1Department of Psychology, University of Colorado, Boulder 80309-0345; and 2Department of Anatomy and Neurobiology, Colorado State University, Fort Collins, Colorado 80523
![]() |
ABSTRACT |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Jones, Michael S., Kurt D. MacDonald, ByungJu Choi, F. Edward Dudek, and Daniel S. Barth. Intracellular Correlates of Fast (>200 Hz) Electrical Oscillations in Rat Somatosensory Cortex. J. Neurophysiol. 84: 1505-1518, 2000. Oscillatory activity in excess of several hundred hertz has been observed in somatosensory evoked potentials (SEP) recorded in both humans and animals and is attracting increasing interest regarding its role in brain function. Currently, however, little is known about the cellular events underlying these oscillations. The present study employed simultaneous in-vivo intracellular and epipial field-potential recording to investigate the cellular correlates of fast oscillations in rat somatosensory cortex evoked by vibrissa stimulation. Two distinct types of fast oscillations were observed, here termed "fast oscillations" (FO) (200-400 Hz) and "very fast oscillations" (VFO) (400-600 Hz). FO coincided with the earliest slow-wave components of the SEP whereas VFO typically were later and of smaller amplitude. Regular spiking (RS) cells exhibited vibrissa-evoked responses associated with one or both types of fast oscillations and consisted of combinations of spike and/or subthreshold events that, when superimposed across trials, clustered at latencies separated by successive cycles of FO or VFO activity, or a combination of both. Fast spiking (FS) cells responded to vibrissae stimulation with bursts of action potentials that closely approximated the periodicity of the surface VFO. No cells were encountered that produced action potential bursts related to FO activity in an analogous fashion. We propose that fast oscillations define preferred latencies for action potential generation in cortical RS cells, with VFO generated by inhibitory interneurons and FO reflecting both sequential and recurrent activity of stations in the cortical lamina.
![]() |
INTRODUCTION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Transient somatosensory stimuli
produce a series of stereotyped cellular events within the cerebral
cortex that are reflected in field-potential recordings as the
somatosensory evoked potential (SEP) complex. The SEP waveform consists
of a surface positive/negative waveform (P1/N1) beginning 10-15 ms
post-stimulus followed by a slower biphasic waveform (P2/N2) of
200-300 ms duration. Because this waveform exhibits a similar
morphology across a variety of species (Allison and Hume
1981) and may be recorded both intra- and extracranially, the
SEP is one of few signals that may be used to relate noninvasive
studies of somatosensory function in humans to detailed information of
the underlying physiological processes obtained from animal models. In
addition, intracranial measurements using multichannel electrode arrays
placed directly on the cortex reveal that the SEP is a focal response
with a spatial organization that closely reflects the organization of
the somatosensory forebrain (Di and Barth 1991
;
Di et al. 1989
, 1994
) and thus provides a means of
studying rapid spatiotemporal interactions that occur among large cell
populations during cortical information processing (Jones and
Barth 1997
).
Although historically considered a slow wave phenomenon, the SEP
exhibits high-frequency (>200 Hz) content that is receiving increasing
interest as to its role in cortical information processing. This
activity is often apparent as a series of small amplitude deflections
superimposed on the underlying slow-wave components of the evoked
potential complex that, following high-pass filtering, are seen to
correspond to a burst of high-frequency field potential oscillations.
These fast oscillations have been observed extracranially in humans
(Curio et al. 1994a,b
, 1997
; Eisen et al.
1984
; Emori et al. 1991
; Gobbelé et
al. 1998
; Green et al. 1986
; Hashimoto et
al. 1996
; Maccabee et al. 1983
; Yamada et
al. 1984
, 1988
) as well as in intracranial recordings in the
awake and in the ketamine-anesthetized rat (Jones and Barth
1999b
; Kandel and Buzsaki 1997
).
Using high-resolution epipial mapping of the vibrissa-evoked response
in rat vibrissa/barrel cortex, our laboratory determined that fast
oscillations triggered by transient displacement of individual
vibrissae are somatotopically organized, propagate rapidly within
somatosensory cortex, and interact in a manner that suggests that they
play a role in precisely timing the arrival of sensory afferent
information (Jones and Barth 1999b). Laminar recordings
demonstrate that vibrissa-evoked fast oscillations exhibit a dipolar
pattern of extracellular field potentials oriented perpendicular to the
cortical surface and extending throughout the cortical layers
(Jones and Barth 1999b
; Kandel and Buzsaki 1997
). These high-frequency dendritic currents have been shown to influence the timing of action potentials at least in the
infragranular layers, where multi-unit activity in layer V has been
correlated with the periodicity of the fast oscillatory response
(Kandel and Buzsaki 1997
). Recent in-vitro results
suggest that small but rapid fluctuations of intracellular potential
may serve as a powerful mechanism for controlling the timing of action
potential generation in individual cells with a resolution in the
sub-millisecond range (Mainen and Sejnowski 1995
). Thus,
sensory-evoked fast oscillations may mediate the precise timing of
interactions between adjacent neural circuits in somatosensory cortex,
providing a mechanism for discriminating slight latency differences in
the activation of the somatotopic cortical map established by rapid
movement of an object across the receptive surface. However, little is known about the neural generators or functional significance of these
sensory-evoked fast oscillations at the cellular level.
In the present study, the cellular correlates of fast oscillations in
rat vibrissa/barrel cortex were investigated in detail with combined
intracellular and epipial field-potential recording in the intact,
ketamine-anesthetized animal. Intracellular responses in rat
somatosensory cortex evoked by transient vibrissae stimulation were
recorded while field potentials at the cortical surface were simultaneously monitored. This allowed comparison of the cellular response with fast oscillatory components of the SEP extracted from the
surface record. Our specific objectives were to determine 1)
the types and laminar distribution of cortical cells participating in
the oscillatory response, 2) whether extracellular fast
oscillations evident within apical pyramidal dendrites are associated
with high-frequency intracellular currents recordable at the cell soma and if these rapid transients might therefore serve as a mechanism for
precisely controlling spike timing, and 3) whether there are distinct subpopulations of cells producing sensory-evoked trains of
action potentials closely associated in frequency, phase, and post-stimulus latency to fast oscillations. Preliminary results of this
study were reported in abstract form (Jones and Barth 1999a).
![]() |
METHODS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Surgical preparation
All procedures were performed in accordance with University of Colorado Institutional Animal Care and Use Committee guidelines for the humane use of laboratory animals in biological research. Male Sprague-Dawley rats (250-350 g) were anesthetized to surgical levels using intramuscular injections of ketamine HCl (100 mg/kg) and xylazine (25 mg/kg) and were secured in a stereotaxic frame. A unilateral craniotomy extending from bregma to lambda and lateral to the temporal bone was performed over the right hemisphere, exposing a wide region of parietotemporal cortex where the dura was reflected. The exposed cortical surface was regularly bathed with physiological saline and body temperature was maintained with a regulated heating pad. Additional ketamine and xylazine were administered as required to maintain a level of anesthesia such that the corneal reflex could barely be elicited. Animals were killed by anesthesia overdose, without regaining consciousness, at the conclusion of the experiment.
Stimulation
The 25 large facial vibrissae of the contralateral (left) mystacial pad were tied together and displaced simultaneously at ~1 cm from their base with a laboratory-built stimulator that converted computer-generated square-wave pulses (duration 0.3 ms) into silent dorsolateral displacements (~0.5 mm) of the stimulator arm.
Surface and laminar recording
Epipial maps of the vibrissa-evoked SEP complex were recorded
with a flat multi-channel electrode array consisting of 64 silver wires
arranged in an 8 × 8 grid (tip diameter 100 µm, inter-electrode spacing 500 µm) covering a 3.5 × 3.5 mm area of the cortical
surface in a single placement (see Fig.
1). Laminar recordings were performed with a linear array of 24 platinum electrodes (20 µm diameter, 125 µm spacing) (EEG KFT, Budapest, Hungary) inserted perpendicular to
the cortical surface in the approximate center of the barrel field
until the top electrode was barely visible at the cortical surface.
Laminar potentials were preamplified with a unity gain buffer (2 fA
input bias current). Both laminar and epipial potentials were
referenced to a silver ball electrode secured over the contralateral frontal bone and were simultaneously amplified (×10,000), analog filtered (band-pass cutoff 6 dB at 0.001-3000 Hz, roll-off 5 dB/octave), and digitized at 10 KHz. Surface and laminar potentials were recorded following 50-100 presentations of vibrissae stimulation, with data from individual trials stored digitally for subsequent analysis. All filtering used in the analysis was performed digitally and employed a zero-phase shift algorithm that processed the input data
in both the forward and backward directions to prevent the introduction
of systematic phase distortion. Band-pass filtering was implemented
using a second-order Butterworth filter exhibiting a maximally flat
passband to avoid frequency-dependent amplitude distortion. The
signal-to-noise ratio of the field potential data was ~20, estimated
by comparing the variance of evoked oscillatory potentials with that of
the prestimulus baseline.
|
Intracellular recording
Intracellular recording was performed using glass micropipettes
pulled from thin-walled aluminosilicate glass on a Flaming-Brown micropipette puller (Sutter Instruments, model P-87) and filled with
K+-acetate (3 M). In some instances,
micropipettes were beveled to improve penetration into smaller cells.
In-vivo impedance of electrodes ranged from 80 to120 mohm. Recording
and current injection was performed with an Axoclamp 2-A amplifier
(Axon Instruments) equipped with a 0.1 gain headstage (Axon
Instruments, model HS-2A). Exposed cortex in the center of the
vibrissa/barrel region was stabilized with a small Kevlar plate
containing a small (1 mm) access hole, through which the microelectrode
was inserted, that was brought into contact with the pia. A silver wire
was permanently mounted at the immediate periphery of the access hole
for simultaneous monitoring of the surface SEP. Micropipettes were
advanced perpendicularly into the cortex in 0.5-µm increments using a
piezo translator equipped with a compensating motor drive
(Märzhäuser PM-10) and micrometer that indicated the depth
of the electrode tip. Criteria for an acceptable cell impalement were a
resting membrane potential of at least 60 mV with overshooting action
potentials and satisfactory characterization of the cell using
depolarizing current pulses as described by McCormick et al.
(1985)
. Intrinsic characterization was also used to infer the
cell morphology, with regular spiking (RS) and intrinsic bursting (IB)
cells presumed to be pyramidal or stellate in morphology and fast
spiking (FS) cells assumed to be aspiny inhibitory interneurons
(Connors and Gutnick 1990
; McCormick et al.
1985
). When a stable cell impalement was obtained, 50-100
200-ms trials were recorded during vibrissa stimulation (100 ms
baseline + 100 ms post-stimulus). Intracellular and surface records
were low-pass filtered at 3 KHz (
6 dB at 3 KHz, roll-off 5 dB/octave), digitized at 10 KHz, and stored for subsequent analysis.
Numerical results are presented as mean ± SE.
![]() |
RESULTS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Surface and laminar extracellular recording
Figure 1 illustrates a typical cortical response recorded using
the 64-channel surface array following transient displacement of the
vibrissae. The averaged evoked-response recorded with the array
consisted of a stereotyped biphasic waveform (Fig. 1B;
n = 50), the morphology and distribution of which was
similar to that obtained in previous studies (Jones and Barth
1997, 1999b
). The SEP was characterized by an early
positive/negative response beginning 10-15 ms after stimulus onset
(Fig. 1C1); the peaks of these events have been designated
P1 and N1, respectively, to indicate their polarity and sequence of
occurrence. In addition to these slow waves (here designated SW),
higher-frequency content was evident as a series of small deflections
superimposed on the underlying SW (Fig. 1C1, arrows).
Following high-pass filtering (>200 Hz), these deflections appeared as
a burst of high-frequency oscillations accompanying the SW (Fig.
1C2; note difference in amplitude scale). Spectral analysis
of the high-pass filtered data consistently demonstrated two discrete
spectral peaks (Fig. 1C, left). Averaged spectra
suggested that the high-frequency activity accompanying the SW be
segregated into two frequency bands, the first including activity
between 200 and 400 Hz (here designated as fast oscillations or FO) and
a second frequency band including activity between 400 and 600 Hz (here
designated as very fast oscillations or VFO). The broadness of these
frequency bands reflects the variability of the fast oscillatory
activity observed across animals and, to a lesser extent, within a
given animal over the recording session. Within a given set of trials, the activity in these bands was almost sinusoidal, as demonstrated by
filtering the SW for these two frequencies as shown in Fig. 1,
C3 and C4. The relative amplitude and latency
demonstrated by fast oscillations in this example was typical, with FO
usually larger in amplitude and earlier in latency than the
accompanying VFO activity. Although this trend repeatedly appeared in
the field potential data, oscillatory bursts were somewhat variable in
latency, duration, and amplitude across trials and between animals.
The SEP complex exhibited considerable variability in post-stimulus latency across trials, as revealed by superimposing the SW responses (Fig. 2A, left). This variability made the construction of averages of fast oscillatory activity difficult because it too exhibited intertrial latency variability (Fig. 2A, right), which resulted in the attenuation of stimulus-locked averages (Fig. 2B, right). However, fast oscillations exhibited a consistent latency relationship to the SW, as demonstrated by using the latency of the P1 peak to align the oscillatory activity across trials (Fig. 2C; subsequently referred to as "P1-aligned"). This resulted in a marked improvement of time-locked averages of FO and VFO activity compared with that obtained using the stimulus onset (compare Fig. 2, B and D). As such, P1 alignment was subsequently applied to all extracellular and intracellular responses included in the study.
|
Figure 3 depicts P1-aligned responses in
SW, FO, and VFO frequency bands recorded at the 24 electrodes of the
laminar array in six animals (Fig. 3, A, B, and
C, respectively; left, individual animal
averages; right, grand average across all animals). Only the
top 18 electrodes of the array were located within cortical lamina. The
surface SW was identifiable at the most superficial electrode (Fig.
3A) and exhibited a complex pattern of polarity reversal in
the depth, which was in agreement with previous laminar studies of
sensory cortex (Abbes et al. 1991; Bode-Greuel et
al. 1987
; Di et al. 1990
; Sukov and Barth
1998
). Depth profiles of fast oscillatory activity were
obtained by filtering the wideband laminar data on a
channel-per-channel basis. Although there was substantial overlap of
the two types of oscillations when results were superimposed across
animals (Fig. 3, B and C, left), the grand average confirmed a general trend of earlier onset of FO activity
in the response (Fig. 3, B and C, right). Both FO
and VFO exhibited the same polarity reversal across all of their
respective oscillatory peaks, with FO reversal occurring at ~900 µm
and VFO at ~750 µm (R in Fig. 3, B and C).
The depth reversal of FO and VFO indicates that these oscillations were
not volume-conducted from extracortical structures but rather that the
activity was generated locally within cortical cell populations. The
laminar results furthermore indicate that fast oscillations primarily reflect activity within infragranular cells because the depth at which
polarity reversal occurs is necessarily located superficial to the
somata of cells participating in the response.
|
Intracellular recording
Stable recordings were obtained from 67 cells in 13 animals. Four
cells were classified as fast spiking (FS), five as intrinsically bursting (IB), and the remainder as regular spiking (RS). RS cells exhibited long-lasting (>0.5 ms) spikes and strong spike-frequency adaptation in response to injection of depolarizing current pulses (0.5-1.0 nA; 100 ms). IB cells were distinguished by their response to
depolarizing current injections with bursts consisting of two or more
action potentials, with spike amplitude within a burst usually
successively decreased. Action potentials in IB cells were otherwise
indistinguishable from those of RS cells and, in some instances,
injection of strong levels of depolarizing current often evoked a
constant firing pattern similar to that of an RS cell. FS cells
exhibited short-duration (<0.5 ms) action potentials caused by a rapid
rate of repolarization. The response of these cells to depolarizing
current injection exhibited a wide dynamic range and little or no spike
frequency adaptation, with strong stimulation evoking a sustained
firing rate in excess of 500 Hz. Representative examples of the
responses of the three cell classes to depolarizing current injection
are shown in Fig. 4,
A-C. Only IB cells exhibited a preferred laminar
distribution and were found exclusively in layer V, whereas RS and FS
cells were uniformly distributed in the cortical lamina (Fig.
4D). The sample size of IB cells (n = 5) may
have been a factor in their limited distribution as these cells have
been observed in more superficial lamina in other studies (see, e.g.,
Contreras et al. 1997).
|
Excitatory postsynaptic potential (EPSP) onset latency varied
according to cell location within the cortical lamina (Fig. 5). Earliest post-stimulus latencies were
observed in middle cortical layers, followed by cells in superficial
layers, and then in deep lamina. Organized into granular,
supragranular, and infragranular responses based on the approximate
laminar extent of these cytoarchitectural regions (Fig. 5, g, s, and i,
respectively), EPSP latency could be ranked as follows: granular cells,
5.1 ± 0.12 ms; supragranular cells, 6.0 ± 0.20 ms;
infragranular cells, 7.6 ± 0.28 ms. These latencies are
consistent with the established functional anatomy of the SEP, in which
the termination of thalamocortical afferents initiates excitation in
middle cortical lamina, which then passes to cells in supragranular and
then infragranular layers (Abbes et al. 1991; Di
et al. 1990
). These results are also in agreement with response
latencies obtained in previous in-vivo intracellular (Carvell
and Simons 1988
; Moore and Nelson 1998
;
Zhu and Connors 1999
) and extracellular
(Armstrong-James et al. 1992
; Simons
1978
) unit studies of barrel cortex.
|
Spike response
RS CELLS. Almost all RS cells responded to vibrissae stimulation in at least some trials with one or more action potentials. In nine cells, the vibrissa-evoked response remained subthreshold regardless of stimulus amplitude; these cells were not included in the analysis. Similar to FO and VFO activity (Fig. 2), the use of P1 alignment greatly reduced the latency variability of the intracellular data and, furthermore, often revealed FO or VFO periodicity in the response of a given cell. An example of the emergence of FO rhythmicity in the intracellular response is shown in Fig. 6. Although in most trials the cell response consisted of a double spike (Fig. 6A, inset), this was not reflected in the post-stimulus time histogram, which was not strongly bimodal. However, when spike latencies were computed relative to the P1 peak of the accompanying surface response in each trial rather than to the stimulus onset, the cell response was more faithfully reproduced in the resulting histogram, which consisted of two large spike clusters (Fig. 6B). When superimposed over the averaged epipial FO from these trials (Fig. 6B, gray trace), the histogram spike clusters are seen to be separated by one period of the surface FO activity.
|
|
|
|
|
FS CELLS. FS cells exhibited a unique response to vibrissae stimulation (Fig. 11) that consisted of a burst of 2-5 action potentials with an average interspike frequency (508 ± 6.7 Hz) similar to the average frequency of surface VFO (515 ± 2.2 Hz). The latency of FS spike-bursts coincided with that of VFO activity, and action potentials within a given burst tended to align with peaks of the surface response (Fig. 11, gray bars). In two of the four cells, phase alignment of the spike and surface responses was remarkably consistent across all spikes in the burst (Fig. 11, A and B). The response in the remaining two FS cells exhibited a similar phase relationship to the surface activity but also included some spikes early or late in the response that were not aligned with VFO peaks (Fig. 11, C and D).
|
IB CELLS. The evoked response of intrinsically bursting cells was qualitatively similar to that of FS cells in that it consisted of multi-spike bursts of two or more action potentials (Fig. 12). The response in three of the five IB cells was highly variable (Fig. 12, C-E, gray traces) and, in contrast with the FO- or VFO-related variability demonstrated by RS cells (cf. Figs. 7-9), was not coordinated with surface oscillatory activity. The interspike frequency of the evoked bursts in IB cells was 372 ± 2.0 Hz, which was approximately midway between that of FO and VFO, and a comparison of intracellular responses with the accompanying surface data demonstrated that spikes in the response were not aligned with a consistent phase of either type of fast oscillation (Fig. 12, gray bars). Thus the IB response appeared to be unrelated to the fast oscillations present in the surface SEP complex.
|
![]() |
DISCUSSION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
These results demonstrate that short-latency extracellular slow waves of the SEP complex recorded at the cortical surface are accompanied by two distinct oscillatory bursts with center frequencies of approximately 300 Hz (FO) and 500 Hz (VFO), respectively. Laminar extracellular potentials associated with both FO and VFO reverse polarity, indicating that they are generated by coherent currents within the parallel apical dendrites of cortical pyramidal cells. Intracellular recordings reveal that these high-frequency dendritic currents are sufficient to produce fast potentials at the soma, serving as a mechanism to precisely control the timing of action potentials in RS cells. FS cells produce distinct bursts of action potentials in response to sensory stimulation that are closely associated in frequency, phase, and post-stimulus latency to VFO. No specific cell population could be identified that exhibited an analogous relationship with FO activity, which suggests that this response may be produced by local circuit interactions.
Physiological interpretation of the evoked-potential complex is based
on several key assumptions regarding the neural generation of
extracellular field potentials in laminar cortex (Creutzfeldt et
al. 1966; Mitzdorf 1985
). The optimal cellular
geometry for the production of field potentials that can be recorded
either extracranially, or intracranially on the cortical surface and across the laminae as considered in the present study, is that of a
long, vertically oriented dendritic process with active synaptic sites
of limited spatial extent. In populations of cells, when such processes
are aligned in parallel and are synchronously activated at
approximately the same location along their lengths, the resulting summed extracellular potentials approximate an open field geometry that
can be recorded at a distance. The radially and symmetrically organized
dendritic processes of most interneurons produce a closed potential
field that cannot be recorded with large extracellular surface and
laminar electrodes (Llinas and Nicholson 1976
). However, the activity of interneurons may be recorded indirectly via their postsynaptic contact with the apical dendrites of pyramidal cells. The
open field of synchronized potentials in large populations of apical
dendrites typically conforms to that expected from an equivalent
current dipole the polarity of which is defined by complementary
source/sink pairs that reflect the locations of underlying
transmembrane currents (Mitzdorf 1985
). The
extracellular slow waves and fast oscillations recorded in the present
study must therefore reflect highly synchronized synaptic input
impinging on the apical dendrites of populations of cortical pyramidal cells.
Laminar recording indicates that the P1 and N1 slow waves of the SEP
are produced by the activation of a vertically oriented current dipole
in the supragranular layers followed by the activation of a dipole
extending from the cortical surface through the infragranular layers
(Fig. 3). This finding replicates previous work that demonstrated that
the P1/N1 wave of the vibrissae-evoked SEP is produced by sequential
activation of supra- and infragranular pyramidal cells in the barrel
field (Di et al. 1990), which reflects a vertical cascade of excitatory activity during the P1/N1 that has been described
for evoked responses in the somatosensory (Abbes et al.
1991
; Di et al. 1990
), auditory (Barth
and Di 1990
; Sukov and Barth 1998
), and visual
(Bode-Greuel et al. 1987
; Mitzdorf 1987
) cortex.
In contrast to slow wave activity, the spatiotemporal characteristics of which suggest the sequential activation of supra- and infragranular pyramidal cells, both FO and VFO exhibit a relatively simple laminar profile characterized by a single reversal point common to all oscillatory peaks that is positioned in middle cortical lamina, which implies that cells in deep cortical lamina dominate the response. Thus laminar data suggest that the presynaptic elements generating fast oscillations are distinct from those responsible for slow wave activity. In addition, both laminar results and surface data suggest that the presynaptic elements generating FO are distinct from those responsible for VFO activity. Not only do the two types of fast oscillations reverse polarity at different depths, but the amplitude and latency differ as well, with FO typically of larger amplitude and earlier in latency than accompanying VFO activity (Figs. 1-3). These characteristics would not be expected if the two types of fast oscillations shared a common neural generator.
Intracellular correlates of slow and fast extracellular field potentials in RS cells
The predominant cell type yielding intracellular records in the present study was RS cells distributed throughout the supra- and infragranular layers. Vibrissa stimulation evoked a long-lasting EPSP at the soma, with a delayed onset in the infragranular, as compared with the supragranular, cells (Fig. 5). The latency and duration of these depolarizing slow waves appear to follow the P1 and N1 and probably reflect intracellular correlates of dendritic currents giving rise to the slow wave components of the SEP. The post-stimulus latency, phase, frequency, and relative amplitude of the fast deflections superimposed on slow intracellular EPSPs in RS cells resemble simultaneously recorded extracellular FO and VFO superimposed on the P1/N1 slow waves of the SEP. We propose that these deflections represent intracellular correlates of fast oscillatory dendritic currents conducted to the soma.
The alignment of action potentials and subthreshold events at common
latencies across trials (Figs. 7-9) suggests that, although small in
amplitude, fast deflections are effective in bringing a cell to
threshold for action potential generation. This observation is
consistent with recent in-vitro results that suggest that small but
rapid fluctuations of intracellular potential may serve as a mechanism
for triggering action potentials in individual cells (Mainen and
Sejnowski 1995; Nowak et al. 1997
). It
is also consistent with previous extracellular unit studies that
demonstrate an association between increases in multi-unit activity
(MUA) in layer V and fast oscillations (400-600 Hz) that occur during
high-voltage population spikes and when evoked by thalamic stimulation
(Kandel and Buzsaki 1997
). MUA was phase-locked to fast
oscillations, which indicates that the rapid synaptic currents
reflected in field potential recordings control the timing of action
potential generation in populations of pyramidal cells with a precision in the millisecond range. An analogous functional arrangement was
demonstrated in the hippocampus, where fast oscillations serve to
quantize spike latencies of CA1 pyramidal cells (Buzsaki et al.
1992
). Action potential bursts of hippocampal inhibitory
interneurons are consistently aligned to peaks of 200 Hz oscillatory
bursts in local field potentials whereas their pyramidal cell targets tend to fire single action potentials aligned with oscillatory troughs
or not at all. Although the phase alignment of RS cells to fast
oscillations demonstrated in the present study is more variable than
that reported for hippocampal pyramidal cells, such variability is not
unexpected given that the fast oscillatory response was extracted from
surface rather than local field-potential data. Nonetheless, in a large
portion of the RS cells studied, fast oscillations reflect a series of
preferred latencies at which a given cell is likely to fire.
Neurogenerator of VFO
The close association of stimulus-evoked activity in fast-spiking cells with the surface VFO suggests that FS cells are the generators of these oscillations. This is supported by three features of the data. 1) The evoked response in fast-spiking cells is coincident with VFO activity. As shown in Fig. 11, FS cells responded to vibrissa stimulation with bursts of 2-6 action potentials at latencies that overlap the post-stimulus latency of the accompanying VFO. 2) The interspike interval of FS bursts was almost identical to the periodicity of the VFO (action potential frequency, 508 ± 6.7 Hz; VFO, 515 ± 2.2 Hz). No other cells in the study were found to exhibit stimulus-evoked spike bursts of this frequency. 3) Action potentials in FS cell bursts exhibited a consistent phase alignment with the surface VFO (Fig. 11, gray bars) and thus with each other, suggesting that as a population their activity would summate coherently in field potentials measured at the cortical surface.
The proposed emergence of VFO as a consequence of within-burst
synchronization of fast-spiking cells implies the coordination of
activity across a large cell population at an extraordinarily fine
temporal scale. However, such results are consistent with accumulating
evidence that FS cells form a functionally continuous network.
Synchronous firing of FS cells to within 1 ms was previously reported
in barrel cortex (Swadlow et al. 1998) and is strongest in cells in middle cortical layers that receive monosynaptic input from
somatosensory thalamus (Swadlow 1995
; Swadlow et
al. 1998
). Synchronization may be sustained as other lamina
become active through a combination of electrical and chemical synaptic
interactions. It was recently demonstrated using dual intracellular
recording that electrotonic coupling acts to promote the synchronous
firing of local (~100 µm) networks of FS cells (Galarreta
and Hestrin 1999
; Gibson et al. 1999
).
Application of the gap junction blocker halothane eliminated
synchronized activity in these preparations; similarly, we found that
the administration of halothane suppresses fast oscillations in the
intact brain (M. S. Jones and D. S. Barth, unpublished
observations). Over longer distances, synchronization of FS cells may
be chemically mediated. Anatomical (Keller and White
1987
) and physiological (Connors et al. 1988
;
Galarreta and Hestrin 1999
; Gibson et al.
1999
) studies have demonstrated that FS cells receive
projections from other FS cells; computational models suggest that
reciprocal inhibitory connections will result in synchronous firing
given a sufficiently narrow spike width (van Vreeswijk et al. 1995
).
Such reciprocal interactions may achieve synchrony only as a
steady-state condition; this may explain why some FS cells recorded in
the present study exhibited early (Fig. 11C) or late (Fig.
11D) spikes that were not aligned with the remainder of the response.
It is unlikely that extracellular VFO is produced by dendritic currents
within FS cells themselves. It is more likely that it reflects the
synaptic activity of the postsynaptic targets of FS cells. FS cells in
neocortex have been identified as sparsely spiny or smooth GABAergic
inhibitory interneurons (Azouz et al. 1997;
Kawaguchi 1993
; McCormick et al. 1985
).
As noted above, interneuron activity cannot contribute directly
to extracellular field potentials because of their closed field
geometry. The model that emerges from this constraint is that burst
firing in FS cells is the source of rhythmic activity in the VFO band
and appears in field potentials recorded at the cortical surface by
virtue of projections of FS cells onto the apical dendrites of
pyramidal cells. The induced synaptic currents in pyramidal cell
targets act to phase-lock action potentials, perhaps as a rebound spike following a rapid inhibitory postsynaptic potential (IPSP) or, more directly, as a transient hyperpolariztion superimposed on concurrent excitation (cf. Mainen and Sejnowski
1995
). This description is admittedly incomplete; for example,
it does not account for additional complexities of neocortical
circuitry such as the reciprocal inhibitory connections described in
the preceding paragraph. However, the proposed circuitry is
consistent with many features of the present data, such as the
similarity of the spike frequency of burst firing in FS cells and VFO
frequency, the VFO rhythmicity observed in the spike responses of
putative pyramidal cells, and the somewhat late latency of VFO relative
to the evoked-potential complex, which suggests that it is at least
disynaptic with respect to thalamic afferents. Experimental
confirmation of this model might be possible through the selective
destruction of GABAergic neurons, which should result in the
disappearance of VFO, although this may be difficult to accomplish
without introducing pathological epileptiform activity.
Although intracellular results indicate that VFO-aligned responses
occur in cells located throughout the cortical lamina (Fig. 10), the
depth profile of VFO activity suggests that it is largely generated by
cells in infragranular layers (Fig. 3). The polarity reversal of the
oscillations, which is necessarily located superficial to the soma of
the cells participating in the response, lies in middle cortical depths
(750 µm). At minimum, this demonstrates that the contribution of the
supragranular cells is occluded by the infragranular response, if it is
present at all. The apparent paradox between the intracellular and
extracellular distribution of VFO-related activity may be explained by
features of cortical anatomy that favor the contribution of
infragranular cells to fast oscillatory field potentials measured at
the cortical surface. The apical dendrites of infragranular cells are
longer than those of pyramidal cells in more superficial lamina, thus
providing a more optimal dendritic geometry for generating large
synaptically induced current dipoles. Furthermore, there is greater
horizontal convergence of GABAergic projections in infragranular layers
(Nicoll et al. 1996; Salin and Prince
1996
). Therefore, large-amplitude field potentials in these
lamina may be a reflection of increased inhibitory synaptic currents.
It is interesting to note that IB cells, although located in
infragranular lamina, probably do not participate in the VFO response
because this cell type receives few projections from GABAergic
interneurons (Nicoll et al. 1996
).
Neurogenerator of FO
Because no specific cell type or subpopulation could be identified
that appeared to be responsible for generating FO activity, it is
believed that these oscillations may reflect local interactions among
populations of cortical cells. The onset latency of FO activity relative to the evoked potential complex is similar to the "initial fast response" observed in some of the earliest recordings of the SEP
(Dempsey and Morison 1943; Perl and Whitlock
1955
; see also Arezzo et al. 1986
). The initial
fast response consists of a series of surface-positive deflections
superimposed on the first component of the primary evoked response
(P1). Whereas the first of these deflections exhibits very short
latency and has been attributed to the discharge of thalamocortical
projection fibers, the remainder are thought to reflect the sequential
activation of cortical elements following arrival of the afferent
volley because they can be elicited using single shocks of the internal capsule (Dempsey and Morison 1943
; Landau and
Clare 1956
; Perl and Whitlock 1955
) and because
they survive extensive lesioning of ventrobasal thalamus (Morin
and Steriade 1981
).
Recent studies in monkey somatosensory cortex suggest that the initial
fast components of the SEP result from the sequential activation of
stellate and then supragranular and infragranular pyramidal cells
(Nicholson-Peterson et al. 1995). Although this activation sequence is consistent with the laminar progression of EPSP
latencies demonstrated in the present study (Fig. 5), several features
of the current data suggest that the generation of FO may not be
entirely based on sequential activation of stations in the cortical
hierarchy. A principal finding of the present study was that, in at
least some cells, action potentials in the vibrissae-evoked response
may align with more than one peak of FO activity (Figs. 6, 7, and 9).
Distributed over a large cell population, these preferred spike
latencies would be passed on to postsynaptic targets and, concurrently,
projections onto apical dendrites of neighboring pyramidal cells would
contribute to different peaks of the oscillatory field potential
measured at the cortical surface. In this manner, a given station in
the cortical hierarchy may contribute to different peaks of the surface
response. Such neural circuit interactions would not only effectively
produce FO in a localized region of somatosensory cortex, but would
also explain how these oscillations rapidly propagate through the
somatotopic cortical map, as noted in previous epipial recordings
(Jones and Barth 1999a
).
A second feature of the present results that is apparently at odds with
the sequential generation of FO is the laminar distribution of this
activity. As shown in Fig. 3, FO was found to exhibit a relatively
simple reversal pattern dominated by activity in the infragranular
layers. Sequential activation would be expected to lead to a more
complex laminar profile similar to that of slow wave activity
(Abbes et al. 1991; Bode-Greuel et al.
1987
; Di et al. 1990
; Sukov and Barth
1998
). Although features of the postsynaptic elements
participating in the response may account for this discrepancy, it
should be noted that because multi-vibrissae stimulation was used, the
evoked response includes both thalamically mediated input from the
principal whisker and intracortically propagated afferents from
surrounding vibrissae, including both slow wave and fast oscillatory
activity (Jones and Barth 1999a
). As such, any
hierarchical activation resulting from the principal response may be
poorly reflected in the laminar profiles of Fig. 3. Whole-field stimulation was employed in the present study because it results in
large-amplitude fast oscillations that exhibit a relatively consistent
morphology across trials; however it likely elicits a disparate
response pattern compared with that evoked by displacement of a single
vibrissa. The alterations in extracellular and unit responses that
occur when using single-vibrissa stimulation will be the subject of a
subsequent study.
Conclusions
Fast oscillations presented in the present study, and the
intracellular events associated with them, extend the classical picture
of cellular processes underlying the evoked potential complex. An
incoming thalamocortical volley may not result merely in a temporally
fused slow depolarization in middle laminar pyramidal cells but, at
least in some cells, may also establish preferred latencies of action
potential generation at the FO frequency. The sequential and recurrent
propagation of action potentials at these latencies throughout the
pyramidal cell population may explain the persistence of FO activity
beyond the P1 peak well after thalamic afferents are waning. Such a
model would account for the oscillatory variations in MUA at FO
frequencies observed in infragranular layers (Kandel and Buzsaki
1997) and in the intracortical propagation of fast oscillations
following displacement of individual vibrissae (Jones and Barth
1999b
). Furthermore, intracellular correlates of surface VFO
activity suggest that the role of inhibitory interneurons may be more
complex than is suggested by the accepted roles of receptive field
modification or excitability suppression in that their burst activity
may provide a mechanism of high-resolution spike timing that can act in
isolation (Fig. 8) or in concert with the slower FO activity (Fig. 9).
Although these events are an order of magnitude faster than current
established models of synchronized cortical information processing
(Bullock 1992
), such time scales are directly relevant
to recent findings that the cortical response arising from
paired-vibrissae stimulation is sensitive to interstimulus interval
changes at the millisecond level (Jones and Barth 1999b
;
Shimegi et al. 1999
).
![]() |
ACKNOWLEDGMENTS |
---|
We thank L. A. Bakel for laminar data collection.
This research was supported by Whitehall Foundation Grant S-97-06 and National Institute of Neurological Disorders and Stroke Grant 1 R01 NS-36981-01.
![]() |
FOOTNOTES |
---|
Address for reprint requests: D. S. Barth, Dept. of Psychology, University of Colorado, Campus Box 345, Boulder, CO 80309-0345 (E-mail: dbarth{at}psych.colorado.edu).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Received 13 March 2000; accepted in final form 31 May 2000.
![]() |
REFERENCES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|