Department of Cell Biology and Anatomy, Neuroscience Research Group, University of Calgary, Calgary, Alberta T2N 4N1, Canada
![]() |
ABSTRACT |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Lemon, N. and R. W. Turner. Conditional Spike Backpropagation Generates Burst Discharge in a Sensory Neuron. J. Neurophysiol. 84: 1519-1530, 2000. Backpropagating dendritic Na+ spikes generate a depolarizing afterpotential (DAP) at the soma of pyramidal cells in the electrosensory lateral line lobe (ELL) of weakly electric fish. Repetitive spike discharge is associated with a progressive depolarizing shift in somatic spike afterpotentials that eventually triggers a high-frequency spike doublet and subsequent burst afterhyperpolarization (bAHP). The rhythmic generation of a spike doublet and bAHP groups spike discharge into an oscillatory burst pattern. This study examined the soma-dendritic mechanisms controlling the depolarizing shift in somatic spike afterpotentials, and the mechanism by which spike doublets terminate spike discharge. Intracellular recordings were obtained from ELL pyramidal somata and apical dendrites in an in vitro slice preparation. The pattern of spike discharge was equivalent in somatic and dendritic regions, reflecting the backpropagation of spikes from soma to dendrites. There was a clear frequency-dependent threshold in the transition from tonic to burst discharge, with bursts initiated when interspike intervals fell between ~3-7 ms. Removal of all backpropagating spikes by dendritic TTX ejection revealed that the isolated somatic AHPs were entirely stable at the interspike intervals that generated burst discharge. As such, the depolarizing membrane potential shift during repetitive discharge could be attributed to a potentiation of DAP amplitude. Potentiation of the DAP was due to a frequency-dependent broadening and temporal summation of backpropagating dendritic Na+ spikes. Spike doublets were generated with an interspike interval close to, but not within, the somatic spike refractory period. In contrast, the interspike interval of spike doublets always fell within the longer dendritic refractory period, preventing backpropagation of the second spike of the doublet. The dendritic depolarization was thus abruptly removed from one spike to the next, allowing the burst to terminate when the bAHP hyperpolarized the membrane. The transition from tonic to burst discharge was dependent on the number and frequency of spikes invoking dendritic spike summation, indicating that burst threshold depends on the immediate history of cell discharge. Spike frequency thus represents an important condition that determines the success of dendritic spike invasion, establishing an intrinsic mechanism by which backpropagating spikes can be used to generate a rhythmic burst output.
![]() |
INTRODUCTION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Burst discharge in sensory circuits
can play an important role in encoding stimulus features or
synchronizing activity across widely disparate regions (Lisman
1997; Ritz and Sejnowski 1997
). Oscillatory
burst discharge consists of the rhythmic generation of spikes grouped
in a burst and a subsequent hyperpolarization of membrane potential.
The depolarizing phase of a burst is often driven by a depolarizing
afterpotential (DAP) which can be generated by any of several inward
currents (Azouz et al. 1996
; Bourque 1986
; Ghamari-Langroudi and Bourque 1998
;
Higashi et al. 1993
; Hoehn et al. 1993
;
Huguenard 1996
). We have shown that a DAP can also be
generated at the soma by backpropagating dendritic
Na+ spikes in electrosensory lateral line lobe
(ELL) pyramidal cells (Lemon and Turner 1999
;
Turner et al. 1994
). These are principal output cells in
an electrosensory system which enables these animals to electrolocate
with respect to external electric fields (Bullock and
Heiligenberg 1986
; Heiligenberg 1991
;
Moller 1995
; Turner et al. 1999
). Sodium
spikes are initiated near the soma and exhibit a substantial increase
in duration as they backpropagate ~200 µm into apical dendrites
(Turner et al. 1994
). As a result, current flow
associated with dendritic spike discharge sources back to the soma to
generate a DAP that follows the somatic spike. During repetitive
discharge the peak amplitude of the DAP increases, resulting in a
depolarizing membrane potential shift that eventually reaches threshold
for a high-frequency spike doublet. The spike doublet is followed by a
burst afterhyperpolarization (bAHP) that temporarily prevents spike
discharge. The process underlying the depolarizing shift and generation
of a spike doublet can thus group cell output into a series of
oscillatory spike bursts.
The ability for backpropagating dendritic spikes to directly influence
somatic membrane potential has become apparent in physiological and
modeling studies (Golding et al. 1999; Larkum et
al. 1999
; Magee and Carruth 1999
; Mainen
and Sejnowski 1996
; Turner et al. 1994
;
Wang 1999
; Yuste et al. 1994
;
Zhang et al. 1993
). Spike doublets have also been
reported in several bursting cell types, although their role in burst
discharge has not been determined (Calvin and Sypert
1976
; Ermentrout and Kopell 1998
; Koch
1999
; Lo et al. 1998
; Paré et al.
1995
; Whittington et al. 1997
). As found for
many sensory neurons, spike bursts in ELL pyramidal cells have a
demonstrated role in feature detection in vivo (Gabbiani et al.
1996
; Lisman 1997
). Furthermore, pyramidal cells
exhibit a frequency selectivity across multiple topographic maps in
vivo that correlates with the properties of spike bursts across these maps in vitro (Shumway 1989
; Turner et al.
1996
). Thus the ELL represents an ideal system to investigate
the underlying basis and significance of burst discharge in sensory neurons.
This study examines the soma-dendritic mechanisms underlying the somatic depolarizing shift and spike doublet generation during oscillatory discharge in ELL pyramidal cells. We describe a novel mechanism by which spike frequency establishes a conditional form of backpropagation that generates an oscillatory pattern of spike bursts.
![]() |
METHODS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Weakly electric Apteronotus leptorhynchus were
obtained from local importers and maintained at 26-28°C in fresh
water aquaria. All chemicals were obtained from SIGMA (St. Louis, MO)
unless otherwise noted. In all cases, recordings were obtained from
separate pyramidal cell somata or apical dendrites using an in vitro
slice preparation. Animals were anesthetized in a small tank containing 0.05% phenoxy-ethanol, and ELL tissue slices of 300-450 µm
thickness prepared as described previously (Kotecha et al.
1997; Turner et al. 1994
). Slices were cut by
Vibratome along either the transverse or longitudinal axis into a
bicarbonate-buffered artificial cerebrospinal fluid (ACSF) consisting
of (in mM): 124 NaCl, 2.0 KCl, 1.25 KH2PO4, 1.5 CaCl2, 1.5 MgSO4, 24 NaHCO3, and 10 D-glucose (pH 7.4). In some cases, slices were cut in a reduced chamber filled with a sucrose-based ACSF consisting of (in mM): 218 sucrose, 25 NaHCO3, 3.25 K Gluconate, 4.5 MgCl2, 0.1 CaCl2, 10 Glucose, and 1 Na pyruvate (Kotecha et al. 1997
) and
transferred to a Petri dish containing regular ACSF. In the latter
slices more cells appeared to be available for recording but there were
no significant differences in the physiological properties of
pyramidal cells in slices cut in either medium. Slices were
maintained as an interface preparation at room temperature in an
oxygenated (95% O2-5%
CO2) in vitro slice chamber.
The majority of pyramidal cell recordings were obtained in the
centromedial segment (CMS) map of the ELL. Since identification of
segmental maps can be difficult in slices cut with a longitudinal orientation, a limited number of recordings may also have been obtained
from the centrolateral (CLS) or lateral (LS) segmental maps. Glass
microelectrodes were backfilled with 2 m K acetate (pH 7.4; ~90
m resistance) and intracellular recordings were obtained from either
pyramidal cell somata or apical dendrites. Unless otherwise indicated,
all dendritic recordings were obtained ~150 µm from the cell body
layer near the border between the tractus stratum fibrosum
and the molecular layer (Maler 1979
). Input resistance in somatic recordings ranged between 23 and 122 M
[73 ± 21.0 (SD) M
] and in dendritic recordings between 31 and 90 M
(74 ± 12.7 M
; n = 10 random samples). The similarity of
these values reflects the fact that dendritic recordings obtained 200 µm from the soma are within 0.36 electrotonic length according to
recent modeling studies (B. Doiron, L. Maler, and A. Longtin,
unpublished data). Resting potential at the soma was
77 ± 10 mV
and in dendrites
76 ± 7.8 mV (n = 10 random
samples). Direct current injection of
0.1 to
0.5 nA was applied
when necessary to reduce spontaneous activity.
Pyramidal cells were antidromically activated using square wave stimulus pulses (0.1 ms) delivered via an isolation unit to a bipolar stimulating electrode consisting of twisted 62 µm nichrome wire. Evoked activity was recorded and stored on a PC for offline analysis (CED, Cambridge, UK).
Tetrodotoxin (Alomone Labs, Israel) was focally ejected over the
dendritic axis using electrodes of 1-2 µm tip diameter and 10-15
psi according to the protocol detailed in Turner et al. (1994). Average values are indicated as mean ± SD and
statistical significance calculated using a Student's
t-test or the Wilcoxon Signed Ranks test. Data plots were
fit by a sigmoidal or Gaussian function using the software Microcal Origin.
![]() |
RESULTS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
The occurrence of burst discharge in ELL pyramidal cells has been
reported under both in vitro and in vivo recording conditions (Gabbiani et al. 1996; Metzner et al.
1998
; Turner et al. 1994
, 1996
). This study
extends this work by examining the mechanisms for generating
oscillatory spike bursts in both the soma and apical dendrites of
pyramidal cells. In all cases, recordings were obtained from separate
somatic or dendritic recordings. Figure
1A illustrates that an
oscillatory series of spike bursts could be evoked in either somatic or
dendritic recordings using depolarizing current injection. Single spike
bursts at the soma consisted of a series of spikes that steadily
increased in frequency until generating a higher frequency spike
doublet and bAHP. Bursts were most often comprised of 5-7 spikes near
threshold but could consist of only spike doublets at the highest
frequencies of burst discharge or at more depolarized membrane
potentials. Since Na+ spike discharge is
initiated at the soma and backpropagates into apical dendrites
(Turner et al. 1994
), the basic pattern and frequency of
oscillatory bursts were similar at the somatic and dendritic level. The
primary difference was that the dendritic bAHP was smaller in amplitude
than the somatic bAHP, or consisted of only a slow return of membrane
potential to baseline over 10-20 ms.
|
Oscillatory discharge in pyramidal cells is thus defined as the grouping of spike discharge into rhythmic bursts, with each burst ending in a spike doublet and bAHP. A spike doublet is defined as the shortest interspike interval within a spike train, which was always followed by a bAHP. This study focuses on the soma-dendritic interaction that promotes a shift in spike afterpotentials during repetitive spike discharge and the manner in which a spike doublet contributes to burst termination.
A schematic diagram depicting how a backpropagating spike influences
somatic membrane excitability is shown in Fig. 1B.
Na+ spike discharge is initiated near the cell
body and is followed by spike backpropagation through ~200 µm
(approximately one-third) of the apical dendritic arborization. The
somatic spike is of narrow duration while the dendritic spike rapidly
increases in duration with distance from the soma. As a result, current
associated with dendritic spike discharge sources back to evoke a DAP
at the soma that is superimposed on the somatic fast AHP (fAHP) and slow AHP (sAHP) (Turner et al. 1994). It is important to
emphasize that this mechanism predicts that the DAP is generated
primarily by electrotonic conduction during backpropagation
of the spike from soma to dendrite, and not by a spike propagating in a
return fashion from dendrite to soma. We have also established that the DAP in pyramidal cells is not generated by a
Ca2+-dependent current or by synaptic potentials
(Azouz et al. 1996
; Bourque 1986
;
Higashi et al. 1993
; Steriade et al.
1998
; White et al. 1989
), because focal somatic
or dendritic ejections of the general
Ca2+-channel blocker Cd2+
(400 µM) do not reduce the DAP (n = 9). In addition,
we have no evidence for the existence of
Ih in ELL pyramidal cells that could
contribute to burst discharge. Rather, the source of current for the
DAP originates with the backpropagating TTX-sensitive Na+ spike (Turner et al. 1994
).
Pyramidal cells respond to depolarizations near spike threshold
with a relatively tonic discharge of Na+ spikes
in the pattern of an initial lag and subsequent increase in frequency
(Fig. 1C) (Mathieson and Maler 1988). This
resulted from a shift in the balance between depolarizing and
hyperpolarizing spike afterpotentials that was most visible for somatic
recordings, where the peak amplitude of the fAHP and sAHP decreased
while the peak amplitude of the DAP increased with successive spikes. This produced a net shift in membrane potential of up to 5.3 mV during
burst discharge at the soma (2.8 ± 1.4 mV; n = 9). As previously reported, dendritic spikes were of smaller amplitude
and longer duration compared with the soma, with only a small sAHP
(Turner et al. 1994
). Repetitive discharge converted the
small dendritic sAHP to a net depolarization of up to 9.8 mV (6.2 ± 2.84 mV; n = 9; Fig. 1C). Following
the bAHP, the amplitude of both fast and slow AHPs partially recovered
by the first spike of the next burst, allowing the process to begin
again (Fig. 1, A and C).
Factors determining burst threshold
The ability for pyramidal cells to generate spike bursts was assessed by constructing frequency/current (F/I) plots of spike discharge (n = 70). Since we found that the pattern and frequency of discharge was entirely equivalent in somatic and dendritic recordings, these numbers have been pooled in the following analyses. Near spike threshold, pyramidal cells responded to current injection with a relatively tonic pattern of single spike discharge (Fig. 2A). As depolarizing current was increased (0.1-1.2 nA), 61% of recordings shifted from tonic to burst discharge, revealing an intensity-dependent shift in the pattern of spike output (n = 43/70). Although some cells exhibited burst discharge on initial spike generation, there was typically a threshold for the transition to burst discharge corresponding to ~70% of the maximum spike frequency attained on F/I plots (Fig. 2B). Burst frequency at this threshold was consistently between 10-60 Hz (34 ± 10.8 Hz; n = 28) and increased to ~50-130 Hz (76 ± 35.3 Hz; n = 28) at the maximum range of F/I plots (Fig. 2B). By comparison, spike frequency (not including the spike doublet) ranged between 25-60 Hz at spike threshold and 123-475 Hz at the highest levels of current injection. An interval histogram of the average interspike interval (ISI) at burst threshold revealed that bursts were evoked when the ISI fell within ~3-7 ms (5.5 ± 1.38 ms; n = 34; Fig. 2C). These studies indicate that pyramidal cell output shifts to an oscillatory series of spike bursts once spike frequency exceeds ~140 Hz.
|
Changes in somatic spike afterpotentials during repetitive discharge
Given the clear change in spike afterpotentials during repetitive discharge (Fig. 1C), we examined more closely the underlying basis for the change in AHP and DAP amplitudes that lead to the generation of a spike doublet.
AHPS.
Superimposition of the DAP on the somatic AHPs prevented us from
quantifying changes in the AHP in intact cells. We therefore sought a
method to "isolate" somatic AHPs from the DAP. We have previously
shown that the somatic DAP can be reduced by focally ejecting TTX in
the mid-dendritic region (Turner et al. 1994). However,
these earlier experiments did not attempt to block discharge of all
dendritic Na+ spikes, including those in the most
proximal dendrites. We therefore expanded on this approach to block all
spike backpropagation to examine somatic AHPs in the absence of a DAP.
To ensure that all dendritic activity was blocked, we continued TTX
ejections until TTX had diffused to the cell layer and affected the
somatic spike. By restricting our measurements to those records
obtained immediately prior to this time, we were able to examine
somatic AHPs in relative isolation.
|
DAP. Based on the mechanism illustrated in Fig. 1B, one can predict that an increase in the degree of dendritic depolarization during a burst will enhance DAP amplitude at the soma. We therefore compared somatic and dendritic spikes during burst discharge. This revealed no significant decrease in somatic spike amplitude (92 ± 7.9% of control; n = 9) and no change in the rate of somatic spike repolarization or duration over the time course of a single burst (Fig. 4, A and C). In contrast, dendritic spikes exhibited a significant decrease in amplitude during a burst (65 ± 11.7% of control; n = 9; P < 0.001, Student's t-test), as well as a substantial change in the rate of repolarization and duration (Fig. 4B). By the third spike of a burst, the rate of repolarization of dendritic spikes could decrease to 66% of control values while spike duration increased up to 135% (Fig. 4, C and D).
|
|
|
Spike doublets and burst termination
Since spike discharge is grouped into bursts by the recurring
generation of a spike doublet and bAHP, the spike doublet may act to
inhibit the soma-dendritic process driving spike discharge. As
indicated above, the second spike of the doublet in dendrites was only
a small partial spike, suggesting a failure of dendritic spike
invasion. In this regard, Turner et al. (1994) reported different refractory periods for somatic and dendritic spikes in ELL
pyramidal cells. This was confirmed in the present study, although we
also noted a wide range of refractory periods for dendritic recordings.
Figure 7A provides
representative recordings of spike discharge at the soma over the
distance spikes backpropagate in pyramidal cell dendrites. This
illustrates a decrease in spike amplitude and an increase in spike
duration with distance from the soma that could alter refractory
period. Indeed, a plot of relative refractory period measured at all
locations revealed progressively longer refractory periods as spike
width increased from ~2 ms (somatic recordings) to 6.8 ms (distal
dendritic recording) (Fig. 7B; n = 27). This
would help account for why dendritic spike amplitude decreased
gradually as antidromic C-T intervals were reduced below ~4.5 ms
(Fig. 5D). Spike discharge in the more distal dendritic
sites that are associated with a long refractory period would be
blocked by relatively long C-T intervals compared with proximal
dendritic sites with shorter refractory periods. The progressive block
of distal dendritic sites would thus remove a percentage of the
dendritic depolarization recorded at more proximal sites.
|
The average value for relative refractory period at the soma was
2.6 ± 0.64 ms (n = 12). By defining the dendritic
refractory period as the C-T interval that evoked the smallest partial
spike in recordings 100 µm from the soma, we arrived at an average value of 4.5 ± 0.93 ms (n = 14). By comparison,
the average ISI for spike doublets measured at the soma was 3.9 ± 1.08 ms (n = 12). These average values are consistent
with the interval of the spike doublet falling inside the dendritic
refractory period. However, given the increase in refractory period
over the cell axis we specifically tested the relationship between
spike doublet interval and spike refractory period in a set of
individual recordings. These experiments compared the spike doublet ISI
recorded during current-evoked burst discharge to the refractory period
measured with antidromic stimulus pairs. This revealed that the ISI for spike doublets at the soma was consistently longer than the measured somatic spike refractory period in a given cell (difference of 0.96 ± 0.78 ms; n = 9; P < 0.01;
Wilcoxon Signed Ranks test). As a result, the somatic spike doublet was
made up of two full amplitude action potentials (Fig.
8, A and B). In
contrast, the spike doublet ISI measured in dendritic recordings
always fell within the range of antidromic C-T intervals that reduced
dendritic spike amplitude (Fig. 8, C and D;
difference of 0.4 ± 0.55 ms; P < 0.05; Wilcoxon
Signed Ranks test; n = 8), suggesting a loss of spike
backpropagation in some portion of the dendritic tree.
|
We tested whether a spike doublet could terminate a burst by simulating a spike doublet during an antidromic stimulus train (Fig. 9). This was accomplished by inserting an additional stimulus at varying latencies within an antidromic stimulus train of fixed ISI. In dendrites we found that the addition of an antidromic stimulus at latencies close to the dendritic refractory period produced only a small partial spike on the falling phase of the previous spike (Fig. 9A). This blocked the discharge of at least one subsequent dendritic spike of the background stimulus train, presumably through the actions of a bAHP. In contrast, an additional stimulus inserted at longer latencies evoked nearly-full sized dendritic spikes and no subsequent inhibition of spike discharge (data not shown). In somatic recordings, the insertion of an additional antidromic stimulus at effective C-T intervals of ~3-5 ms evoked a spike doublet comprised of two full spikes followed by inhibition of subsequent spike discharge (Fig. 9B). The use of C-T intervals near the somatic spike refractory period resulted in failure of the imposed spike but no interruption of spike discharge in response to subsequent stimuli in the train. These experiments are important in identifying the dendritic spike refractory period as a key factor controlling spike backpropagation at the end of a burst.
|
Mechanism of burst generation
Figure 10 illustrates our working hypothesis of how spike frequency establishes a process of conditional backpropagation to generate burst discharge. During repetitive spike discharge at ISIs greater than ~8 ms, there is a faithful backpropagation of each spike from soma to dendrite. Spike ISIs of ~3-7 ms promote a selective broadening of dendritic spikes, allowing dendritic spikes to summate and exert a progressively greater influence on the soma as a DAP. As a result, the DAP increases in amplitude during a spike burst, ultimately reaching threshold to trigger a spike doublet at the soma. The spike doublet ISI falls outside of the somatic refractory period, allowing two full action potentials to be generated. The first of these spikes backpropagates into dendrites, but the second fails to backpropagate as it falls within the refractory period of dendritic spike discharge. The dendritic depolarization is thus abruptly removed from one spike to the next, allowing the cell to repolarize and terminate the burst. Repetition of this process acts to generate a rhythmic series of spike bursts separated by bAHPs.
|
![]() |
DISCUSSION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
This study describes a novel mechanism by which the success of dendritic spike backpropagation can generate oscillatory spike bursts in a central neuron. Specifically, we have shown that spike frequency imparts a process of conditional backpropagation which alternately increases or removes the influence of backpropagating spikes on somatic membrane potential to produce an oscillatory burst output.
Frequency-dependent spike broadening
A critical feature of this process is a frequency-dependent
potentiation of the DAP. This was associated with a change in the rate
of repolarization, duration, and temporal summation of dendritic
spikes. Frequency-dependent broadening and summation of dendritic
spikes is maximal for spike frequencies between 140-330 Hz (3-7 ms
ISI), and controls burst threshold by regulating the transition from
tonic to burst output (Fig. 2). The exact spike frequency capable of
reaching this transition point depends on the number of spikes invoking
dendritic spike summation (Fig. 6, C and D). As a
result, the immediate history of spike discharge can influence burst
threshold. We would thus expect burst output to begin at even lower
spike frequencies during the spontaneous activity ordinarily found in
vivo (Gabbiani et al. 1996; Metzner et al.
1998
). This is supported by the finding that spike bursts can
be recorded in vivo without sensory stimulation at spontaneous discharge frequencies of ~50 Hz (Metzner et al. 1998
).
Frequency-dependent spike broadening has previously been reported in
cell bodies, axons, and presynaptic terminals (Aldrich et al.
1979; Bourque 1991
; Jackson et al.
1991
; Ma and Koester 1995
; Mathes et al.
1997
; Quattrocki et al. 1994
; Shao et al. 1999
). This can involve a cumulative inactivation of A-type,
delayed rectifying, or Ca2+-activated
K+ currents underlying spike repolarization,
often with an associated increase of Ca2+ current
with each Na+ spike (Aldrich et al.
1979
; Borst and Sakmann 1999
; Bourque
1991
; Jackson et al. 1991
; Klemic et al.
1998
; Ma and Koester 1995
; Quattrocki et
al. 1994
; Shao et al. 1999
). Backpropagating
Na+ spikes can also activate dendritic
Ca2+ currents in a manner dependent on spike
frequency and conduction along the dendritic axis (Callaway and
Ross 1995
; Larkum et al. 1996
; Spruston
et al. 1995
).
The selective change in dendritic spike repolarization found in ELL
pyramidal cells during repetitive discharge may be due to differences
in the density or subtype of K+ channels in
somatic versus dendritic regions, or their inactivation by the very
different action potential waveforms in these two compartments. We can
state that Ca2+-dependent currents do not appear
to contribute to either spike broadening or the depolarizing shift in
pyramidal cell dendrites, as 400 µM Cd2+
ejections in the dendritic region had no effect on these potentials (n = 3). Dendritic spike broadening thus results
primarily from actions on TTX-sensitive dendritic
Na+ spikes and
non-Ca2+-dependent repolarizing
K+ currents. The most likely factor we have
identified thus far is a gymnotid homologue of the mammalian Kv3.3
K+ channel which is found throughout the
dendritic arbor of pyramidal cells (Lemon et al. 1998;
Morales et al. 1998
; Rashid and Dunn 1998a
,b
). Given the established role for Kv3 channels in spike repolarization (Rudy et al. 1999
), it will be important
to determine the ability for Kv3 channels to contribute to dendritic
spike broadening.
Burst termination
The generation of a somatic spike doublet provides an intrinsic
mechanism to pattern spike discharge into bursts when the dendritic
refractory period blocks spike backpropagation. A failure of
backpropagating spikes to invade distal dendritic branches has been
observed in other cells when repetitive discharge dramatically reduces
dendritic spike amplitude (Colbert et al. 1997;
Jung et al. 1997
; Spruston et al. 1995
).
Synaptic inhibition can also regulate the degree of spike
backpropagation (Buzsáki et al. 1996
;
Tsubokawa and Ross 1996
). However, the mechanism we
describe here differs substantially from these cases in representing an all-or-none failure of dendritic spike invasion at the level of proximal apical dendrites from one spike to the next. This conditional backpropagation is thus a novel soma-dendritic interaction, and the
first consideration of the role for refractory periods in regulating
dendritic spike discharge.
The exact location of dendritic spike failure cannot be identified.
However, we have recorded dendritic spike failure in locations as close
as ~50 µm from the cell layer, suggesting that dendritic refractory
period increases at a minimal distance from the soma. The difference in
somatic and dendritic spike refractory periods may result from a
comparatively low density of Na+ and
K+ channels in dendritic regions. The refractory
period of spike discharge has in fact been shown to depend on
K+ conductances involved in spike repolarization
(Massengill et al. 1997). This is further supported by
the correlation between spike width and refractory period shown in Fig.
7. Failure of spike backpropagation would thus become more likely if
cumulative inactivation of dendritic K+ channels
during a burst increased dendritic refractory period with respect to
the spike doublet ISI. The failure of backpropagation would be further
assured by a cumulative inactivation of dendritic Na+ channels (Colbert et al. 1997
;
Jung et al. 1997
; Mickus et al. 1999
), as
suggested by the decline in dendritic spike amplitude during spike bursts.
The spike doublet was always followed by a bAHP that temporarily hyperpolarized the membrane after a spike burst. We have not fully determined the ionic mechanisms underlying this potential but find that the bAHP can be partially reduced by focal ejections of Cd2+ (400 µM) in the somatic (n = 4) but not dendritic region (n = 3). This would be consistent with an elevated Ca2+ entry on generation of a high-frequency spike doublet that is restricted to the somatic compartment. We are uncertain as to whether such a Ca2+ current accumulates during the spike train, or is generated only by the spike doublet. In either case, the magnitude of the bAHP will be more apparent upon the sudden loss of the DAP when spike backpropagation fails upon the second spike of the doublet. Given the potential role of the bAHP in regulating burst frequency, the ionic basis and distribution of channels underlying this response warrants further investigation.
Functional significance
Many neurons initiate Na+ spike discharge
near the soma which is followed by spike backpropagation over extensive
regions of the dendritic arbor (Chen et al. 1997;
Johnston et al. 1996
; Larkum et al. 1996
;
Spruston et al. 1995
; Stuart and Sakmann
1994
; Stuart et al. 1997
; Turner et al.
1991
). The potential for dendritic spikes to directly
depolarize somatic membrane during backpropagation was considered in
some of the earliest reports of intracellular recordings (Calvin
and Hartline 1977
; Calvin and Sypert 1976
; Granit et al. 1963
; Nelson and Burke
1967
). A relationship between backpropagating spikes and a
somatic DAP has also been shown in hippocampus and cortex, although
this differs from the mechanism we describe here by incorporating
dendritic Ca2+ current or a somatic
INaP (Golding et al.
1999
; Larkum et al. 1999
; Magee and
Carruth 1999
; Mainen and Sejnowski 1996
;
Wang 1999
; Yuste et al. 1994
;
Zhang et al. 1993
). Spike doublets have been reported in
bursting cells ranging from receptors to cortical pyramidal cells
(Calvin and Hartline 1977
; Calvin and Sypert
1976
; Granit et al. 1963
; Koch
1999
; Lo et al. 1998
; Nelson and Burke 1967
; Paré et al. 1995
; Steriade et
al. 1993
). Spike doublets have been shown to contribute to
synchronizing neuronal discharge, but no specific role in burst
formation has been identified (Ermentrout and Kopell
1998
; Koch 1999
; Whittington et al.
1997
). The conditional backpropagation described in the present
study may then contribute to the generation of burst output in other
cell types.
Spike bursts in cortical cells have become recognized for their role in
encoding stimulus location, orientation, and spatial frequency
selectivity (Gray and McCormick 1996; Lisman
1997
). Similarly, burst discharge in ELL pyramidal cells has a
direct role in feature extraction in vivo (Gabbiani et al.
1996
; Metzner et al. 1998
). Pyramidal cell spike
bursts in vivo occur with interspike intervals in the range of 8-25 ms
and a lower limit of ~4-5 ms (Metzner et al. 1998
; J. Bastian, unpublished data). Close examination of published records in
Metzner et al. (1998)
would indicate that not all
pyramidal cell spike bursts in vivo show the progressive increase in
intraburst spike frequency we found in vitro. Similarly, unit
discharges resembling spike doublets are apparent in in vivo recordings, although not all bursts recorded end with a spike doublet.
Therefore conditional backpropagation is not necessary to account for
all burst discharge of pyramidal cells in vivo. This would be expected
for short-lasting depolarizations that do not sufficiently drive
pyramidal cells to incorporate the entire process of dendritic spike
broadening and conditional backpropagation. Rather, conditional
backpropagation likely represents one of several factors controlling
burst output during electrosensory processing.
![]() |
ACKNOWLEDGMENTS |
---|
This research was supported by the Canada Medical Research Council and the Alberta Heritage Foundation for Medical Research. N. Lemon was supported by a MRC Studentship, and R. W. Turner is an AHFMR Senior Scholar.
![]() |
FOOTNOTES |
---|
Address for reprint requests: R. W. Turner, Neuroscience Research Group, University of Calgary, 3330 Hospital Dr. N.W., Calgary, Alberta T2N 4N1, Canada (E-mail: rwturner{at}ucalgary.ca).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Received 17 February 2000; accepted in final form 16 May 2000.
![]() |
REFERENCES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|