| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Department of Physiology and Neuroscience Program, Michigan State University, E. Lansing, Michigan 48824-3320
Submitted 6 November 2002; accepted in final form 6 March 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
450 ms) followed by
a linear decline (–0.02 to –6.3 imp ·
s–1 · s–1);
discharge frequency was unrelated to current trajectory. Phasic-firing cells
(108/285; 38%) responded with a burst discharge having an initial rapid,
exponential decrease ( 30 ms) and subsequent linear decline
(–1 to –78 imp · s–1 ·
s–1). Phasic cells were activated preferentially
by fast current ramps (slope, 70 pA/s–2.2 nA/s) with the number and
frequency of impulses increasing with current slope. Delayed-firing cells
(44/285; 15%), responded to current steps with an accelerating firing
following a substantial latent period (0.5–4 s) and discharged during
current ramps with slopes less than 100 pA/s. Intracellular staining
revealed a significant association between electrophysiological profile and
neuronal morphology. A majority of presumed projection cells (22/30; 73%)
exhibited tonic firing to step-wise activation. The preponderance of phasic
and delayed firing cells, 93% (42/45) and 71% (12/17), respectively, were
interneurons with local or intersegmental terminations. Differential
sensitivity to static and time-varying components of membrane current suggest
differences in neuronal signaling properties that may have important
implications for integration of mechanosensory information in the deep spinal
dorsal horn. |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
) and is therefore
important for initial transformation of mechanical sensory information from
the body. The region also serves as the origin for segmental reflex pathways
(Hongo et al. 1989) and
several ascending tracts that transmit noxious and innocuous sensory
information (see rev. by Willis and
Coggeshall 1991), indicating an involvement in diverse
somatosensory and motor functions.
Integration of mechanosensory information in the deep dorsal horn is
influenced by complex factors that include neuronal electrophysiological
properties and synaptic mechanisms underlying cell-to-cell connectivity.
Experiments using both in vivo and in vitro spinal cord preparations have
presented clear evidence that dorsal horn neurons display considerable
heterogeneity in their responses to direct membrane depolarization. King et
al. (1988) first reported that
neurons in laminae III–VI respond to depolarizing current with
sustained, repetitive firing. Subsequent studies identified a population of
cells with rapidly adapting discharge
(Jiang et al. 1995;
Lopez-Garcia and King 1994;
Russo and Hounsgaard 1996b),
reportedly associated with activation of a low-threshold, transient calcium
conductance (Murase and Randic
1983; Russo and Hounsgaard
1996b; Ryu and Randic
1990). More recently, another class of neurons was described that
responds to membrane depolarization with accelerated firing after a
substantial time delay from the stimulus onset, involving activation of
high-threshold L-type calcium channels and a slowly inactivating potassium
conductance (Morisset and Nagy
1998; Russo and Hounsgaard
1996a).
In spite of this work, fundamental questions remain about the functional
importance of neuronal properties to sensory processing at the first spinal
relay. Past investigations have used a variety of paradigms to classify firing
patterns of dorsal horn neurons. Most studies categorized neuronal firing
during the first second or less of discharge
(King et al. 1988;
Lopez-Garcia and King 1994;
Thomson et al. 1989;
Yoshimura and Jessell 1989).
Studies of deep dorsal horn neurons that examined discharge patterns during
extended activation produced evidence of unsuspected complexity (e.g.,
Jiang et al. 1995;
Russo and Hounsgaard 1996a).
However, without the use of a consistent testing protocol, it is difficult to
assess what proportion of intrinsic circuitry is represented by each response
class and, hence, the relative contribution to integrative function. Second,
little is known about the role of physiologically defined neurons within the
functional architecture of laminae III–V. This is important because the
region contains a substantial population of interneurons that contribute to
local and segmental processing of sensory input
(Mannen 1975;
Mannen and Suguira 1976;
Ramon y Cajal 1995;
Scheibel and Scheibel 1968;
Schneider 1992;
Schneider et al. 1995) in
addition to projection cells (Brown and
Fyffe 1981; Brown et al.
1976; Carstens and Trevino
1978; Geisler et al.
1979). Third, the functional significance of multiple,
physiological classes relative to somatosensory processing is unclear. Because
mechanosensory afferents (see, e.g., rev. by
Burgess and Perl 1973) signal
both dynamic and steady-state information (e.g., rate and amount of skin
indentation, motion parallel to the surface), membrane depolarizations of
laminae III–V neurons during cutaneous stimulation are considerably more
complex than those activated by current pulses. It is therefore important to
explore how neurons defined by step-wise activation behave when
depolarizations are imposed that vary in the time domain. This information
could give important clues about how different physiological classes function
in the analysis of complex mechanosensory information.
I attempted to fill these gaps by performing comprehensive, systematic
analyses of spike-frequency adaptation for an extensive population of deep
dorsal horn neurons and making comparisons with axonal geometry. The aims were
accomplished using tight-seal, whole cell recordings in isolated preparations
of spinal cord from young Syrian hamsters that permitted high-quality,
intracellular labeling of axonal projections during the recording process
(Schneider 1992;
Schneider et al. 1995).
Responses to static and time-variant membrane depolarization were investigated
using standardized current commands of sufficient duration to reveal
differences in steady-state firing. I describe here three classes of deep
dorsal horn neurons distinguished by spike-frequency adaptation and discharge
timing that evidenced differential sensitivity to rate of membrane
depolarization. Recordings were obtained from interneurons and a smaller
population of presumptive projection cells in spinal laminae III–V. Some
aspects of the results have previously appeared in abstract form
(Schneider and McNaughton
1995).
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
All procedures involving the use of animals were in accordance with
Guide for the Care and Use of Laboratory Animals (National Institutes
of Health Publication No. 865-23, Bethesda, MD) and were approved by the
All-University Committee on Animal Use and Care. Three- to 4-wk-old male
Syrian hamsters (40–55 g) were anesthetized with urethan (1.5 mg/g ip)
(Bagust et al. 1982;
Schneider 1992). After animals
were rendered unresponsive to noxious pinching of a skin fold, the vertebral
column and thoracolumbar spinal cord were removed and placed in ice-cold
(4–8°C) dissection solution containing (in mM) 216 sucrose, 2.5 KCl,
0.25 CaCl2, 10 MgCl2, 1.25
NaH2PO4, 26 NaHCO3, and 10 glucose (pH
7.35–7.45, 290–310 mosm/l) equilibrated with 95% O2-5%
CO2. Animals were subsequently killed by exanguination after
severing the descending aorta.
Spinal cord slices
After dissecting vertebrae and meninges, a 5- to 6-mm block of spinal cord from approximately T10–L5 was glued to a block of 8% agar with cyanoacrylate, mounted on the stage of a Vibratome (TPI, St. Louis, MO) and immersed in cold dissection solution. Two or three sections (350- to 500-µm thick) were cut slowly (<1 mm/min) in the sagittal plane with a stainless steel razor blade at the maximum available vibration frequency. Sections were collected in an oxygenated artificial cerebrospinal fluid (ACSF) containing (in mM) 125 NaCl, 2.5 KCl, 2.5 CaCl2, 1.5 MgCl2, 1.25 NaH2PO4, 26 NaHCO3, and 10 glucose (pH 7.35–7.45; 290–310 mosm/l) and incubated for 1 h at room temperature. For some experiments, 500-µm-thick sections were prepared in the transverse plane using similar procedures.
Individual slices were transferred to a recording chamber (volume, 1.5 ml) mounted on a fixed-stage microscope and superfused with ACSF (5–6 ml/min, 25°C). Slices were held in place under a grid of parallel nylon monofilaments and viewed with a x10 objective. Under transillumination, the substantia gelatinosa (Rexed's lamina II) was visible as a thin translucent band and served as a convenient landmark to visually guide placement of recording pipettes within deeper spinal laminae.
Spinal cord hemisections
A series of experiments was also performed on thoracolumbar spinal cord sectioned once in the median sagittal plane from animals of the same age and weight range as those from which slices were obtained. These experiments investigated the extent to which neuronal properties are altered by the slicing procedure and provided more complete reconstruction of lengthy axons than is practical in tissue slices. After dissection, tissue blocks were placed in a recording chamber perfused with ACSF (8–10 ml/min, 25°C) and oriented with the medial surface upward. A pressor foot constructed from an electron microscope grid (0.5-mm grid size) was placed in contact with the cord surface to improve mechanical stability.
Electrophysiological recording
Recording pipettes were fabricated from borosilicate glass (N-51A; Drummond
Scientific, Broomall, PA) and filled with internal solution containing (in mM)
130 K-gluconate, 5 NaCl, 1 CaCl2, 1 MgCl2, 11 EGTA, 10
HEPES, 2 ATP (equine, magnesium salt), and 0.1 GTP (lithium salt; pH 7.3,
280–285 mosm/l; DC resistances, 5–7 M
). In some
experiments, pipettes were backfilled with internal solution containing 2%
biocytin (free base, Sigma Chemical, St. Louis, MO) to label cells for
morphological analyses. Tight-seal (1–2.5 G), whole cell
patch-clamp recordings in current-clamp mode
(Blanton et al. 1989) were
obtained blindly from neurons in laminae III–V, 
100–250 µm
from the slice surface. Signals were amplified (0–5 kHz bandwidth) using
Axoclamp 2 or Axopatch 1D amplifiers (Axon Instruments, Union City, CA),
passed through a digital data recorder and stored on magnetic tape or computer
analyzed on-line with a Digidata 1200 interface and pClamp 6 software (Axon
Instruments). Junctional current was nulled prior to establishment of gigohm
seals. Series resistance ranged from 15 to 40 M after attaining
cell-attached recording configuration and was compensated before collecting
data. Whole cell recordings were more successful (1–3 cells per
experiment) and maintained for longer periods than previous sharp micropipette
studies of deep dorsal horn neurons in this preparation
(Schneider 1992;
Schneider et al. 1995). Search
times were reduced and fewer tracks were required to obtain usable data from a
given experiment.
Electrical stimulation and data analysis
Neuronal firing patterns and steady-state current-voltage relations were
examined during application of constant current pulses through the recording
pipette. Rheobase (IRh) was measured by determining the
minimum current needed to evoke spike discharges. Small hyperpolarizing
voltage changes in response to anodal current were used to measure membrane
time constant (m) and neuronal input resistance
(Rin). The slowest time constant of the voltage response
was used as an estimate of m. Discharge patterns were
characterized in relation to magnitude of stimulating current expressed as
multiples of IRh (xT). Instantaneous firing frequency was
computed from interspike interval measurements using custom software running
in Axobasic (Axon Instruments). Plots of firing frequency as a function of
current were fitted using commercial graphical software (SigmaPlot version
4.0; SPSS, Chicago, IL) employing the Marquardt-Levenburg algorithm.
Statistical differences between data groups were analyzed with t-test
(2-tailed probability), 
2 (1 sample), or one-way ANOVA with
Neuman-Keul posttest comparisons (GBStat, Dynamic Microsystems, Silver Spring,
MD). Numerical data are presented as means ± SD.
Identification of neurons and axon projections
Neurons were filled with biocytin contained in the internal solution by
passive diffusion. After experiments slices were fixed overnight in cold
(4°C) phosphate-buffered 4% paraformaldehyde/4% sucrose (pH 7.4), and
washed for 12–48h(4°C) in 0.1 M phosphate-buffered 30% sucrose (pH
7.35). Tissue was frozen-sectioned (40 µm thickness) parallel to the slice
surface, pretreated with ethanol (Metz et
al. 1989) and processed according to the ABC method using standard
procedures. Sections were mounted serially on gelatin-coated glass slides,
dehydrated, cleared, and coverslipped. On occasion biocytin-filled neurons
were prepared for fluorescence histochemistry by incubating tissue slices in
avidin D conjugated with Texas Red (12 µg/ml; Vector Laboratories,
Burlingame, CA). Stained neurons were examined under transmitted light or
epifluorescence (Ex: 595 nm) and photographed. Representative examples were
reconstructed using a drawing tube attachment (x100).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
(n = 94), and m was 56 ± 32 ms
(n = 83). Cell capacitance (C), calculated as
m/Rin, was 167 ± 75 pF. Estimates
of Rin and m are 6- to 10-fold higher than
values reported for deep dorsal horn neurons in previous micropipette studies
(Jiang et al. 1995;
King et al. 1988;
Schneider 1992), probably
because of reduced current shunt afforded by the whole cell recording
configuration (Staley et al.
1992). There were no significant differences in spike morphology
(rise time or half-width), m, and IRh for
neurons recorded in spinal slices and hemisections. Neurons generated
overshooting action potentials, although spike amplitude was significantly
greater (P < 0.01) for cells in hemisections than in slices [67
± 16 mV (n = 72) and 59 ± 14 mV (n = 48),
respectively]. Neurons in slice preparations had greater
Rin [468 ± 281 M (n = 47) vs. 287
± 163 M (n = 47), P < 0.01] and lower cell
capacitance [129 ± 44 pF (n = 43) vs. 210 ± 80 pF
(n = 38), P < 0.0001] than those in hemisections,
suggesting that neuronal dendrites lying close to the surface may have been
pruned by the sectioning procedure. Neurons recorded from slices also were
slightly hyperpolarized relative to those in hemisections [–56 ±
6 mV (n = 187) vs.–54 ± 6 mV (n = 90),
P < 0.05] and exhibited less spontaneous activity. Not
surprisingly, measurements of some membrane properties were temperature
dependent. Elevating the bath temperature to 30–35°C decreased
Rin (25 ± 12%, n = 6), m
(26 ± 15%, n = 6), spike amplitude (25 ± 7%, n
= 5), and half-width (26 ± 6%, n = 5) but was without
consistent effect on resting membrane potential or pattern of discharge
activated by direct membrane depolarization. (Sustained elevation of bath
temperature was associated with reduced tissue viability and was therefore
avoided.)
|
Discharge patterns
As shown by the examples in Fig. 1, three categories of neurons were distinguished on the basis of discharge timing and characteristics of spike frequency adaptation in response to step-wise depolarizing current (5-s duration) applied through recording pipette at resting potential. Electrophysiological properties for each class are compared in Table 1. It should be noted that neuronal properties summarized in Table 1 pertain to both slices and hemisections and the subsequent presentation of results do not distinguish between the two preparations.
|
Tonic cells
Almost half of the neurons recorded [n = 133, 47% of sample (76 in
slices; 57 in hemisections)] discharged continuously during application of
depolarizing current at near-threshold strength (e.g., Figs.
1A and
2, A–C),
resembling deep dorsal horn neurons described previously in rat
(Jiang et al. 1995;
King et al. 1988). Tonic cells
had a mean Rin of 428 ± 279 M (n =
47) and mean m of 65 ± 39 ms (n = 39).
IRh measured at resting potential was 45 ± 41 pA
(n = 92).
|
The initial discharge frequency of tonic cells averaged 25 ± 14
imp/s (n = 20) measured at a pulse strength of 2T. Firing was maximal
at the onset of the current and was graded with pulse strength (0.5 ±
0.3 imp · s–1 ·
pA–1). In about half of tonic cells (48%,
n = 26), plots of instantaneous firing frequency were well-fitted by
a single linear regression (R2 0.55–0.92;
Fig. 2A). For these
neurons, firing slowed during current application (2–5T) by 32 ±
20% (n = 19) to 15 ± 7 imp/s (n = 48), a decrease of
–0.14 to –4.3 imp · s–1
· s–1. As shown in
Fig. 2A, both the
magnitude (percent decrease) and rate of adaptation increased with increasing
pulse strength. For the remaining neurons (n = 28), adaptation was
biphasic (Fig. 2, B and
C), resembling behavior of motoneurons activated by
steady depolarizing current (Granit et al.
1963; Kernell and Monster
1982; Sawczuk et al.
1995). During the initial 0.5–2 s of stimulating current
(1.5–5T), plots of firing frequency could be fitted by a single
exponential with a time constant of 450 ± 365 ms (range, 47 ms to 2.1
s, n = 23). This period was followed by a slower, linear decline in
firing (R2 0.52–0.99) in which discharge
decreased at a rate of 1.0 ± 1.4 imp ·
s–1 · s–1
(n = 22). The time constant of the exponential phase decreased with
increasing current (–35 ± 62 ms/pA, n = 15; e.g.,
Fig. 2, B and
C). The magnitude of adaptation was 51 ± 18%
(n = 29) which was significantly greater (P < 0.01) than
for tonic cells with monophasic linear firing behavior.
Phasic cells
Phasic cells constituted a group of lamina III–V neurons [n
= 108, 38% of sample (86 in slices; 22 in hemisections)] in which direct
depolarization elicited firing characterized by pronounced spike frequency
adaptation (Fig. 1B)
(see also Lopez-Garcia and King
1994; Russo and Hounsgaard
1996b; Thomson et al.
1989). The average peak discharge frequency measured at 2T was 140
± 63 imp/s (n = 25), significantly greater than for tonic
neurons at similar currents (P < 0.0001), and was graded with
pulse strength (range, 1.5–4T). Action potentials were often
superimposed on a transient depolarizing membrane potential (9 ± 4 mV,
n = 17) that was followed by a sag in membrane voltage
(Fig. 1Ba,
middle).
Phasic cells could also be distinguished from tonic neurons by their
electrophysiological properties (see Table
1). The mean resting potential of this class was –57
± 5.7 mV (n = 104), slightly hyperpolarized relative to tonic
cells. They exhibited significantly lower m (44 ± 20
ms, n = 33) and higher IRh (75 ± 58 pA,
n = 61) than tonic cells. Phasic cell Rin
averaged less than tonic neurons (302 ± 183 vs. 428 ± 279
M), although the difference was not statistically significant
(P > 0.05), and there was no difference in cell capacitance
between the two classes. When stimulated at higher pulse strengths
(2–4T), a few phasic cells (n = 9) exhibited significantly
(P < 0.05) slower firing adaptation (time constant 368 ±
292 ms; n = 6) compared with the others (not shown). In these cells,
average initial firing frequency (2T) was 30 ± 14 imp/s (n =
6) significantly less than phasic neurons with rapid adaptation (P
< 0.0001) but not different from tonic cells (P = 0.5). Such
neurons may represent a transitional type with properties intermediate to the
two groups.
Spike-frequency adaptation of phasic cells exhibited multiple phases
(Fig. 2,
D–F). Frequency plots could be fitted by a
single exponential (mean =30 ± 30 ms, n = 19) followed
by a linear regression (R2 0.91–0.99) with
slope of –20 ± 19 imp · s–1
· s–1 (n = 24; e.g.,
Fig. 2D). Increasing
the pulse strength above threshold (1.2–2T) evoked a later period of
low-level, sustained firing (Figs.
1Bb and
2E) in two-thirds of
phasic cells (73/108). During this period, discharge frequency decayed slowly
(–1.4 ± 1.0 imp · s–1 ·
s–1, n = 10) to a steady-state level of
10 ± 8 imp/s (n = 32). As can be seen by comparing
Fig. 2, C and
E, the adapted firing level was significantly lower
(P < 0.01) and the overall magnitude of spike-frequency adaptation
(89 ± 10%, n = 32) was greater (P < 0.0001) than
for tonic cells at comparable activating currents.
To determine whether the differences between tonic and phasic cells could be accounted for by variations in resting membrane potential, responses to step activation were studied while injecting constant depolarizing current through the pipette (n = 5). As shown by the experiments illustrated in Fig. 3 (A–C), depolarizing phasic cells by 10-mV prolonged firing and reduced the peak discharge frequency. However, the time constant of adaptation was relatively unaffected (Fig. 3, B and C). By contrast, hyperpolarizing tonic cells (n = 2) had the effect of decreasing average discharge frequency without altering the rate of adaptation (Fig. 3D), arguing that differences in spike frequency adaptation between the two classes were not simply due to variations in resting membrane potential. The three-dimensional plot in Fig. 3E demonstrates that tonic and phasic neurons were reliably differentiated by the testing procedures employed. It can be seen that phasic cells overall had a higher initial discharge frequency (fini) and greater firing adaptation (% f decrease) than tonic cells when tested at resting potential over a similar range of pulse strengths (stimulus, xT). The distribution of data points in Fig. 3E is consistent with the finding that phasic and tonic cells exhibit fundamentally different responses to step-wise membrane depolarization but also suggests that response properties of the two classes may form a continuum.
|
Delayed-firing cells
A third class of laminae III–V neurons exhibited responses that were
characterized by an exceptionally long delay to the first action potential,
resembling responses reported previously in mammalian
(Dekin et al. 1987; Morrisset
and Nagy 1998; Storm 1988) and
invertebrate neurons (Byrne et al.
1979; Getting
1983). Delayed-firing cells were infrequently encountered in these
preparations [n = 44, 15% of sample (33 in slices; 11 in
hemisections)]. For reasons that are not entirely clear, whole cell recordings
from these cells were more difficult to maintain than those from phasic or
tonic neurons, thereby limiting the time available for study. Estimated values
of Rin, m, and IRh for
delay cells were 407 ± 231 M (n = 11), 58 ± 20
ms (n = 11), and 66 ± 54 pA (n = 30), respectively,
indistinguishable from tonic and phasic neurons
(Table 1), suggesting that the
unusual firing properties of these cells are not attributable to cell
damage.
Figure 1C shows a
representative response to rectangular current pulses. At resting potentials
between –50- and –60-mV latency to the first spike ranged from 0.5
to 4 s, considerably longer than m
(Table 1), and was maximal for
near-threshold activation (1–1.2T). The membrane potential showed a
gradual depolarization (0.6–6.8 mV/s) preceding the first impulse (e.g.,
Figs. 1C and
4A), similar to the
behavior reported for hippocampal CA1 cells
(Storm 1988). The slope of the
depolarization increased and firing delay decreased (e.g., Figs.
1C and
2, G and H) a
pulse strength was increased. As shown in
Fig. 4, delayed excitation was
dependent on resting potential; depolarizing the neuron by 5–10 mV
decreased the response delay and increased discharge frequency (n =
4).
|
Discharge of delayed-firing cells accelerated (40 ± 24%, n = 14) during application of depolarizing current (Figs. 1Ca and 2, G and H) and firing tended to be irregular compared with tonic and phasic cells, (e.g., compare scatter of instantaneous firing plots shown in Fig. 2G with A and E). This distinction was more evident for neurons recorded from spinal slices than from hemisections, perhaps due to differences in spontaneous activity between the two preparations. Some cells (n = 7) exhibited adaptation at higher pulse strengths (1.2–2.5T), as illustrated by the decrease in instantaneous firing shown in Fig. 2H after 3 s of activation by 120-pA current. For a few delay cells (Fig. 2I, n = 6), firing patterns underwent a pronounced shift at higher stimulus currents (>1.5T), suggesting that factors regulating firing in the other two classes were also present. Plots of instantaneous discharge frequency versus time for delayed-cells could not be fit consistently using standard regression analyses.
Responses to ramp-hold current injection
Having established systematic differences in responses to step-wise
membrane depolarization, the next goal was to find out how lamina III–V
neurons respond to time-varying membrane depolarizations. The approach was to
depolarize neurons with computer-generated families of current commands (3-s
duration) consisting of a variable slope "ramp" (10 pA/s to
2.2 nA/s) followed by a constant "hold" period (duration,
1.4–2.9 s). Current ramps applied to the somatic membrane produced
voltage trajectories that were well-fit by a linear regression model (slope,
6–350 mV/s; R2 = 0.88 ± 0.06) over the range
of the ramp phase (Fig. 5).
|
Phasic cells (n = 23; 17 in slices, 6 in hemisections) responded
with a brief burst of impulses when the membrane potential was depolarized
rapidly but relatively poorly when the slope of the command current was
slowed. This property was unique among lamina III–V neurons in these
preparations. Ramp commands (60–200 pA/s; 1–1.25T) that
imposed voltage trajectories of 13–34 mV/s at resting potentials between
–55 and –65 mV activated in phasic cells a depolarizing potential
whose amplitude was graded with the ramp slope
(Fig. 6A2). The
amplitude and time course of this response were similar to the transient
depolarization induced by rectangular current pulses. Increasing the ramp
slope further (70 pA/s to 2.2 nA/s) resulted in a transient discharge in
which the number and frequency of action potentials were positively correlated
with the current trajectory (Fig. 7,
A–C). Maximum discharge frequency (64–133
imp/s) was reached at ramp slopes between 500 pA/s and 1.4 nA/s (producing
depolarizations of 35–170 mV/s; Fig.
5B). Increasing the magnitude of the hold phase (range,
1.25–2.5T) prolonged the firing of some phasic cells, activating
impulses at a frequency that was similar to step-wise current stimulation (not
shown).
|
|
The robust rate sensitivity of phasic cells was not shared by tonic
neurons. This class typically responded to ramp-hold commands (1–1.25T)
with repetitive firing during the steady-state phase of the stimulus
(n = 20; 13 in slices, 7 in hemisections; examples shown in
Fig. 7, D–F). At
higher currents (1.25–6T), firing increased and tonic cells discharged
during the ramp phase (40–100 pA/s), but discharge frequency was
unrelated to trajectory of the current or membrane depolarization as
illustrated by the relatively flat appearance of frequency plots depicted in
Fig. 7. Two of the tonic cells
tested with ramp-hold commands exibited rapid initial firing adaptation (
<400 ms, e.g., Fig.
2C) to rectangular pulses. For these cells, responses to
current ramps were positively correlated with current trajectory (not shown),
providing another suggestion that tonic and phasic neurons may form a
functional continuum.
The responses of delayed-firing cells to ramp-hold commands were similar to
tonic neurons, but with several differences (see
Fig. 7, G–I).
Like tonic cells, ramp-hold stimuli delivered at near-threshold intensities
(1–1.2T) activated repetitive firing during the hold phase (n =
16; 10 in slices, 6 in hemisections; Fig.
6B2). However, discharge frequency slowed (Figs.
6B2 and
7, G and H)
as the ramp slope increased (11/13 cells, 85%). This was associated with a
gradual membrane depolarization (slope, 7–27 mV/s) that resembled the
ramp-like membrane response observed with rectangular pulses (cf. Figs.
1Ca,
4A, and
6B2, 
). At
higher stimulating currents (1.2–5T), impulses were activated during the
ramp phase for slopes less than 30 pA/s.
Figure 8 compares responses of
tonic and delay cells as a function of current trajectory. Increasing ramp
slope over 100-fold caused a small but significant decrease (unpaired
t-test, P < 0.005, 2-tailed) in discharge frequency of
delayed-firing cells (Fig.
8B1) compared with those in the tonic class
(Fig. 8A1). Increase
in ramp slope also caused an increase in response latency of delay cells
relative to onset of the hold phase (Fig.
8B2) but not for tonic cells
(Fig. 8A2) as
indicated by the positive slope of plots in B2.
|
To reiterate, tonic cells appeared to respond to the absolute magnitude of
the depolarizing stimulus; their discharge frequency being unrelated to
current trajectory during the ramp phase. In comparison, phasic cells were
activated selectively by fast current ramps, exceeding 70 pA/s for
threshold stimuli. When stronger depolarizing current was applied, discharges
could be activated during the hold phase, although at lower rates than tonic
cells under similar conditions. Like tonic neurons, cells in the delay class
fired during the hold phase but became less responsive to the stimulus as the
ramp slope increased.
Morphological features
An extensive population of neurons (n = 117; 52 from slice
experiments, 65 from hemisections) was recovered after intracellular staining
and histological processing. Labeled cell bodies were distributed throughout
Rexed's laminae III–V and exhibited major and minor diameters of 18.4
± 6.0 and 11.4 ± 3.5 µm (n = 99), respectively. Soma
configuration was uncorrelated with firing pattern, and there were no
significant differences in spike amplitude, Rin,
m, or IRh between labeled neurons and
those recorded with standard pipette internal solution.
Labeled neurons were classified into two principle groups based on axon
trajectory and branching pattern as summarized in
Table 2. The majority of cells
(74%, n = 87) were interneurons with extensive, longitudinally
oriented terminations within laminae III–V in agreement with previous
morphological studies of hamster deep dorsal horn interneurons
(Schneider 1992;
Schneider et al. 1995).
Interneurons could be further subdivided into two groups. Sixty-seven (77%)
gave rise to local axons with highly branched terminations overlapping the
cell body and dendrites (Fig. 9, A
and E). Twenty others (23%) were deep axon cells having
ventral-going stem axons that bifurcated into lengthy rostrocaudal fibers with
collateral branches in laminae IV–VI,
(Fig. 9, B and
F). The remaining cells (26%, n = 30) gave rise
to sparsely branched axons that projected ventromedially toward laminae VII
and X (Fig. 9D) or
entered the dorsolateral white matter (Fig.
9C). These neurons were classified as presumptive
projection cells.
|
|
As summarized in Table 2,
interneurons could display tonic, phasic, or delay firing characteristics with
no consistent differences in the response properties of cells with local
versus deep axons (2; P = 0.7). Relative to
interneurons, presumptive projection cells were significantly more homogeneous
(2; P < 0.001), with about three-fourths (22/30)
of this type displaying tonic firing to step current application. A strong
association between firing pattern and axonal projection was noted for one
class of cell. It was found that the preponderance of neurons exhibiting
phasic firing patterns (93%, 42/45) presented morphology typical of
interneurons with most of these (79%, 33/42) being of the local axon type.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Functional properties of deep dorsal horn neurons
The differences in firing properties of deep dorsal horn neurons do not appear to be related to the procedures used to isolate the tissue. Intrinsic properties of lamina III–V neurons were found to be relatively independent of the preparation from which they were recorded. However, neurons in hemisections had slightly depolarized resting potentials, greater spike amplitude, lower Rin, greater cell capacitance, and greater spontaneous synaptic activity than neurons in sagittal slices. These differences could be primarily attributed to interruption of synaptic inputs and pruning of neuronal processes during the slicing procedure.
The results are in overall agreement with previous investigations reporting
considerable diversity in the firing patterns of dorsal horn neurons based on
characteristics of spike frequency adaptation
(Jiang et al. 1995;
King et al. 1988;
Lopez-Garcia and King 1994;
Thompson et al. 1989) and discharge timing
(Morisset and Nagy 1998).
However, other studies showing that some deep dorsal horn neurons exhibit a
degree of firing plasticity under the control of modulatory inputs
(Morisset and Nagy 1996;
Russo and Hounsgaard 1994;
Russo et al. 1998) have
questioned the usefulness of previous categorization schemes based on these
criteria. I conclude from quantitative analyses of spike frequency adaptation
that differences in firing properties among deep dorsal horn neurons can be
substantial and relatively independent of recording conditions and
experimental preparations. Most phasic and tonic cells contrasted strongly in
their responses to current steps and ramp-hold stimuli. Initial discharge
frequency and firing adaptation exhibited some degree of voltage dependency,
but there was no compelling evidence that a neuron's resting potential
influenced its assignment to a category. To some extent, magnitude and rate of
adaptation were dependent on injected current amplitude. However, rate of
adaptation and level of adapted firing for most phasic neurons were
distinguishable from tonic cells, suggesting that these classes are relatively
independent of stimulating current over the range tested. On the other hand,
the responses of delayed-firing cells were quite sensitive to membrane
potential: depolarizing the membrane potential by 5–10 mV shifted firing
to a pattern of maintained discharge characteristic of tonic neurons (see also
Morisset and Nagy 1998).
Therefore the present results may underestimate the number of neurons in this
class.
The extensive population of neurons sampled by the present study provides
additional insight into the composition of neural networks within the deep
spinal dorsal horn. In hamster, tonic and phasic cells comprises 47 and 38% of
the neurons in lamina III–V, respectively, whereas delayed-firing cells
accounted for only 15% of the population. Recent studies in young and neonatal
rats reported that up to two-thirds of the neurons in laminae III–V
exhibited tonic firing patterns with relatively fewer neurons having rapidly
adapting, phasic-type responses (Hochman
et al. 1997; Ruscheweyh and
Sankühler 2002). Delayed firing neurons were absent in the
deep dorsal horn of young rats similar in age to hamsters used in the present
study (Ruscheweyh and Sankühler
2002) and were only infrequently encountered in neonatal animals
(Hochman et al. 1997). The
differences between the studies could reflect variations in species,
classification, and sampling procedures. In addition, it has recently been
reported that the discharge pattern of phasically firing deep dorsal horn
neurons are sensitive to modulatory actions of biogenic amines
(Garraway and Hochman 2001).
These differences notwithstanding, the results are in general agreement that
neurons of the tonic and phasic types are abundant in laminae III–V,
with delayed-firing cells being relatively scarce.
Characteristics of spike frequency adaptation
Tonic neurons have been reported recently in rat spinal lamina I with
firing that undergoes monoexponential decay during one-second membrane
depolarizations (Prescott and De Koninck
2002). The present study identified two populations of
tonic-firing neurons in hamster laminae III–V: one with monophasic
linear adaptation and another the discharge of which decayed exponentially,
followed by a period of linear decline. Biphasic firing adaptation has been
reported for tonic-firing neurons in the deep spinal dorsal horn but not
described quantitatively (Jiang et al.
1995). It appears that firing adaptation of phasic neurons can
exhibit a similar complexity. The time course of the decline in phasic cell
discharge may exhibit up to three distinct phases described by combinations of
exponential and linear functions. Initial adaptation was limited to the first
few spikes in the discharge, the time course being fitby a single exponential,
and was followed by a subsequent linear decline lasting 
1 s. A later
period of slow, linear adaptation was apparent in some phasic cells if
activated by a strong depolarization.
The nature of curve fits suggest that initial and later phases of
adaptation have different underlying ionic mechanisms. Previous studies in
dorsal horn suggest that activation of low-threshold transient (T-type)
calcium channels may mediate the burst response of phasic cells to step
depolarization (Russo and Hounsgaard
1994,
1996b;
Ryu and Randic 1990). A
hyperpolarization-activated cation current
(Mayer and Westbrook 1983) is
evidenced by some dorsal horn neurons
(Grudt and Perl 2002;
Jiang et al. 1995;
Yoshimura and Jessell 1989)
and may also shape phasic firing patterns (e.g.,
Erikson et al. 1993). Although
these processes may contribute to the initial phase of adaptation, later
phases could involve activation of several outward currents, including
Na+-activated potassium current
(Safronov and Vogel 1996;
Schwindt et al. 1989),
Ca+-activated potassium current
(Lancaster and Nicoll 1987;
Madison and Nicoll 1984;
Sawchuck et al. 1997), and M current
(Madison and Nicoll 1984).
Alternatively, a persistent inward sodium current, which has been shown to
influence repetitive firing in several neuron types
(Fleidervish et al. 1996;
Nishimura et al. 1989;
Stafstrom et al. 1985), may
also contribute to late adaptation in tonic and phasic cells. It is not yet
known if significant currents of these types are present in deep dorsal horn
neurons and should be investigated in future experiments. Regardless of the
underlying mechanisms, spike firing of laminae III–V neurons is clearly
regulated by complex processes that probably have important functional
consequences on their signal processing characteristics and ultimately
influencing integration of complex somatosensory information.
Responses to time-varying membrane depolarizations
An important finding of the present study is that many neurons were
differentially responsive to ramps of depolarizing current. This surprising
result suggests that deep dorsal horn neurons are tuned differently to static
and dynamic components of afferent sensory input, influencing spinal
integration of sensory information. Current ramps applied directly to the cell
membrane produced linear voltage responses (6–350 mV/s) similar to
the trajectory of postsynaptic responses generated in deep dorsal horn neurons
by noxious and innocuous cutaneous mechanical stimuli both in vitro
(Lopez Garcia and King 1994;
Schneider and Perl 1994;
Schneider, unpublished observations) and in vivo
(Woolf and King 1989). The
finding that phasic cells were strongly and selectively activated by fast
membrane depolarizations means that these cells might be especially sensitive
to inputs from rapidly adapting mechanoreceptors. This property could be
associated with activation of the same low-threshold, T-type calcium
conductance that generates characteristic burst responses of phasic cells
(Murase and Randic 1983;
Russo and Hounsgaard 1996b;
Ryu and Randic 1990). A
transient, inward calcium current (IT) underlying
low-threshold calcium potentials in cat lateral geniculate neurons also is
activated selectively by depolarizations exceeding 30 mV/s (cf.
Crunelli et al. 1989,
Fig. 9). This value is very
close to the threshold for ramp activation of phasic cells in the present
study, suggesting that a mechanism similar to IT
contributes to rate sensitivity in deep dorsal horn neurons.
The present results also showed that the responses for a majority of
delayed-firing cells to depolarizing current were inversely related to
membrane voltage trajectory, declining with increasing ramp slope. The delayed
excitation and accelerated firing that are characteristic of neurons of this
class may be shaped by several ionic mechanisms. L-type calcium channels have
been associated with acceleration of firing in motoneurons
(Carlin et al. 2000;
Hounsgaard and Kiehn 1989) and
also in some deep dorsal horn interneurons
(Morisset and Nagy 1999;
Russo and Hounsgaard 1996a).
Furthermore, a slowly inactivating potassium current, similar to the
ID type described in hippocampal CA1 neurons
(Storm 1988), has also been
implicated in the behavior of deep dorsal horn neurons with delayed,
accelerating discharges (Morisset and Nagy
1998). The results of the present experiments using ramp-hold
current injection suggest that one or both of these mechanisms are involved in
producing the inverse relationship between discharge frequency and trajectory
of membrane depolarization depicted in Fig.
8. This idea is consistent with the suggestion that an
ID-like mechanism suppresses responsiveness of deep dorsal
horn neurons to brief, transient inputs
(Morisset and Nagy 1998) and
receives additional support from the recent observation that delayed-firing
neurons in lamina I fail to respond to fast, afferent-evoked EPSPs
(Prescott and De Koninck
2002). The rate sensitivity of delay cells appears to be strongly
influenced by the prevailing membrane potential, as evidenced by the shift in
discharge pattern to step-wise currents induced by depolarizing current
injection. It remains to be determined the extent to which this property
influences integration of afferent activity evoked by natural stimuli.
Morphology of recorded neurons and functional considerations
The present findings point to substantial differences in the functional composition of interneurons and projection cells that form networks in laminae III–V. Interneurons in the deep dorsal horn are characterized by a striking heterogeneity in their responses to membrane depolarization, whereas neurons having morphology consistent with a projection function appear to be more homogeneous.
Labeled neurons in laminae III–V fell into two principle groups,
interneurons and projection cells, based on trajectory of axonal
ramifications. The location, dendritic distribution, and axonal arrangements
of interneurons confirmed prior descriptions of neuronal morphology in hamster
deep dorsal horn (Schneider
1992; Schneider et al.
1995) and are consistent with extensive reports in other species
(Mannen 1975;
Mannen and Suguira 1976;
Ramon y Cajal 1909;
Scheibel and Scheibel 1968).
In the present study, the ratio of interneurons making localized connections
(i.e., over an area comparable to the separation between adjacent dorsal
roots) outnumbered those with presumed intersegmental projections by 3:1,
approximately twice the previously published ratio of 1.5:1
(Schneider 1992). The
discrepancy is probably due to sampling differences associated with the use of
sharp microelectrodes (Schneider
1992; Schneider et al.
1995) versus whole cell recording methodology in the present
experiments. Regardless, these two classes dominate the deep dorsal horn. They
display considerable diversity in firing characteristics and sensitivity to
time-varying membrane depolarizations, apparently reflecting the complexity of
local circuits in which these neurons take part. This functional
heterogeneity, however, may not be characteristic of interneurons in other
dorsal horn laminae. A recent finding that nonprojection neurons in hamster
lamina I consistently responded to step-current activation with sustained,
repetitive firing (Grudt and Perl
2002) may be evidence of fundamental differences in organization
between the superficial and deep dorsal horn.
The vast majority of phasic cells (>90%), those uniquely signaling transient membrane depolarizations, were identified as interneurons on the basis of axonal morphology. Of these, an exceptionally high number (33/42 or almost 80%) formed densely branched terminations within the laminae III–V neuropil in vicinity of the soma and dendritic tree. This observation is significant for two reasons. First, it suggests that an important function of local circuits in the deep dorsal horn is the initial processing of afferent information signaling rapid stimulus motion or displacement and identifies a likely class of neuron responsible. Second, the results provide a useful tool for identifying a major morphological class of lamina III–V interneurons during recordings, assisting future studies of dorsal horn functional organization.
About one-fourth of labeled neurons in laminae III–V were classified
as presumptive projection cells on the basis of ventromedial trajectory of the
axon or funicular location of a stained process that could be traced back to
the labeled cell. This result is consistent with previous work showing that
several ascending sensory pathways originate from this area in rat and cat
(Brown and Fyffe 1981;
Brown et al. 1976;
Carstens and Trevino 1978;
Geisler et al. 1979;
Rustioni and Kaufman 1977).
The ultimate destination of labeled axons could not be determined by the
present experiments, and it is possible that some neurons join propriospinal
projection systems (Chung and Coggeshall
1983; Chung et al.
1984,
1987). The response profile of
presumptive projection cells was more homogeneous than interneurons with the
bulk exhibiting tonic firing patterns. This is in agreement with a previous
report that projection neurons acutely isolated from rat dorsal horn discharge
repetitively in response to sustained depolarization
(Huang 1987) although
responses of spinal projection neurons to DC stimulation have not been
adequately examined in vivo to permit meaningful comparisons with in vitro
data. Together these results suggest that, in the deep dorsal horn, functional
diversity is a hallmark of interneurons rather than projection cells. Based on
the present data, it would appear that deep dorsal horn neurons giving rise to
long distance projections may have a surprisingly limited repertoire for
encoding changes in membrane potential.
Concluding remarks
The present study provides an alternate perspective from which to view integrative function in the deep dorsal horn. It seems reasonable to conclude that differences in sensitivity to rate of membrane depolarization among lamina III–V neurons, such as those presented here, will have important consequences on the transformation of primary mechanosensory information. Phasic cells were able to respond selectively to relatively fast membrane depolarizations, indicating they should favor activation by brief, rapidly adapting inputs signaling transient skin displacement or motion, while being less responsive to static or slowly changing stimuli. At depolarized membrane potentials, phasic cells can exhibit low-level repetitive firing to maintained depolarizations, thereby switching to a mode that enables signaling of slowly adapting inputs while preserving rate-sensitive information. Phasic cells may therefore emulate variable "high-pass" filters within the network. On the other hand, tonic cells discriminated poorly between fast and slow current ramps and so would be hypothesized to respond with comparable fidelity to slowly and rapidly adapting inputs. Accordingly, tonic cells may function in a "broad band" capacity, signaling both static and dynamic components of sensory stimuli. Delayed-firing cells might be tuned to excitation by slowly adapting inputs, as suggested by the decline in discharge frequency with increasing rate of membrane depolarization, indicating that this class complements the high-pass characteristics of phasic neurons.
The present study leaves open the question of whether lamina III–V neurons are assembled into functionally distinct circuits having different input-output relations. Such circuits would be important substrates for integration and serial processing of sensory information in the deep dorsal horn. It would, therefore, be of considerable importance to determine the constituent neurons of these hypothetical circuits, functional properties of their afferent input, and the organization and nature of their synaptic connectivity.
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
This work was supported by Grant NS-25771 from the National Institute for Neurological Disorders and Stroke.
|
FOOTNOTES |
|---|
, n =
31) and tonic (
, n = 48) neurons. Overall, phasic cells
exhibited higher initial firing frequency and greater spike frequency
adaptation than tonic cells for similar activating currents.
) and
increased firing frequency, producing a continuous repetitive discharge for
duration of the current pulse.
, axonal branches. Cell body and dendrites are drawn in shading.
Neurons in C and D were classified as presumptive projection
cells on the basis of initial axonal trajectory (see text). Process indicated
(
) in C was traced into the dorsal columns. Neuron depicted in
E was classified as a local axon interneuron, whereas the cell shown
in F has morphology representative of deep axon cells with
inter-segmentally projecting daughter fibers. Scale bar in C applies
to D–F.