Howard Hughes Medical Institute, Division of Biology 216-76, California Institute of Technology, Pasadena, California 91125
![]() |
ABSTRACT |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Dvorak-Carbone, Hannah and
Erin M. Schuman.
Long-term depression of temporoammonic-CA1 hippocampal synaptic
transmission. The temporoammonic pathway, the direct projection from layer III of the entorhinal cortex to area CA1 of the hippocampus, includes both excitatory and inhibitory components that are positioned to be an important source of modulation of the hippocampal output. However, little is known about synaptic plasticity in this pathway. We
used field recordings in hippocampal slices prepared from mature (6- to
8-wk old) rats to study long-term depression (LTD) in the temporoammonic pathway. Low-frequency (1 Hz) stimulation (LFS) for 10 min resulted in a depression of the field response that lasted for 1
h. This depression was saturable by multiple applications of LFS. LTD
induction was unaffected by the blockade of either fast
(GABAA) or slow (GABAB) inhibition.
Temporoammonic LTD was inhibited by the presence of the
N-methyl-D-aspartate (NMDA) receptor antagonist
AP5, suggesting a dependence on calcium influx. Full recovery from
depression could be induced by high-frequency (100 Hz) stimulation
(HFS); in the presence of the GABAA antagonist bicuculline,
HFS induced recovery above the original baseline level. Similarly, HFS
or
-burst stimulation (TBS) applied to naive slices caused little
potentiation, whereas HFS or TBS applied in the presence of bicuculline
resulted in significant potentiation of the temporoammonic response.
Our results show that, unlike the Schaffer collateral input to CA1, the
temporoammonic input in mature animals is easy to depress but difficult
to potentiate.
![]() |
INTRODUCTION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
The hippocampus is a brain structure that plays a
critical role in learning and memory (Eichenbaum et al.
1992; Press et al. 1989
; Squire and
Zola-Morgan 1991
; Zola-Morgan and Squire 1990
). The fundamental information processing pathway in the hippocampus is
usually considered to be the trisynaptic circuit, in which the
entorhinal cortex sends a perforant path projection to the granule
cells of the dentate gyrus, which send mossy fibers to the pyramidal
cells of the CA3 region, which send Schaffer collaterals to the CA1
pyramidal cells, which then project back to the entorhinal cortex
(Andersen et al. 1966
; Brown and Zador
1990
). These projections are all glutamatergic and excitatory
(Andersen 1975
; Misgeld 1988
). Within
each region there are local GABAergic interneurons that provide
feedforward and feedback inhibition (e.g., see Freund and
Buzsáki 1996
; Lacaille et al. 1989
;
Ribak and Seress 1983
; Woodson et al.
1989
).
However, there is more to hippocampal processing than this simple
trisynaptic loop. In particular, there is a direct projection from
entorhinal cortex to area CA1, effectively bypassing the first two
stages of the conventional circuit. This projection is referred to as
the temporoammonic pathway (Maccaferri and McBain 1995;
Ramón y Cajal 1911
) because of its origins in the
entorhinal cortex (in the temporal lobe) and its termination in CA1,
part of the cornu ammonis of the hippocampus. Unlike the perforant path, which consists of axons from stellate excitatory neurons of layer
II of the entorhinal cortex, the temporoammonic pathway consists of
axons from pyramidal cells of layer III of the entorhinal cortex
(Steward and Scoville 1976
). These temporoammonic axons terminate preferentially in the area of the distal dendrites of the
pyramidal cells, stratum lacunosum-moleculare (SLM). These axons make
asymmetric (and hence probably excitatory) synapses, >90% of which
are onto the spines of CA1 pyramidal cells (Desmond et al.
1994
). Temporoammonic axons also synapse onto the inhibitory basket and chandelier cells of CA1 (Kiss et al. 1996
)
and are likely to innervate the interneurons of SLM (Lacaille
and Schwartzkroin 1988
; Vida et al. 1998
).
There is some controversy as to whether the temporoammonic input to the
hippocampus is primarily excitatory or primarily inhibitory (see
Soltesz and Jones 1995). Field recordings in vitro
(Colbert and Levy 1992
; Doller and Weight
1982
) and in vivo (Leung et al. 1995
;
Yeckel and Berger 1990
, 1995
) reveal a population
excitatory postsynaptic potential (EPSP) in SLM after stimulation of
the temporoammonic pathway, supporting the ultrastructural evidence for
an excitatory input onto the distal dendrites of CA1 neurons. However,
intracellular recordings (Empson and Heinemann 1995
) show a mixed response, including a monosynaptic glutamatergic EPSP and
a disynaptic inhibitory postsynaptic potential with both GABAA and GABAB components.
Various forms of short- and long-term plasticity have been extensively
studied at all three synapses in the trisynaptic circuit of the
hippocampus (Bliss and Collingridge 1993). As yet,
however, relatively little is known about plasticity in the
temporoammonic pathway, whose activity can modulate plasticity in the
Schaffer collateral pathway to CA1 (Levy et al. 1998
)
and whose position in the hippocampal circuitry suggests a potent role
in the modulation of hippocampal output from CA1. A previous study
reported induction of long-term potentiation (LTP) in this pathway in
vitro only when fast inhibition was blocked by addition of the
GABAA antagonist bicuculline (Colbert and Levy
1993
); in vivo, LTP of the excitatory current sink in the
distal dendrites of area CA1 evoked by stimulation of the perforant
path (including temporoammonic axons) was reported after tetanic
stimulation (Leung et al. 1995
). In this study we report
that long-term depression (LTD) can readily be induced in the
temporoammonic pathway, in slices prepared from mature rats, by means
of low-frequency stimulation (LFS). Some of these results previously
appeared in abstract form (Dvorak and Schuman 1996
).
![]() |
METHODS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Tissue preparation
Slices were prepared from 6- to 8-wk old male Sprague-Dawley
rats. All use of animals was performed according to the guidelines of
the Caltech Institutional Animal Care and Use Committee. Rats were
decapitated after halothane anesthesia, and the brain was rapidly
removed to ice-cold, oxygenated artificial cerebrospinal fluid (ACSF)
containing (in mM) 119 NaCl, 2.5 KCl, 1.3 MgSO4, 2.4 CaCl2, 1.0 NaH2PO4, 26.2 NaHCO3, and 11.0 glucose. The dorsal surface of the
posterior half of each hemisphere was glued onto the stage of a cooled
oscillating tissue slicer (OTS-3000-04; FHC) and covered with chilled
ACSF; 500-µm slices were cut, with the optimal slices (as assessed
visually, by ease of identification of distinct layers, as well as
electrophysiologically, by the presence of robust field potentials)
generally found 4-4.5 mm below the ventral surface. The extraneous
cortical and subcortical tissue was gently dissected away with the
small end of a spatula. (A small number of early experiments were done
on slices prepared with a Stoelting tissue chopper. However, slices
prepared on the tissue slicer had larger temporoammonic responses.) The
slices were then allowed to recover in an interface chamber at room
temperature for 1 h before experiments were started. Further
microdissection was performed either in ice-cold ACSF immediately after
slice preparation or in the recording chamber before the start of the experiment. All electrophysiology was done with the slices submerged and constantly perfused with oxygenated ACSF at room temperature.
To clearly isolate the temporoammonic response, it was necessary
to further dissect the slice (see Fig.
1). The entire dentate gyrus was removed
to eliminate the possibility of activation of the trisynaptic pathway
and to prevent contamination of a field response recorded in SLM by the
much larger field elicited in dentate gyrus by concurrent activation of
the perforant path. In most experiments, including all those in which
bicuculline was used, CA3 was also removed to prevent induction of
seizure-like activity as well as to eliminate the possibility of
disynaptic activation via the perforant path projection to CA3. Also, a
cut was made through stratum radiatum (SR) in distal CA1 (near the subiculum) perpendicular to the cell body layer to prevent antidromic activation of Schaffer collaterals by the stimulating electrode in SLM
(see Fig. 1). Schaffer collaterals do not enter SLM to any appreciable
extent (Amaral and Witter 1989), so this cut cleanly isolates temporoammonic axons.
|
Electrophysiology
Bipolar tungsten electrodes, either concentric or paired
needles, were used for stimulation. One electrode was placed in SR to
stimulate the Schaffer collaterals; the other was placed in SLM to
stimulate the temporoammonic pathway (Fig. 1). Stimulus pulses were 100 µs long, monophasic, and ranged from 10 to 100 µA in the Schaffer
collateral pathway and 100 to 200 µA in the temporoammonic pathway.
Stimulus intensities were selected to produce submaximal responses with
no population spike. Field recordings were made with low-resistance
micropipettes filled with 3 M NaCl. The Schaffer collateral response
was recorded in SR, and the temporoammonic response was recorded in
SLM. Separation of the two pathways was further confirmed by the
observation of a positive-going field potential in the other layer
(Fig. 1) (Colbert and Levy 1992).
The following stimulation paradigms were used: high-frequency
stimulation (HFS) = 100 Hz for 1 s, repeated four times at 20- or
30-s intervals; -burst stimulation (TBS) = four bursts of five
pulses at 100 Hz, 200 ms between bursts, repeated four times at 20- or
30-s intervals; LFS = 1 Hz stimulation for 10 min. All stimulus
pulses were of the same length and amplitude as test pulses. Test
pulses were applied once every 20 or 30 s to each pathway. The
initial slope of the field potential after the end of the fiber volley
was measured.
Drugs were applied by dilution of concentrated stock solutions into the perfusion medium. Stock solutions were made up in water, with the exception of nifedipine, which was prepared in DMSO (×1,000) and stored protected from light. Experiments with nifedipine were performed in low light. CGP 55845A was a kind gift from Novartis (Basel, Switzerland); all other drugs were obtained from Sigma (St. Louis, MO).
Data analysis
Data were collected directly onto an IBM-compatible computer with in-house software. All numerical values are listed as means ± SE unless otherwise stated. Depression and potentiation were measured by taking an average of the initial slopes of the field EPSPs (fEPSPs) over 10-min periods immediately before and 20-30 or 50-60 min after the end of LFS, HFS, or TBS. Student's paired t-test was used to determine statistical significance for within-group comparisons; the unpaired t-test was used between groups. Results from each experimental manipulation were compared with the same control group. P values >0.05 are reported in the text as not significant (NS). Points in figures represent means ± SE across all experiments; each point is the average of data taken over 5 min. Representative traces, shown in insets, are averages of five consecutive sweeps from a representative experiment, taken 5 min before LFS, HFS, or TBS and 25 min after the end of LFS, HFS, or TBS.
![]() |
RESULTS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Temporoammonic field response is depressed by LFS
When LFS (see METHODS) was applied to the
temporoammonic pathway in normal ACSF, the response was significantly
depressed (Fig. 2A) (mean
percent of baseline at 30-40 min, 75.9 ± 3.4%, n = 26, P < 0.0001). This synaptic
depression persisted for 1 h (mean percent of baseline at 60-70 min,
73.7 ± 5.0%, n = 7, P < 0.01)
and was not accompanied by any significant changes in the size of the
presynaptic fiber volley (Fig. 2A).
|
During the application of LFS to the temporoammonic pathway, the Schaffer collateral pathway was not stimulated; during the rest of the experiment, test pulses were applied to the Schaffer collateral pathway at the same frequency as to the temporoammonic pathway. Synaptic strength in the Schaffer collateral pathway was not affected, whereas the temporoammonic response was depressed by LFS (Fig. 2A) (mean percent of baseline at 30-40 min, 95.0 ± 4.7%, n = 9, NS).
LTD has also been observed after LFS of the Schaffer collateral
pathway (Mulkey and Malenka 1992) but only in slices
prepared from younger animals (Dudek and Bear 1993
;
Wagner and Alger 1995
). When the LFS protocol was
applied to the Schaffer collateral pathway in this study, little or no
depression was seen. On average, a small but not statistically
significant depression of the response was observed (Fig.
2B) (mean percent of baseline at 30-40 min, 87.3 ± 6.2%, n = 8, NS); this trend was due only to
results from two slices that were depressed to 55 and 65% of baseline.
To determine whether LTD of the temporoammonic pathway can be saturated, the LFS protocol was applied repeatedly for 10 min every 30 min either four or five times. The response asymptotically approached a level of ~30-50% of the original baseline (Fig. 2C), reaching its maximally depressed level after three or four epochs of LFS.
GABA-mediated synaptic transmission is not required for temporoammonic LTD
Synaptic depression of the temporoammonic response could be
induced in the absence of fast GABAergic inhibition. After the GABAA antagonist bicuculline (20 µM) was added to the
perfusion solution, a slight but not significant increase in the
temporoammonic field response was generally observed (mean percent of
original response 20-25 min after bicuculline application, 105 ± 6%, n = 4, NS). The response was allowed to reach a
steady baseline for 20 min before application of LFS. LFS in the
presence of bicuculline induced depression of the field response (Fig.
3) (mean percent of baseline at 30-40
min, 75.7 ± 2.6%, n = 20, P < 0.0001) that was not significantly different from that observed in
control slices. This depression lasted
1 h (mean percent of baseline
at 60-70 min, 79.9 ± 3.2%, n = 14, P < 0.001).
|
Depression of the temporoammonic response was also possible in the absence of slow, GABAB-mediated inhibition. Addition of the GABAB antagonist CGP 55845A (1 µM) had no significant effect on the baseline field potential elicited by temporoammonic stimulation (mean percent of original response 20-25 min after CGP 55845A application, 112 ± 7%, n = 5, NS). When LFS was applied in the presence of CGP 55845A, the field response was still significantly depressed (Fig. 3) (mean percent of baseline at 30-40 min, 72.7 ± 3.5%, n = 5, P < 0.01). This amount of depression was also not significantly different from that seen in control slices.
Blockade of NMDA receptor-mediated transmission reduces temporoammonic LTD
What are the early signaling events important for establishing
temporoammonic LTD? Calcium ion is an important initiator of many
short- and long-term plasticity processes in neurons (Bliss and
Collingridge 1993; Delaney et al. 1989
;
Katz and Miledi 1968
; Neveu and Zucker
1996
). LTD induction in other hippocampal pathways has been
shown to be NMDA receptor dependent (Cummings et al. 1996
; Mulkey and Malenka 1992
; Thiels et
al. 1996
). When LFS was applied to the temporoammonic pathway
in the presence of the NMDA receptor antagonist
D,L-2-amino-5-phosphonovaleric acid (AP5; 50 µM), LTD was significantly reduced relative to control
(P < 0.05) (Fig. 4)
(mean percent of baseline at 30-40 min, 90.1 ± 5.4%,
n = 15, P < 0.05; 60-70 min,
91.2 ± 4.8%, n = 5, P < 0.05), although a small, but significant amount of residual depression was
still observed. It is worth noting that there was variability between
experiments; in about half of the experiments AP5 treatment appeared to
block LTD, whereas in the other half it had little effect. LTD was not
blocked further when higher concentrations of AP5 were used; in the
presence of 100 µM AP5, LTD was similarly reduced but not completely
blocked (data not shown) (mean percent of baseline at 30-40 min,
88.3 ± 1.5%, n = 3, NS different from depression
in 50 µM AP5).
|
In the hippocampus, some forms of heterosynaptic LTD are dependent on
L-type calcium channel activation (Christie and Abraham 1994; Wickens and Abraham 1991
). However,
homosynaptic temporoammonic LTD was not blocked by the presence of the
L-type calcium channel blocker nifedipine (20 µM) (data not shown;
mean percent of baseline at 30-40 min, 82.6 ± 1.7%,
n = 5, P < 0.05). The combination of nifedipine and AP5 produced slightly, although not significantly, greater inhibition of temporoammonic LTD than AP5 alone (Fig. 4) [mean
percent of baseline at 30-40 min, 93.4 ± 4.3% of baseline, n = 7, NS different from AP5 alone, significantly
different from baseline (P < 0.05)]. The block of
temporoammonic LTD by AP5 and nifedipine was reversible; when LFS was
applied again to slices 30 min after washout of the drugs, significant
depression was observed (data not shown; mean percent of baseline at
30-40 min, 70.7 ± 4.4%, n = 7, P < 0.0001).
Blockade of muscarinic receptors does not affect temporoammonic LTD
In addition to the temporoammonic projection, SLM of CA1 also
receives a substantial cholinergic input from the medial septum (Matthews et al. 1987). Activation of the muscarinic ACh
receptor (mAChR) is implicated in other forms of hippocampal synaptic
plasticity (Auerbach and Segal 1996
). To determine
whether mAChRs are involved in temporoammonic LTD, we applied LFS in
the presence of 1 µM atropine. Atropine itself had no significant
effect on the temporoammonic field response. In the presence of
atropine, the temporoammonic field response was depressed to the same
extent as in control ACSF (Fig. 5) (mean
percent of baseline at 30-40 min, 75.1 ± 5.0%, n = 4, NS different from control).
|
Reversal of temporoammonic LTD
To test whether LTD in the temporoammonic pathway is reversible,
HFS (see METHODS) was applied either 30 or 60 min after the end of LFS. In experiments conducted in normal ACSF, HFS induced a
significant recovery of the depressed synaptic response (Fig. 6A, ) (mean percent of
original baseline at 20-30 min, 89.4 ± 3.6%, n = 22, P < 0.05). The increase relative to the
depressed baseline was 116.3 ± 3.6% of baseline
(n = 22, P < 0.001). The reversal of
LTD was even greater when HFS was applied in the presence of 20 µM
bicuculline (Fig. 6A,
) (mean percent of original
baseline at 20-30 min, 115.3 ± 4.8%, n = 18, P < 0.05). The increase relative to the depressed
baseline was 150.1 ± 9.1% of baseline (n = 18, P < 0.0001). This difference in response to HFS after
LTD is similar to that seen when HFS was applied to naive slices.
|
Complete reversal of temporoammonic LTD was achieved by repeated application of HFS. In normal ACSF, three applications of HFS at 5-min intervals resulted in a full recovery to the original baseline response, which persisted for 1 h after the last HFS (Fig. 6B) [mean percent of original baseline at 30-40 min after first HFS, 106.8 ± 9.0%, n = 5, significantly different from depressed level of 76.4 ± 4.5% (P < 0.01), NS different from baseline; mean percent of original baseline at 60-70 min, 103.3 ± 6.4%, n = 5, significantly different from depressed level (P < 0.01), NS different from baseline].
Long-term potentiation
In several hippocampal pathways, HFS or TBS can induce long-term
potentiation (LTP) (Bliss and Collingridge 1993). When
HFS was applied to the temporoammonic pathway in naive slices, little potentiation of the field response was observed (Fig.
7A,
) (mean percent of
baseline at 20-30 min, 107.8 ± 7.5%, n = 7, NS). Furthermore, no potentiation was observed when TBS was applied
(Fig. 7B,
; mean percent of baseline at 20-30 min,
96.3 ± 3.7%, n = 5, NS). For LTP to be induced,
a certain level of postsynaptic depolarization must be reached in
response to the excitatory input; in some pathways, this requires
overcoming the inhibition that is concurrently activated (e.g., see
Steward et al. 1990
; Wigstrom and Gustafsson
1983
). The idea that fast inhibitory transmission opposes the
induction of LTP in this pathway was tested by using the
GABAA antagonist bicuculline. When HFS was applied to the
temporoammonic pathway in the presence of 20 µM bicuculline,
significant potentiation was observed (Fig. 7A,
) (mean
percent of baseline at 20-30 min, 134.2 ± 6.8%,
n = 10, P < 0.05). Delivery of TBS in
the presence of bicuculline also resulted in significant potentiation
(Fig. 7B,
) (mean percent of baseline at 20-30 min,
118.9 ± 7.9%, n = 5, P < 0.05),
as was previously observed (Colbert and Levy 1993
).
|
LTP induced by HFS in most hippocampal pathways is dependent on the
activation of NMDA receptors (Bliss and Collingridge
1993). We tested whether the potentiation of the temporoammonic
pathway seen in the presence of bicuculline required NMDA receptor
activation by applying HFS in the presence of both bicuculline (20 µM) and AP5 (50 µM). Under these conditions, HFS did not induce LTP
(Fig. 7A,
) (mean percent of baseline at 20-30 min,
104.1 ± 3.3%, n = 4, NS).
![]() |
DISCUSSION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
We examined the capacity for long-term synaptic modification of
the temporoammonic CA1 synapse in the hippocampus. LTD was consistently
induced by simple LFS of the temporoammonic pathway in slices taken
from 6- to 8-wk-old animals; the same protocol applied to the Schaffer
collateral pathway resulted in little or no depression (Fig.
2B). Other studies reported an age dependence of LTD
induction in the Schaffer collateral pathway, with little or no LTD
induced by LFS in slices from older animals (Dudek and Bear
1993; Wagner and Alger 1995
). Unlike LTD of the
Schaffer collateral response, LTD of the temporoammonic response is
robust in slices prepared from adult animals. It should be noted,
however, that LTD of the commissural input to CA1 has been shown in
adult animals in vivo, although only by means of application of a
paired-pulse LFS protocol (Thiels et al. 1994
).
Repeated application of LFS to the temporoammonic pathway resulted in
saturation of depression at a maximal level of ~40-50% of the
original baseline, similar to or lower than that seen in the Schaffer
collateral pathway (Dudek and Bear 1993; Mulkey
and Malenka 1992
). However, after cessation of LFS, the
response rebounded somewhat, suggesting that there may be a "floor"
below which the temporoammonic response can be pushed only temporarily.
This transient, larger depression may also be analogous to the
short-term potentiation seen after tetanic stimulation, which then
decays away to reveal LTP of lesser magnitude.
LTD in the temporoammonic pathway is independent of GABAA
or GABAB receptor activation. This also contrasts with
Schaffer collateral LTD, where LTD induced by LFS in slices from adult animals is enhanced in the presence of the GABAA antagonist
bicuculline, and LTD in young animals is inhibited by the
GABAB antagonist CGP 35348 (Wagner and Alger
1995). Although stimulation of the temporoammonic pathway does
clearly activate interneurons, which in turn make both
GABAA- and GABAB-mediated synapses onto CA1 pyramidal cells (Dvorak and Schuman 1997
; Empson
and Heinemann 1995
), depression of the monosynaptic, excitatory
component of this pathway is neither enhanced nor reduced by
interneuron activity. The difference between Schaffer collateral and
temporoammonic LTD in terms of the involvement of inhibition may be due
to the fact that stimulation in SLM may activate GABAergic pathways
only disynaptically, in a feedforward fashion (Empson and
Heinemann 1995
), whereas stimulation in SR to activate the
Schaffer collaterals can also directly activate axons of CA1
interneurons (e.g., see Arai et al. 1995
; Lambert
et al. 1991
). The differences in temporal patterning of
excitation and inhibition in the Schaffer collateral and temporoammonic
pathways may play a role in the differential responses to LFS shown by
these two pathways.
The induction of homosynaptic LTD in other pathways requires an
increase in intracellular Ca2+ concentration, either by
influx through NMDA receptors (Cummings et al. 1996;
Mulkey and Malenka 1992
) or voltage-gated calcium channels (Bolshakov and Siegelbaum 1994
; Christie
et al. 1997
) or by release from intracellular stores
(Reyes and Stanton 1996
). Homosynaptic LTD of the
Schaffer collateral pathway is sometimes fully blocked by the NMDA
receptor antagonist AP5 (Dudek and Bear 1992
;
Mulkey and Malenka 1992
), but in other cases only a
partial block is seen (Bolshakov and Siegelbaum 1994
;
Kemp and Bashir 1997
). Homosynaptic temporoammonic LTD
was significantly inhibited, relative to control slices, in the
presence of 50 µM AP5, although there were some individual
experiments in which AP5 did not block LTD. Similarly, in the presence
of nifedipine and AP5 together, a complete block of LTD was sometimes
observed, whereas at other times less complete inhibition was observed.
NMDA receptors are found in the distal dendrites of CA1 pyramidal
cells, although not as densely in SLM as in SR (Jacobson and
Cottrell 1993
; Jarvis et al. 1987
;
Monaghan and Cotman 1985
), and an NMDA receptor-mediated response to temporoammonic stimulation was observed physiologically (Colbert and Levy 1992
; Empson and Heinemann
1995
); furthermore, LTP of the temporoammonic pathway was fully
blocked by 50 µM AP5 (Fig. 7). It therefore seems unlikely that the
incomplete block of temporoammonic LTD by AP5 is due to an absence or
paucity of NMDA receptors in SLM. Conversely, although calcium imaging
studies show that some voltage-dependent calcium channels are clearly present in the distal dendrites of hippocampal pyramidal cells, L-type
calcium channels are most abundant close to the soma, with much lower
densities in the more distal dendrites (see Johnston et al.
1996
). This may account for the relatively small effect of
nifedipine on temporoammonic LTD. The mechanism of the induction of the
residual, non-NMDA component of temporoammonic LTD remains to be
elucidated. Other studies found a contribution of calcium release from
intracellular stores (Reyes and Stanton 1996
;
Wang et al. 1997
), activation of T-type calcium channels
(Christie et al. 1997
; Oliet et al. 1997
;
Wang et al. 1997
), and metabotropic glutamate receptor
activation (Oliet et al. 1997
; O'Mara et al. 1995
) to homosynaptic LTD at other hippocampal synapses; these calcium sources may also contribute to temporoammonic LTD.
One possible explanation for the variability between experiments in the
sensitivity of temporoammonic LTD to blockade of NMDA receptors may be
a difference between temporoammonic fibers arising in the lateral and
medial entorhinal cortex. There are distinct physiological differences
between the lateral and medial perforant path (LPP and MPP) projections
to dentate gyrus (McNaughton 1980; McNaughton and
Barnes 1977
). The two projections may also differ in their
dependence on NMDA receptor activation for LTP, at least in vivo
(Bramham et al. 1991
), although in vitro studies show no
difference in NMDA receptor dependence of LTP induction (Colino and Malenka 1993
). The LPP and MPP can be activated and
recorded independently in vivo or in vitro by virtue of the laminar
segregation of their axon terminals along the proximodistal axis of the
dendrites of the granule cells of the dentate gyrus; the LPP terminates in the outer third of the molecular layer, whereas the MPP terminates in the middle third of the molecular layer (Witter et al.
1989
). The temporoammonic projection, on the other hand, maps
along the transverse axis of CA1, with medial fibers terminating closer to CA2 and lateral fibers closer to the subiculum (Witter et al. 1989
). It should therefore in theory be possible to selectively record responses to either medial or lateral inputs by varying the
position of the electrode along the transverse axis of the hippocampal
slice. We did not systematically monitor this, but the bulk of our
experiments was performed with the recording electrode somewhere in the
middle of the CA1 area, where it would likely record inputs from both
pathways. We did perform a small number of experiments with two
recording electrodes in SLM at either end of CA1 but observed no
consistent differences in LTD expression or AP5 sensitivity
(Dvorak-Carbone and Schuman, unpublished data). However, this question
would likely best be resolved by in vivo experiments where the lateral
and medial temporoammonic fibers could more unambiguously be stimulated independently.
SLM, the terminal field of the temporoammonic pathway, receives a
strong cholinergic input from the septum (Matthews et al. 1987), and mAChR activation has been implicated in some forms of hippocampal LTD (Auerbach and Segal 1996
). Because
stimulation directly in SLM could have activated septal axons remaining
in the slice, it was important to show that the LTD observed was not
due to release of ACh from septal afferents. Temporoammonic LTD was
unaffected by application of the mAChR antagonist atropine (Fig. 5),
showing that the observed depression was not due to the activation of
cholinergic inputs.
When studying synaptic depression, it is important to show that the
effect of LFS on the synaptic response does not simply reflect damage
to the synapses or general degradation of the slice. To monitor the
health of the slice, we applied test stimuli to the Schaffer collateral
pathway alternately with the test stimuli applied to the temporoammonic
pathway and observed no change in the Schaffer collateral response
while the temporoammonic response was depressed (Fig. 2A).
The ability to reverse synaptic depression with HFS can also
counterindicate synaptic rundown or poor slice health (Dudek and
Bear 1993; Mulkey and Malenka 1992
). When we applied HFS to a depressed temporoammonic pathway, we observed a
partial recovery of the response to a level between the depressed level
and the original baseline; repeated application of HFS brought the
response back up to the original baseline, suggesting that the
depressed temporoammonic pathway had not suffered some nonspecific damage (Fig. 6). Furthermore, HFS applied to the depressed pathway in
the presence of bicuculline resulted in complete recovery and potentiation of the response above the original baseline level (Fig.
6).
The requirement for multiple applications of HFS to bring the depressed
temporoammonic response back to baseline after LTD is consistent with
the lack of LTP seen in naive slices. In agreement with a previous
report (Colbert and Levy 1993), we were able to potentiate the temporoammonic response in naive slices by HFS or TBS
only in the presence of bicuculline. Compared with the Schaffer
collateral response, the temporoammonic field response is fairly small
to begin with and might therefore require disinhibition before it can
be potentiated; indeed, it has been shown that LTP cannot be induced in
this pathway, even in the presence of bicuculline, if the initial
response is too small (Colbert and Levy 1993
). [LTP of
the temporoammonic pathway in the slice preparation in the presence of
intact inhibition has been shown only in a situation where the
temporoammonic input was stimulated so as to elicit a population spike
recorded in stratum pyramidale (Doller and Weight 1985
);
we never observed population spike activity in response to
temporoammonic stimulation.] The LTP that was induced by HFS of the
temporoammonic pathway was dependent on the activation of NMDA
receptors (Fig. 7), just like LTP in many other hippocampal pathways
(Bliss and Collingridge 1993
).
The entorhinal cortex is not the only brain region to send a projection
to SLM of area CA1. Fibers from nucleus reuniens thalami (Dolleman-Van Der Weel and Witter 1996;
Wouterlood et al. 1990
), the amygdala (Petrovich
et al. 1997
), and area TE of inferotemporal cortex
(Yukie and Iwai 1988
) also terminate in SLM, raising the possibility that activity in these areas might also serve to modulate or gate information flow through the trisynaptic circuit. It is worth
noting that extracellular stimulation electrodes placed in SLM may well
activate these fibers as well as temporoammonic axons.
The function of the temporoammonic pathway and of other inputs to SLM
is not yet well understood. In vivo studies suggest that the
temporoammonic input plays a role in the generation of oscillations
(Buzsáki et al. 1995
) and 40-Hz oscillations
(Charpak et al. 1995
); models of the hippocampus as a
heteroassociative learning network include the temporoammonic pathway
and the Schaffer collateral pathway as two distinct information-bearing
inputs to CA1 (Hasselmo and Schnell 1994
). The
inhibition activated by temporoammonic input may serve to gate the
output of the hippocampus (Dvorak and Schuman 1997
;
Empson and Heinemann 1995
). Temporoammonic activity has
also been shown to be capable of modulating the induction of LTP at the
Schaffer collateral input to CA1 (Levy et al. 1998
); it
will be interesting to examine the impact of plasticity of the
temporoammonic pathway on this heterosynaptic modulatory effect.
![]() |
ACKNOWLEDGMENTS |
---|
This work was supported by the Howard Hughes Medical Institute, including a predoctoral fellowship to H. Dvorak-Carbone.
![]() |
FOOTNOTES |
---|
Address reprint requests to E. M. Schuman.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Received 13 August 1998; accepted in final form 5 November 1998.
![]() |
REFERENCES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|