| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J Neurophysiol (November 1, 2002). 10.1152/jn.00085.2002
Submitted on 7 February 2002
Accepted on 18 July 2002
1Department of Psychiatry and 2Department of Physiology, New York Medical College, Valhalla 10595; and 3Department of Physiology and Biophysics, Albert Einstein College of Medicine, Bronx, New York 10461
 |
ABSTRACT |
|---|
 TOP
 ABSTRACT
 INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
Granata, Antonio R. and
Morton I. Cohen.
Rhythmic Properties of Neurons in the Rostral Ventrolateral
Medulla of the Rat In Vitro: Effects of Clonidine.
J. Neurophysiol. 88: 2262-2279, 2002.
The rostral
ventrolateral medulla (RVLM) is thought to be the main central site for
generation of tonic sympathetic activity. In the rat in vitro slice
preparation, we used intracellular recordings to identify different
populations of neurons in the RVLM: 43 spontaneously active neurons
with regular (R) or irregular (I) patterns of spike firing and 10 silent neurons. The degree of regularity was quantified by the
coefficient of variation (CV = SD/mean) of interspike interval durations, as well as by the rhythmic properties of the spike autospectrum and autocorrelation. The distribution of CVs was clustered: R and I neurons were defined as those with CVs 
12% (n = 21) or >12% (n = 22),
respectively. The R-type and I-type neurons resemble the type II and
type I neurons, respectively, which were previously characterized in
the RVLM in vivo as barosensitive and bulbospinal. Both types may be
important in generation of sympathetic tone. Clonidine (1-100 µM)
was applied to 10 R-type neurons and 16 I-type neurons. The firing of
21/26 was depressed to the point of silence. However, 18/26 neurons
were excited earlier in the perfusion. The later depression of firing
occurred in both I and R neurons and in different cases was associated
with either hyperpolarization or depolarization.
| |
INTRODUCTION |
|---|
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
The rostral
ventrolateral medulla (RVLM) is generally considered as the site of
generation of tonic sympathoexcitatory activity (Sun
1995
). Barosensitive neurons in the RVLM selectively project to
regions in the spinal cord involved in autonomic control (Barman and Gebber 1985). Furthermore, these neurons are considered to form a major contributor of vasomotor efferent output and to regulate cardiovascular sympathetic activity, mainly by monosynaptic projections to sympathetic preganglionic neurons (Zagon and Smith
1993). Using an in vivo intracellular recording technique, we
demonstrated the presence of at least two types of barosensitive
bulbospinal neurons in the RVLM (Granata and Kitai
1992). Neurons of the first group, characterized as
barosensitive type II, showed a very regular pattern of discharge
during baroreceptor inactivation elicited by hypotension, whereas at
normal values of arterial blood pressure, they were inhibited by
chloride-dependent inhibitory postsynaptic potentials (IPSPs)
synchronized to the cardiac cycle via baroreceptor inputs. A second
group of RVLM barosensitive neurons, characterized as type I, showed an
irregular pattern of discharge with excitatory postsynaptic potentials
(EPSPs) locked to the cardiac cycle. For both groups, aortic nerve
stimulation elicited polysynaptic IPSPs (Granata 1994,
1995). The existence of bulbospinal barosensitive neurons in
the RVLM with an irregularly firing pattern and responding with
polysynaptic IPSPs to aortic nerve stimulation was corroborated in
another in vivo study (Zagon and Spyer 1996).
Other investigators, using extracellular recording techniques in vivo,
found a group of neurons in the RVLM that had a very regular firing
pattern when glutamatergic receptors in the medulla were blocked
(Sun et al. 1988a). Furthermore, they also recorded from
similarly firing neurons in the in vitro medullary slice preparation
and concluded that, in the absence of baroreceptor and other inputs,
these neurons assume a regular firing pattern and could be considered
pacemaker-like (Sun et al. 1988a). However, the
characterization of sympathoexcitatory neurons in slice preparations is
seriously limited because the recorded neuronal activity cannot be
correlated with signal(s) clearly recognized as sympathetic outflow
activity, nor is it possible to evoke identifiable cardiovascular or
other sympathetic-related afferent inputs to those neurons. In an
effort to get around these limitations, we attempted in this study to
identify and characterize groups of neurons in the RVLM in vitro having
electrophysiological characteristics similar to those previously shown
in vivo by barosensitive neuron types I and II (Granata and
Kitai 1992). This method provides an indirect way for
identification of sympathetic-related neurons in vitro. Another
approach relied on histochemical techniques showing that the recorded
neurons project to the spinal cord (Sun et al. 1988b). However, this criterion is rather limited, since it is well known that
sympathoexcitatory neurons comprise only a very minor proportion of
neurons having descending axons. Although the in vitro slice preparations are at a disadvantage for electrophysiological
characterization of medullary sympathetic neurons, they are more
suitable than the in vivo preparation for investigating the
pharmacological properties of such neurons, because it is possible to
accurately control the administration of pharmacological agents during
stable intracellular recording.
It has long been accepted that adrenergic mechanisms are involved in
the central control of cardiovascular function (Kobinger 1978). Moreover, the RVLM has been depicted as an important
site of action of different pharmacological agents that affect
catecholaminergic transmission (Granata et al. 1986) and
that are used to treat vascular hypertension, such as the
2-adrenergic agonist clonidine, which is a well-accepted centrally
acting antihypertensive agent that binds to 2-adrenergic receptors
(Ernsberger et al. 1987). Furthermore, different
laboratories have investigated the effect of clonidine on barosensitive
bulbospinal neurons in the RVLM, but the results so far have been
divergent. For example, in the cat, clonidine applied by iontophoresis
inhibited only very few barosensitive bulbospinal neurons in the RVLM
(Clement and McCall 1991). On the other hand, in the
rat, clonidine inhibited only a subpopulation of bulbospinal
barosensitive neurons (Allen and Guyenet 1993).
Moreover, the latter research group, using an in vitro slice
preparation that included the RVLM, found that clonidine (1 µM) did
not affect the activity of neurons defined as putative sympathoexcitatory by those authors (Sun and Guyenet
1990).
The pharmacological mechanism responsible for the sympathoinhibitory
effect in the RVLM of clonidine and related drugs is still a matter of
discussion. Some investigators support the idea that these drugs reduce
central sympathetic activity by activating 2-adrenergic receptors
(Schmitt et al. 1973). On the other hand, for clonidine
as well as other imidazoline derivatives like moxonidine that are
effective for treating different forms of the hypertensive syndrome
(Prichard et al. 1997; Van Zwieten 1999;
Ziegler et al. 1996), it was speculated that they
mediate their central sympatholytic effect by activating imidazoline
receptors in the RVLM (Ernsberger et al. 1990).
In general, bulbospinal barosensitive neurons in the RVLM are
considered to be the sites of action of adrenergic agents mediating central sympathoinhibitory responses (Reis 1996). Yet a
critical question to be addressed is: what neuronal type(s) in the RVLM mediate these effects? Adrenergic neurons of the C1 group have direct
monosynaptic projections to sympathetic preganglionic neurons in the
spinal cord (Milner et al. 1988). Although there is no clear evidence that these neurons are directly involved in generation of tonic sympathetic activity, some investigators have proposed these
neurons as the targets of the sympathoinhibitory effect of hypotensive
agents like clonidine and other 2-adrenergic agonists (Ruffolo et al. 1993). In addition, it has been proposed
that clonidine affects neuronal transmission in the RVLM by activating presynaptic 2-adrenergic receptors (Tingley and Arneric
1990).
The aims of this series of experiments were as follows: first, to
further investigate in vitro the electrophysiological characteristics (including the firing properties) of neurons within the RVLM that have
similar characteristics to those defined in the same medullary region
in vivo; and second, to determine in these characterized neurons the
actions of the 2-adrenergic agonist clonidine, as well as the
possible block of clonidine effects by the selective nonimidazole
2-adrenergic antagonists rauwolscine and yohimbine. For this
purpose, we recorded intracellularly in vitro from neurons located in a
coronal medullary slice preparation that included the RVLM.
To quantify the firing patterns of the neurons, in particular with respect to regularity or irregularity, we used time-domain analysis (interspike intervals timing, auto- and cross-correlations) and frequency-domain analysis (autospectra). We formulated quantitative criteria that allowed us to distinguish groups of RVLM neurons having different properties.
| |
METHODS |
|---|
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
Experimental preparation
The experiments were done on male Sprague-Dawley rats (90-120 g) anesthetized with ether and decapitated. The brain was rapidly removed from the skull and placed in a dish containing cold Ringer solution (4°C), equilibrated with 95% O2-5% CO2, and containing the following (in mM): 124 NaCl, 5.0 KCl, 1.3 MgSO4, 1.25 KH2PO4, 2.0 CaCl2, 26 NaHCO3, and 10.0 D-glucose. The medulla with the caudal part of the pons was dissected and fixed with cyanoacrylic glue in a frontal plane to the cutting chamber of a vibratome (Oxford Instruments). The chamber was then filled with cold (4°C) oxygenated Ringer solution. The medulla was cut into 400-500 µm coronal slices containing the RVLM, which were transferred to an incubation chamber (Medical Systems) and maintained in oxygenated Ringer solution at 25-27°C for approximately 1 h. After that, one slice was transferred to a recording chamber; the bottom surface of the slice was placed on a filter paper held in place on a fine nylon mesh in the chamber, which was continuously superfused with oxygenated Ringer solution at a rate of 0.3-1.0 ml/min at 34 ± 1°C.
Recordings
Intracellular recordings were performed with microelectrodes
made from borosillicate (2.0 mm OD) filamented glass capillary tubing
and filled with a solution of 4-6% biocytin (Sigma) in 0.05 M of KCl
or KAcetate, with pH 7.0-7.6. Electrode resistances were 80-170 M
.
From neurons located between the rostral pole of the RVLM and a plane
500 µm caudally, intracellular potentials were recorded with a
low-noise, high-input-resistance bridge amplifier and displayed on a
Tektronix (8000 series) oscilloscope for on-line analysis. The signals
were also fed into a chart-recorder for monitoring the membrane
potential and membrane input resistance, and into a digital
oscilloscope (Nicolet), at sampling intervals of 2-50 µs with 12-bit
precision, at a maximum of 16K data points per record. Continuous
recordings were taken into a video cassette recorder via a
digital interface (NeuroData) at a sampling rate of 22 or 44 kHz.
Current application
During the recording of the continuous data stream, we made intermittent applications of trains (repetition rate, 0.5-1.0 Hz) of depolarizing or hyperpolarizing current pulses (380-ms pulse duration) with graded magnitudes. The neuronal membrane input resistance was calculated from the magnitudes of the current pulses and of the evoked intracellular voltage changes, which determined the current/voltage relationship (as displayed in Fig. 3F).
Off-line analysis
Signals were digitized by an A/D converter (RC Electronics) at a sampling rate of 2,500 or 5,000 Hz (0.4-ms or 0.2-ms bin duration, respectively) and transferred into the hard disk of a PC. Pulses (duration 0.4 ms or 0.2 ms) were derived from the spikes by a time-amplitude discriminator (WPI). By means of a customized program, portions of the digitized data stream were viewed on the monitor, and sections that did not include current pulses were marked to be used for data analysis. In long recordings obtained during drug perfusion, segments occurring at various times after start of perfusion were marked for later analysis of the time course of drug effects. The digitized data in these time slices (duration range, 30-100 s) were subjected to time-domain and frequency-domain analysis.
Interspike interval analysis
The interspike intervals in each data segment were measured, and the values of successive intervals were plotted (Fig. 1) to verify the stationarity of the sample. If the display indicated nonstationarity, as shown by the existence of a trend, another sample was chosen from the original data. The regularity of discharge of the spikes in the sample was ascertained by computation of the coefficient of variation (CV = SD/mean) of the interval distribution. Obviously, a nonstationary sample would furnish an incorrect value of the CV, because of the continuous change with time of the interval durations.
|
Correlation analysis
Autocorrelation histograms (ACHs) of spike-derived pulses and
cross-correlation histograms (CCHs) of pulses versus membrane potential
(MP) were computed. The correlograms were normalized using the square
root of the variance of each signal (Bendat and Piersol
1986). The spike-MP CCH furnished information on the temporal relation between spike occurrence and the slower depolarizing and
hyperpolarizing change of MP; for this purpose, the inclusion of both
positive and negative lags in the display was suitable.
Spectral analysis
The autospectrum of a neuron's spike activity was computed from
a data array derived by subjecting the spike-derived pulses to a
low-pass filtering operation (effective sampling rate 500 Hz) by means
of the sinc function (Christakos et al. 1991). The final
spectra were obtained by averaging the spectra computed from data
windows of 2.048-s duration, using a standard fast Fourier transform algorithm (Bendat and Piersol 1986).
The number of windows (range, 5-100) was determined by the duration of
the data sample. With these parameters, the spectral frequency
resolution was 0.49 Hz/bin. The autospectral strength was initially
quantified as relative power, defined as the power in each frequency
bin divided by total power (Bendat and Piersol 1986).
Several variables obtained from the autospectra were used to
characterize rhythmicity of firing: peak autospectral (fundamental)
frequency, number of harmonic peaks, and the rhythmicity index.
Rhythmicity index
For each data sample, this index was calculated as follows (Fig.
1): The baseline power was calculated as the mean of the power in the
flat portion of the autospectrum (i.e., the portion after decay of
rhythmicity). The peak power was defined as the amplitude of the peak
autospectral (fundamental) frequency, and the rhythmicity index was
defined as (peak power)/(baseline power). This index is similar to the
k ratio used by Sears and Stagg (1976) for
analysis of strength of cross-correlation.
Identification of the recorded neurons
Some of the recorded neurons were intracellularly labeled with
biocytin (n = 52) by passing 4- to 6-nA negative
rectangular pulses of 150-ms duration at 3.3 Hz for 6-10 min. After
recording, the slice containing the injected neuron(s) was transferred
to a fixative solution of 4% paraformaldehyde and 1.5% picric acid in
phosphate buffer saline (PBS) at pH 7.4, and stored at 4°C overnight.
After that, the slice was transferred to a solution of 30% sucrose in
0.15 M PBS for 4 h and then sectioned on a freezing microtome into
50-µm-thick sections, which were then rinsed several times in 0.1 M
phosphate buffer (pH 7.4) and processed for histofluorescence in the
intracellularly labeled neuron(s). Histological sections were incubated
for 4 h in avidin-Texas Red 1:200 in phosphate buffer solution
plus 0.1% triton X100 and 1% sodium azide. At the end of the
incubation, the tissue was rinsed several times in a fresh solution of
phosphate buffer for 1 h and the sections were mounted in a
solution of glycerol in phosphate buffer (1:3) plus 1%
n-propylgallate. Sections were examined under an epifluorescence microscope with a filter combination for rhodamine (G-2A). The location
of a filled neuron was determined and marked on a page from a
histological atlas (Paxinos and Watson 1986).
Drug administration
Drugs used in this study [clonidine hydrochloride (1-100 µM), rauwolscine hydrochloride (10 µM), and yohimbine hydrochloride (10 µM)] were dissolved in Ringer solution, which was applied by a gravity perfusion system.
| |
RESULTS |
|---|
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
Characterization of different neuronal types
In in vitro slice preparations of the rat (n = 17), intracellular recordings were obtained from 43 spontaneously
firing neurons and 10 silent neurons. These were found in a region of
the RVLM extending from the caudal border of the facial nerve nucleus
to a plane 500 µm caudal, and extending from the medial border of the
spinal nucleus of the trigeminal nerve to the lateral border of the
nucleus paragigantocellularis (Fig. 2).
The average number of neurons recorded per preparation was about three
(range, 1-6). The recorded neurons had resting MP between 
43 and
85 mV. The membrane input resistance was measured by injections of
intracellular current and ranged from 32 to 218 M.
|
The spontaneously active neurons could be further classified into two
groups: 1) neurons with a very regular (R) pattern of action
potential discharge and 2) neurons discharging action
potentials in an irregular (I) fashion, which had substantial
spontaneous postsynaptic potentials. The criterion for designating a
neuron as R-type or I-type was the CV of interspike intervals as
described in detail below. The mean MP for the R group was 54.74 ± 1.82 (SE) mV, and for the I group was 52.31 ± 1.25 mV; this
difference was not statistically significant.
The analysis of properties of a typical regularly firing R-type neuron is shown in Fig. 3. Neurons of this type had a spike afterhyperpolarization consisting of an initial rapid component, followed by a slower hyperpolarized component, which in turn was followed by a slow depolarizing potential that drove the membrane potential to the threshold for action potential generation (Fig. 3, A and B). This pattern can be seen in the spike-MP CCH of Fig. 3D.
|
When the firing of R-type neurons was suppressed by a hyperpolarizing
pulse, after cessation of the pulse the rhythm resumed with regular
interspike intervals such as those occurring before current injection
(Fig. 3A); this effect is similar to that observed in vivo
for barosensitive type II neurons (Granata and Kitai
1992). Furthermore, EPSPs were not observed during
hyperpolarization, whether it was produced by negative current pulses
(Fig. 3A) or by continuous negative current sufficient to
bring the membrane below the spike threshold level.
Another feature commonly found in R-type neurons (14/18 tested with
hyperpolarization pulses) was an initial relatively fast-developing anomalous rectification, indicated by a decay in the hyperpolarization (Fig. 3A). In this sample of R-type neurons, the
current/voltage relationship measured at the initial peak of the
hyperpolarization (before activation of the time-dependent inward
rectification) was found to be linear up to a range of 30 to 50 mV
(as in Fig. 3F).
The regularity of spike firing is indicated in the spike ACH (Fig. 3C) and autospectrum (Fig. 3E) by the occurrence of numerous narrow peaks (fundamental and harmonics) and was quantified by the CV of interspike intervals, which was 3% for this neuron.
Similar analyses for a typical irregularly firing (I-type) neuron are
shown in Fig. 4. The pattern of spike
afterhyperpolarization is shown in the spike-MP CCH (Fig.
4D). During application of hyperpolarizing current pulses,
the neuron's MP had spontaneous postsynaptic potentials (Fig. 4,
A and B) that increased in magnitude when the
neuron was hyperpolarized (Fig. 4B); at the offset of the
hyperpolarizing pulses, irregular firing resumed. During constant hyperpolarization (applied to allow increased magnitude of spontaneous EPSPs), depolarizing current pulses elicited spike firing (Fig. 4B). I-type neurons only rarely (4/19 tested) exhibited
anomalous rectification when the membrane potential was hyperpolarized
by negative current pulses of magnitude comparable to those applied to
R-type neurons (Fig. 4, A and B). In 13 of 18 I-type neurons tested, the current/voltage relationship showed an
inward rectification when the membrane was hyperpolarized more than
15 to 20 mV from resting membrane potential (Fig. 4F).
|
The limited rhythmicity of spike firing for this I-type neuron is indicated in the unit ACH (Fig. 4C) and the autospectrum (Fig. 4E) by the occurrence of only a single autospectral peak (i.e., absence of harmonics) and by the relatively high value (31%) of the CV of interval durations.
The CV of interspike intervals was adopted as a criterion of degree of regularity of a neuron's firing. To verify the stationarity of the interval distribution during the spike train, a plot was made of successive values of interval duration, as in Fig. 1, where the flatness (lack of trend) indicates stationarity.
For the sample of 43 spontaneously firing neurons, the CVs ranged from 3% to 40%. The plots of Fig. 5 show the relation between several discharge variables and the CV for the sampled population. As seen in Fig. 5A, the rhythmicity index, calculated from the unit autospectrum as in Fig. 1, for the 43 neurons ranged from 21.0 to 1.5, a higher index being associated with smaller CV. The plot seems to indicate separation between regions of lower and higher CV near CVs of 10-12%.
|
To obtain a more exact indicator of such separation, we plotted the
cumulative distribution (range, 0.0-1.0) of the rhythmicity index
(CUM. RHY. INDEX) versus CV values, as shown in Fig. 5B. There seems to be an inflection point (change of slope) at about CV = 12%. Therefore two linear regression lines were computed for
values of CV 12% versus >12%, respectively. The slopes, calculated as CUM. RHY. INDEX/CV, were as follows: left = 0.0679 and
right = 0.0078. It can be seen that there is a marked difference
of slope between the two regression lines (left/right slope = 8.70). Therefore the criterion for separation of two regions was set at
12% CV (indicated by vertical lines), with neurons having CV 12%
being designated as regular (R; n = 21), and neurons
having CV >12% being designated as irregular (I; n = 22).
An inverse relation also existed between CV and number of autospectral peaks (fundamental and harmonics; Fig. 5C), as well as between CV and the frequency value of the main (fundamental) autospectral peak (range, 4-34 Hz; Fig. 5D). However, the latter distribution was more scattered than the other distributions (Fig. 5, A and C). Nevertheless, the difference of spectral frequencies between the R and I groups (15.60 ± 1.58 vs. 10.42 ± 1.18 Hz, respectively) was statistically significant (P < 0.02).
The mean firing rate (number of spikes/sample time) was usually close (within 0.5 Hz) to the frequency value of the fundamental autospectral peak. This was due to the presence of a modal interval corresponding to the peak autospectral frequency.
The silent neurons (n = 10) were induced to fire action
potentials (with irregularly distributed intervals) by injecting
depolarizing current (Fig.
6B). These neurons had resting
membrane potentials that ranged from 50 mV to 72 mV, with a mean
value of 61.5 mV. This value was more negative than that observed in
either R-type or I-type neurons, as mentioned earlier in this section, and it was statistically significant for both types (t-test,
P < 0.001). The silent neurons had longer action
potential durations (Fig. 6A) than the spontaneously firing
neurons (Fig. 3B): 4.20 ± 0.30 ms vs. 1.68 ± 0.50 ms, respectively. This difference was statistically significant
(t-test, P < 0.001). There was no
significant difference of action potential duration between the R-type
and I-type neurons. The current/voltage relationship in silent neurons (Fig. 6C) was linear 20 mV below the spontaneous membrane
potential in nearly all neurons tested.
|
The silent neurons were localized in an area close to the ventral medullary surface (<100 µm), whereas spontaneously firing neurons were broadly distributed in the RVLM. The morphology of five biocytin-labeled silent neurons showed somata predominantly polygonal in shape. On the other hand, the labeled cell bodies of spontaneously active neurons of either R-type (n = 6) or I-type (n = 7) were fusiform, oval, and polygonal in shape. The size of the cell bodies was similar for both silent and spontaneously active groups, with ranges of 15-55 µm in the long axes and 10-28 µm in the short axes.
Effects of clonidine on neurons of different types
After an 8- to 12-min period of recording from spontaneously firing neurons (to verify stability of recording conditions), clonidine (1-100 µM) was applied by superfusion while continuously recording from 26 RVLM neurons (10 of R-type and 16 of I-type). The onset of change of membrane potential produced by clonidine varied between different neurons (range, 55-220 s from the start of perfusion). For 21/26 neurons, there was eventually almost complete suppression of firing after a time of perfusion of 180-2,100 s. For 18 neurons, there was an initial excitation (increase of discharge frequency and increased rhythmicity of discharge). Several patterns of response were observed, as summarized in Table 1. Several different patterns of response to clonidine could be found in each individual slice.
|
Regularly firing neurons
For R-type neurons (Table 1, left pair of columns), the most
common pattern of response to clonidine (1 µM or 10 µM), found in
six neurons (example in Figs.
7-9),
was first a short-lasting moderate hyperpolarization (3-5 mV from the
control membrane potential) and increase in input resistance. After
that, the neuron started to depolarize, the discharge frequency
increased, and the membrane input resistance decreased (Figs.
7B and 8B). Subsequently, the neuron became
further depolarized (10-18 mV from the control membrane potential),
and now the discharge frequency was reduced (Fig. 7, C and
D). Finally, the neuron became more depolarized (8-12 mV
from the previous level shown) and almost ceased firing (Figs. 7E and 8C). At this point, the anomalous
rectification was abolished, and the membrane input resistance was
moderately increased (Fig. 8D, 
) over the control level
(Fig. 8D, 
). After washout with normal Ringer solution,
the changes were partially reversed (Fig. 7F).
|
|
|
The effects of clonidine on the neuron's rhythmicity over the time course of perfusion are shown in the unit autospectra and spike-MP CCHs of Fig. 9. The fundamental autospectral frequency in the control state was 14.2 Hz (Fig. 9A), with prominent rhythmicity, as indicated by the presence of many harmonic peaks and a value of 7% for the CV of interval duration. The excitation in the earlier portion of clonidine perfusion encompassed 1) increase of autospectral frequency to 21.0 Hz after 410 s (Fig. 9B) and to 23.4 Hz after 683 s (Fig. 9C) and 2) increase of rhythmicity index from 6.4 in Fig. 9A to 9.4 in Fig. 9B and to 10.4 in Fig. 9C. Thereafter, during the continued depolarization, there was depression of activity, indicated in Fig. 9D (1,206 s after start of perfusion) by reduction of spectral frequency to 5.4 Hz and reduction of rhythmicity index to 2.9. After the sample of Fig. 9D, with continued clonidine infusion the neuron became silent (Figs. 7C and 8C). Finally, after washout with normal Ringer solution, the changes were partially reversed, as indicated in Fig. 9E by resumption of discharge, with a peak autospectral frequency of 6.3 Hz and rhythmicity index of 4.5. The spike-MP CCHs (Fig. 9, right panels) show a MP pattern similar to that in Fig. 3: postspike afterhyperpolarization followed by a slow depolarizing potential.
A somewhat different pattern of response to clonidine was shown by
two other R-type neurons: there was no excitation, but rather only
hyperpolarization was elicited (12 to 18 mV), and there was a
marked decrease of firing rate and a moderate decrease or no change of
membrane input resistance (750-960 s after clonidine infusion).
Finally, for another R-type neuron, clonidine elicited only
depolarization and excitation; and for still another R-type neuron,
clonidine had no effect. It might be relevant that this last neuron had
a spontaneous autospectral peak (at 34 Hz), which was the highest in
the sampled population.
Irregularly firing neurons
Several patterns of response were observed. In six I-type neurons (Table 1, right columns; example in Figs. 10-12), clonidine (1 µM or 10 µM) first produced a transient (520-650 s after start of clonidine infusion) membrane depolarization (3-5 mV from the control membrane potential), with increased discharge frequency and a 10-20% increase in membrane input resistance (Figs. 10B and 11B). After that (1400-1650 s after start of clonidine infusion), the membrane potential changed in the opposite direction, becoming hyperpolarized (4-6 mV from the control membrane potential) with substantial reduction of discharge frequency and a prolonged augmentation in input resistance (Fig. 10C). Finally, we observed a further hyperpolarization and reduction in firing frequency while the membrane resistance remained elevated (Fig. 10, D and E, and Fig. 11, C and D).
|
|
|
The effect of clonidine on the neuron's rhythmicity over the time course of perfusion are shown in the unit autospectra and spike-MP CCHs of Fig. 12. The neuron had relatively weak rhythmicity in the control state (Fig. 12A), as indicated by the presence of only one autospectral peak at 16.6 Hz. Clonidine initially (Fig. 12B, 540 s after start of perfusion) produced only a moderate excitation (spectral frequency increase to 22.0 from 16.6 Hz) with little change of rhythmicity index (from 3.4 to 3.8). Thereafter (Fig. 12C, 1,500 s after start of infusion) there was depression of activity due to hyperpolarization, with decrease of peak spectral frequency (from 22.0 to 12.3 Hz) and rhythmicity index (from 3.8 to 2.2). After further infusion (Fig. 12D, 1,700 s after start), with additional hyperpolarization there was complete loss of rhythmicity, as indicated by the flatness of the autospectrum, together with further decrease of mean firing rate (to 2.2/s from 12.1/s in Fig. 12C). Finally, after 450 s of drug washout (Fig. 12E), there was partial recovery: increase of mean firing rate (from 2.2/s to 3.9/s) but no recovery of rhythmicity. The spike-MP CCHs (Fig. 12, right panels) show a MP pattern similar to that in Fig. 4: spike afterhyperpolarization followed by recovery of membrane potential.
In another type of response, shown by three I-type neurons, clonidine (1 µM or 10 µM) also initially elicited depolarization with increase in firing rate and spectral frequency, but thereafter, instead of hyperpolarization, there was further depolarization accompanied by a dramatic decrease of firing frequency and decrease of membrane input resistance. This kind of response was evoked in two additional I-type neurons by infusion of 100 µM clonidine.
In another four I-type neurons, clonidine (1 µM or 10 µM) did not elicit an early excitation but did elicit the hyperpolarizing response with a marked decrease of firing rate and increase of input resistance. A similar type of response was evoked in one additional I-type neuron by infusion of 100 µM clonidine. In two I-type neurons, clonidine (1 µM or 10 µM) evoked only a moderate depolarization and increase in firing rate. Finally, in one I-type neuron, clonidine did not produce any effect.
For both R-type and I-type neurons, clonidine did not change the duration or shape of the action potentials.
For 21/26 neurons, clonidine perfusion (1-100 µM) eventually produced depression of firing to the point of silence. In addition, 18/26 neurons were excited earlier in the perfusion. The later depression occurred in both R-type and I-type neurons and was associated with hyperpolarization/inhibition in 12 neurons and with depolarization in 9 neurons. Thus the predominant effect of clonidine was eventually depression of firing to the point of silence.
Effects of -adrenoceptor antagonists on clonidine
responses
We studied the actions of the 2-adrenoceptor
antagonists rauwolscine (4 neurons) and yohimbine (2 neurons) on the
effects produced by clonidine. In all cases, the blockers were perfused alone for a period of 15-20 min before perfusing the slice with a
solution containing both the agonist and the antagonist.
For three R-type neurons (example in Figs. 13 and 14), rauwolscine (10 µM) applied in the superfusate elicited a moderate depolarization and increase of firing rate without change of membrane input resistance. This moderate excitation is indicated by the changes in spectral variables seen in Fig. 14A (control) versus Fig. 14B (910 s after start of rauwolscine perfusion): increase of peak autospectral frequency from 6.8 to 9.8 Hz and moderate increase of rhythmicity index from 5.6 to 6.5. Subsequent application of clonidine (1 µM) in the presence of rauwolscine did not produce the usual inhibitory effect on firing (Fig. 13, A and B), i.e., rauwolscine blocked the inhibition. However, in the presence of rauwolscine, there was still a moderate excitation by clonidine, as indicated in Fig. 14, B-D: increase of rhythmicity index from 6.5 in Fig. 14B to 7.9 in Fig. 14C (286 s after clonidine) to 8.8 in Fig. 14D (after an additional 428 s of clonidine perfusion). Finally, after 1,200 s of washout of both agents (Fig. 14E), there was a tendency toward reversal of the moderate excitatory effects of both agents, as can be seen by comparison of Fig. 14E (after washout) with Fig. 14A (before drug application). The spike-MP CCHs (not displayed) for this neuron show that there was little difference of MP pattern (except for a moderate change of spectral frequency) between the condition after rauwolscine infusion (Fig. 14A) and after subsequent clonidine infusion (Fig. 14B).
|
|
Finally, for one I-type neuron, the inhibitory effect of clonidine (1 µM) was not blocked by prior infusion of rauwolscine (10 µM).
The effects of the antagonist yohimbine were analyzed for two I-type neurons. The example of Fig. 15 shows that yohimbine (10 µM) did not block the inhibitory effects of clonidine (1 µM), since the increased negativity of membrane potential, increase of membrane resistance, and decrease of firing rate produced by clonidine still occurred after prior application of yohimbine. Similar results were obtained in the second neuron.
|
| |
DISCUSSION |
|---|
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
Types of neuronal discharge
The results presented in this study support our earlier proposal
(Granata 1995; Granata and Kitai 1992)
that the RVLM contains two different types of spontaneously active
neurons. A first group exhibits a very regular pattern of action
potential discharge, whereas a second group has an irregular pattern of
firing and shows spontaneous postsynaptic potentials. A third group of
neurons is normally silent, and action potentials occur only during
neuronal depolarization elicited by intracellular injection of positive current.
In the study of sympathetic neural discharge, the existence of rhythms
in population and unit discharges, at both peripheral and central
levels, has merited the attention of researchers (Barman and
Gebber 2000). These rhythms are pertinent as 1)
markers of sympathetic discharge and 2) features that could
aid the study of mechanisms producing sympathetic tone. In studies of
the rat in vivo, the presence of rhythms locked to the cardiac cycle
(presumably via baroreceptor afferents) in brain stem unit discharges
has served to identify those units as being involved in sympathetic activity (Granata and Kitai 1992; Sun et al.
1988a; Zagon and Spyer 1996), but there have
been no reports of other types of fast rhythm, such as the "10-Hz"
rhythm found in the cat (Barman and Gebber 2000).
Another type of rhythm in RVLM neurons of the rat is a regular
discharge that was found after kynurenate application (Sun et
al. 1988a) and was designated as "pacemaker-like". In
addition, similar unitary rhythms have been found in in vitro
preparations, such as brain-stem/spinal cord preparations
(Oshima et al. 2000) and the medullary slice preparation
(Granata 1995; Kangrga and Loewy 1995;
Lewis and Coote 1993; Piguet and Schlichter
1998; Sun and Reis 1994; Sun et al.
1988b).
In the original report on regularly firing neurons (Sun et al.
1988a), a neuron's discharge was characterized as regular if there was marked periodicity in its spike-triggered interval histograms (a variant of autocorrelation). The designation of regularity was also
based on examination of superimposed spike-triggered oscilloscope
sweeps (Granata and Kitai 1992) as well as on plots of
the n'th versus the (n + 1)'th interval duration (Lewis and Coote 1993). However, in other reports (Oshima et al.
2000; Piguet and Schlichter 1998), no
quantitative criterion for regularity was stated. Moreover, none of the
published reports quantify the degree of firing irregularity.
Therefore, in this study, we developed methods for quantifying the
degree of regularity or irregularity of a neuron's discharge.
We used as the primary criterion for regularity the CV of the neuron's interspike interval duration during a stationary period of discharge. (The requirement for stationarity led us to use only time slices where there was no secular trend in the data.) The CV was related to several other variables that indicate rhythmicity. 1) Rhythmicity index (Fig. 5A), which consists of the normalized power amplitude of the fundamental frequency peak (peak amplitude/baseline amplitude), was inversely related to the CV of interval durations. The two variables are intrinsically related in a reciprocal manner, because a low CV means that adjacent interval durations are close in value, which results in stronger rhythmicity as indicated in the autocorrelation and the autospectrum. 2) Number of autospectral peaks (fundamental plus harmonics; Fig. 5C) was also inversely related to CV, since strength of rhythmicity is reflected in the Fourier transform as number of harmonics. 3) Peak spectral frequency (Fig. 5D) had a more dispersed relation to CV, even though there was still a significant tendency for an inverse relation between frequency and CV values. This observation indicates that neurons with similar spectral frequencies may differ somewhat in their CVs.
The cumulative distribution of rhythmicity index versus CV values (Fig.
5B) had a shape that was neither linear nor sigmoidal. However, there seemed to be an inflection point near CV = 12% that suggested a separation of the curve into two segments, indicating existence of distinct ranges having lower or higher CV values. To
evaluate this apparent separation, we computed and plotted (Fig.
5B) two linear regression lines, using values of the index in different ranges of CV: 1) CV values 12% (left line)
or 2) CV value >12% (right line). Comparison of the two
regression lines showed that the slope of the left line was 8.70 times
the slope of the right line. Therefore we chose a CV of 12% (indicated
by vertical lines in Fig. 5) as the borderline value for designation of
firing as regular (R; CV 12%) versus irregular (I; CV > 12%). Of 43 neurons recorded, 21 were classified as R-type and 22 as I-type. We consider this a useful distinction, even though there is
some inexactitude of type assignment at the borderline region between
the two subdistributions.
The R-type neurons are characterized by the presence of an
afterhyperpolarizing potential, followed by a slow depolarizing potential. This observation suggested that various conductances could
play a critical role in generating regular firing, both in vitro and in
vivo (Granata 1995; Granata and Kitai
1992). In addition, many of these neurons had a rapidly
developing anomalous rectification that could also contribute to
generation of nonsynaptically mediated R firing, as reported for
neurons in other systems (Yarom and Llinas 1987). During
hyperpolarization below the threshold for action potential generation,
EPSPs were not observed, and the R firing pattern resumed after release
of activity suppression at the end of a hyperpolarizing pulse. The
presence of this type of neuron in the RVLM of the in vitro slice
preparation has been reported by other investigators (Kangrga
and Loewy 1995; Lewis and Coote 1993; Sun
et al. 1988b). It was also found that the regular firing of
some of these neurons was not dependent on synaptic inputs, since it
persisted after
low-Ca2+/high-Mg2+
application, in both the slice preparation (Sun and Reis
1994) and in the brain stem/spinal cord preparation
(Oshima et al. 2000).
The irregularly firing neurons produced spontaneous postsynaptic potentials that increased in magnitude with greater hyperpolarization; thus some of these potentials could have been EPSPs. Only a few neurons in this group displayed anomalous rectification when the neuron was hyperpolarized by pulses of negative current. In addition, in this group we found that the current/voltage curve had an inward rectification.
Two major differences between the characterized R and I neurons were as
follows. 1) There was a higher incidence of anomalous rectification in R neurons (14/18) than in I neurons(4/19); this was
highly significant (P < 0.001 by the
2 test). 2) The mean autospectral
frequency was greater for the R than for the I neurons (15.6 and 10.4 Hz, respectively); this was highly significant (P < 0.02 by the t-test). However the relation between frequency
and CV was dispersed (Fig. 5C), i.e., neurons with similar
CV values could have considerably different frequencies. Finally, there
was no significant difference between R and I types for two other
variables: action potential duration and resting membrane potential.
Since in the in vitro slice preparation there is no certain way of
ascribing sympathetic-related function to the recorded RVLM neurons, it
is necessary to compare their firing properties with those of RVLM
neurons recorded in vivo. In a rat in vivo preparation (Zagon
and Spyer 1996), intracellular recordings showed the presence
in RVLM of bulbospinal barosensitive neurons with membrane potential
oscillations and spike firing rhythms locked to the cardiac cycle, in
the form of either a rhythmic hyperpolarization or depolarization. In
addition, aortic depressor nerve stimulation produced in those neurons
polysynaptic IPSPs as well as biphasic EPSP-IPSP sequences. Similar
results were found in another study using in vivo rat preparations
(Granata 1995; Granata and Kitai 1992),
where bulbospinal barosensitive neurons were found in the RVLM that
were designated as type I and type II, having electrophysiological characteristics that were similar to those of the I and R type neurons,
respectively, of the present study. Of particular interest was the
observation that some type II neurons had a very regular firing
patterns during a period of reduced baroreceptor input (hypotension),
but when normotension was restored the discharges became modulated in
synchrony with the cardiac cycle (Figs. 7A and 9A
in Granata and Kitai 1992). In that study, we analyzed 8 type II neurons and 17 type I neurons. We also analyzed 12 bulbospinal neurons (Fig. 5 in Granata and Kitai 1992) with a very
regular discharge that had no temporal relation to the cardiac cycle. Thus in both in vitro and in vivo preparations there was heterogeneity of neuronal discharge types.
Another group of investigators (Lipski et al. 1996),
using intracellular recordings in the in vivo rat preparation, has
reported conflicting results: they could not find regularly firing
bulbospinal barosensitive neurons. This negative result could have been
due to technical deficiencies, since R-type neurons tend to be smaller than I-type neurons, as reported by Oshima et al.
(2000), for the brain stem/spinal cord preparation. The
small-sized neurons in the study of Lipski et al. (1996)
could have been preferentially damaged, because even some of the larger
(I-type) neurons in that study showed signs of damage (Fig.
8C in Lipski et al. 1996), thus indicating
that the technique used was not suitable for penetration of the smaller
R-type neurons in the RVLM.
Our positive results show that there is a significant population of
R-type neurons in the RVLM of both in vivo preparations (Granata
and Kitai 1992) and slice preparations (present study). Confirming evidence has been supplied in a recent study by
Oshima et al. (2000), who used a whole cell patch-clamp
technique in the neonatal rat brain stem/spinal cord preparation, which
preserves the sympathetic neural network(s). They demonstrated the
presence in the RVLM of two types of spontaneously active neurons
(regularly and irregularly firing) as well as of silent neurons; these
neuronal types resemble those of the present study.
At present we are not in a position to provide a detailed explanation
of the mechanisms of generation of tonic sympathoexcitatory activity in
RVLM presympathetic neurons. A well-known hypothesis is that
sympathetic activity is generated by neuronal networks in the medulla
(Barman and Gebber 2000). Another hypothesis is that
sympathetic activity is generated by pacemaker-like neurons in the RVLM
(Sun et al. 1988a). Our own data suggest that a more neutral term, such as "endogenous oscillators," be used for this population. We might speculate that the I neurons are more involved in
synaptically mediated network interactions, whereas the R neurons function to provide background excitation to the network(s). Moreover, since both the R and I groups are found in the RVLM, it seems reasonable that interactions between the two populations may promote generation of sympathetic tone, as was proposed in earlier publications (Granata 1995; Granata and Kitai 1992).
Our quantitative criteria for regularity of firing may be useful for
evaluation of possible interactions between different groups of
neurons. 1) Although we distinguish between R and I firing
by means of the location of the inflection point in the CV distribution
(12% vs. >12%, respectively), there was a range of values around
this point, indicating a possible transition zone between the two
patterns. 2) It is noteworthy that even the most irregular
neurons still had a moderately low value of CV (maximum CV in the
sample was 40%); and 42/43 of the neurons had at least one
autospectral peak during the control state. Thus the distribution of
intervals was almost never random. (By comparison, for a Poisson
distribution the ratio variance/mean of the intervals has a value = 1.0.) Only when a neuron was nearly silent (e.g., as in Fig.
12D) did all rhythmicity disappear. This lack of complete randomness in firing of I-type neurons could be due to a component of
endogenous oscillation based on membrane conductances or weak synaptic
input from R-type neurons. 3) The increased firing and increased rhythmicity of a neuron's discharge during the initial period of clonidine perfusion was usually associated with a reduction of CV (e.g., as in Fig. 1, from 29.0% to 4.5%). This lability indicates that the neuronal type can change with physiological conditions.
The present observations in vitro cannot determine the proportion of
neurons in our sample that correspond to presympathetic neurons in
vivo. The same limitation is applicable to other studies using similar
preparations, despite a study where histochemical observations
indicated a bulbospinal projection (Hayar and Guyenet 2000), since in that study there was no specification of firing pattern.
Effects of clonidine on putative sympathoexcitatory neurons in the RVLM
Our present results in vitro indicate that the 2-adrenergic
agonist clonidine inhibits RVLM neurons having electrophysiological characteristics similar to those of RVLM populations in vivo
(Granata and Kitai 1992). This suggests that clonidine
applied in vivo could elicit reduction of arterial blood pressure by
inhibiting different types of presumed barosensitive RVLM neurons that
are thought to produce sympathoexc
, silent
neurons. IO, inferior olive; NA, nucleus ambiguus; NTS, nucleus tractus
solitarius; RO, nucleus raphe obscurus; RVL, rostral ventrolateral
medulla; STN, spinal trigeminal nucleus; STT, spinal trigeminal tract;
X, dorsal vagal motor nucleus; XII, hypoglossal motor nucleus.
