge
recovery of timed evoked events during a series of predetermined voltage jumps
(Hausser and Roth 1997;
Pearce 1993). The Strathclyde
whole cell program V3.2.6 (courtesy of J. Dempster) was used to generate
voltage jumps from the reversal potential to –60 mV. Three to four
repetitions of a series of 11 trials with stimulation times from –150 to
+100 ms relative to the voltage jump were recorded and traces uncontaminated
by spontaneous events were averaged and subtracted from traces without
stimulation. The charge recovery was determined as the area-over-curve (AOC)
of subtracted traces and plotted against time relative to the voltage jump. A
single exponential was fitted to the curve from 150 to 15 ms before the
voltage jump and the time constant compared with the time constant from events
recorded 100 ms after the voltage jump.
Data sampling
Field DC potentials were amplified 100 times through the headstage and again 10 times through a BrownLee 440 precision amplifier (San Jose, CA) and band-pass filtered (0.1–2 kHz) before digitization and storage on a Pentium-processor-based PC using the in-house designed event detection and analysis (EVAN) software package (www.EV-AN.Thotec.com). Analog whole cell recordings were low-pass filtered at 2 kHz (–3 dB, 8-pole Bessel, Frequency Devices 9002) and digitized on-line at 8 kHz. Back-ups of field and whole cell recordings were digitized through a Neurocorder (Neurodata) at 88 kHz and stored on videotape.
Drugs
The sources of drugs were as follows: ifenprodil (Tocris, stock dissolved in DMSO to100 mM), D-APV (Tocris), R-(–)-3-(2-carboxypiperazin-4-yl)propanephosphonic acid (D-CPP, Tocris), lidocaine N-ethyl bromide (QX-314, Alomone lab), tetraethylammoniumchloride (TEA, Sigma), 6,7-dinitroquinoxaline-2,3(1H,4H)-dione (DNQX, Sigma), picrotoxin (Sigma) and D-serine (Sigma).
Detection, selection, and analysis of NMDA receptor-mediated events
NMDA receptor-mediated events were identified in 300-s-long continuous
recordings based on the following criteria: a baseline (BL) of 50 ms followed
by a downward deflection of 3 x BL SD (
BL) lasting
for >0.5 ms was identified as an event and periods of 100 ms before and 300
ms after were sampled around the detection time. A dead time of 15 ms and
baseline standard deviation (SD) criteria of BL < 10 pA
were used to select events only once. Events were counted for frequency
measurement but rejected for further analysis if any other events were
detected in –50 ms prior or 250 ms after the detection point for a given
event. Gaussian distributions were fitted to 10–90% RT, peak, 67% decay
time, and total charge carried by the event (AOC, from detection till +250
ms). The number of Gaussians in the overall distribution for each parameter
was determined by the F value for two distributions with n
and n + 1 peaks by F = [(SSn –
SSn+1)/(fn
–
fn+1)]/(SSn/fn+1)
and the associated P value by P = 1 – invf[F,
(fn –
fn+1),
fn+1], using P
< 0.05 as level of significance where SS is sum of squares, f is
degrees of freedom, and invf is the inverse of the F distribution
function. The inter-event intervals were binned in log-scaled bins and the
mean (µ) for the exponential distribution was determined by fitting the
function N = l*exp{ln (x/ µ) – exp[ln (x/µ)]} to
the data. The weighted decay time constant (
weighted) of
spontaneous and evoked events was estimated from the mean EPSC trace by
dividing the AOC measured from the peak (AOCpeak) by the peak
amplitude or by fitting a single exponential to the 20–80% decay-phase.
An estimate of mean-channel conductance, 
, was obtained by
nonstationary fluctuation analysis (NSFA) applied to groups of 60–90
events comprised in the major Gaussian (mean ± SD) of the peak
distribution. The variance around the mean in the period 5 240 ms after
the mean peak were divided into 25 bins and plotted against the mean
single-channel current and fitted with the relation
in which 2 is the bin-variance of traces around the mean,
is the baseline variance,
i is the single-channel current and I is the mean current
passed by the ensemble of N channels
(De Koninck and Mody 1994).
The conductance is given by = i/(Vm
– Vrev) in which Vm is the
membrane (holding) potential and Vrev is the reversal
potential of NMDA receptors in this preparation. A Vrev =
3 mV was obtained from I-V plots of evoked events at holding
potentials ranging from –60 to +40 mV. An estimate of the average number
of channels open at the peak is given by the ratio of peak current to the
single-channel current, i, obtained by NSFA. Spectral analysis was
performed on 39 discontinuous 4.096-s-long (32,768 points) traces (low-pass
filtered at 3 kHz, 8-pole Bessel) of a baseline period and a similar period
recorded during D-APV or D-CPP. The data were
Blackman-windowed and the one-sided power spectrum for each trace was obtained
by fast Fourier transformation (FFT) and all FFTs averaged for each period.
The subtracted power spectrum (baseline – APV or CPP) was plotted in a
log-log diagram and the corner-frequencies, fc, were
obtained by fitting the data to a Lorentzian of the general form:
,
in which S(0) is the lowest frequency noise intercept of the spectral
density and fc is the corner-frequency. The corner
frequencies relate to the relaxation time constant of a single exponential by
= 1/(2
· fc).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
Pharmacological isolation of NMDA receptor-mediated events recorded at
34–35°C was obtained through the removal of currents through
GABAA, AMPA, and kainate receptors, the application of the
glycine-site agonist D-serine, and recording in a low
Mg2+-medium to relieve the voltage-dependent block of
NMDA receptors. Postsynaptic Na+ and K+ channels were
blocked by QX-314, Cs, and TEA in the intracellular solution. While holding
cells at –60 mV, the holding current decreased to a stable value
(–35 ± 10.7 pA, n = 17,
Table 1) 3–5 min after
establishment of the whole cell configuration. Application of 50 µM
D-APV completely blocked the synaptic events, decreased the
baseline holding current by 18.8 ± 3 pA (n = 5) and decreased
the SD of the baseline current (Fig. 1,
A–C). Therefore the spontaneous events were
mediated solely by NMDA receptors while the openings of NMDA channels also
significantly contributed to the baseline variance. NMDA receptor-mediated
events are characterized by slow rise time, long decay time, and a mean peak
of –13.2 ± 0.8 pA and baseline SD (BL) 2.7
pA (Table 1, means ± SE
of 17 granule cells). A set signal to noise ratio of 3 was used as detection
limit (ratio of lowest peak detected and BL). Distributions
of peak amplitude and of rise-time were in all cells best fitted to two or
three Gaussian distributions (Figs.
2B and
3B). Similarly, 67%
decay times were best described by two or three Gaussian distributions
(Fig. 3C), whereas in
the majority of cells the distribution of charge carried by the events (AOC,
fC) was adequately described by a single-Gaussian with a mean of –615
± 39 (SE) fC (Fig.
2C, Table
1). Overall, there is an apparent proportionality among rise time,
67% decay time and amplitude of the events
(Fig. 3D).
|
|
|
|
Spectral analysis of 160-s-long traces (average of 39 traces of 4.096 s
length) in four cells resolved two corner frequencies in the range
2.6–3.2 and 210–288 Hz, corresponding to exponential relaxation
time constants of 49–61 and 0.6–0.8 ms respectively. The
low-frequency component is in good agreement with the decay time constant of
the average event recorded during control conditions
(Fig. 4, A and
B), while the latter frequency most likely reflects the
average lengths of openings of NMDA channels in these neurons
(Lieberman and Mody 1994).
|
NMDA channel conductance
An estimate of the synaptically activated NMDA-channel conductance was
obtained through nonstationary fluctuation analysis (NSFA,
Fig. 2D and
Table 1). To reduce the
variation due to possible differences in synaptic size, number of channels,
and dendritic filtering (De Koninck and
Mody 1994; Traynelis et al.
1993), NSFA was performed on 60–90 events selected around
the mean ± 1 SD in the dominant Gaussian of the peak-amplitude
distribution (Fig. 2B,
shaded area). Arguably, scaling of individual peaks to the mean peak before
subtraction would be preferred to account for the variable number of channels
shaping the individual event if the popen is high.
However, NMDA channels possess a high intrinsic variance in the time from
binding of agonist to first opening
(Lester et al. 1990) and a
relatively low popen. Therefore, even after selecting
events as described in the preceding text, a significant bias is introduced
with peak scaling because of differing peak times. Without peak scaling, the
analysis of 17 granule cells yielded an estimated average unitary current
i = –3.42 pA corresponding to a conductance, = 60
± 5 pS (Table 1). With
an average peak amplitude of –13.2 pA
(Table 1), this led to an
estimate of approximately four NMDA channels open at the peak of the average
spontaneous NMDA EPSC.
Comparison of evoked and spontaneous events
Events were evoked by a stimulating electrode in the perforant path at the subiculum-entorhinal cortex border (Fig. 5A, site 1) or in the hippocampal fissure at the site from which field potentials were evoked (Fig. 5B, site 2) every 20 s. Individual slices were stimulated at one site only. Events evoked from both sites had longer rise-times and larger peak-currents and AOC than the spontaneous events (Tables 2 and 3). However, the decay time constant of site-1-evoked events was not significantly larger than that of spontaneous events (paired t-test, n = 17, P = 0.10). Evoked events from stimulation site 2 differed from those evoked from site 1 by higher peak amplitudes and by longer decay time constants, significantly different from that of spontaneous events (table 3, P < 0.05, paired t-test) recorded during the same period.
|
|
|
Charge recovery of evoked events
Stimulation at site 1 of cells held at +3 mV (the reversal potential of
currents through NMDA receptors) did not cause detectable events. By timing
the stimulation relative to a jump to a holding potential of –60 mV from
the reversal potential, the remaining charge caused by the event can be
recovered by the shift to the latter holding potential when corrected for the
capacitive current. The charge recovery in a series of 11 trials with
stimulation times ranging from –150 to +100 ms relative to the voltage
jump is shown in Fig.
6A. An event recorded 100 ms after the voltage jump is
shown in Fig. 6B. The
average rise time (RT0–100), for these events of four cells
is 16.2 ± 2.4 (SE) ms. The fitted decay time constant, is 52.3
± 6.3 ms. Fitting a single exponential to the charge recovery graph
from 150 to 15 ms before the jump (Fig.
6B) yields an average time constant of 48.1 ± 7.6
ms.
|
Effect of ifenprodil on spontaneous and evoked events in whole cell recordings
Ifenprodil (10 µM, in final 0.1 
DMSO) had no effect on RT
10–90%, 67% decay time, peak-current, weighted or fitted tau measured in
the decay phase of EPSC's charge carried by the events (AOC, pA*ms), baseline
holding current, baseline SD, or unit channel conductance of spontaneous
events (Table 2, Fig. 7A–F).
However, ifenprodil significantly reduced the frequency of spontaneous events
(Table 2,
Fig. 7C). Evoked
events from site 1 recorded simultaneously during the same period had
significantly reduced peak currents and AOC
(Fig. 7D) which was
similar for site 2 stimulation (Table
3).
|
Field potentials
Field potentials of pharmacologically isolated events were evoked with a stimulating electrode in the perforant path close to the hippocampal fissure and to the stratum lacunosum moleculare. With a current stimulation adjusted to yield 60% of the maximal response, field-responses remained stable for >1 h. As seen with evoked events in whole cell configuration, ifenprodil significantly (paired t-test, P < 0.05, n = 6) reduced the peak amplitude and AOC of NMDA receptor-mediated field EPSPs. In addition, ifenprodil also reduced the decay time of the field potentials (table 3, n = 6, P < 0.05, paired t-test, 5–10 min after application).
Dorsoventral differences in synaptic bursting of granule cells
We identified a sharp transition zone along the dorsoventral axis
separating two regions of the hippocampus with different levels of synaptic
bursting of granule cells. Ventral to this zone, lying 3 mm above the
interaural line (Fig.
8A), five of five granule cells displayed spontaneous
bursts (2–3/min) with peak currents 2 orders of magnitude larger
than the mean peak current of normal events in ventral and dorsal slices
(Fig. 8C). This type
of large event was never observed in dorsal slices. The presence of
spontaneous field events with similar frequency (2–3/min) recorded in
the same region is most likely a reflection of intracellular bursts. Evoked
field and whole cell events in this ventral region were characterized by
double peaks, 2–300 ms apart, both sensitive to inhibition by
D-APV (not shown) and ifenprodil
(Fig. 8D). In whole
cell configuration, these bursts reversed polarity when recorded at +40 mV
(Fig. 8, C and
D), excluding the possibility of unclamped
Ca2+ currents as the underlying mechanism. In field
recordings, we investigated the possibility that the double peaks observed
were due to reverberating activity or antidromic stimulation between
entorhinal cortex and dentate gyrus by cutting away the entorhinal cortex
(Fig. 8B) while
recording field potentials. In each of the four slices studied, the double
peak disappeared after a cut severing the entorhinal cortex from the dentate
gyrus (Fig. 8F).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
;
Hestrin et al. 1990;
Traynelis et al. 1993). Here
we characterized the spontaneous and evoked events purely mediated by NMDA
receptors in the same cell, which allowed us to discern considerable
differences between the activation of NMDA receptors by stimulus evoked and
spontaneous glutamate release in hippocampal slices. We characterized the
basic parameters of pharmacologically isolated spontaneous NMDA receptor
mediated events in rat dentate gyrus granule cells (Figs.
2 and
3,
Table 1). We also demonstrated
that spontaneous openings of NMDA receptors significantly contribute to the
holding current of the cell shown by its sensitivity to D-APV or
D-CPP (Fig. 1, B and
C). Furthermore we have established that the NR2B subunit
containing NMDA channels do not participate in the shaping of EPSCs activated
by spontaneously released transmitter but regulate the frequency of these
events (Fig. 7C). In
contrast, events evoked in the same cell by stimulating the perforant path
were significantly diminished by ifenprodil
(Table 2,
Fig. 7D), an effect
quite similar to that described for NMDA receptor-mediated field-potentials
(Gordey et al. 2001). Finally
we show that granule cells from ventral brain slices
(Fig. 8, A and
B) display spontaneous NMDA receptor-mediated bursts
(Fig. 8C) and evoked
events (Fig. 8, D and
E, whole cell) that are 2 orders of magnitude larger
than their counterparts recorded in dorsal slices. Characterization of spontaneous events
The basic characteristics (RT 10–90%, 67% decay time, peak-currents, decay time constant, and channel conductance) have not been previously measured directly from the analysis of spontaneous NMDA receptor-mediated events. In every cell, the distributions of peak amplitudes, 10–90% rise times, and 67% decay time were best described by a sum of 2 or 3 Gaussians (Figs. 2B and 3, B and C), whereas the charge carried by the event (AOC) was adequately described by a single Gaussian (Fig. 2C).
The decay time constant of 42.1 ± 2.1 ms (n = 17,
Table 1) corresponds well to
the mean activation unit (super-cluster duration) of NMDA channels during a
synaptic activation, and we note that it is comparable to the super-cluster
duration of NR2A channels after single fast exposure (1 ms) to low glutamate
(5–100 nM, 36 ms) and to the fast component resolved in the
response to high glutamate (1 mM, 70 ms)
(Wyllie et al. 1998). Spectral
analysis of 4.096-s-long epochs during control perfusion and during perfusion
with D-CPP show that the first corner frequency of 2.9 Hz
corresponding to a decay time constant of 54.9 ms is in good agreement with
the actual observed decay time constant of 49.2 ms
(Fig. 4, A and
B). A second corner frequency in the low-millisecond
range relates to the channel kinetics of openings during super-cluster
duration (events) and also of NMDA channel noise in the baseline and is
slightly lower than the mean open time of NMDA channels (1.3 ms) that has been
previously described in cell-attached recordings
(Lieberman and Mody 1994).
The NMDA receptor subtype expressed in adult rat dentate gyrus granule
cells is predominantly NR1, NR2A, and NR2B with a lower expression of NR2D
while NR2C is absent (Dunah et al.
1996; Fritschy et al.
1998; Monyer et al. 1994;
Watanabe et al. 1993;
Wenzel et al. 1995). NR2A and
NR2B receptors are kinetically indistinguishable—they possess both high
(50–60 pS) and low (38 pS) conductance states, dependent on
temperature, extracellular Mg2+ and
Ca2+ concentrations and recording configuration (whole
cell vs. single channel/patch) (Clark et
al. 1997; Cull-Candy et al.
1998; Gibb and Colquhoun
1992; Lieberman and Mody
1994; Stern et al.
1994). NR2D receptors possess two lower conductance-levels at 42
and 19 pS (Cull-Candy et al.
1998). The application of NSFA presumes that two conditions are
met regarding the composition of the channel-type and kinetics underlying the
events. These basic assumptions derived from the binomial formulation of NSFA
(Sigworth 1980) are that the
kinetics of the N channels shaping the event are independent from each other
and these N channels have a single conductance level during openings. The
second of these conditions are not necessarily met here because NMDA channels
possess distinct subconductance levels
(Clark et al. 1997;
Cull-Candy et al. 1998;
Gibb and Colquhoun 1992;
Stern et al. 1994). The
estimate of the conductance obtained here is therefore only a weighted average
of the different conductance levels of a presumably heterogeneous NMDA
receptor population, which nevertheless remains a reasonable characterization
of the weighted mean channel conductance. Our estimate of the channel
conductance 60 pS (table
1, n = 17) is similar to that determined at room
temperature by Traynelis et al.
(1993) of 56–65 pS and
to that of Clark et al. (1997)
of 59 pS both using whole cell recordings of cerebellar granule cells, a
preparation mainly containing NR2B receptors
(Cull-Candy et al. 1998).
Considering the increased channel conductance with temperature, our estimate
recorded at 34°C appears comparatively low. The reason for the discrepancy
will have to be determined by future studies, but it may be due to the
synaptic localization of the channels. The ratio of average peak to single
channel current yields an estimate of 4 channels open at the peak of
spontaneous events (Table 1), a
small number compared with 40–50 channels open at the peak of
spontaneous GABAA receptor-mediated events
(De Koninck and Mody
1994).
Differences between evoked and spontaneous events
Stimulus evoked events had two to three times larger peak amplitudes than
spontaneous events but showed more cell-to-cell variation in peak amplitude
than spontaneous events. This is most likely due to variations in required
stimulus strength and to a variable connection between EC afferents en passage
the site of stimulation and the dendrites of the recorded cell. The higher
peak amplitude of evoked (–46.5 pA) as compared with spontaneous events
(–13.2 pA) most likely results from the activation of an additional
number of release sites by the stimulated fiber(s) on the recorded cell. The
long rise time of stimulus evoked events (peak-smoothing) may arise from the
addition of several embedded components (see also
Traynelis et al. 1993): evoked
stimuli cause vesicle release through the activation of multiple release sites
and possibly also by asynchronous release of multiple quanta from single
release sites. Both situations would lead to an asynchrony in activation of
postsynaptic receptors and thus a skew toward longer rise times. However,
staggering of spontaneous events to match rise time and peak amplitude of the
evoked event in the same cell show that when evoked from site 1, but not site
2 (different slice), kinetics of spontaneous and evoked NMDA EPSCs are
comparable (Fig. 5, A and
B). Although it is possible to closely replicate an
average evoked event from the staggered addition of average spontaneous events
(Fig. 5A), it should
not be assumed that evoked events are merely a scale-up (summation) of
spontaneous events. For example, the population of synapses giving rise to
spontaneous events may not be identical to the synapses activated by
stimulation since the perforant path contacts granule cells monosynaptically
only in the outer two thirds of the dendritic tree. Furthermore, an average
event is an average of single spontaneous events from a distribution usually
best described as being log-normal (see also Figs.
2, B and C,
and 3, B and
C). It is probably an oversimplification to assume that
events from activated synapses along the outer two-thirds of the granule cell
dendritic tree follow a similar distribution. As outlined in the preceding
text, the synapses giving rise to spontaneous and evoked events might be
anatomically distinct, and for evoked events, on average may be more distal.
Voltage escape at distant synapses and slow onset-kinetics (like NMDA
channels) can lead to serious distortion of rise time, peak current, and decay
time measured at the soma, whereas the charge recovered at the soma is much
less affected. To ascertain that the evoked events are not merely a filtered
version of the spontaneous events, we employed the voltage jump/charge
recovery method (Hausser and Roth
1997; Pearce
1993). Figure 6, A and
B, shows how the charge recovery can reconstruct the true
decay time constant of evoked events, yielding a value comparable to that of
spontaneous events. Therefore the kinetics of evoked and spontaneous events
faithfully reflect the true time course of synaptic NMDA channel
conductance.
Evoked events from stimulation in site 2
(Fig. 6B, recorded in
whole cell) have similar rise times to the events evoked from site 1 but
larger peak amplitudes (–413 ± 113 pA) and significantly longer
decay time constants compared with spontaneous events in the same cell
(Table 3,
weighted = 66.3 ± 9.4). This larger decay time constant
may be due to the activation of a set of receptors by spillover of glutamate
or perhaps stimulation of synapses in the inner molecular layer (i.e., from
secondarily activated mossy cells) not activated during spontaneous activity
or by "minimal" stimulation from the more distant site 2.
Role of NR2B receptors
Ifenprodil is a noncompetitive inhibitor of NMDA receptors with a
>400-fold selectivity for NR2B over that of NR2A
(Williams 1993) with a
mechanism of action related to the proton-mediated inhibition of NMDA
receptors. We found that ifenprodil had no effect on any measured parameter of
spontaneous events, except their frequency
(Table 2,
Fig. 7, A–F). In
field potentials and evoked whole cell currents ifenprodil caused a 40%
reduction in peak amplitude and charge carried by the event
(Table 2). The differential
effect of an NR2B antagonist on spontaneous and evoked events raises some
questions about the differential distribution of NR2A and NR2B receptors.
Ultrastructural studies employing immunogold labeled antisera recognizing
extracellular epitopes concluded that NR1 is localized on the spine of
98–99% of asymmetric profiles at Schaffer-collateral synapses in the CA1
(Racca et al. 2000; Takumi et
al.
1999a,b)
and in the dentate gyrus outer molecular layer
(Milner and Drake 2001). These
studies also reported NR1 labeling of a few presynaptic profiles in both
regions, but in the dentate, these were shown to be mainly associated with
µ-opiod receptor containing fibers establishing symmetric contacts on
granule cell dendrites (Milner and Drake
2001). However, because it has been argued that a detailed
analysis of presynaptic NR1 localization requires the use of antibodies
recognizing intracellular epitopes (Paquet
and Smith 2000; Racca et al.
2000), a cautious conclusion is that NR1 is either not expressed
or is expressed at below detection levels (using extracellular epitope
recoginizing antisera) in terminals at asymmetric synapses in the molecular
layer. Unfortunately, ultrastructural analyses of NR2 receptors are not
subtype specific, but functional studies from various synaptic preparations
including hippocampal CA3–CA1 Schaffer-collaterals supports a picture of
predominantly synaptically localized NR2A receptors with a largely
extrasynaptic spread of NR2B receptors
(Chen and Diamond 2002;
Diamond 2001;
Isaacson 1999; Kullman et al.
1999; Tovar and Westbrook;
1999) in which the timing of their activation is proportional to
their distance from the presynaptic release site
(Barbour and Hausser 1997;
Takumi et al. 1999). However, two functional studies also support a
presynaptic localization of NMDA receptors at hippocampal excitatory synapses
(Breukel et al. 1998) and more
specifically of NR2B receptors in a localized excitatory circuit in the
entorhinal cortex (Woodhall et al.
2001). The possibility exists that heterotrimeric receptors (for
example NR1/NR2A/NR2B) are expressed at synapses in adult rat dentate gyrus
granule cells because there is good evidence for such receptors in recombinant
expression systems (Brimecombe et al.
1997; Cheffings and Colquhoun
2000; Chen et al.
1997; Vicini et al.
1998). At present, evidence for the existence of heterotrimeric
receptors in vivo comes from an NR1/NR2B/ NR2D receptor combination expressed
in P0 dentate granule cells (Pina-Crispo and Gibb 2002). Although the NR2B and
NR2D receptor subunits are both considered to be predominantly extrasynaptic,
we cannot rule out the existence of a NR1/NR2A/NR2B receptor combination at
dentate granule cell synapses. This might be supported by comparing the decay
kinetics of NMDA receptors in recombinant systems with the synaptic currents
recorded by us. Although NR2A and NR2B receptors display similar channel
conductance, activation studies of expressed receptors show marked differences
in decay kinetics (Brimecombe et al.
1997; Chen et al.
1997; Vicini et al.
1998). Moreover, presumed NR1/NR2A/NR2B receptors displayed decay
kinetics and pharmacological sensitivity intermediate between those of NR2A
and NR2B receptors. Based on the kinetic model of Chen et al.
(2001), we have calculated the
time constant at 32°C of the decay of a population of NR2A ( = 16.5
ms) and NR2B ( = 122.6 ms) receptors (rate constants are as described by
Chen et al. 2001)
(Table 1, using a Q10 of 2.5).
Both values are about threefold off from the decay of the
synaptic events of 41.1 ms measured by us, and it may be that the difference
stems from a mixed population of synaptic NR1/NR2A and NR1/NR2A/NR2B
receptors. However, synaptic NMDA receptors are subject to modulation by a
number of intracellular proteins, kinases and phosphatases, cytoskeleton
anchoring proteins and Ca2+-activated modulators, each
of which have been shown to affect NMDA receptor properties. Such
intracellular regulation at the synapse is not reflected by the modeled data
or by recombinant expression systems thus making difficult the accurate
comparisons with the properties of receptors at intact synapses.
In line with the potential NMDA receptor localization and composition at
granule cell synapses, there might be several explanations for the effect of
ifenprodil: the activity of layer II stellate neurons in the EC, the neurons
that target granule cells, may be under the strict control of an NR2B receptor
controlled excitatory circuit (Woodhall et
al. 2001) leading to an ifenprodil-induced diminished activity in
the entire EC-dentate gyrus circuitry. Ifenprodil may act presynaptically on
NR2B receptors or N-type Ca2+ channels
(Bath et al. 1996) to reduce
presynaptic Ca-influx and hence release probability leading to a diminished
frequency of spontaneous events. However, the reduction of presynaptic
Ca2+ entry through voltage-gated channels by 10 µM
ifenprodil is unlikely given the lower affinity of the drug for Ca2
channels (Bath et al. 1996)
than for NR2B receptors (Williams
1993). In light of the overlapping amplitude distributions of the
spontaneous events recorded before and after ifenprodil administration
(Fig. 7B), it is
unlikely that the decreased frequency seen in ifenprodil was the result of
smaller amplitude events having been driven into the noise. Furthermore, based
on the longer deactivation times of NR2B compared with NR2A receptors, which
would result in a longer decay of the synaptic currents, the unaltered decay
times recorded under control conditions and in the presence of ifenprodil
(Fig. 7A) argue
against the presence of synaptic NR2B receptors. Based on the unaltered
kinetics of the spontaneous events in the presence of ifenprodil, glutamate
release by spontaneous action potentials is not sufficient to spillover and
activate extrasynaptic NR2B receptors, i.e., NR2B receptors do not participate
in shaping the spontaneous NMDA EPSCs at granule cell synapses. The presence
of NR2B receptors at postsynaptic densities may also be regulated, and such
receptors may not be in use during spontaneous activity. Stimulus-evoked
events and field potentials cause the near-simultaneous activation of multiple
synapses and may spillover glutamate to neighboring synapses. Surprisingly,
the decay time constant of evoked responses from sites 1 and 2 (whole cell)
was unchanged by ifenprodil (Tables
2 and
3), suggesting that the
increased glutamate release during these two types of evoked responses does
not selectively activate extrasynaptic NR2B receptors. However, ifenprodil
caused a reduction in both the peak and decay time constant of NMDA field
EPSPs, suggesting that under these conditions, extrasynaptic NR2B receptors
are activated by overspill of glutamate. The discrepancy between data in these
three different preparations presumably stems from the much higher stimulus
strengths and the proximity of the stimulating electrode required for evoking
field EPSPs rather than EPSC. We cannot exclude the possibilities that NR2B
receptor-mediated responses specifically ran down in whole cell recordings or
that an ifenprodil effect at presynaptic EC layer II afferents will likely
increase the number of failures during stimulation and thus reduce the number
of activated granule cell synapses.
Synaptic bursting in ventral slices
Spontaneous and evoked events in ventral slices with amplitudes of 2
orders of magnitude larger than mean peak in dorsal slices are NMDA receptor
mediated based on their reversal at positive holding potentials
(Fig. 8, C and
D) and APV and ifenprodil sensitivity
(Fig. 8E). Ventral
slices also displayed spontaneous field potentials, while evoked fields were
composed of double peaks dependent on an intact perforant path
(Fig. 8F). Future
studies will need to examine this phenomenon in more detail. At present, we
note that the ventral hippocampus and EC possess several characteristics, each
of which may contribute to the generation of these events. First, the presence
of supragranular mossy fibers is frequently seen in control animals in this
region (Buckmaster and Dudek
1997), which may increase excitability under these recording
conditions. Second, horizontal slices from the ventral rather than dorsal
hippocampus presumably contain more intact connections between layer two-three
EC neurons and dentate granule cells (i.e., perforant path). Yet the angle of
slicing may not be the only explanation because it is well established that
the ventral hippocampus in vivo is much more excitable than the dorsal
hippocampus, e.g., during kindling (Lerner-Natoli 1984).
Summary
We have provided the first characterization of spontaneous NMDA receptor-mediated synaptic events at granule cell synapses and demonstrated a dorsoventral difference in these events. The two stimulation sites investigated here show that evoked events from the EC/subiculum border better approximate kinetic parameters of spontaneous events than those evoked by stimulating near the hippocampal fissure. Using ifenprodil we have shown that NR2B receptors do not contribute to the spontaneous NMDA receptor-mediated EPSCs at granule cell synapses but regulate their frequency. The synaptic NMDA receptor aggregates predominantly contain NR2A with a possible contribution of NR2A/NR2B receptors. It remains to be determined whether the NR2B antagonist ifenprodil exerts its frequency-reducing effect at synapses through a reduction in the NR2B receptor controlled excitability in a local EC circuit thus reducing the output of the EC or alternatively via a presynaptic NR2B receptor on EC layer II afferents terminating in the molecular layer.
|
DISCLOSURES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
Present address of N. O. Dalby: H. Lundbeck, Ottiliavej 9, 2500 Valby, Denmark.
|
FOOTNOTES |
|---|
Address for reprint requests: I. Mody, Depts. of Neurology and Physiology, The David Geffen School of Medicine at UCLA, 710 Westwood Plaza, RNRC 3-155, Los Angeles, CA 90095-1769 (E-mail: mody{at}ucla.edu).
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
Bath CP, Farrell LN, Gilmore J, Ward MA, Hicks CA, O'Neill MJ, and Bleakman D. The effects of ifenprodil and eliprodil on voltage-dependent Ca2+ channels and in gerbil global cerebral ischaemia. Eur J Pharmacol 299: 103–12, 1996.[ISI][Medline]
Brimecombe JC,
Boeckman FA, and Aizenman E. Functional consequences of NR2 subunit
composition in single recombinant N-methyl-D-aspartate
receptors. Proc Natl Acad Sci USA
94: 11019–11024,
1997.
Breukel AI, Besselsen E, Lopes da Silva FH, and Ghijsen WE. A presynaptic N-methyl-D-aspartate autoreceptor in rat hippocampus modulating amino acid release from a cytoplasmic pool. Eur J Neurosci 10: 106–114, 1998.[ISI][Medline]
Buckmaster PS and Dudek FE. Neuron loss, granule cell axon reorganization, and functional changes in the dentate gyrus of epileptic kainate-treated rats. J Comp Neurol 385: 385–404, 1997.[ISI][Medline]
Chan SF and
Sucher NJ. An NMDA receptor signaling complex with protein phosphatase 2A.
J Neurosci 21:
7985–7992, 2001.
Cheffings CM and Colquhoun D. Single channel analysis of a novel NMDA channel from
Xenopus oocytes expressing recombinant NR1a, NR2A and NR2D subunits.
J Physiol 526:
481–491, 2000.
Chen N,
Moshaver A, and Raymond LA. Differential sensitivity of recombinant
N-methyl-D-aspartate receptor subtypes to zinc inhibition.
Mol Pharmacol 51:
1015–1023, 1997.
Chen N, Ren J,
Raymond LA, and Murphy TH. Changes in agonist concentration dependence
that are a function of duration of exposure suggest
N-methyl-D-aspartate receptor nonsaturation during
synaptic stimulation. Mol Pharmacol
59: 212–219,
2001.
Chen S and
Diamond JS. Synaptically released glutamate activates extrasynaptic NMDA
receptors on cells in the ganglion cell layer of rat retina. J
Neurosci 22:
2165–2173, 2002.
Clark BA,
Farrant M, and Cull-Candy SG. A direct comparison of the single-channel
properties of synaptic and extrasynaptic NMDA receptors. J
Neurosci 17:
107–116, 1997.
Clark BA and
Cull-Candy SG. Activity-dependent recruitment of extrasynaptic NMDA
receptor activation at an AMPA receptor-only synapse. J
Neurosci 22:
4428–4436, 2002.
Cull-Candy SG, Brickley SG, Misra C, Feldmeyer D, Momiyama A, and Farrant M. NMDA receptor diversity in the cerebellum: identification of subunits contributing to functional receptors. Neuropharmacology 37: 1369–1380, 1998.[ISI][Medline]
De Koninck Y and Mody I. Noise analysis of miniature IPSCs in adult rat brain slices:
properties and modulation of synaptic GABAA receptor channels.
J Neurophysiol 71:
1318–1335, 1994.
Diamond JS.
Neuronal glutamate transporters limit activation of NMDA receptors by
neurotransmitter spillover on CA1 pyramidal cells. J
Neurosci 21:
8328–8338, 2001.
Dunah AW, Yasuda RP, Wang YH, Luo J, Davila-Garcia M, Gbadegesin M, Vicini S, and Wolfe BB. Regional and ontogenic expression of the NMDA receptor subunit NR2D protein in rat brain using a subunit-specific antibody. J Neurochem 67: 2335–2345, 1996.[ISI][Medline]
Fritschy JM, Weinmann O, Wenzel A, and Benke D. Synapse-specific localization of NMDA and GABA(A) receptor subunits revealed by antigen-retrieval immunohistochemistry. J Comp Neurol 390: 194–210, 1998.[ISI][Medline]
Gibb AJ and
Colquhoun D. Activation of N-methyl-D-aspartate
receptors by L-glutamate in cells dissociated from adult rat
hippocampus. J Physiol 456:
143–179, 1992.
Gordey M, Mekmanee L, and Mody I. Altered effects of ethanol in NR2A(DeltaC/DeltaC) mice expressing C-terminally truncated NR2A subunit of NMDA receptor. Neuroscience 105: 987–997, 2001.[ISI][Medline]
Hardingham GE, Fukunaga Y, and Bading H. Extrasynaptic NMDARs oppose synaptic NMDARs by triggering CREB shut-off and cell death pathways. Nat Neurosci 5: 405–414, 2002.[ISI][Medline]
Hausser M and
Roth A. Estimating the time course of the excitatory synaptic conductance
in neocortical pyramidal cells using a novel voltage jump method. J
Neurosci 17:
7606–7625, 1997.
Hestrin S,
Nicoll RA, Perkel DJ, and Sah P. Analysis of excitatory synaptic action in
pyramidal cells using whole cell recording from rat hippocampal slices.
J Physiol 422:
203–225, 1990.
Isaacson JS. Glutamate spillover mediates excitatory transmission in the rat olfactory bulb. Neuron 23: 377–384, 1999.[ISI][Medline]
