Download SHORT-TERM SYNAPTIC PLASTICITY: A COMPARISON OF TWO

REVIEWS
SHORT-TERM SYNAPTIC PLASTICITY:
A COMPARISON OF TWO SYNAPSES
Dawn M. Blitz, Kelly A. Foster and Wade G. Regehr
During physiological patterns of activity, synaptic activity is regulated by many forms of shortterm plasticity. Here, we compare the functional consequences of such plasticity at the
synapse from the climbing fibre to the Purkinje cell in the cerebellum and at the synapse
between the retinal ganglion cell and the thalamocortical relay neuron in the lateral geniculate
nucleus. Despite superficial similarities between these two powerful synapses, they have
distinctive synaptic plasticity. The climbing fibre synapse is highly reliable but accomplishes
this through many synaptic specializations. However, the retinogeniculate synapse
dynamically regulates the flow of visual information by using two types of receptor that have
different types of plasticity. These synapses illustrate the important functional consequences
of synaptic plasticity.
READILY RELEASABLE POOL
A pool of synaptic vesicles that is
available for rapid fusion with
the presynaptic membrane on
arrival of a nerve impulse. The
vesicles are docked to the
membrane and have been
biochemically primed for
release.
Department of
Neurobiology, Harvard
Medical School, 220
Longwood Avenue, Boston,
Massachusetts 02115, USA.
Correspondence to W.G.R.
e-mail: wade_regehr@
hms.harvard.edu
doi:10.1038/nrn1475
630
Although synapses throughout the brain share many
features, they also have distinct properties. In most cases
the same general sequence of events leads to neurotransmitter release: an action potential is initiated in the
axon near the cell body, it propagates down the axon,
voltage-gated calcium channels in the presynaptic
terminal open and admit calcium, and this triggers
vesicle fusion. Liberated neurotransmitter then binds to
receptors on the postsynaptic cell and ultimately this
influences the firing of the postsynaptic neuron1,2. This
complex series of events is regulated in many ways,
which makes each type of synapse distinct. Synapses
can differ in size, probability of transmitter release and
complement of postsynaptic receptors3–6. In addition,
physiological activity patterns, in which synaptic inputs
are often activated at high frequencies, sometimes in
bursts, lead to alterations in synaptic strength at all types
of synapse7–11. How the spiking pattern of a presynaptic
cell ultimately influences the firing of its targets depends
on the properties of short-term plasticity that are specific
to the activated synapses7,12–14.
Many mechanisms can lead to use-dependent
alterations in synaptic strength during high-frequency
activation15,16. At some synapses, a reduction in neurotransmitter release leads to short-term depression either
by reducing the probability of release or by depleting the
| AUGUST 2004 | VOLUME 5
of vesicles17,18 (FIG. 1a). At other
synapses, repeated activation increases the probability of
neurotransmitter release19, either by saturating a local
calcium buffer20,21 or by increasing calcium concentration
in the presynaptic terminal22–24 (FIG. 1b).
The properties of postsynaptic receptors can also
contribute to short-term plasticity. Desensitization of
postsynaptic receptors, in which exposure to neurotransmitter results in receptors entering a non-responsive
state, can reduce synaptic responses during repeated
activation9,25–28 (FIG. 1c). In addition, repetitive activity
can lead to a decrease in synaptic response amplitude
owing to receptor saturation29–31. Receptor saturation
means that fewer receptors are available for neurotransmitter to bind to on subsequent stimulation (FIG. 1d).
However, currents mediated by channels with slow
kinetics, such as NMDA (N-methyl-D-aspartate) channels, can still be large despite receptor saturation. This is
due to summation of excitatory postsynaptic currents
(EPSCs) (FIG. 1d), as occurs at the retinogeniculate
synapse (see below)9. Saturation of receptor/channel
complexes with faster kinetics, such as AMPA (α-amino3-hydroxy-5-methyl-4-isoxazole propionic acid) receptors, can have quite different consequences, as found
at the synapse between the climbing fibre and the
Purkinje cell (see below)30. This is only a partial list of
READILY RELEASABLE POOL
www.nature.com/reviews/neuro
©2004 Nature Publishing Group
REVIEWS
Similarities of two powerful synapses
Presynaptic
a Depression
b Facilitation
Postsynaptic
c Desensitization
d Saturation
Figure 1 | Presynaptic and postsynaptic mechanisms of short-term plasticity.
Schematized voltage-clamp traces illustrate the influence of two presynaptic mechanisms
(a, b) and two postsynaptic mechanisms (c, d) of plasticity on a pair of excitatory postsynaptic
currents (EPSCs). Cartoons of presynaptic boutons illustrate possible explanations for
presynaptic depression (a) and facilitation (b) and cartoons of postsynaptic spines illustrate
desensitization (c) and saturation (d). a | Top, presynaptic depression results in a smaller
second EPSC. Bottom, fewer vesicles are available for release on the second (right) stimulus
than on the first (left). b | Top, facilitation results in a larger second EPSC. Bottom, a residual
elevation in intracellular calcium (green shading), combined with the influx of calcium in
response to the second stimulus, results in enhanced release. c | Top, similar to depression,
desensitization results in a smaller second EPSC. Bottom, under prolonged exposure to
transmitter, some receptors can enter a non-responsive state (red crosses, right) and be unable
to respond to transmitter released during a second stimulus. d | Top, channels with slow
kinetics (such as NMDA (N-methyl-D-aspartate) channels) that experience saturation can
produce a large amount of current following a second stimulus, despite the smaller incremental
amplitude of the second EPSC, owing to summation with the previous EPSC. Bottom, for
receptors with high affinity for the transmitter, a population of receptors can remain bound with
transmitter (red circles) and therefore be unavailable to respond to the transmitter released in
response to a second stimulus.
the mechanisms that lead to short-term synaptic plasticity, and at most synapses multiple mechanisms are
present that interact and lead to complex responses
during realistic patterns of synaptic activation.
Much attention is focused on understanding the
mechanisms that underlie such short-term synaptic
plasticity and its functional consequences. Here, rather
than providing a comprehensive overview, we explore
the functional consequences of short-term synaptic
plasticity by comparing two synapses: the synapse
between retinal ganglion cells and thalamocortical
relay neurons (the retinogeniculate synapse) and
the synapse between climbing fibres and cerebellar
Purkinje cells. Despite superficial similarities,
use-dependent plasticity at these synapses leads to
profound functional differences during realistic
patterns of activity.
NATURE REVIEWS | NEUROSCIENCE
Climbing fibre synapses are made by neurons in the
inferior olive onto Purkinje cells in the cerebellum32.
In the adult, there is only one climbing fibre input for
each Purkinje cell, and it forms several hundred synaptic
contacts onto proximal Purkinje cell dendrites33,34
(FIG. 2a). Activation of climbing fibres releases glutamate,
which in turn activates AMPA receptors on the
Purkinje cell. The size of this synaptic response can be
quantified under voltage-clamp conditions, where the
postsynaptic cell is maintained at a given potential,
enabling synaptic currents to be measured without
interference from postsynaptic activity (FIG. 2b). The
EPSC that is elicited by a single climbing fibre is large
and has a maximal conductance of 300–500 nS. If the
response is measured in current-clamp mode and
the cell is allowed to respond naturally, climbing fibre
activation leads to a series of regenerative responses,
collectively known as a complex spike35 (FIG. 2c). So, the
climbing fibre provides a powerful and reliable signal to
Purkinje cells.
The retinogeniculate synapse is similar in many ways.
Between one and three retinal ganglion cells (RGCs)
provide a powerful synaptic drive to each thalamocortical
relay neuron in the lateral geniculate nucleus (LGN) of
the thalamus36,37. As with climbing fibres, each retinogeniculate input consists of many synaptic contacts onto
the proximal dendrites of a thalamocortical neuron38
(FIG. 2d). Voltage-clamp recordings reveal that the retinogeniculate synaptic response is also a large synaptic
current (FIG. 2e) with a maximal conductance of 10–40 nS.
In current-clamp mode, a single stimulus reliably elicits a
spike in the thalamocortical neuron (FIG. 2f). This finding
indicates that the retinogeniculate synapse can reliably
activate relay neurons in the LGN and thereby provides
a secure means of relaying visually evoked responses
from the retina to the thalamus and, ultimately, to the
visual cortex.
These anatomical features and the fact that both
synapses can reliably evoke spikes in their postsynaptic
targets indicate that they are well suited to provide
powerful and reliable synaptic drive to their postsynaptic targets. However, this view is based on
responses to stimulation at low frequencies and does
not reflect the contributions of synaptic plasticity that
could occur under higher-frequency, physiological
activity patterns. As we discuss below, differences in
presynaptic firing patterns and short-term synaptic
plasticity at these synapses result in vastly different
activation responses under physiological conditions.
The differences at these two synapses indicate
that short-term synaptic plasticity leads to distinctive
characteristics that reflect the specific roles of each
type of synapse.
The climbing fibre synapse
Reliability is important for the climbing fibre synapse
because climbing fibre activity is suggested to provide
an error signal that is important in cerebellar motor
learning. This process is thought to occur, at least in
part, in the cerebellar cortex through long-term
VOLUME 5 | AUGUST 2004 | 6 3 1
©2004 Nature Publishing Group
REVIEWS
d
a
20 µm
20 µm
b
e
1 nA
1 nA
10 ms
10 ms
c
f
0
2 ms
Vm (mV)
Vm (mV)
0
10 ms
–55
–60
Figure 2 | A comparison of the climbing fibre and retinogeniculate synapses. a | Confocal
image of a Purkinje cell labelled with the fluorescent dye Alexa 568 hydrazide, and a calcium
green dextran-labelled climbing fibre. This indicates the relationship of the climbing fibre to the
Purkinje cell dendrite. Reproduced, with permission, from REF. 129 © (2000) Elsevier Science.
b,c | A voltage-clamp recording illustrates that a single climbing fibre stimulus elicits a large inward
current (b), while a current-clamp recording shows that the same stimulus elicits a complex spike
in the Purkinje cell (c). d | A reconstruction of a thalamocortical neuron, with the locations of retinal
ganglion cell (RGC) synaptic contacts indicated by red dots. Reproduced, with permission, from
REF. 38 © (1987) John Wiley & Sons, Ltd. e,f | A trace recorded in voltage-clamp mode shows that
a single stimulus to an RGC axon in the optic tract elicits a large inward current in a postsynaptic
thalamocortical neuron (e). In current-clamp mode, the same single RGC activation elicits an
excitatory postsynaptic current that is sufficiently large to elicit an action potential (f).
PAIRED-PULSE DEPRESSION
A decrease in the amplitude of
the second of two closely timed
excitatory postsynaptic currents.
It can result presynaptically from
a decrease in the amount of
neurotransmitter released or
postsynaptically as a result of
desensitization.
632
depression (LTD) of granule cell synapses onto
Purkinje cells39–41. In addition to the excitatory input
provided by climbing fibres, each Purkinje cell
receives about 100,000 inputs from granule cells34,42,43.
Climbing fibre activation initiates a complex spike
that depolarizes the dendritic arbour of Purkinje
cells and leads to calcium entry 44,45. This widespread
electrical and calcium signal contributes to LTD of
granule cell synapses that are activated at about the
same time as the climbing fibre46,47. To elicit LTD
repeatedly, the climbing fibre must powerfully and
reliably activate Purkinje cells during patterns of
activity that are encountered in vivo.
| AUGUST 2004 | VOLUME 5
Typically, the climbing fibre fires at about 1 Hz, but it
can also fire bursts of 3 action potentials at between
5 and 15 Hz in response to sensory stimuli48–50. The
response of the Purkinje cell to this type of activity
pattern can be evaluated in cerebellar slice preparations
by recording from a Purkinje cell in current-clamp
mode. When a 15-Hz train of 3 pulses is delivered, the
Purkinje cell fires a complex spike in response to each
stimulus30 (FIG. 3a). Although there are slight differences
between these complex spikes, overall they are remarkably consistent. The consistency of the Purkinje cell
response indicates that the climbing fibre provides a
powerful input to the Purkinje cell and maintains its
efficacy during physiological activity patterns.
To understand how the climbing fibre drives the
Purkinje cell so reliably, it is useful to measure the synaptic currents that give rise to the postsynaptic response
under voltage-clamp conditions. These currents are
mediated solely by AMPA receptors, as NMDA receptors
are eliminated from the climbing fibre synapse early in
postnatal development51,52. When a similar pattern of
excitation is applied to the climbing fibre while the
Purkinje cell is held in voltage-clamp, the resulting
EPSCs undergo little depression (FIG. 3b). This indicates
that, on this timescale, short-term plasticity does
not make an important contribution to transmission.
Three main factors explain this lack of plasticity. First,
AMPA receptor desensitization, which can contribute to
depression at other synapses, is not prominent at the
climbing fibre synapse53–55. Second, although climbing
fibres show pronounced depression on a short timescale,
recovery from depression is accelerated by presynaptic
accumulation of calcium30,54. Third, multivesicular
release and receptor saturation at this synapse make
the Purkinje cell relatively insensitive to changes in the
amount of transmitter released29–31. By examining these
three aspects of transmission, we can better understand
how the climbing fibre synapse is specialized to drive the
Purkinje cell so reliably.
Desensitization. Although climbing fibre synapses seem
to have a high probability of release, which leads to
activation of a large fraction of postsynaptic AMPA
receptors, desensitization of these receptors does not
contribute to PAIRED-PULSE DEPRESSION at this synapse53–55.
This is interesting in light of the fact that receptors in
patches pulled from the soma or dendrites do undergo
desensitization in response to puffs of glutamate29,56. The
fact that desensitization of AMPA receptors contributes
little to synaptic depression might result in part from the
structure of climbing fibre synapses. Although climbing
fibres make hundreds of synaptic contacts onto Purkinje
cells, each contact and its associated postsynaptic density
on the Purkinje cell is isolated from neighbouring
synapses by BERGMANN GLIAL ensheathment57. Bergmann
glia express a high density of glutamate transporters58,59,
so they can probably terminate glutamate signals rapidly
and prevent glutamate from pooling and spilling over to
neighbouring sites. In addition, recovery from desensitization is relatively rapid29,56, which might allow recovery
from desensitization between stimuli.
www.nature.com/reviews/neuro
©2004 Nature Publishing Group
REVIEWS
a
Vm (mV)
0
50 ms
–20
–40
–60
b
1 nA
50 ms
c
1.0
EPSC2/EPSC1
Calcium-dependent recovery from depression. The
second factor that contributes to the reliability of climbing fibre synapses is that elevated presynaptic calcium
accelerates recovery from depression30,54. This is illustrated
by examining the time dependence of the depression that
occurs for pairs of EPSCs that are separated by different
time intervals. Recovery of paired-pulse depression at the
climbing fibre synapse occurs in two phases — a fast
component with a time constant of 100 ms that accounts
for 45% of recovery and a slow component with a time
constant of 2.5 s that accounts for the remaining 55%
(REF. 30)(FIG. 3c, green circles). These two phases of recovery indicate that some sites become release-competent
much faster than others. Decreasing the concentration
of extracellular calcium decreases the amplitude of
the rapid component of recovery, as does chelating
presynaptic calcium with EGTA (FIG. 3c, blue circles).
Therefore, accelerated recovery is driven by presynaptic
calcium that enters during the action potential. This
accelerated recovery helps minimize depression during
ongoing activity and might have a similar role at other
synapses60–62.
0.5
Control
0.6
EGTA
0.2
0.2
0.0
BERGMANN GLIA
Astrocytes that are located in the
cerebellum with their cell bodies
close to a Purkinje cell. They
extend radial fibres along the
dendritic tree of the Purkinje cell
and ensheath synapses made by
the climbing fibre and parallel
fibres.
Receptor saturation. The third factor that allows
the climbing fibre to evoke a consistent response in the
Purkinje cell is that the release of several vesicles at each
release site in response to an action potential provides
a glutamate signal that is sufficient to saturate the postsynaptic receptors29. This was shown using the lowaffinity AMPA receptor antagonist γ-D-glutamylglycine
(γDGG), which has exceptionally rapid kinetics. The key
observation was that γDGG reduced the EPSC to a
greater extent when the synapse was depressed after
high-frequency stimulation than it did otherwise29
(FIG. 3d). To interpret these experiments it is necessary to
consider that γDGG and glutamate compete for binding
sites on the AMPA receptor.Although synaptically evoked
glutamate is rapidly removed from the synaptic cleft
after release (within milliseconds), the fast kinetics of
γDGG allow it to compete with glutamate during this
time. The efficacy of γDGG depends on the concentration of glutamate in the cleft. The smaller the glutamate
signal, the more effectively γDGG can compete and the
larger the effect of γDGG on the EPSC. Therefore
the more effective block of AMPA receptors by γDGG
when depression is prominent is the result of a lower
concentration of glutamate in the synaptic cleft. One
possible explanation is that spillover from neighbouring
sites is prominent when the probability of release is high
and that this increases the amount of glutamate at a
given release site. However, each release site seems to be
well isolated from its neighbours by glial ensheathment57,
so this explanation is unlikely. Instead, the differential
effects of γDGG indicate that more vesicles are released
at an individual site on the first stimulus than on the
second, when the EPSC is depressed. The difference in
the amount of depression with and without γDGG also
indicates that under control conditions, when γDGG is
not present, postsynaptic AMPA receptors are saturated.
Saturation of the postsynaptic receptors renders the
Purkinje cell less sensitive to changes in the amount of
NATURE REVIEWS | NEUROSCIENCE
0.0
0
5
10
Time (s)
d
1 nA
1 nA
30 ms
30 ms
Control
DGG
Figure 3 | Recovery from depression and saturation at the
climbing fibre to Purkinje cell synapse. a | Current-clamp
recording from a Purkinje cell. The climbing fibre elicits three
similar complex spikes in the Purkinje cell when stimulated with
three pulses at 15 Hz. b | Voltage-clamp recording from a
Purkinje cell reveals that the postsynaptic currents elicited by
stimulating the climbing fibre with this pattern are similar in
amplitude. c | The time-course of recovery from depression
under control conditions (green circles) and in the presence of
EGTA to chelate intracellular calcium (blue circles). Recovery
occurs with an initial rapid phase and a second slower phase.
The inset shows these same curves on an expanded timescale.
d | Voltage-clamp recordings from a Purkinje cell showing a pair
of excitatory postsynaptic currents generated by stimulating the
climbing fibre twice at 50-ms intervals under control conditions
(left) and in the presence of γ-D-glutamylglycine to relieve AMPA
(α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid)
receptor saturation (right). Modified, with permission, from
REF. 30 © (2002) Elsevier Science.
glutamate that is released by the climbing fibre, which
reduces the effects of depression at this synapse. This is
because under physiological conditions the receptors are
overwhelmed by the amount of glutamate released,
resulting in an EPSC that is an underestimate of the
amount of neurotransmitter released. As a result,
subsequent depressed EPSCs that result from smaller
glutamate transients are closer to the size of the first
EPSC than if there were a linear relationship between
VOLUME 5 | AUGUST 2004 | 6 3 3
©2004 Nature Publishing Group
REVIEWS
the EPSC amplitude and the amount of glutamate
released. Furthermore, the rate of recovery from depression is increased because recovery to the maximum
EPSC amplitude occurs before complete recovery of
the glutamate transient. Overall, receptor saturation
minimizes the effects of decreased transmitter release on
the postsynaptic cell.
Complex synaptic mechanisms. Many forms of synaptic
plasticity work together at the climbing fibre synapse to
ensure a roughly constant synaptic response, even though
it seems that some depression is unavoidable at synapses
where the probability of release is high. High-frequency
activity that tends to depress transmission is counteracted
by accelerated recovery from depression that is driven by
a buildup of presynaptic calcium concentrations. Multivesicular release and receptor saturation also minimize
the effects. These mechanisms combine to ensure that the
postsynaptic conductance is relatively constant even in
the face of reduced neurotransmitter release. So, the initial apparent simplicity of the climbing fibre synapse is
misleading. The ability to maintain a roughly constant
synaptic response under physiological conditions reflects
the interaction of many synaptic specializations.
The retinogeniculate synapse
Despite similarities in the ability of a single stimulus to
elicit a postsynaptic spike at the climbing fibre and the
retinogeniculate synapses, there are important differences
between these synapses that are consistent with specializations that are suited to different functional roles.
Specifically, the function of the retinogeniculate synapse is
to transfer visual information contained in the firing patterns of RGCs to thalamocortical neurons in the LGN.
Although the large amplitude of the RGC-elicited excitatory postsynaptic potentials (EPSPs) and their ability
to reliably evoke firing in thalamocortical neurons in
response to low-frequency activation indicate that the
retinogeniculate synapse simply relays RGC activity to
thalamocortical neurons, this is not the case.
The resting potential of thalamocortical neurons can
greatly influence the transfer of visual signals across the
retinogeniculate synapse. At hyperpolarized membrane
potentials, thalamocortical neurons are in burst mode,
and excitatory inputs activate low-threshold calcium
channels that can elicit bursts of action potentials63–66. At
more depolarized potentials, the low-threshold calcium
channels are inactivated, and thalamocortical neurons
are in a tonic response mode64,67,68. Although visual
stimuli can elicit responses in both tonic and burst
mode, neurons in the awake animal are more commonly in tonic mode69–71. Here we focus on how synaptic
plasticity influences retinogeniculate transmission when
thalamocortical neurons are in tonic mode.
Paired extracellular recordings of RGCs and thalamocortical neurons during visual stimulation show that
even at apparently strong synaptic connections, an individual RGC spike does not always elicit a postsynaptic
action potential37,72,73. This is illustrated in simultaneous
extracellular recordings of a RGC and a thalamocortical
neuron (FIG. 4a). In the small interval shown from such a
634
| AUGUST 2004 | VOLUME 5
recording in FIG. 4a, there are fourteen RGC spikes and
nine thalamocortical neuron spikes. Seven of the RGC
spikes (yellow triangles) trigger no response, five of the
RGC spikes (blue triangles) trigger single short-latency,
precisely timed spikes, and two of the RGC spikes
(black triangles) trigger a burst of two spikes each
(FIG. 4a). The appearance of multiple spikes per presynaptic action potential indicates that this synapse can
amplify incoming information from the retina, which
may be important for activating cortical neurons73,74.
The prominence of bursts of spikes in thalamocortical
neurons is often taken to be a hallmark of a cell in burst
mode. However, as we will explore below, the properties
of the retinogeniculate synapse can also lead to bursts
of action potentials when thalamocortical neurons are
in tonic mode.
Besides variability in the number of spikes elicited,
there is also variability in the timing of elicited spikes
relative to the presynaptic spikes. In addition to shortlatency, precisely timed spikes, thalamocortical neurons
can also have delayed responses, indicating that the
synapse can alter the timing of visual information that
is relayed to the cortex37,73,75. Furthermore, in vivo, the
timing between RGC spikes has been shown to influence their efficacy in eliciting thalamocortical neuron
spikes73,75–77. Unlike the physiological activity patterns
of climbing fibres, RGCs can fire over a large range of
frequencies with a great deal of variety in their patterns
of activity78–82. A dependence of efficacy on RGC spike
timing could relay additional information about the
patterns of RGC activity. So, in vivo observations indicate
that this synapse does not simply relay visual signals but
instead processes them and helps to control the visual
information that reaches the cortex.
AMPA and NMDA currents. Unlike the climbing
fibre-to-Purkinje cell synapse, transmission at the retinogeniculate synapse is mediated by both AMPA and
NMDA receptors83–87. To understand how properties of
the retinogeniculate synapse contribute to the complex
transformation of RGC spikes into thalamocortical neuron activity, it is necessary to consider both the AMPA
and NMDA components. These two receptor types
mediate currents with different properties88–90. Currents
elicited by AMPA receptors at the retinogeniculate
synapse have a linear current–voltage relationship and
rapid kinetics91,92. However, currents through NMDA
receptors have a nonlinear current–voltage relationship
because, at hyperpolarized membrane potentials, the
channel is blocked by magnesium92–95. NMDA currents
also have slower kinetics than AMPA currents96,97.
These properties make NMDA currents suited to their
established role in several forms of long-term synaptic
plasticity98–100. NMDA receptors are also important in
eliciting postsynaptic firing92,101–104. In vivo experiments
using receptor antagonists in the LGN showed that both
AMPA and NMDA currents contribute to visually
evoked responses in thalamocortical neurons83,84,86,105
(FIG. 4b). Although NMDA channels often require the
postsynaptic cell to be depolarized by AMPA receptors
before they can be activated sufficiently to elicit spikes,
www.nature.com/reviews/neuro
©2004 Nature Publishing Group
REVIEWS
c Stimulus
a
NMDA
0.1 s
AMPA
Control
Control
Spikes s–1
300
0
300
500 pA
0
1.0
0
300
Spikes s–1
Spikes s–1
1.0
Time (s)
No plasticity
30
30
0
0
0
0
0
–55
–55
1.0
Recovery
300
Depression
only
30
1.0
0
1.0
Recovery
Desens +
depression
d
g (nS)
Spikes s–1
Spikes s–1
d-APV
300
0
100 ms
1.0
CNQX
Vm (mV)
Spikes s–1
b
300
0
Time (s)
0
–55
100 ms
Figure 4 | AMPA and NMDA receptor contributions to retinogeniculate transmission. a | Thalamocortical neuron (bottom)
and retinal ganglion cell (RGC; top) spike trains in vivo illustrate that an individual RGC action potential can elicit zero (white triangles),
one (blue triangles) or multiple spikes (black triangles) in a thalamocortical neuron. Data courtesy of W. M. Usrey, J. B. Reppas and
R. C. Reid, Harvard Medical School. b | In vivo the lateral geniculate nucleus (LGN) neuron responses to visual stimuli are
represented as peri-stimulus time histograms under control conditions (top), in the presence of the AMPA and NMDA receptor
antagonists CNQX (6-cyano-7- nitroquinoxaline-2,3-dione) (middle, left) and d-APV (D(–)-2-amino-5-phosphonovaleric acid)
(middle, right), and during recovery from drug application (bottom). Lines below graphs indicate time of light presentation. CNQX
and d-APV were applied by iontophoresis into the LGN. Modified, with permission, from REF. 86 © (1991) American Physiological
Society. c | Voltage-clamp traces illustrate isolated NMDA (N-methyl-D-aspartate, middle) and AMPA (α-amino-3-hydroxy-5-methyl4-isoxazole propionic acid, bottom) currents in response to RGC axon stimulation with the stimulus train shown above recordings.
Isolated NMDA and AMPA currents were recorded in the presence of 5 µM NBQX (6-nitro-7-sulphamoylbenzo(f)quinoxaline-2,3dione, an AMPA receptor antagonist) and 5 µM CPP (3-+-(2-carboxypiperazin–4-yl)propyl-1 phosphonic acid, an NMDA receptor
antagonist), respectively. Each trace is the average of five trials. Modified, with permission, from REF. 9 © (2002) Elsevier Science.
d | Thalamocortical neuron responses (bottom) to dynamic-clamp AMPA conductance waveforms (top) recorded in a slice
preparation are shown. Conductance waveforms with the influence of desensitization and depression present (left), with the
influence of only depression present (middle), and with no plasticity present were injected (right). Modified, with permission, from
REF. 95 © (2003) American Physiological Society.
NMDA currents alone elicit some action potentials in
response to visual stimuli (FIG. 4b). This is probably due to
the current–voltage relationship of NMDA currents at the
retinogeniculate synapse. When thalamocortical neurons
are in tonic mode, despite the magnesium block of
NMDA channels, NMDA current is elicited by activation
of the retinogeniculate synapse in the absence of any
depolarization93,95. Blockade of AMPA receptors reduces
the initial component of the responses more than blockade of NMDA receptors (FIG. 4b). However, NMDA receptor blockade has more influence on the later portion of
the response, where AMPA receptor blockade has little
effect. These data show that AMPA receptors make a
small contribution late in a train of RGC activity 86. This is
surprising given the large amplitude of individual AMPA
currents in response to a single stimulus (FIG. 2e).
An examination of EPSCs that are evoked by physiological patterns of activity helps to explain the observed
contributions of NMDA receptors and AMPA receptors
to visually evoked responses in thalamocortical neurons.
The retinogeniculate synapse has been studied in a
slice preparation that includes the optic tract, making it
possible to activate single RGC inputs. The properties of
NATURE REVIEWS | NEUROSCIENCE
the retinogeniculate synapse can be studied without
interference of active conductances by studying the
response under voltage-clamp, and without the complications of inhibitory synapses by including GABA
(γ-aminobutyric acid) receptor antagonists. In response
to a physiological stimulation pattern, AMPA currents in
thalamocortical neurons are highly depressed during
higher frequency stimuli, but recover substantially
during long interstimulus intervals9 (FIG. 4c). The small
AMPA receptor contribution during sustained highfrequency activity reflects activity-dependent plasticity
and helps to explain why AMPA receptors apparently
become less effective during sustained visual stimulation
(FIG. 4b). The response of the NMDA component is quite
different. Although NMDA receptor EPSCs are also
depressed during trains, the much longer duration of
NMDA currents causes summation and, as a result, a
large NMDA current remains despite depression9 (FIG. 4c).
This indicates that the properties of the NMDA currents
can overcome the influence of short-term plasticity and
enable NMDA receptors to continue to mediate visually
evoked responses in thalamocortical neurons during
sustained visual stimulation (FIG. 4b).
VOLUME 5 | AUGUST 2004 | 6 3 5
©2004 Nature Publishing Group
REVIEWS
DYNAMIC-CLAMP
A technique to introduce
artificial synaptic or voltagegated conductance into a
neuron. The time course, voltage
dependence and reversal
potential are measured under
voltage-clamp conditions and
are used to determine the
appropriate current to be
injected to mimic the synaptic
conductance by the dynamicclamp technique, in currentclamp recording mode.
636
AMPA receptor desensitization contributes prominently to the use-dependent plasticity of the AMPA
component9. The probability of release at the retinogeniculate synapse has not been quantified to the
degree that it has at the climbing fibre synapse. Based
on the amount of presynaptic depression, it is likely to
be approximately 0.3 (REF. 9), which is much lower than
at the climbing fibre synapse. However, anatomical
specializations probably account for the contributions
of AMPA receptor desensitization to use-dependent
plasticity at this synapse. Unlike at the climbing fibre
synapse, at the retinogeniculate synapse there are
many release sites in close proximity without glial
separation106. This arrangement also occurs at other
synapses at which desensitization contributes to
depression, including auditory nerve fibre synapses
onto nucleus magnocellularis neurons, and cerebellar
mossy fibre synapses onto granule cells27,28. Many
release sites in close proximity allow glutamate pooling, and enable glutamate from one release site to bind
to and cause desensitization of AMPA receptors at
a neighbouring release site. As a result, glutamate
release at this neighbouring site will be less effective
if it occurs before these receptors have recovered from
desensitization (FIG. 1c). Because desensitization recovers within around 100 ms (REFS 9,27,28), this form
of plasticity attenuates AMPA currents at high firing
frequencies.
The anatomy of the retinogeniculate synapse has
different consequences for NMDA currents during
high-frequency activity. Because NMDA receptors
have a high affinity for glutamate, spillover of glutamate
that can occur with many closely spaced release sites
can effectively activate a particularly large fraction
of NMDA receptors and lead to their partial saturation.
As a result of the slow kinetics of NMDA receptors, glutamate will still be bound to many receptors if a second
stimulus occurs a short time later (FIG. 1d). This partial
saturation of NMDA receptors results in fewer receptors
being available to mediate additional current, as
indicated in FIG. 4c by the smaller incremental currents
during a train of RGC action potentials9. However, the
slow kinetics of NMDA receptors lead to a large total
NMDA conductance during high-frequency activity.
This differs from AMPA currents, which are minimal
during high-frequency activity owing to desensitization9
(FIG. 4c). These differences in postsynaptic plasticity
of the two components probably contribute to the differences in their roles during the early and late portions
of visual stimulation86 (FIG. 4b).
Functional consequences of plasticity. To understand
how plasticity at this synapse contributes to shaping
thalamocortical neuron responses, it is useful to allow
the neurons to respond naturally in current-clamp
mode and to manipulate the components of plasticity
individually. However, pharmacological isolation of
different forms of short-term plasticity in current-clamp
is impractical. Presynaptic depression cannot be manipulated without altering other neuronal properties.
Similarly, pharmacological manipulation of AMPA
| AUGUST 2004 | VOLUME 5
receptor desensitization has prominent secondary
effects, such as changing the time course of synaptic
currents, which can alter thalamocortical neuron
responses independently of plasticity owing to desensitization9,10,27. Therefore, a different approach must
be used.
The DYNAMIC-CLAMP technique provides a powerful
means of manipulating individual aspects of shortterm plasticity and examining their influence on
neuronal responses107. Dynamic-clamp allows the
researcher to mimic synaptic inputs and record
the neuronal responses. The amplitude and kinetics
of the synaptic conductances can be controlled and
manipulated, and properties such as the nonlinear
current–voltage relationship of NMDA receptors can
be incorporated. Furthermore, mechanisms of synaptic
plasticity can be included and manipulated individually95,107. Voltage-clamp analysis of the short-term
plasticity of AMPA and NMDA currents at the retinogeniculate synapse provided the information that was
necessary for successful application of the dynamicclamp approach9,95.
Dynamic-clamp studies of the AMPA component
of the retinogeniculate synapse revealed that depression and desensitization severely limit the efficacy of
AMPA currents during a train of RGC input. The large
amplitude of an initial AMPA excitatory postsynaptic
conductance (EPSG) is always effective at eliciting an
action potential in a thalamocortical neuron. However,
subsequent AMPA EPSGs are ineffective when both
depression and desensitization shape EPSG amplitudes
(FIG. 4d, left). In the absence of desensitization, more
AMPA EPSGs are effective at eliciting spikes (FIG. 4d,
middle). In the absence of desensitization and depression, AMPA EPSGs are effective throughout a stimulus
train (FIG. 4d, right). So, activity-dependent plasticity
severely limits the efficacy of the AMPA component
at the retinogeniculate synapse during sustained RGC
activity95. Consequently, AMPA currents can make
a much greater contribution to the postsynaptic
response after quiescent periods than during sustained
high-frequency activity. These data explain in vivo
findings that AMPA receptors are not very effective at
conveying the late component of a visually evoked
response (FIG. 4b). By contrast, the summation of NMDA
currents allows them to make greater contributions
during periods when the AMPA component contributes
little to the response86.
Distinctions between AMPA and NMDA components also result in different contributions of AMPA
receptors and NMDA receptors to the responses of
thalamocortical neurons to individual presynaptic
action potentials. Pharmacological experiments in vitro
reveal that the AMPA component elicits one shortlatency, precisely timed action potential for each RGC
spike. However, the NMDA component elicits action
potentials with longer, more variable latencies and can
elicit multiple action potentials per presynaptic spike95.
The properties of the AMPA and NMDA components,
therefore, explain why thalamocortical neuron responses
are not simple relays of RGC input.
www.nature.com/reviews/neuro
©2004 Nature Publishing Group
REVIEWS
a
Cell 1
Cell 2
Cell 3
Cell 4
Cell 5
Cell 6
Cell 7
0
500
1,000
1,500
2,000
2,500
Time (ms)
b
Vm (mV)
0
–55
0
0.6
0
Time (s)
c
A20, N10
0
0.6
0.6
Time (s)
Time (s)
A40, N10
A30, N40
Vm (mV)
0
–55
Cell 1
Cell 2
Cell 3
Cell 4
Cell 5
Cell 6
0
0.6
0
Time (s)
0.6
Time (s)
0
0.6
Time (s)
Figure 5 | AMPA and NMDA component amplitudes contribute to variability in response
spike number and timing between different thalamocortical neurons. a | The responses of
seven thalamocortical neurons to multiple presentations of the same visual stimulus in vivo are
shown as raster plots. Visual stimulus onset is at time zero and is ongoing for the duration
illustrated. Modified, with permission, from REF. 108 © (2002) Society for Neuroscience. b |
Responses of three thalamocortical neurons in a slice preparation to retinal ganglion cell axon
stimulation are shown. Stimulus pattern is indicated above the traces. Raster plots of 2–3 trials
per cell are plotted below. c | Responses of a single thalamocortical neuron to injection of AMPA
and NMDA conductances with dynamic-clamp are shown. AMPA (α-amino-3-hydroxy-5-methyl4-isoxazole propionic acid, A) and NMDA (N-methyl-D-aspartate, N) conductance amplitudes are
indicated above traces. Raster plots of four trials each for six neurons for each conductance
combination are plotted below. Parts b and c modified, with permission, from REF. 95 © (2003)
American Physiological Society.
NATURE REVIEWS | NEUROSCIENCE
Response reproducibility and variability. The responses
of individual thalamocortical neurons in vivo to
repeated presentations of a visual stimulus are highly
reproducible108 (FIG. 5a). In contrast, spike number and
timing vary between thalamocortical neurons108 (FIG. 5a).
These characteristics are also found in the slice preparation, even when all other influences, such as cortical feedback and inhibitory circuitry, are eliminated. A train of
stimuli delivered to an RGC input in a slice preparation
elicits reproducible responses during repeated trials in a
single thalamocortical neuron and different responses
in different thalamocortical neurons (FIG. 5b). These
responses vary from failures during the response (FIG. 5b,
left), to a faithful representation of the pattern
(FIG. 5b, middle) or a pattern that includes additional
action potentials95 (FIG. 5b, right). The similarities
between in vivo findings and in vitro findings when
thalamocortical neurons are isolated from other influences indicates that differences in responses between
thalamocortical neurons are due to properties of the
retinogeniculate synapse.
Differences in the magnitude of AMPA and NMDA
currents between synapses contribute to differences
in thalamocortical neuron responses95. The range of
responses that are elicited by synaptic inputs is replicated
by injecting different combinations of artificial AMPA
and NMDA conductance amplitudes using dynamicclamp. Responses vary from containing failures (FIG. 5c,
left), to being faithful representations of the presynaptic
pattern (FIG. 5c, middle), to containing additional spikes
(FIG. 5c, right), in a manner that is similar to visually
evoked thalamocortical neuron responses observed
in vivo. The response of an individual thalamocortical
neuron can be readily switched between these responses
by altering the combination of AMPA and NMDA conductances that is injected (FIG. 5c, top). For example,
a larger NMDA component can amplify RGC input by
eliciting multiple postsynaptic spikes per presynaptic
spike95 (FIG. 5c, right). This indicates that RGC input can
be amplified in thalamocortical neurons even when they
are not in burst mode. However, in tonic mode the
bursts are mediated by NMDA currents whereas in
burst mode they are mediated by low-threshold calcium
channels66,109. Such differences in responses owing to the
combination of AMPA and NMDA conductances are
probably important for how visual information is passed
onto cortical neurons, as these neurons respond more
strongly to the second of two closely spaced stimuli110–112.
These experiments show that a single thalamocortical
neuron can generate various responses simply by altering
the AMPA and NMDA conductances. So, although other
factors such as intrinsic properties of thalamocortical
neurons and local circuitry might contribute to differences in responses between thalamocortical neurons,
these data support the idea that differences in responses
between thalamocortical neurons are a consequence of
the properties of the retinogeniculate synapse.
The properties of the retinogeniculate synapse result
in a dynamic transfer of visual information. The plasticity
of AMPA currents severely limits their efficacy during
sustained high-frequency activity, and the long duration
VOLUME 5 | AUGUST 2004 | 6 3 7
©2004 Nature Publishing Group
REVIEWS
of the NMDA currents enhances their contributions
during sustained activity. This can change the nature of
thalamocortical neuron responses from precisely timed,
short-latency action potentials after a quiescent period
to longer-latency bursts of multiple action potentials per
presynaptic spike during sustained activity. Furthermore, the relative balance of AMPA and NDMA
currents and their maximal conductances can result in
different responses to the same RGC activity pattern.
The transfer ratio at this synapse can range from low to
high depending on the contributions of these two
synaptic currents. Consequently, how RGC input is
transformed into thalamocortical neuron firing will
have important implications for the visual information
that is transferred to cortical neurons.
General implications
Although the climbing fibre and retinogeniculate
synapses are both powerful synapses, they have many
specializations that result in functional differences. The
climbing fibre synapse uses multiple mechanisms to
produce a reliable response to repetitive activation that
is stereotyped in different Purkinje cells. The plasticity of
the retinogeniculate synapse, however, transforms the
presynaptic activity pattern into a postsynaptic response
that is dynamically regulated throughout a train and that
can endow different retinogeniculate synapses with a
range of properties. These two synapses illustrate the
importance of synaptic plasticity and synaptic specializations in determining the effect of a cell on the firing
of its targets.
Short-term plasticity also has important functional
consequences at other synapses12–14,113–115. Synapses with
a high probability of release, such as the climbing fibre
and retinogeniculate synapses, tend to undergo shortterm depression, which can have important behavioural
consequences. For example, depression at thalamocortical synapses in barrel cortex underlies the behavioural
adaptation to whisker deflections115. In the visual cortex,
synaptic depression at synapses between layer 4 and
layer 2/3 equalizes the postsynaptic response to changes
in the firing frequency of rapidly and slowly firing
inputs12. Depression is also important in the crustacean
pyloric network. Here, the extent of depression at a
synapse between a circuit neuron and the pacemaker
neurons controls the frequency of oscillation of the
pacemaker neurons and, therefore, of the rhythm that is
generated by this circuit113. On the other hand, the calyx
of Held might be more similar to the climbing fibre in
1.
2.
3.
4.
5.
638
Kandel, E. R., Schwartz, J. H. & Jessell, T. M. Principles of
Neural Science (Elsevier, New York, 2000).
Nicholls, J. G., Martin, A. R. & Wallace, B. G. From Neuron
to Brain: A Cellular and Molecular Approach to the Function
of the Nervous System (Sinauer Associates, Sunderland,
1992).
Cowan, W. M., Sudhof, T. C. & Stevens, C. F. Synapses
(Johns Hopkins Univ. Press, Baltimore, Maryland, 2001).
Craig, A. M. & Boudin, H. Molecular heterogeneity of central
synapses: afferent and target regulation. Nature Neurosci. 4,
569–578 (2001).
Xu-Friedman, M. A. & Regehr, W. G. Structural contributions
to short-term synaptic plasticity. Physiol. Rev. 84, 69–85
(2004).
6.
7.
that it reliably evokes postsynaptic action potentials at
physiological firing frequencies in spite of depression in
the EPSC amplitude at these frequencies116.
It can be difficult to determine the functional consequences of short-term plasticity at many other synapses.
In contrast to the synapses discussed in this review,
synapses are often small and influence firing only by
acting with other synaptic inputs and by interacting
with intrinsic conductances in the postsynaptic
cell117–119. In hippocampal CA1 pyramidal neurons, for
example, EPSPs are amplified by their interaction with
intrinsic currents that can lead to the generation of
plateau potentials. The influence of EPSPs on postsynaptic firing therefore depends on the properties of
the intrinsic currents that are activated115, and the consequences of short-term plasticity will be complicated
by how it affects the activation of these currents.
In considering how such small synaptic inputs
activate their targets, it is often necessary to consider the
spatial location of inputs on the dendritic arbour as well
as voltage-gated channels120–122. In some neurons, the
expression of hyperpolarization (Ih) channels differs
along the dendrite and results in differences in the
impact of EPSPs depending on where on the dendrite
they are initiated123,124. In addition, the location of
voltage-gated sodium currents can have important
implications for how short-term plasticity shapes
postsynaptic firing. Dendritic sodium channels can
enable synaptic inputs to activate dendritic action
potentials125,126. A region of high sodium channel
expression that requires little synaptic current to reach
threshold for a dendritic action potential could decrease
the influence of synaptic depression at those sites in a
manner similar to the climbing fibre synapse. Moreover,
to consider the interactions of synaptic inputs requires
precise knowledge of the timing of the inputs127,128.
Despite barriers to the study of many synapses, it is
clear that synaptic plasticity is of general importance in
shaping how neurons influence their targets. However, as
is evident from the examples discussed here, the influence
of synaptic plasticity can be distinct at different synapses,
even ones that have some synaptic properties in common.
A detailed understanding of the functional consequences
of short-term plasticity therefore depends on understanding many features of pre- and postsynaptic neurons.
The complement of plasticity mechanisms and the
synaptic specializations of different synapses will be
important in determining the functional consequences of
short-term plasticity at each synapse.
Atwood, H. L. & Karunanithi, S. Diversification of synaptic
strength: presynaptic elements. Nature Rev. Neurosci. 3,
497–516 (2002).
Fuhrmann, G., Segev, I., Markram, H. & Tsodyks, M. Coding
of temporal information by activity-dependent synapses.
J. Neurophysiol. 87, 140–148 (2002).
This modelling study used experimentally determined
parameters of depressing and facilitating cortical
synapses to examine the dependence of
postsynaptic firing on the history of presynaptic
firing. Depressing and facilitating synapses produce
maximum information in their postsynaptic
responses at different presynaptic firing frequency
ranges.
| AUGUST 2004 | VOLUME 5
8.
9.
Kreitzer, A. & Regehr, W. Modulation of transmission during
trains at a cerebellar synapse. J. Neurosci. 20, 1348–1357
(2000).
Chen, C., Blitz, D. M. & Regehr, W. G. Contributions of
receptor desensitization and saturation to plasticity at
the retinogeniculate synapse. Neuron 33, 779–788
(2002).
This voltage-clamp analysis revealed different shortterm plasticity of the NMDA and AMPA components
of the EPSC. During presynaptic bursts, pronounced
AMPA receptor desensitization reduced the efficacy
of the AMPA component, whereas slow kinetics and
receptor saturation accentuated the efficacy of the
NMDA component during such bursts.
www.nature.com/reviews/neuro
©2004 Nature Publishing Group
REVIEWS
10. Brenowitz, S. & Trussell, L. O. Minimizing synaptic
depression by control of release probability. J. Neurosci. 21,
1857–1867 (2001).
11. Varela, J. A. et al. A quantitative description of short-term
plasticity at excitatory synapses in layer 2/3 of rat primary
visual cortex. J. Neurosci. 17, 7926–7940 (1997).
12. Abbott, L. F., Sen, K., Varela, J. A. & Nelson, S. B. Synaptic
depression and cortical gain control. Science 275, 220–222
(1997).
13. Dittman, J. S., Kreitzer, A. C. & Regehr, W. G. Interplay
between facilitation, depression, and residual calcium at
three presynaptic terminals. J. Neurosci. 20, 1374–1385
(2000).
14. Markram, H., Lubke, J., Frotscher, M. & Sakmann, B.
Regulation of synaptic efficacy by coincidence of postsynaptic
APs and EPSPs. Science 275, 213–215 (1997).
15. Zucker, R. S. & Regehr, W. G. Short-term synaptic plasticity.
Annu. Rev. Physiol. 64, 355–405 (2002).
16. von Gersdorff, H. & Borst, J. G. Short-term plasticity at the
calyx of held. Nature Rev. Neurosci. 3, 53–64 (2002).
17. Eccles, J. C., Katz, B. & Kuffler, S. W. Nature of the
‘endplate potential’ in curarized muscle. J. Physiol. (Lond.)
124, 574–585 (1941).
18. Feng, T. P. Studies on the neuromuscular junction.
Chin. J. Physiol. 16, 341–372 (1941).
19. Katz, B. & Miledi, R. The role of calcium in neuromuscular
facilitation. J. Physiol. (Lond.) 195, 481–492 (1968).
20. Felmy, F., Neher, E. & Schneggenburger, R. Probing the
intracellular calcium sensitivity of transmitter release during
synaptic facilitation. Neuron 37, 801–811 (2003).
There was no change in the calcium sensitivity of
transmitter release during facilitation. Along with
reference 21, this study provides compelling evidence
for local buffer saturation as a mechanism underlying
facilitation at some synapses.
21. Blatow, M., Caputi, A., Burnashev, N., Monyer, H. &
Rozov, A. Ca2+ buffer saturation underlies paired pulse
facilitation in calbindin-D28k-containing terminals. Neuron
38, 79–88 (2003).
This study and reference 20 established that
facilitation can result from saturation of endogenous
buffers. A particularly compelling observation was
that the properties of facilitation were altered
markedly in mice in which the calcium-binding protein
calbindin-D28k was knocked out.
22. Sippy, T., Cruz-Martin, A., Jeromin, A. & Schweizer, F. E.
Acute changes in short-term plasticity at synapses with
elevated levels of neuronal calcium sensor-1. Nature
Neurosci. 6, 1031–1038 (2003).
This study indicates that at some synapses the highaffinity calcium sensor NCS-1 is responsible for
facilitation. This study, along with references 23 and
24, suggests that multiple mechanisms can lead to
facilitation.
23. Atluri, P. P. & Regehr, W. G. Determinants of the time course
of facilitation at the granule cell to Purkinje cell synapse.
J. Neurosci. 16, 5661–5671 (1996).
24. Kamiya, H. & Zucker, R. S. Residual Ca2+ and short-term
synaptic plasticity. Nature 371, 603–606 (1994).
25. Jones, M. V. & Westbrook, G. L. The impact of receptor
desensitization on fast synaptic transmission. Trends
Neurosci. 19, 96–101 (1996).
26. Sun, Y. et al. Mechanism of glutamate receptor
desensitization. Nature 417, 245–253 (2002).
27. Trussell, L. O., Zhang, S. & Raman, I. M. Desensitization of
AMPA receptors upon multiquantal neurotransmitter
release. Neuron 10, 1185–1196 (1993).
28. Xu-Friedman, M. A. & Regehr, W. G. Ultrastructural
contributions to desensitization at cerebellar mossy fiber to
granule cell synapses. J. Neurosci. 23, 2182–2192 (2003).
29. Wadiche, J. I. & Jahr, C. E. Multivesicular release at climbing
fiber-Purkinje cell synapses. Neuron 32, 301–313 (2001).
A low-affinity AMPA receptor antagonist was used to
show, for the first time, that multivesicular release and
postsynaptic receptor saturation take place at
individual release sites at the climbing fibre synapse.
30. Foster, K. A., Kreitzer, A. C. & Regehr, W. G. Interaction of
postsynaptic receptor saturation with presynaptic
mechanisms produces a reliable synapse. Neuron 36,
1115–1126 (2002).
This study showed that postsynaptic receptor
saturation, multivesicular release and calciumdependent recovery from depression cause the
climbing fibre to evoke a reliable response in
the Purkinje cell under physiological conditions.
31. Harrison, J. & Jahr, C. E. Receptor occupancy limits
synaptic depression at climbing fiber synapses. J. Neurosci.
23, 377–383 (2003).
32. Eccles, J. C., Llinas, R. & Sasaki, K. The excitatory synaptic
action of climbing fibers on the Purkinje cells of the
cerebellum. J. Physiol. (Lond.) 182, 268–296 (1966).
33. Ramón y Cajal, S. Histologie du Systeme Nerveux de
l’Homme et des Vertebres (Maloine, Paris, 1911).
34. Palay, S. L. & Chan-Palay, V. Cerebellar Cortex (Springer,
New York, 1974).
35. Eccles, J., Llinas, R. & Sasaki, K. Excitation of cerebellar
Purkinje cells by the climbing fibers. Nature 203, 245–246
(1964).
36. Chen, C. & Regehr, W. G. Developmental remodeling of the
retinogeniculate synapse. Neuron 28, 955–966 (2000).
37. Usrey, W. M., Reppas, J. B. & Reid, R. C. Specificity and
strength of retinogeniculate connections. J. Neurophysiol.
82, 3527–3540 (1999).
38. Hamos, J. E., Van Horn, S. C., Raczkowski, D. & Sherman,
S. M. Synaptic circuits involving an individual
retinogeniculate axon in the cat. J. Comp. Neurol. 259,
165–192 (1987).
39. Ito, M. Cerebellar long-term depression: characterization,
signal transduction, and functional roles. Physiol. Rev. 81,
1143–1195 (2001).
40. Carey, M. & Lisberger, S. Embarrassed, but not depressed:
eye opening lessons for cerebellar learning. Neuron 35,
223–226 (2002).
41. Bloedel, J. R. & Bracha, V. Current concepts of climbing
fiber function. Anat. Rec. 253, 118–126 (1998).
42. Napper, R. M. A. & Harvey, R. J. Number of parallel fiber
synapses on an individual purkinje cell in the cerbellum of
the rat. J. Comp. Neurol. 274, 168–177 (1988).
43. Napper, R. M. A. & Harvey, R. J. Quantitative study of the
purkinje cell dendritic spines in the rat cerebellum. J. Comp.
Neurol. 274, 158–167 (1988).
44. Llinas, R. & Sugimori, M. Electrophysiological properties of
in vitro purkinje cell somata in mammalian cerebellar slices.
J. Physiol. (Lond.) 305, 171–195 (1980).
45. Llinas, R. & Sugimori, M. Electrophysiological properties of
in vitro Purkinje cell dendrites in mammalian cerebellar
slices. J. Physiol. (Lond.) 305, 197–213 (1980).
46. Ito, M., Sakurai, M. & Tongroach, P. Climbing fibre induced
depression of both mossy fibre responsiveness and
glutamate sensitivity of cerebellar Purkinje cells. J. Physiol.
(Lond.) 324, 113–134 (1982).
47. Konnerth, A., Dreessen, J. & Augustine, G. J. Brief dendritic
calcium signals initiate long-lasting synaptic depression in
cerebellar Purkinje cells. Proc. Natl Acad. Sci. USA 89,
7051–7055 (1992).
48. Armstrong, D. M. & Rawson, J. A. Activity patterns of
cerebellar cortical neurones and climbing fibre afferents in
the awake cat. J. Physiol. (Lond.) 289, 425–448 (1979).
49. Schwarz, C. & Welsh, J. P. Dynamic modulation of mossy
fiber system throughput by inferior olive synchrony: a
multielectrode study of cerebellar cortex activated by motor
cortex. J. Neurophysiol. 86, 2489–2504 (2001).
50. Bloedel, J. R. & Ebner, T. J. Rhythmic discharge of climbing
fibre afferents in response to natural peripheral stimuli in the
cat. J. Physiol. (Lond.) 352, 129–146 (1984).
51. Llano, I., Marty, A., Armstrong, C. M. & Konnerth, A.
Synaptic- and agonist-induced excitatory currents of
Purkinje cells in rat cerebellar slices. J. Physiol. (Lond.)
434, 183–213 (1991).
52. Farrant, M. & Cull-Candy, S. G. Excitatory amino acid
receptor-channels in purkinje cells in thin cerebellar slices.
Proc. R. Soc. Lond. B 244, 179–184 (1991).
53. Hashimoto, K. & Kano, M. Presynaptic origin of paired-pulse
depression at climbing fiber-Purkinje cell synapses in the rat
cerebellum. J. Physiol. (Lond.) 506, 391–405 (1998).
54. Dittman, J. S. & Regehr, W. G. Calcium dependence and
recovery kinetics of presynaptic depression at the climbing
fiber to Purkinje cell synapse. J. Neurosci. 18, 6147–6162
(1998).
55. Dzubay, J. A. & Jahr, C. E. The concentration of synaptically
released glutamate outside of the climbing fiber-Purkinje cell
synaptic cleft. J. Neurosci. 19, 5265–5274 (1999).
56. Hausser, M. & Roth, A. Dendritic and somatic glutamate
receptor channels in rat cerebellar Purkinje cells. J. Physiol.
(Lond.) 501, 77–95 (1997).
57. Xu-Friedman, M. A., Harris, K. M. & Regehr, W. G. Threedimensional comparison of ultrastructural characteristics at
depressing and facilitating synapses onto cerebellar Purkinje
cells. J. Neurosci. 21, 6666–6672 (2001).
58. Chaudhry, F. A. et al. Glutamate transporters in glial plasma
membranes: highly differentiated localizations revealed by
quantitative ultrastructural immunocytochemistry. Neuron
15, 711–720 (1995).
59. Rothstein, J. D. et al. Localization of neuronal and glial
glutamate transporters. Neuron 13, 713–725 (1994).
60. Stevens, C. F. & Wesseling, J. F. Activity-dependent
modulation of the rate at which synaptic vesicles become
available to undergo exocytosis. Neuron 21, 415–424
(1998).
61. Wang, L.-Y. & Kaczmarek, L. K. High-frequency firing helps
replenish the readily releasable pool of synaptic vesicles.
Nature 394, 384–388 (1998).
NATURE REVIEWS | NEUROSCIENCE
62. Sakaba, T. & Neher, E. Calmodulin mediates rapid
recruitment of fast-releasing synaptic vesicles at a calyx-type
synapse. Neuron 32, 1119–1131 (2001).
Previous studies had shown that elevations of
presynaptic calcium accelerated recovery from
depression. This study shows that this effect is
mediated by calmodulin at the calyx of Held.
63. Lu, S. M., Guido, W. & Sherman, S. M. Effects of membrane
voltage on receptive field properties of lateral geniculate
neurons in the cat: contributions of the low-threshold Ca2+
conductance. J. Neurophysiol. 68, 2185–2198 (1992).
64. McCormick, D. A. Cellular mechanisms underlying
cholinergic and noradrenergic modulation of neuronal firing
mode in the cat and guinea pig dorsal lateral geniculate
nucleus. J. Neurosci. 12, 278–289 (1992).
65. Steriade, M., Jones, E. G. & McCormick, D. A. Thalamus
(Elsevier Science, Oxford, 1997).
66. Sherman, S. M. Dual response modes in lateral geniculate
neurons: mechanisms and functions. Vis. Neurosci. 13,
205–213 (1996).
67. Sherman, S. M. Tonic and burst firing: dual modes of
thalamocortical relay. Trends Neurosci. 24, 122–126 (2001).
68. Lu, S. M., Guido, W. & Sherman, S. M. The brain-stem
parabrachial region controls mode of response to visual
stimulation of neurons in the cat’s lateral geniculate nucleus.
Vis. Neurosci. 10, 631–642 (1993).
69. Guido, W. & Weyand, T. Burst responses in thalamic relay
cells of the awake behaving cat. J. Neurophysiol. 74,
1782–1786 (1995).
70. Weyand, T. G., Boudreaux, M. & Guido, W. Burst and tonic
response modes in thalamic neurons during sleep and
wakefulness. J. Neurophysiol. 85, 1107–1118 (2001).
71. Ramcharan, E. J., Gnadt, J. W. & Sherman, S. M. Burst and
tonic firing in thalamic cells of unanesthetized, behaving
monkeys. Vis. Neurosci. 17, 55–62 (2000).
72. Cleland, B. G. & Lee, B. B. A comparison of visual
responses of cat lateral geniculate nucleus neurones with
those of ganglion cells afferent to them. J. Physiol. (Lond.)
369, 249–268 (1985).
73. Usrey, W. M. Spike timing and visual processing in the
retinogeniculocortical pathway. Phil. Trans. R. Soc. Lond. B
357, 1729–1737 (2002).
74. Swadlow, H. A. & Gusev, A. G. The impact of ‘bursting’
thalamic impulses at a neocortical synapse. Nature
Neurosci. 4, 402–408 (2001).
75. Mastronarde, D. N. Two classes of single-input X-cells in cat
lateral geniculate nucleus. II. Retinal inputs and the
generation of receptive-field properties. J. Neurophysiol. 57,
381–413 (1987).
76. Usrey, W. M., Reppas, J. B. & Reid, R. C. Paired-spike
interactions and synaptic efficacy of retinal inputs to the
thalamus. Nature 395, 384–387 (1998).
Paired RGC and thalamocortical neuron recordings
in vivo showed the dependence of postsynaptic
spikes on the timing of RGC spikes. This study also
established that presynaptic spike timing influenced
the synchronous firing of thalamocortical neurons.
77. Rowe, M. H. & Fischer, Q. Dynamic properties of retinogeniculate synapses in the cat. Vis. Neurosci. 18, 219–231
(2001).
78. Rodieck, R. W. & Stone, J. Response of cat retinal ganglion
cells to moving visual patterns. J. Neurophysiol. 28,
819–832 (1965).
79. Kara, P., Reinagel, P. & Reid, R. C. Low response variability
in simultaneously recorded retinal, thalamic, and cortical
neurons. Neuron 27, 635–646 (2000).
80. Meister, M. & Berry, M. J. 2nd. The neural code of the retina.
Neuron 22, 435–450 (1999).
81. Kuffler, S. W. Discharge patterns and functional
organization of mammalian retina. J. Neurophysiol. 16,
37–68 (1953).
82. Barlow, H. B. Summation and inhibition in the frog’s retina.
J. Physiol. (Lond.) 119, 69–88 (1953).
83. Turner, J. P., Leresche, N., Guyon, A., Soltesz, I. & Crunelli, V.
Sensory input and burst firing output of rat and cat
thalamocortical cells: the role of NMDA and non-NMDA
receptors. J. Physiol. (Lond.) 480, 281–295 (1994).
84. Funke, K., Eysel, U. T. & FitzGibbon, T. Retinogeniculate
transmission by NMDA and non-NMDA receptors in the cat.
Brain Res. 547, 229–238 (1991).
85. Sillito, A. M., Murphy, P. C., Salt, T. E. & Moody, C. I.
Dependence of retinogeniculate transmission in cat on
NMDA receptors. J. Neurophysiol. 63, 347–355 (1990).
86. Kwon, Y. H., Esguerra, M. & Sur, M. NMDA and non-NMDA
receptors mediate visual responses of neurons in the cat’s
lateral geniculate nucleus. J. Neurophysiol. 66, 414–428
(1991).
87. Hohnke, C. D., Oray, S. & Sur, M. Activity-dependent
patterning of retinogeniculate axons proceeds with a
constant contribution from AMPA and NMDA receptors.
J. Neurosci. 20, 8051–8060 (2000).
VOLUME 5 | AUGUST 2004 | 6 3 9
©2004 Nature Publishing Group
REVIEWS
88. Zorumski, C. F. & Thio, L. L. Properties of vertebrate
glutamate receptors: calcium mobilization and
desensitization. Prog. Neurobiol. 39, 295–336 (1992).
89. Mayer, M. L. & Westbrook, G. L. The physiology of
excitatory amino acids in the vertebrate central nervous
system. Prog. Neurobiol. 28, 197–276 (1987).
90. Monaghan, D. T., Bridges, R. J. & Cotman, C. W. The
excitatory amino acid receptors: their classes,
pharmacology, and distinct properties in the function of the
central nervous system. Annu. Rev. Pharmacol. Toxicol. 29,
365–402 (1989).
91. Harata, N., Katayama, J. & Akaike, N. Excitatory amino acid
responses in relay neurons of the rat lateral geniculate
nucleus. Neuroscience 89, 109–125 (1999).
92. Esguerra, M., Kwon, Y. H. & Sur, M. Retinogeniculate
EPSPs recorded intracellularly in the ferret lateral geniculate
nucleus in vitro: role of NMDA receptors. Vis. Neurosci. 8,
545–555 (1992).
93. Ramoa, A. S. & McCormick, D. A. Enhanced activation of
NMDA receptor responses at the immature retinogeniculate
synapse. J. Neurosci. 14, 2098–2105 (1994).
94. Mayer, M. L. & Westbrook, G. L. Permeation and block of
N-methyl-D-aspartic acid receptor channels by divalent
cations in mouse cultured central neurones. J. Physiol.
(Lond.) 394, 501–527 (1987).
95. Blitz, D. M. & Regehr, W. G. Retinogeniculate synaptic
properties controlling spike number and timing in relay
neurons. J. Neurophysiol. 90, 2438–2450 (2003).
This follow up to reference 9 used current- and
dynamic-clamp recordings to study the effects of
single RGC inputs on the firing of thalamocortical
neurons in brain slice. The AMPA component of the
EPSC elicited precisely timed spikes whereas the
NMDA component allowed a single presynaptic spike
to evoke brief bursts of postsynaptic spikes.
96. Jonas, P. & Spruston, N. Mechanisms shaping glutamatemediated excitatory postsynaptic currents in the CNS. Curr.
Opin. Neurobiol. 4, 366–372 (1994).
97. Sprengel, R. & Seeburg, P. H. The unique properties of
glutamate receptor channels. FEBS Lett. 325, 90–94 (1993).
98. Collingridge, G. L. & Bliss, T. V. Memories of NMDA
receptors and LTP. Trends Neurosci. 18, 54–56 (1995).
99. Collingridge, G. L. & Bliss, T. V. NMDA receptors: their role
in long-term potentiation. Trends Neurosci. 10, 288–293
(1987).
100. Bear, M. F. Progress in understanding NMDA-receptordependent synaptic plasticity in the visual cortex. J. Physiol.
(Paris) 90, 223–227 (1996).
101. Armstrong-James, M., Welker, E. & Callahan, C. A. The
contribution of NMDA and non-NMDA receptors to fast and
slow transmission of sensory information in the rat SI barrel
cortex. J. Neurosci. 13, 2149–2160 (1993).
102. Binns, K. E. & Salt, T. E. Excitatory amino acid receptors
participate in synaptic transmission of visual responses in
640
103.
104.
105.
106.
107.
108.
109.
110.
111.
112.
113.
114.
115.
116.
117.
the superficial layers of the cat superior colliculus.
Eur. J. Neurosci. 6, 161–169 (1994).
Daw, N. W., Stein, P. S. G. & Fox, K. The role of NMDA
receptors in information processing. Annu. Rev. Neurosci.
16, 207–222 (1993).
Scharfman, H. E., Lu, S. M., Guido, W., Adams, P. R. &
Sherman, S. M. N-methyl-D-aspartate receptors contribute
to excitatory postsynaptic potentials of cat lateral geniculate
neurons recorded in thalamic slices. Proc. Natl Acad. Sci.
USA 87, 4548–4552 (1990).
Sillito, A. M., Murphy, P. C. & Salt, T. E. The contribution of
the non-N-methyl-D-aspartate group of excitatory amino
acid receptors to retinogeniculate transmission in the cat.
Neuroscience 34, 273–280 (1990).
Famiglietti, E. V. Jr & Peters, A. The synaptic glomerulus and
the intrinsic neuron in the dorsal lateral geniculate nucleus of
the cat. J. Comp. Neurol. 144, 285–334 (1972).
Prinz, A. A., Abbott, L. F. & Marder, E. The dynamic clamp
comes of age. Trends Neurosci. 27, 218–224 (2004).
Reinagel, P. & Reid, R. C. Precise firing events are
conserved across neurons. J. Neurosci. 22, 6837–6841
(2002).
McCormick, D. A. & Feeser, H. R. Functional implications of
burst firing and single spike activity in lateral geniculate relay
neurons. Neuroscience 39, 103–113 (1990).
Usrey, W. M., Alonso, J. M. & Reid, R. C. Synaptic
interactions between thalamic inputs to simple cells in cat
visual cortex. J. Neurosci. 20, 5461–5467 (2000).
Usrey, W. M. The role of spike timing for thalamocortical
processing. Curr. Opin. Neurobiol. 12, 411–417 (2002).
Kara, P. & Reid, R. C. Efficacy of retinal spikes in driving
cortical responses. J. Neurosci. 23, 8547–8557 (2003).
Nadim, F., Manor, Y., Kopell, N. & Marder, E. Synaptic
depression creates a switch that controls the frequency of
an oscillatory circuit. Proc. Natl Acad. Sci. USA 96,
8206–8211 (1999).
Fortune, E. S. & Rose, G. J. Roles for short-term synaptic
plasticity in behavior. J. Physiol. (Paris) 96, 539–545 (2002).
Chung, S., Li, X. & Nelson, S. B. Short-term depression at
thalamocortical synapses contributes to rapid adaptation of
cortical sensory responses in vivo. Neuron 34, 437–446
(2002).
Whole-cell patch-clamp recordings in vivo provided a
convincing demonstration that short-term depression
of thalamocortical synapses contributes to the
adaptation of cortical neurons in response to whisker
deflections.
Taschenberger, H. & von Gersdorff, H. Fine-tuning an
auditory synapse for speed and fidelity: developmental
changes in presynaptic waveform, EPSC kinetics, and
synaptic plasticity. J. Neurosci. 20, 9162–9173 (2000).
Axmacher, N. & Miles, R. Intrinsic cellular currents and the
temporal precision of EPSP-action potential coupling in CA1
pyramidal cells. J. Physiol. (Lond.) 555, 713–725 (2004).
| AUGUST 2004 | VOLUME 5
118. Fricker, D. & Miles, R. EPSP amplification and the precision
of spike timing in hippocampal neurons. Neuron 28,
559–569 (2000).
119. Magee, J. C. Dendritic integration of excitatory synaptic
input. Nature Rev. Neurosci. 1, 181–190 (2000).
120. Magee, J. C. & Johnston, D. Synaptic activation of voltagegated channels in the dendrites of hippocampal pyramidal
neurons. Science 268, 301–304 (1995).
121. Williams, S. R. & Stuart, G. J. Role of dendritic synapse
location in the control of action potential output. Trends
Neurosci. 26, 147–154 (2003).
122. Takagi, H. Roles of ion channels in EPSP integration at
neuronal dendrites. Neurosci. Res. 37, 167–171 (2000).
123. Magee, J. C. Dendritic hyperpolarization-activated currents
modify the integrative properties of hippocampal CA1
pyramidal neurons. J. Neurosci. 18, 7613–7624 (1998).
124. Williams, S. R. & Stuart, G. J. Voltage- and site-dependent
control of the somatic impact of dendritic IPSPs.
J. Neurosci. 23, 7358–7367 (2003).
125. Hanson, J. E., Smith, Y. & Jaeger, D. Sodium channels and
dendritic spike initiation at excitatory synapses in globus
pallidus neurons. J. Neurosci. 24, 329–340 (2004).
126. Golding, N. L. & Spruston, N. Dendritic sodium spikes
are variable triggers of axonal action potentials in
hippocampal CA1 pyramidal neurons. Neuron 21,
1189–1200 (1998).
127. Fuentealba, P., Crochet, S., Timofeev, I. & Steriade, M.
Synaptic interactions between thalamic and cortical inputs
onto cortical neurons in vivo. J. Neurophysiol. 91,
1990–1998 (2004).
128. Carter, A. G. & Regehr, W. G. Quantal events shape
cerebellar interneuron firing. Nature Neurosci. 5, 1309–1318
(2002).
129. Kreitzer, A. C., Gee, K. R., Archer, E. A. & Regehr, W. G.
Monitoring presynaptic calcium dynamics in projection fibers
by in vivo loading of a novel calcium indicator. Neuron 27,
25–32 (2000).
Acknowledgements
D.M.B. and K.A.F. contributed equally in the preparation of this review.
Competing interests statement
The authors declare no competing financial interests.
Online links
FURTHER INFORMATION
Encyclopedia of Life Sciences: http://www.els.net/
synaptic plasticity: short term | AMPA receptors | long-term
depression and depotentiation
Regehr’s laboratory homepage:
http://neuro.med.harvard.edu/site/faculty/regehr.html
Access to this interactive links box is free online.
www.nature.com/reviews/neuro
©2004 Nature Publishing Group