| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
-Frequency Excitatory Input to Granule Cells Facilitates Dendrodendritic Inhibition in the Rat Olfactory Bulb
Department of Neurosciences, Case Western Reserve University, Cleveland, Ohio 44106
Submitted 6 March 2003; accepted in final form 30 March 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
-frequency stimulation of glutamatergic axons in the granule
cell layer. Long-range excitatory axon connections from mitral cells
innervated by different subpopulations of olfactory receptor neurons may
provide a gating input to granule cells, thereby facilitating the mitral cell
lateral inhibition that contributes to odorant encoding. |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
; Laurent
1999; Macrides and Chorover
1972; Mori et al.
1999). In vivo intracellular recordings have demonstrated that
distinct inhibitory and excitatory synaptic inputs underlie the complex mitral
cell firing patterns elicited by natural sensory stimuli
(Hamilton and Kauer 1989;
Wellis et al. 1989).
Perturbation of inhibitory function in the mammalian olfactory bulb
(Nusser et al. 2001), or the
equivalent region in the honey bee brain
(Stopfer et al. 1997),
disrupts the temporal patterning of principal cell action potentials as well
as the ability of the animal to discriminate different odors.
Inhibitory synaptic inputs to mitral cells are mediated by a complex
collection of axo- and dendrodendritic microcircuits. The dominant form of
inhibition onto mitral cells is from GABAergic granule cells that mediate
recurrent and lateral inhibition (Ezeh et
al. 1993; Shepherd and Greer
1998). These unusual interneurons lack an axon and instead release
their neurotransmitter from dendritic spines at specialized reciprocal
synaptic contacts with the secondary dendrites of mitral cells
(Rall et al. 1966;
Shepherd and Greer 1998). The
inhibitory function of this dendrodendritic microcircuit was proposed by Rall
et al. (1966) on the basis of
extracellular recordings and demonstrated in the isolated turtle olfactory
bulb by Jahr and Nicoll (1980,
1982) using intracellular
methods. Dendrodendritic inhibition was subsequently demonstrated in the
mammalian olfactory bulb in vitro by several groups
(Chen et al. 2000;
Isaacson and Strowbridge 1998;
Schoppa et al. 1998).
Inhibitory granule cells have functional
N-methyl-D-aspartate (NMDA) and non-NMDA receptors
(Wellis and Kauer 1994), which
can be colocalized at individual dendrodendritic synapses
(Sassoe-Pognetto and Ottersen
2000). However, recent studies using acute slices of rodent
olfactory bulb have shown a critical role for NMDA receptors during
dendrodendritic inhibition while AMPA receptor blockade has generally little
effect on dendrodendritic inhibition (DDI) in Mg2+-free
conditions (Chen et al. 2000;
Halabisky et al. 2000;
Isaacson and Strowbridge 1998;
Schoppa et al. 1998). This
result sets the olfactory bulb apart from most other CNS regions in which AMPA
receptors play a dominant role in activating local inhibitory circuits. In
granule cells, Ca2+ influx through both
voltage-dependent Ca2+ channels (VDCC)
(Isaacson 2001;
Isaacson and Strowbridge 1998)
and NMDA receptors (Chen et al.
2000; Halabisky et al.
2000) can trigger GABA release, suggesting that the involvement of
NMDA receptors in DDI may reflect this additional source of presynaptic
Ca2+ influx. Alternatively, the prolonged time course of
NMDA receptor-mediated postsynaptic potentials may preferentially activate
presynaptic voltage-gated Ca2+ channels
(Schoppa and Westbrook
1999).
Under physiological conditions, the NMDA receptors that control DDI in
olfactory bulb slices are tonically blocked by Mg2+
(Isaacson and Strowbridge
1998). However, granule cells receive other types of excitatory
input onto their proximal dendrites (Kishi
et al. 1984; Orona et al.
1984). One potential function of these proximal glutamatergic
synapses may be to depolarize granule cell dendrites sufficiently to unblock
gemmular NMDA receptors and thereby modulate DDI. In the present study, we
demonstrate that brief tetanic stimulation in the granule cell body layer
facilitates recurrent inhibition onto mitral cells evoked by single action
potentials in the presence of physiological levels of
Mg2+ ions. The tetanic input itself produces a modest
feedforward inhibitory response in mitral cells that is blocked by AMPA
receptor antagonists. This unusual form of gated recurrent inhibition has
functional effects on mitral cells and can modulate the response of mitral
cells to simulated excitatory postsynaptic potentials (EPSPs).
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
resistance) typically contained (in mM) 140 K-methylsulfate, 8 NaCl, 10 HEPES,
0.2 EGTA, 4 MgATP, 0.3 Na3GTP, and 10 phosphocreatine. In some
experiments, 20 mM
bis-(o-aminophenoxy)-N,N,N',N'-tetraacetic
acid (BAPTA) was added to the internal solution to increase the intracellular
calcium-buffering capacity. The Na+-channel blocker lidocaine
N-ethyl bromide (QX-314, 5 mM) was added to the internal solution in
some granule-cell recordings to block the generation of fast action potentials
(Fig. 3D). Patch
electrodes used for voltage-clamp recordings
(Fig. 6) contained (in mM) 115
CsCl, 25 TEA·Cl, 5 QX-314, 0.2 EGTA, 4 MgATP, 0.3 Na3GTP,
and 10 phosphocreatine.
|
|
|
Voltage and current records were low-pass filtered at 2 kHz and then
digitized at 5 kHz using a 16-bit A/D converter (ITC-18, Instrutech). Input
conductance was determined by recording the voltage response to a 100-ms
duration current pulse. The amplitude of the current step was adjusted at the
beginning of the experiment to generate a 10- to 15-mV hyperpolarization.
Resting input conductance was estimated by dividing the current step amplitude
by the average hyperpolarization during the last 60 ms of the current pulse.
The change in input conductance following a single action potential was
determined by recording the responses to two 100-ms test pulses— one
pulse preceding the action potential and one pulse that was initiated 20 ms
after the action potential. We estimated the magnitude of recurrent inhibitory
postsynaptic responses by determining the
GInPost/GInPre conductance ratio;
increases in this ratio represent a spike-evoked increase in input
conductance. While underestimating the true change in input conductance—
because the underlying inhibitory postsynaptic current (IPSC) decays during
the test pulse and the test pulses are therefore likely to outlast the
inhibitory postsynaptic potential (IPSP)-related conductance change—this
measure has the advantage of being relatively insensitive to changes in the
resting membrane potential (see Fig.
1E). Inhibitory responses were quantified using the input
conductance ratio, with the resting input conductance determined before each
stimulus to control for slow changes in the intrinsic properties of the mitral
cells during the experiment and after drug applications. Because our estimates
of input conductance are based on relatively long-duration current steps, we
cannot determine whether the changes in the input conductance we observe
reflected alterations in the amplitude and/or kinetics of the evoked
inhibitory responses. (The long duration of the current steps used in this
study was reflected the large membrane time constant of mitral cells. Greater
temporal resolution would be possible by using hybrid patch-clamp amplifier
and voltage-clamp steps to assay input conductance.) The functional effect of
recurrent IPSPs also were examined by iteratively determining the current
needed to trigger an action potential using repeated 5-ms current steps
(Fig. 7). The current step
amplitude was varied by 20-pA increments until an action potential was
triggered by a depolarizing step. Current threshold ratios were calculated by
dividing the injected current needed to trigger an action potential when a
granule cell layer (GCL) stimulus and/or conditioning action potential was
evoked by the injected current without any conditioning treatments. In some
experiments, a current waveform consisting of a train of six temporally
overlapping EPSPs was injected into mitral cells to produce a stereotyped
firing pattern (Fig. 7). Each
simulated EPSP in the train, which was modeled after changes in
intracellular-free Ca2+ in mitral cells imaged in rats
in response to olfactory stimulation in vivo
(Charpak et al. 2001), was
generated using a single alpha function with a decay time constant of 100 ms.
The relative amplitude of each simulated EPSP in the train was adjusted
manually to mimic previously reported mitral cell intracellular responses
during sensory stimulation.
|
Sharpened tungsten microelectrodes (FHC) connected to a battery-operated stimulus isolation unit (A360, WPI) were used for extracellular stimulation. Receptor antagonists were applied by switching the bath perfusion solution. Glutamate receptor agonists were applied by focal pressure application using a picospritzer II (General Valve). Electrophysiological data were recorded and analyzed using custom software written in Visual Basic (Microsoft) and Origin 6.1 (Microcal). In most figures, action potentials were truncated to show responses to the test pulses used to measure the conductance ratio. Membrane potentials indicated are not corrected for the liquid junction potential. All chemicals were obtained from Sigma (St. Louis, MO). Data are shown as means ± SE. Significance was performed using paired t-test or ANOVA.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
;
Friedman and Strowbridge 2000)
as well as by extracellular Mg2+
(Fig. 1B1).
Magnesium-sensitive afterpotentials reverse polarity near the chloride
equilibrium potential (Fig.
1B2), suggesting that they are mediated by
GABAA receptors (Bormann et al.
1987). We quantified the functional effect of GABAA
receptor activation in mitral cells by measuring the change in input
conductance immediately after a single action potential (see
METHODS; Fig.
1C). Using this method to assay recurrent DDI, we found
that inhibitory afterpotentials following single action potentials are
associated with an approximately twofold increase in the input conductance
ratio (mean Gin-Post/Gin-Pre = 2.1
± 0.10; range: 1.0–4.2; n = 85;
Fig. 1D1). This
spike-evoked increase in input conductance ratio was blocked by extracellular
Mg2+ (mean conductance ratio = 1.1 ± 0.01; range:
0.9–1.3; n = 65; Fig.
1D2). Single action potentials evoked a large increase in
conductance ratio (conductance ratio >1.25) in 74% of mitral cells in
Mg2+-free ACSF. By contrast, only 1 of 65 mitral cells
recorded in physiological concentrations of Mg2+ (1.2
mM) had conductance ratios >1.25. Unlike changes in IPSP amplitude, which
vary dramatically near rest, the input conductance ratios associated with
recurrent inhibition remained relatively constant as the membrane potential
was varied over a 20-mV range (Fig.
1E). Mechanism of Mg2+-sensitive spike-evoked DDI
The ability of Mg2+ in the extracellular solution to
block spike-evoked DDI suggests that this form of recurrent inhibition
requires the activation of NMDA receptors on granule cells. A similar
requirement has been shown for recurrent and lateral DDI evoked by
voltage-clamp steps in mitral cells
(Isaacson and Strowbridge
1998) and extracellular stimulation
(Schoppa et al. 1998). We
tested the ability of specific NMDA and non-NMDA receptor antagonists to block
spike-evoked DDI as assayed by changes in the input conductance ratio in
Mg2+-free ACSF. As shown in
Fig. 1F, we found that
blockade of non-NMDA receptors with 1,2,3,4-tetrahydro-6-nitro-2,3-dioxo-benzo
[f] quinoxaline-7-sulfonamide disodium (NBQX; 5 µM) did not affect
spike-evoked DDI (conductance ratio = 1.9 ± 0.14). However,
spike-evoked DDI was abolished by 50 µM
D-2-amino-5-phosphopentanoic acid (D-APV), a specific
antagonist of NMDA receptors (conductance ratio = 1.0 ± 0.05).
Co-application of both APV and NBQX did not result in any further reduction in
spike-evoked DDI (conductance ratio = 1.1 ± 0.03). The suppression of
spike-evoked DDI achieved by blocking NMDA receptors alone, or in combination
with non-NMDA receptor antagonists, paralleled the inhibitory effect of
extracellular Mg2+. As expected, picrotoxin (PTX; 50
µM), an antagonist of GABAA receptors, also abolished
spike-evoked DDI (conductance ratio = 1.1 ± 0.02). None of the receptor
antagonists tested altered mitral cell input conductance in the absence of an
evoked action potential (data not shown). We also measured DDI to determine
whether intracellular BAPTA can block the glutamate release evoked by single
action potentials in mitral cells. As reported for voltage-clamp DDI (Issacson
and Strowbridge 1998), we find that spike-evoked DDI decayed rapidly after
making a whole cell recording with an internal solution that contained BAPTA
(20 mM; n = 8) while spike-evoked DDI was stable in control
recordings (internal solutions containing 0.2 mM EGTA; n = 8) from
the same slices (Fig.
1G). These results suggest that the NMDA receptors play a
critical role during spike-evoked DDI as well as voltage-clamp DDI.
DDI in physiological Mg2+ concentrations
We next sought to test under what conditions the tonic blockade of granule
cell NMDA receptors by extracellular Mg2+ ions could be
relieved. Under control conditions (1.2 mM Mg2+ ACSF),
the conductance change evoked by a suprathreshold depolarizing step
(conductance ratio = 1.02 ± 0.02; n = 25) was not different
from that evoked by a just-subthreshold stimulus (conductance ratio = 1.01
± 0.02; n = 25), indicating that a single action potential did
not trigger a measurable inhibitory response. During sensory stimulation,
-frequency oscillations are induced in many olfactory bulb neurons
(Adrian 1950). We sought to
test whether -frequency synaptic stimulation might activate inhibitory
granule cells and prime them to respond to glutamate released by mitral cells
at dendrodendritic synapses. We found that tetanic stimulation at -band
frequencies in the GCL (5 stimuli at 50 Hz) stimulated GABA release and
increased mitral cell input conductance (conductance ratio = 1.25 ±
0.02; n = 25; Fig.
2B1). More significantly, single action potentials
following a GCL tetanus now evoked recurrent inhibition that caused a further
increase in input conductance (conductance ratio = 1.43 ± 0.02;
n = 25; Fig.
2B2). These two effects of the GCL tetanus on mitral
cells (activation of feedforward inhibition and facilitation of recurrent
inhibition) are distinct and can be separated by varying the stimulus
protocol. The facilitating effect of the GCL tetanus on recurrent inhibition
was transient and did not lead to any build-up in the inhibitory response to
subthreshold stimuli (Fig.
2C). The spike-evoked increase in input conductance ratio
after GCL stimulation was abolished by the GABAA receptor
antagonist picrotoxin (PTX; 50 µM; n = 4). These results suggest
that the increases in input conductance following single action potentials
shown in Fig. 2, B and
C, reflect the openings of GABAA receptors and
not the augmentation of an intrinsic afterpotential response in mitral
cells.
|
Tetanic stimulation in the GCL evokes synaptic responses in both mitral and granule cells. The direct response of mitral cells to GCL stimulation was variable but often generated a slight depolarization. We did not observe any correlation between the magnitude or sign of the synaptic response in mitral cells evoked by GCL stimulation and the facilitation of recurrent inhibition. We attempted to mimic the stimulation-induced depolarization by preceding the mitral cell action potential with a long-duration depolarizing step (Fig. 2D). This current injection protocol consistently failed to reveal an inhibitory response following the action potential. GCL tetani also potentiated lateral IPSPs evoked using by electrical stimulation in the mitral cell layer (MCL; Fig. 2E). Results for this group of experiments are summarized in Fig. 2F and suggest that GCL stimulation gates recurrent DDI by depolarizing granule cells.
We next examined the receptor pharmacology of gated DDI in Mg2+. The non-NMDA receptor antagonist NBQX (10 µM) reduced but did not block the recurrent inhibition evoked by single mitral cell action potentials after a GCL tetanic stimulation (Fig. 3A). While reduced in amplitude, single action potentials still evoked a statistically significant increase in conductance ratio in NBQX (suprathreshold + GCL stim: 1.15 ± 0.02 / subthreshold + GCL stim: 1.10 ± 0.02; n = 6; P < 0.05) that was eliminated by picrotoxin. NBQX also eliminated most of the mitral cell depolarization during the GCL tetanus (see example traces in Fig. 3A1). In contrast with the effects of NBQX, bath application of the NMDA receptor antagonist D-APV (50 µM) completely blocked recurrent inhibition (Fig. 3B). This action was reversible on washout of APV and suggests that NMDA receptor activation is necessary in tetanus-gated DDI. Interestingly, we found that NBQX reduced the direct (feedforward) inhibition produced by the GCL tetanus more than did APV (see summary plot in Fig. 3C), suggesting that the gating input depolarizes granule cells preferentially by activating AMPA receptors.
Our results thus far suggest a model in which GCL stimulation gates recurrent inhibition by transiently depolarizing granule cells and thereby temporarily relieving the tonic blockade of NMDA receptors located in reciprocal dendrodendritic synapses. To test this model, we first recorded the response of granule cells to a GCL tetanus that was similar to that used to gate mitral cell inhibition (5 x 50 Hz; 40–100 µA). This stimulus evoked an excitatory response in all granule cells tested (n = 13). In granule cells recorded using our standard internal solution, tetanic stimulation in the GCL layer evoked multiple action potentials (Fig. 3D1; n = 5). This response resulted primarily from the activation of AMPA receptors because D-APV (50 µM) had little effect, whereas NBQX (5 µM) greatly reduced the response (Fig. 3D2; n = 3). We determined the timing of the peak granule cell depolarization associated with GCL tetanus in granule cells loaded with QX-314 to block voltage-gated Na+ channels. In most granule cells tested, the EPSP evoked by a 50-Hz GCL tetanus occurred after the last shock in the tetanus; the average latency from the last GCL stimulus to the EPSP peak was 14.3 ± 1.4 ms (n = 514 responses from 8 cells; Fig. 3D2).
We next sought to define the GCL stimulation parameters that effectively modulated recurrent inhibition. We varied the delay between the end of the GCL tetanus and the mitral cell action potential from 10 to 80 ms (Fig. 4A). We found that both the direct mitral cell inhibition evoked by the GCL stimulation alone and the facilitation of recurrent inhibition were maximal at short latencies (10 and 20 ms) and were not evident with latencies >60 ms. The period of maximal recurrent inhibition correlated well with timing of EPSPs recorded in granule cells in response to similar stimuli (see Fig. 3D2). Recurrent inhibition was facilitated, albeit to reduced levels, at intermediate latencies. At a latency of 60 ms, GCL stimulation facilitated recurrent inhibition following a single mitral cell action potential but did not evoke any measurable feedforward inhibition itself (without an action potential in the mitral cell). Both the direct mitral cell inhibition and the DDI gating effect required trains of GCL stimuli and were not evident when a single GCL stimulus preceded a mitral cell action potential (Fig. 4B). Facilitation of recurrent inhibition was most pronounced with trains of five or eight stimuli at 50 Hz. Experiments in which the stimulus frequency was varied indicate that inhibition is not apparent at GCL stimulus frequencies <25 Hz either alone or in combination with a mitral cell action potential (AP). Stimulus trains at higher frequencies (>75 Hz) strongly activated granule cells and occluded recurrent inhibition evoked by single APs in mitral cells (Fig. 4C). We also examined the effectiveness of different gating stimulus locations. We found that the most reliable results were obtained when the stimulating electrode was placed in the superficial GCL layer, within 250 µm of the MCL, or in the MCL itself, slightly off-beam to the intracellular recording electrode (displaced laterally 100–350 µm). In five cells in which the gating electrode position was varied systematically, we found that moving off-beam >500 µm from the recording electrode abolished gated DDI while stimulating in the deep GCL (>250 µm from the MCL) was only moderately effective (facilitating DDI in 2/5 mitral cells tested.)
|
We have demonstrated that GCL stimulation facilitates recurrent inhibition
in mitral cells through a NMDA receptor-dependent pathway. In the absence of a
gating stimulus, glutamate released from mitral cell dendrites would be
expected to remain bound to NMDA receptors on granule cells for many
milliseconds while the pore of those receptors remains blocked by
Mg2+ ions (Lester et
al. 1990). If the gating effect of the GCL stimulus is due to a
depolarizing synaptic input that unblocks NMDA receptors in granule cell
gemmules, the gating stimulus should also facilitate GABA release when applied
immediately following a mitral cell AP. To test this hypothesis, we triggered
an AP in mitral cells with increasing delays before the GCL tetanus
(Fig. 5A). We found
that a mitral cell AP that preceded the first GCL stimulus by 20 ms (and the
input conductance test step by 120 ms) facilitated recurrent DDI
(suprathreshold + GCL stim: 1.36 ± 0.02 / subthreshold + GCL stim: 1.18
± 0.02; n = 5; P < 0.01). As summarized in
Fig. 5B, the magnitude
of the DDI enhancement with this "reverse" protocol was similar to
that observed when the mitral cell AP occurred 20 ms after the last GCL
stimulus (suprathreshold + GCL stim: 1.39 ± 0.02 / subthreshold + GCL
stim: 1.17 ± 0.004; n = 5; P < 0.01). No DDI was
observed with the "reverse" protocol if the delay from the mitral
cell AP to the first GCL stimulus was increased to 100 ms (suprathreshold +
GCL stim: 1.15 ± 0.01 / subthreshold + GCL stim: 1.16 ± 0.02;
n = 3).
|
We next asked if glutamatergic excitation of granule cells was responsible
for the DDI gating. We examined this issue by employing focal application of
exogenous AMPA to attempt to mimic the gating effect of the GCL tetanus. We
recorded inhibitory responses using voltage-clamp recordings from CsCl-loaded
mitral cells to facilitate detection of unitary IPSCs. Under these conditions,
voltage-clamp steps failed to evoke DDI and instead triggered a rapidly
decaying inward current (Fig.
6A) that was unaffected by PTX
(Friedman and Strowbridge
2000). AMPA applications to the GCL evoked a slight increase in
IPSC frequency in eight mitral cells tested
(Fig. 6A). However, a
large DDI response was evoked when the AMPA application was paired with a
voltage-clamp step (Fig.
6B). This response was significantly larger than that
evoked by either the voltage-clamp step or AMPA puff by itself (GCL AMPA: 30.1
± 7.0 nA·s; step: 39.9 ± 13 nA·s; step + GCL AMPA:
124 ± 30 nA·s; n = 8; P < 0.01). The
increase in current integral associated with the GCL AMPA application and the
facilitated response to a voltage-clamp step were eliminated when the
experiment was repeated in PTX (50 µM;
Fig. 6D), indicating
that the mitral cell response facilitated by AMPA receptor activation was
GABAergic. In contrast with these results, AMPA application to the external
plexiform layer (EPL) caused very little increase in spontaneous IPSC
frequency and did not facilitate DDI responses to voltage-clamp steps
(Fig. 6C; step: 24.5
± 6.3 nA·s; step + EPL AMPA: 36.1 ± 13.2 nA·s;
n = 4; P > 0.1).
Functional effects of gated inhibition of mitral cells
We next examined the functional effects of gated DDI and asked if gated DDI would alter the response of mitral cells to a simulated excitatory synaptic input. We first tested if pairing a GCL tetanus with a mitral cell AP alters the depolarizing current needed to trigger a second AP. We iteratively determined the amplitude of a 5-ms duration current step needed to reach AP threshold when evoked in isolation and after a conditioning AP evoked 20 ms earlier. In physiological Mg2+, the effect of the conditioning AP on the current threshold was not significant (current threshold ratio: 1.13 ± 0.1; P > 0.1; n = 4; Fig. 7A). Similar experiments performed in Mg2+-free ACSF (not shown) generated a large increase in current threshold due to the conditioning spike (current threshold ratio: 2.6 ± 0.1) that was eliminated by PTX. In Mg2+ ACSF, tetanic GCL stimulation resulted in only a minor increase in current threshold ratio in Mg2+ (1.11 ± 0.03; n = 4). However, pairing of the conditioning AP with GCL stimulation increased the amplitude of the current step required to reach AP threshold (1.48 ± 0.04; significantly different from control, P < 0.01; Fig. 7B).
Finally we asked whether gated DDI altered mitral cell firing patterns. In
these experiments, we injected a train of six simulated EPSPs at 2 Hz designed
to mimic the natural response of mitral cells during sniffing
(Charpak et al. 2001). We found
that phasic EPSPs produced more reproducible firing patterns in mitral cells
than did rectangular current pulses. Under control conditions, APs were
triggered reliably by the second through fifth simulated EPSPs (sEPSPs,
Fig. 7C1). Pairing
these sEPSPs with GCL tetanus immediately before the second sEPSP
significantly reduced the number of APs generated by sEPSP2 (from
5.2 ± 0.3 to 3.7 ± 0.2; n = 17 trials from 4 mitral
cells; P < 0.01). Interestingly, the GCL tetanus did not affect
the latency to the first AP evoked by sEPSP2
(Fig. 7C2). Rather,
the tetanus selectively increased the latency to the second and later APs in
sEPSP2. The decrease in the total number of spikes evoked by
sEPSP2 appeared to be a consequence of the increased interspike
intervals. The observation that the latency to the first spike was not
significantly affected by the GCL tetanus suggests that the dominant effect of
the tetanus was to facilitate recurrent inhibition rather than to evoke GABA
release directly. If the predominant effect of the GCL tetanus had been to
trigger feedforward inhibition, the resulting IPSP would be expected to
increase the first spike latency. Such an effect on the first spike latency
was observed but only at higher stimulus intensities (data not shown). Results
similar to these were observed in four mitral cells and are summarized in
Fig. 7, E and
F. These observations demonstrate that a brief GCL tetanus can prime
granule cells to release GABA in response to the glutamate released by spiking
mitral cells.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
-frequency electrical stimulation in the GCL. Although
tetanic stimulation appears to gate DDI through activation of granule cell
AMPA receptors located in the GCL, the gated DDI response still requires
activation of NMDA receptors on granule cells. This model provides a mechanism
for the local control of recurrent and lateral DDI through activity of mitral
cells innervated by different subclasses of olfactory receptor neurons.
Additionally we have shown that DDI gated by tetanic GCL stimulation can
inhibit mitral cell AP generation and can modulate mitral cell firing
patterns. In this study, we used spike-induced changes in input conductance to assay recurrent inhibition. This method is possible to use with mitral cells because these neurons do not have long-lasting intrinsic afterpotentials that would affect delayed measurements of input conductance. We presented the effects of recurrent inhibition as changes in input conductance rather than input resistance so as to provide a more intuitive measure of DDI (increased inhibition generates a positive changes in the input conductance ratio.) A disadvantage of this method is that we underestimate the true magnitude of the IPSP response because the underlying inhibitory conductance is decaying during the test step. However, this disadvantage is offset by the insensitivity of the conductance ratio method to small changes in membrane potential (see Fig. 1E).
Mechanism of spike-evoked DDI in Mg2+-free ACSF
We find that single-spike DDI shares many properties with those found
during more commonly studied voltage-clamp DDI. Intracellular studies in the
turtle olfactory bulb by Jahr and Nicoll
(1980,
1982) first demonstrated
directly that single APs in mitral cells were followed by a prolonged
afterhyper-polarization mediated by reciprocal dendrodendritic synaptic
connections. These studies were followed by more recent explorations of the
cellular mechanism of reciprocal DDI in the mammalian olfactory bulb that
employed either extracellular electrical stimulation
(Schoppa et al. 1998) or
voltage-clamp pulses (Isaacson and
Strowbridge 1998) to trigger glutamate release from mitral cells.
These studies have found that DDI requires the activation of NMDA receptors on
inhibitory interneurons. In the present study, we have also found a
requirement for NMDA receptor activation when single APs are used as the
stimulus for glutamate release and show directly that extracellular
Mg2+ tonically inhibits spike afterpotentials in mitral
cells. Our results on the APV sensitivity of spike-evoked recurrent inhibition
in Mg2+-free ACSF are consistent with those from Chen et
al. (2000).
Gating of DDI in physiological Mg2+
While there has been common agreement that recurrent DDI requires the
activation of NMDA receptors in Mg2+-free ACSF, the role
of AMPA receptors in DDI is less clear. Also unresolved is the question of how
the tonic blockade of these receptors by extracellular
Mg2+ is relieved under physiological conditions. One
possibility, proposed by Schoppa and Westbrook
(1999) suggests that recurrent
inhibition is effectively blocked by transient K+ currents located
in the granule cell gemmule. This model provides one potential explanation for
the failure of colocalized AMPA receptors
(Sassoe-Pognetto and Ottersen
2000) to activate recurrent DDI because intrinsic K+
currents in granule cell dendrites may effectively shunt the AMPA
receptor-mediated EPSP before it can activate the local VDCCs that control
GABA exocytosis (Halabisky et al.
2000; Isaacson and Strowbridge
1998). In this model, NMDA receptor-mediated EPSPs occurring at
the same synapse can trigger GABA release because their time course is
sufficiently long so as to outlast the transient K+ currents in
granule cell dendrites and activate VDCCs. The finding by Isaacson
(2001) that AMPA
receptor-dependent DDI can be revealed by artificially prolonging AMPA
receptor-mediated EPSPs with cyclothiazide is consistent with this model. In
the behaving animal, another source of near-coincident granule cell
depolarization presumably would be necessary to inactivate dendritic
K+ channels, enabling AMPA receptor-dependent GABA release.
We find that -frequency GCL stimulation effectively gates recurrent
inhibition in mitral cells in the presence of Mg2+.
Gating required multiple GCL stimuli at a frequency near 50 Hz. The frequency
sensitivity of the gating signal closely matches the frequency of synchronous
oscillations recorded in vivo in the olfactory bulb during sensory processing
(Adrian 1950;
Eeckman and Freeman 1990),
suggesting that inhibition may be modulated directly by network oscillations.
Although the gating input by itself evoked some GABA release, significantly
more GABA was released following near coincident extracellular GCL and
intracellular mitral cell stimulation. The gating effect of tetanic GCL
stimulation was not limited to recurrent inhibition—we also observed an
enhancement in lateral IPSPs evoked by off-beam stimulation in the mitral cell
layer. The time window for gating of recurrent DDI by a GCL tetanus
(Figs. 4A) coincides
well with the latency to the peak of intracellularly recorded EPSP in granule
cells (Fig. 3D). This
result is consistent with the hypothesis that the depolarization of the
granule cells due to the tetanic EPSP could directly relieve the
Mg2+ blockade of NMDA receptors that controls DDI
microcircuits. It is not known, however, to what extent this excitatory
synaptic response would be attenuated by the cable properties and active
conductances present in granule cell dendrites. Our results do not exclude the
possibility that active Na+ and Ca2+ currents
recruited by the gating EPSP assist in depolarizing granule cell gemmules.
Cellular mechanism of DDI gating
One of the most enigmatic results from the recent studies of olfactory bulb
circuitry (Isaacson and Strowbridge
1998; Schoppa et al.
1998; this study) is that recurrent inhibition—a
near-universal feature in most brain regions—is relatively ineffective
in olfactory bulb slices bathed in physiological levels of
Mg2+. Our results suggest a model by which other
excitatory inputs onto granule cells facilitate recurrent inhibition in the
olfactory bulb. In this model, the gating input initially depolarizes the soma
and proximal dendrites of granule cells, primarily through activation of AMPA
receptors. This initial depolarization itself, perhaps after amplification by
intrinsic currents in the granule apical dendrite, probably has at least two
effects on distal granule cell dendrites that facilitate GABA release. First,
the depolarization will reduce the transient K+ currents present in
granule cells that prevent AMPA receptors from triggering GABA release
(Schoppa and Westbrook 1999).
Second, the gating input would relieve the Mg2+ block of
NMDA receptors once they are activated by glutamate, increasing the
depolarization of the gemmule and providing another source
Ca2+ influx. The resulting depolarization could then
open the local voltagegated Ca2+ channels that govern
GABA exocytosis (Halabisky et al.
2000; Isaacson and Strowbridge
1998), producing a recurrent IPSP in the mitral cell.
Interestingly, we found that gated DDI was still dependent on NMDA receptor
activation, suggesting that the relief of the Mg2+
blockade might be the critical effect of the gating input.
Surprisingly we found that we could mimic the gating effect of the electrical tetanus by applying exogenous AMPA in the GCL but not in the EPL. This finding suggests that gating of DDI depends critically on the location of the activated AMPA receptors; recurrent inhibition is facilitated best by activation of AMPA receptors on the proximal dendrites of granule cells. It is unlikely that exogenous AMPA gated DDI by activating AMPA receptors located in granule cell gemmules because more of these receptors should be activated by the EPL applications than the applications in GCL. Rather, exogenous AMPA likely modulated recurrent DDI by depolarizing large dendritic regions in granule cells. The greater effectiveness of GCL AMPA applications may be due to a higher density of AMPA receptors along the proximal dendrites of granule cells or to intrinsic mechanisms in granule cells that selectively amplify proximal excitatory synaptic inputs. These results also suggests that the source of the gating input to granule cells are glutamatergic axons that synapse on the proximal dendrites of granule cells. The other possibility for the gating input—lateral excitation through antidromic activation of other dendrodendritic synapses—appears to be less likely because AMPA applications in the EPL, which should mimic these inputs, do not appear to gate DDI.
There are two major sources of the glutamatergic axonal synapses in the
GCL: mitral cell axon collaterals (Kishi
et al. 1984; Orona et al.
1984) and centrifugal axons from neurons extrinsic to the
olfactory bulb (Shepherd and Greer
1998). These two categories of synaptic input are difficult to
distinguish in acute slices where axon collaterals from most mitral cells are
severed. The location we found best for facilitating DDI response with
exogenous AMPA, the upper GCL, is consistent with the termination pattern of
mitral cell axon collaterals (Kishi et al.
1984). Our present results and this correlation, although
consistent with mitral cell axon collaterals providing the gating input, do
not exclude a role for centrifugal axons in gating DDI. Also it is possible
that in the intact olfactory bulb—with ongoing spontaneous activity and
sensory input to synchronize synaptic responses onto granule cells—DDI
gating might occur over wider range of stimulus parameters than found in this
in vitro study. Similarly, we cannot exclude the possibility that a
sufficiently strong source of dendrodendritic excitation to granule cells in
the behaving animal might relieve the Mg2+ blockade of
local NMDA receptors without requiring activation of proximal axodendritic
synapses on granule cells.
Functional implications for olfactory sensory processing
Odor stimuli evoke a complex series of excitatory and inhibitory
postsynaptic potentials in mitral cells
(Hamilton and Kauer 1989;
Wellis et al. 1989). Mitral
cell firing patterns—the output of the olfactory bulb—is
determined by these synaptic inputs in combination with the intrinsic
properties of the mitral cell. The inhibitory components to odor responses
must be generated by activity in local circuits within the bulb because the
input from receptor neurons is purely excitatory
(Aroniadou-Anderjaska et al.
2000; Isaacson and Strowbridge
1998). Our results suggest a new mechanism that regulates the
effectiveness of recurrent and lateral inhibition and therefore could
contribute to the temporal patterning of IPSPs in mitral cells during sensory
stimulation. These local axon projections may function to coordinate
homologous glomeruli in each hemibulb that are innervated by projections from
sensory neurons that express the same olfactory receptor protein
(Mombaerts et al. 1996).
|
DISCLOSURES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
|
FOOTNOTES |
|---|
Address for reprint requests: B. W. Strowbridge, Dept. of Neurosciences, Case Western Reserve University, 10900 Euclid Ave., Cleveland, OH 44106, bens{at}po.cwru.edu (email)
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
Aroniadou-Anderjaska V, Zhou FM, Priest CA, Ennis M, and Shipley
MT. Tonic and synaptically evoked presynaptic inhibition of sensory
input to the rat olfactory bulb via GABA(B) heteroreceptors. J
Neurophysiol 84:
1194–1203, 2000.
Bormann J,
Hamill OP, and Sakmann B. Mechanism of anion permeation through channels
gated by glycine and gamma-aminobutyric acid in mouse cultured spinal neurons.
J Physiol 385:
243–286, 1987.
Charpak S,
Mertz J, Beaurepaire E, Moreaux L, and Delaney K. Odor-evoked calcium
signals in dendrites of rat mitral cells. Proc Natl Acad Sci
USA 98:
1230–124, 2001.
Chen WR, Xiong W, and Shepherd GM. Analysis of relations between NMDA receptors and GABA release at olfactory bulb reciprocal synapses. Neuron 25: 625–633, 2000.[ISI][Medline]
Eeckman FH and Freeman WJ. Correlations between unit firing and EEG in the rat olfactory system. Brain Res 528: 238–244, 1990.[ISI][Medline]
Ezeh PI, Wellis
DP, and Scott JW. Organization of inhibition in the rat olfactory bulb
external plexiform layer. J Neurophysiol
70: 263–274,
1993.
Friedman D and
Strowbridge BW. Functional role of NMDA autoreceptors in olfactory mitral
cells. J Neurophysiol 84:
39–50, 2000.
Friedrich RW and Laurent G. Dynamic optimization of odor representations by slow
temporal patterning of mitral cell activity. Science
291: 889–894,
2001.
Halabisky B,
Friedman D, Radojicic M, and Strowbridge BW. Calcium influx through NMDA
receptors directly evokes GABA release in olfactory bulb granule cells.
J Neurosci 20:
5124–5134, 2000.
Hamilton KA and Kauer JS. Patterns of intracellular potentials in salamander mitral/tufted cells in response to odor stimulation. J Neurophysiol 56: 1571–1589, 1989.
Isaacson JS and Strowbridge BW. Olfactory reciprocal synapses: dendritic signaling in the CNS. Neuron 20: 749–761, 1998.[ISI][Medline]
Isaacson JS. Mechanisms governing dendritic
gamma-aminobutyric acid (GABA) release in the rat olfactory bulb.
Proc Nat Acad Sci USA 98:
337–342, 2001.
Jahr CE and
Nicoll RA. Dendrodendritic inhibition: demonstration with intracellular
recording. Science 207:
1473–1475, 1980.
Jahr CE and
Nicoll RA. An intracellular analysis of dendrodendritic inhibition in the
turtle in vitro olfactory bulb. J Physiol
326: 213–234,
1982.
Kishi K, Mori K, and Ojima H. Distribution of local axon collaterals of mitral displaced mitral and tufted cells in the rabbit olfactory bulb. J Comp Neurol 225: 511–526, 1984.[ISI][Medline]
Laurent G.
A systems perspective on early olfactory coding.
Science 286:
723–728, 1999.
Laurent G, Stopfer M, Friedrich RW, Rabinovich MI, Volkovskii A, and Abarbanel HD. Odor encoding as an active dynamical process: experiments computation and theory. Annu Rev Neurosci 24: 263–297, 2001.[ISI][Medline]
Lester RA, Clements JD, Westbrook GL, and Jahr CE. Channel kinetics determine the time course of NMDA receptor-mediated synaptic currents. Nature 346: 565–567, 1990.[Medline]
Macrides F and
Chorover SL. Olfactory bulb units: activity correlated with inhalation
cycles and odor quality. Science
175: 84–87,
1972.
Margrie TW,
Sakmann B, and Urban NN. Action potential propagation in mitral cell
lateral dendrites is decremental and controls recurrent and lateral inhibition
in the mammalian olfactory bulb. Proc Natl Acad Sci
USA 98:
319–324, 2001.
Mombaerts P, Wang F, Dulac C, Chao SK, Nemes A, Mendelsohn M, Edmondson J, and Axel R. Visualizing an olfactory sensory map. Cell 87: 675–686, 1996.[ISI][Medline]
Mori K, Nagao
H, and Yoshihara Y. The olfactory bulb: coding and processing of odor
molecule information. Science
286: 711–715,
1999.
Nusser Z, Kay
LM, Laurent G, Homanics GE, and Mody I. Disruption of GABA(A) receptors on
GABAergic interneurons leads to increased oscillatory power in the olfactory
bulb network. J Neurophysiol
86: 2823–2833,
2001.
Orona E, Rainer EC, and Scott JW. Dendritic and axonal organization of mitral and tufted cells in the rat olfactory bulb. J Comp Neurol 226: 346–356, 1984.[ISI][Medline]
Rall W, Shepherd GM, Reese TS, and Brightman MW. Dendrodendritic synaptic pathway for inhibition in the olfactory bulb. Exp Neurol 14: 44–56, 1966.[ISI][Medline]
Sassoe-Pognetto M and Ottersen OP. Organization of ionotropic glutamate receptors at dendrodendritic synapses in the rat olfactory bulb. J Neurosci 15: 2192–2101, 2000.
Schoppa NE, Kinzie JM, Sahara Y, Sergerson TP, and Westbrook GL. Dendrodendritic inhibition in the olfactory bulb is driven by NMDA receptors. J Neurosci 18: 7690–6802, 1998.
Schoppa NE and Westbrook GL. Regulation of synaptic timing in the olfactory bulb by an A-type potassium current. Nat Neurosci 2: 1106–1113, 1999.[ISI][Medline]
Shepherd GM and Greer CA. Olfactory bulb. In: Synaptic Organization of the Brain (4th ed.), edited by Shepherd GM. New York: Oxford, 1998, p. 159–204.
Stopfer M, Bhagavan S, Smith BH, and Laurent G. Impaired odour discrimination on desynchronization of odour-encoding neural assemblies. Nature 390: 70–74, 1997.[Medline]
Wehr M and Laurent G. Odor encoding by temporal sequences of firing in oscillating neural assemblies. Nature 384: 162–166, 1996.[Medline]
Wellis DP and
Kauer JS. GABAergic and glutamatergic synaptic input to identified granule
cells in salamander olfactory bulb. J Physiol
475: 419–430,
1994.
Wellis DP,
Scott JW, and Harrison TA. Discrimination among odorants by single neurons
of the rat olfactory bulb. J Neurophysiol
61: 1161–1177,
1989.
This article has been cited by other articles:
|
|
 |
|
  C. E. Reisenman, T. Heinbockel, and J. G. Hildebrand Inhibitory Interactions Among Olfactory Glomeruli Do Not Necessarily Reflect Spatial Proximity J Neurophysiol, August 1, 2008; 100(2): 554 - 564. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. Chiang and B. W. Strowbridge Diversity of Neural Signals Mediated by Multiple, Burst-Firing Mechanisms in Rat Olfactory Tubercle Neurons J Neurophysiol, November 1, 2007; 98(5): 2716 - 2728. |
) or
bis-(o-aminophenoxy)-N,N,N',N'-tetraacetic
acid (BAPTA)-containing (20 mM; 
) internal solution (8 mitral cells for
each condition). Dialysis with BAPTA causes a rapid reduction in spike-evoked
DDI, whereas spike-evoked DDI is constant with dialysis using the control
internal solution containing 0.2 mM EGTA.
) of a granule cell. D2,
top: D-APV (50 µM) only slightly reduces the EPSP evoked by
a GCL tetanus in (lidocaine N-ethyl bromide) QX-314-loaded GCs. The
subsequent application of the AMPA receptor antagonist NBQX (5 µM) greatly
reduces the remainder of the response. Bottom: summary histogram of
latency to granule cell EPSP peak in (lidocaine N-ethyl bromide)
QX-314-filled granule cells measured from the last stimulus in the GCL tetanus
[n = 514 excitatory postsynaptic potentials (EPSPs) from 8 granule
cells].
1.5 times control current) if the test step
is preceded by both GCL tetanus and a conditioning AP. B: summary of
effect of a conditioning action potential and GCL stimulation on the current
step amplitude required to reach action potential threshold. Results are
expressed as a ratio compared with the amplitude required to reach AP
threshold of a current step alone. C1: mitral cell APs evoked by a
temporally summating train of 6 simulated EPSPs (sEPSP1–6).
APs are initiated reliably by sEPSP2–5 under control
conditions (control). The number of APs evoked by sEPSP2 is reduced
when a GCL tetanus (5 x 50 Hz; 50 µA; last stimulus 20 ms before
onset of sEPSP2) was applied between sEPSP1 and
sEPSP2 (GCL Stim). The current injection waveform and GCL stimulus
timing shown below the voltage traces. C2: expansion of response to
sEPSP2 showing that GCL stimulation preferentially delayed the 2nd
and later APs without altering the timing of the 1st AP. D: summary
of effects of GCL stimulation on the number of APs generated by each sEPSP.
N's indicated in bar graphs reflect total number of responses analyzed
from 4 mitral cells (**P < 0.01; Student's t-test).
E1: summary of effects of GCL stimulation the latency of 1st, 2nd,
and 3rd APs evoked by sEPSP2. Time measurements are from the onset
of sEPSP2. Data from 4 mitral cells. E2: Same data in
E1 reploted as relative shifts in AP latency with GCL stimulation for
the 1st–3rd APs (compared with control.)
) and suprathreshold (
)
responses with near-coincident GCL stimulation are significantly different
(**P < 0.01; paired t-test).
60 ms. Supra- and
subthreshold conductance ratios significantly different at *P <
0.05 and **P < 0.01. No recurrent inhibition was observed when GCL
tetanus occurred 80 ms before the mitral cell action potential. Example
responses shown above. Data from 3 to 5 mitral cells. B: effect of
varying the number of GCL stimuli. Recurrent inhibition was observed when
trains of 5 or 8 GCL stimuli preceded the mitral cell action potential (supra-
and subthreshold conductance ratios significantly different; P <
0.05). Recurrent inhibition was occluded by the direct inhibition evoked by
trains of 10 GCL stimuli. No DDI was evident after suprathreshold test pulses
with only a single GCL stimulus. Data from 4 mitral cells. C: effect
of varying the frequency of GCL stimulation. Recurrent inhibition was evident
with a stimulation frequency of 50 Hz (P < 0.05), whereas no
inhibition was apparent with 10 Hz GCL stimulation. Recurrent inhibition was
occluded by the direct response to GCL stimulus trains >50 Hz. Data from 3
mitral cells. All responses recorded in 1.2 mM Mg2+
ACSF.