| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1Laboratoire de Neurobiologie des Réseaux Sensorimoteurs, Centre National de la Recherche Scientifique, Unité Mixte de Recherche 7060, Université Paris 5, Centre Universitaire des Saints-Pères, 75270 Paris Cédex 06, France; and 2Laboratoire de Neurosciences, Université de Mons-Hainaut, 7000 Mons, Belgium
Submitted 18 December 2002; accepted in final form 12 March 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
;
Dieringer 1995;
Smith and Curthoys 1989;
Vibert et al. 1997). The
static deficits, which include postural distortions and a spontaneous ocular
nystagmus, disappear over 3–4 days in the guinea pig and rat as in most
mammalian species (Darlington et al.
2002; de Waele et al.
1989; Sirkin et al.
1984; Smith and Curthoys
1989). The dynamic deficits include a reduced gain and abnormal
timing of the vestibuloocular and vestibulospinal synergies. They partially
improve over several weeks, but this recovery is limited to low acceleration
and to the low and middle frequency range (0.1 to 
10 Hz) of head movement
(Broussard et al. 1999;
Gilchrist et al. 1998;
Hamann et al. 1998;
Lasker et al. 2000;
Vibert et al. 1993).
After unilateral labyrinthectomy (UL), the ipsilesional vestibular nucleus
neurons (VNn) lose the excitatory drive provided by labyrinth afferents and
become silent. In contrast, the spontaneous discharge of the contralateral VNn
increases. This imbalance is responsible for the deficits triggered by the
lesion. During the first week after UL, the recovery of a normal resting
discharge by the ipsilesional medial VNn (MVNn) plays a key role in the
disappearance of the static syndrome by restoring the balance between the
activity of neurons in both vestibular nuclei (Ris et al.
1995,
1997). Because the
ipsilesional labyrinth afferents stay silent
(Jensen 1979;
Sirkin et al. 1984), this
recovery is a model of plasticity in the CNS.
Vestibular compensation involves modification of both the intrinsic
properties of the ipsilesional MVNn and the vestibular-related networks in
which they are embedded (Cameron and Dutia
1997; Darlington and Smith
1996; Darlington et al.
2002; Vibert et al.
1999a,
2000;
Yamanaka et al. 2000). We
recently suggested that in rodents, the spontaneous discharge recovered in
vivo by the ipsilesional MVNn might be more and more sustained by changes in
the intrinsic properties of MVNn themselves as the time of compensation
increases (Vibert et al.
1999b). Extracellular recordings on slices taken from
previously labyrinthectomized animals show an increase of the spontaneous
discharge of the ipsilesional MVNn compared with control slices, and/or to
contralesional neurons during the first week of compensation
(Cameron and Dutia 1997;
Ris et al. 2001a;
Vibert et al. 1999b; for
review, see Darlington et al.
2002). In particular, a significant increase in the firing rate of
MVN neurons can be detected as early as 4 h after the lesion in the rostral
third of the nucleus in rats (Cameron and
Dutia 1997). In the guinea pig, we and others have shown that this
increase of the spontaneous firing rate of ipsilesional neurons becomes
stronger when the slices are taken after 1 or 2 mo instead of 1 wk
(Darlington et al. 1989;
Vibert et al. 1999b). This
late in vitro change is not directly involved in the initial recovery of a
normal discharge rate by the ipsilesional MVNn observed in vivo, which is
achieved by the end of the first week after UL. However, it is concomitant to
the recovery of the dynamic synergies triggered by low-acceleration stimuli
that has been described a few weeks after the lesion
(Gilchrist et al. 1998;
Vibert et al. 1993) (see
preceding text).
In vitro intracellular recordings have led to the identification
of two main types of MVNn, the type A and type B neurons, according to their
membrane properties (Gallagher et al.
1985; Him and Dutia
2001; Johnston et al.
1994; Serafin et al.
1991a,b).
It is generally admitted that MVNn represent a continuum of cells whose
properties are distributed between those of two canonical types of neurons,
the type A and B MVNn (Du Lac and
Lisberger 1995a). Studies using intracellular or whole cell
patch-clamp recordings have confirmed that some membrane and response
properties of the ipsilesional MVNn were modified 7–10 days after UL
(see also DISCUSSION). Godaux and Ris
(2001) and Him and Dutia
(2001) have reported an
increase in the proportion of type B MVNn displaying low-threshold calcium
spikes. Him and Dutia (2001)
demonstrated that the average resting membrane potential of type B MVNn was
depolarized by 3 mV compared with control cells, while their input resistance
was increased by 10–15%. Ris et al.
(2001c,
2002) reported that the
nonlinear "overshoot" induced by ramp-like currents was increased
in type B but not in type A ipsilesional MVN neurons. In contrast, the
sensitivity of MVNn to steady-state current injection was not modified.
To determine how the properties of the deafferented MVNn were modified after longer times of compensation, we performed intracellular recordings of MVNn on slices taken from animals 1 mo after a UL and determined their static and dynamic membrane properties. We measured the responses of the ipsilesional MVNn to steps, ramps, and sinusoidal currents of various amplitudes and frequencies. The results were compared with data obtained using the same stimuli for MVNn recorded on control slices.
On all previous publications, MVN neurons have been categorized into type A and B neurons using only qualitative criteria. To assess reliably the long-term effects of UL on this heterogeneous population of cells and remove possible biases in the classification of MVNn, quantitative, objective criteria were developed to characterize the intracellularly recorded MVNn. These criteria were set from the sample of 89 cells recorded in control slices, and then the neurons recorded on the ipsilesional side of slices taken from labyrinthectomized animals were classified in exactly the same way.
We assume that a large majority (>80%) of the MVNn we record on slices
are second-order vestibular neurons, which, in previously labyrinthectomized
animals, have lost labyrinthine input at the time of the lesion. Indeed,
80–85% of the central vestibular neurons recorded in the MVN area of the
isolated whole brain of guinea pig, using similar electrodes to those used on
slices, could be identified as second-order vestibular neurons
(Babalian et al. 1997). This
high proportion of second-order cells is in agreement with previous anatomical
(Carleton and Carpenter 1983;
Sato and Sasaki 1993) and
physiological studies (Chen-Huang et al.
1997; Goldberg et al.
1987).
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Experiments were carried out on pigmented guinea pigs of both genders
(Elevage de la Garenne, Saint-Pierre d'Exideuil, France). The animals were
handled in accordance with the European Communities Council Directive of
November 24, 1986, and following the procedures issued by the French
Ministèredel'Agriculture. The guinea pigs used to obtain control slices
(i.e., intact animals, n = 60) had a mean age of 5 wk (range:
4–8 wk) and a mean weight of 250 g (range: 150–400 g).
Because of the compensation time, the slices obtained 1 mo after UL were from
slightly older animals aged 7–9 wk (n = 26), whose mean weight
was 320 g (range: 250–450 g). Compared with the rat, guinea pig is
a precocious species where the CNS is almost mature at birth (after a
9-wk-long gestation period) and postnatal maturation of the CNS is
minimal (Dobbing and Sands
1970; Nacher et al.
2000). Besides, there was no difference in the membrane properties
of the intracellularly-recorded MVNn obtained from the smallest (150–200
g) versus the largest (300–400 g) intact guinea pigs, which proves that
the properties of MVNn recorded in control slices are not significantly
modified within the age range of the animals that we used.
ULs were performed under halothane anesthesia with the help of an operating
microscope as described in Vibert et al.
(1999a,b).
The semicircular canals, utricle, and saccule were exposed via a
retroauricular approach. The bony labyrinth was drilled, and the ampullae of
all three canals and the otolithic maculae were removed using suction. The
guinea pigs were allowed to compensate in a normal visual environment until
their brain was removed to prepare the slices.
Intracellular electrophysiological recordings
Thick (500 µm) coronal brain stem slices were cut and maintained using
standard techniques (Gallagher et al.
1985; Serafin et al.
1991a; Vibert et al.
1999b). Intracellular electrophysiological recordings were
obtained with sharp, 3 M potassium acetate-containing glass microelectrodes
from neurons within the medial vestibular nucleus (MVN), taking the border of
the IVth ventricle as a landmark.
Given that MVN neurons constitute a heterogeneous population, it was
critical to be sure that we sampled the same populations of neurons before and
after labyrinthectomy. Otherwise, the differences reported in the following
text could be an artifact of having recorded from different populations of
cells. In particular, we have observed that the proportions of type A and B
MVNn were varying along the rostrocaudal extent of the nucleus (N. Vibert, M.
Serafin, M. Mühlethaler, unpublished data). Other authors have
demonstrated that the cellular changes associated with vestibular compensation
can be different in the caudal and rostral parts of the MVN
(Cameron and Dutia 1997;
Yamanaka et al. 2000). To
control for this variability, we decided to restrict as much as possible our
recordings to the two 500-µm coronal slices corresponding to the middle
third of the guinea pig MVN, at the level of the cerebellar peduncles. Only a
few cells (<10%) were recorded in more caudal slices corresponding to the
last third of the MVN, both in normal and previously labyrinthectomized
animals. The same investigator, who always used the same experimental setup
and similar electrodes, obtained all data from previously labyrinthectomized
animals as well as the majority of the control data. The other part of the
control data were from a previously published study
(Ris et al. 2001b) and was
obtained by a different investigator (see RESULTS). There was no
significant difference between the two data sets.
All measurements were done with an Axoclamp 2A system (Axon Instruments,
Union City, CA) in either the bridge or switching discontinuous current-clamp
(DCC) mode (Moore et al.
1993). The electrode resistance varied from 80 to 150 M
.
Both series resistance (bridge balance) and capacitance compensation were
checked throughout the recording of each individual neuron
(Ris et al. 2001b). Part of
the current injections and all data acquisition were performed with a
PC-compatible computer using the "Acquis 1" program (version 4.0,
Bio-logic S.A., Gif-sur-Yvette, France). The sampling rate used for
acquisition varied between 2,000 and 5,000 Hz, depending on the length of the
data-acquisition sequence. Consequently, the amplitudes of the digitized
spikes were variable; however, oscilloscope traces verified that the size of
the action potential was constant at any given membrane potential. The data
were analyzed using program scripts with Mathematica 4.0 (Wolfram Research,
Champaign, IL), or MATLAB 6.5 (The MathWorks, Natick, MA). To minimize the
possibility that experimenter biases might affect the results, the same
scripts were used to quantify the properties and responses of MVNn recorded on
control slices and slices taken from previously labyrinthectomized
animals.
Basic membrane and firing properties of MVNn
Because most MVNn are spontaneously active on slices, the potential was filtered with a 1-Hz low-pass filter to obtain an estimate of its average resting level that was taken as the "mean resting membrane potential" of each neuron. For each cell, this membrane potential value was corrected by measuring and subtracting the extracellular voltage offset found after removal of the electrode from the neuron. No correction was made for liquid junction potentials, but this can be assumed to be constant between slices taken from control and previously labyrinthectomized animals given that both sets of MVNn were recorded using similar electrodes.
The same criteria were used to evaluate the quality of intracellular recordings and select the neurons used for statistical analysis in control slices and in slices taken from previously labyrinthectomized animals. All cells that had resting membrane potentials more negative than –50 mV and spike amplitudes >50 mV were automatically retained. In both types of slices, we also included in the sample the MVNn whose membrane potential ranged from –50 to –40 mV if they displayed spike amplitudes >50 mV and a normal spike width. The number of such cells was 5 of 89 MVNn in control slices and 17 of 78 MVNn in slices taken from previously labyrinthectomized animals. Spike width at threshold was considered to be normal if it stayed within the range of the spike widths (0.70–2.20 ms) measured for the neurons that had resting potentials more negative than –50 mV and spike amplitude >50 mV.
Recordings of the neurons at rest (i.e., with no holding current being
injected through the recording electrode) were used to calculate their mean
spontaneous firing rate, its coefficient of variation (CV) expressed as a
percentage, and to measure the mean amplitude of the spike. For each neuron,
an average of the spike shape and following inter-spike interval profile was
obtained by averaging successive spontaneous spikes taken either at the
resting membrane potential or while the cell was slightly depolarized (for the
few neurons that were silent at rest). The spikes (mean number of 120)
were synchronized to their thresholds, taken at the point on the rising phase
of the action potential where the slope of the potential trace reached an
arbitrary threshold of 10 V/s (Krawitz et
al. 2001). The averaged spike shape was used to determine the
amplitude of the afterhyperpolarization (AHP) and the width of the spike
(taken at threshold). The AHP amplitude was calculated as the membrane
potential difference between spike threshold and the membrane potential
minimum after the falling phase of the spike.
The cell's firing threshold, i.e., the membrane potential for which the
cells begin to fire action potentials (in mV), was assessed as the potential
reached by the neuron at the threshold of the first spike triggered by a slow,
depolarizing current ramp (see following text and
Fig. 2D). For each
cell, we determined whether long-lasting, subthreshold plateau potentials
could be triggered by low-amplitude (0.1–0.2 nA), short-duration (10 ms)
current pulses delivered while the neuron was maintained just below its firing
threshold (Babalian et al.
1997; Serafin et al.
1991a,b)
and measured their mean duration.
|
Quantitative determination of the neuronal type
In previous publications, MVN neurons have been categorized into type A and
B neurons using qualitative criteria. Neurons were characterized as a type A,
type B, or type B with low-threshold calcium spikes (B+LTS) MVN neurons
according to their action potential profile
(Serafin et al. 1991a). Type A
neurons were characterized by their single, deep AHP and in general have a
wider action potential than type B neurons. They display an A-like
rectification when released from hyperpolarization or in response to
depolarizing current pulses given from a hyperpolarized level, which is
visible as an inflection point delaying the depolarization of the neuron
during inter-spike intervals. Type B neurons showed a biphasic, significantly
smaller AHP and narrower action potentials. Neurons that displayed
intermediate properties, or did not clearly fit into any of these two
categories, were grouped as type C MVN neurons.
To assess reliably the long-term effects of UL on this heterogeneous
population of cells and remove possible experimenter biases in the
classification of MVNn, quantitative, objective criteria were
developed to characterize the intracellularly recorded neurons. For each
neuron, the averaged spike profile obtained during spontaneous firing and its
first derivative were used (as already done by
Johnston et al. 1994) to
assess the presence of the A-like rectification and double AHP, the two main
criteria used previously for the qualitative classification.
The presence of an A-like rectification characterizing type A MVNn is
always visible as an inflection of the voltage trace (V) within the
inter-spike interval (see Fig.
1A for the following explanation). This inflection is
better seen on the first derivative of the voltage trace
(dV/dt) as a sudden decrease of the rate of the
depolarization leading to the next spike. The A-like rectification always
begins ≥2–3 ms after the end of the spike whatever the spontaneous
discharge rate of the cell, and the derivative of the voltage trace remains
positive. The strength of the A-like rectification was quantified as the
algebraic decrease 
(dV/dt) in the rate of the
inter-spike depolarization (dV/dt in V/s) associated with
this phenomenon. In the absence of any A-like rectification, this parameter
was set at zero.
|
The presence of an early fast AHP followed by a delayed slow one, i.e., of
the double-component AHP characterizing type B MVNn, was assessed on the
averaged spike profile and then confirmed using the first derivative (see
Fig. 1B for the
following explanation). When a double AHP is present, its second component is
seen as a transient zeroing or negativity of the rate of depolarization, which
always occurs within 2 ms of the end of the spike. The strength of
each double AHP was quantified as the algebraic decrease
(dV/dt) in the rate of the inter-spike depolarization
(dV/dt in V/s) associated to the second component of the
phenomenon. The strength of the double AHP was set at zero when no double AHP
was present.
While most type B MVNn display a clear double-component AHP when spikes are triggered by current steps delivered from a hyperpolarized potential, the double AHP is often not visible on the averaged spontaneous spike profile obtained at rest or during a slight depolarization. A third parameter had to be used to unambiguously characterize type B MVNn recorded at their resting membrane potential, whatever their level of membrane polarization. All type A MVNn display a mostly concave voltage trace during the inter-spike interval following the peak of the AHP, because of the presence of the A-like rectification (Fig. 1C). In contrast, the averaged inter-spike profile obtained for B MVNn is always convex because the velocity of the inter-spike depolarization increases with time once the peak of the second component of the double AHP has been reached (Fig. 1D). Thus the maximum convexity of the average voltage trace obtained during the inter-spike interval was taken as a third parameter. The averaged spike profile obtained for each neuron was used to draw the line (chord) joining the peak of the AHP to the endpoint of the profile (Fig. 1, C and D). The maximum convexity of the inter-spike trace was then measured as the maximum difference (in mV) observed between the voltage trace and the chord in the direction of convexity (i.e., toward hyperpolarizing potentials) during the inter-spike interval. As shown on Fig. 1C, typical type A MVNn have null or low convexity values, whereas type B MVNn display high convexity values (Fig. 1D).
Altogether, a set of quantitative values describing the strength of the A-like rectification, the strength of the double AHP, and the convexity of the voltage trace during the inter-spike interval was obtained for each of the MVNn recorded in control slices. These values were used to plot a three-dimensional graph of the distribution of the neurons according to these parameters, from which quantitative criteria for classification of MVNn in type A, B, or C neurons were obtained (see RESULTS). The same criteria were then used to categorize the sample of neurons recorded on the ipsilesional side of slices taken from previously labyrinthectomized animals.
After the assessment of its basic membrane and firing properties, each neuron was submitted to the stereotyped stimulation protocol described in the following text. The instantaneous firing rate of the cell (IF in spikes/s) was estimated at any time with a Mathematica script that measured the time intervals between two successive action potentials. The time at the end of each interval between action potentials was used to indicate the time for each IF value.
Assessment of the passive input resistance of MVNn using current steps
The passive input resistance of each neuron was assessed using series of hyperpolarizing current steps of decreasing amplitudes. The cell was maintained by steady-state hyperpolarization at a few milli-volts (0–10) below its threshold for discharge, to suppress spikes. The whole cell resistance for each MVNn (input resistance = voltage deflection/current input) was estimated from the final steady-state value of the hyperpolarizing steps.
Injection of depolarizing ramp-like currents
Increasing ramp currents of 0.3-nA amplitude were applied at five different
slopes up to a final steady-state value, as described in detail in Ris et al.
(2001b); while the cell was
maintained at 10 mV below its firing threshold
(Fig. 2). In other words, for
both control and ipsilesional MVNn, the membrane potential at which the ramps
were delivered was set relative to the firing threshold of each cell. The five
slopes corresponded to times to reach the plateau of current of 5,000 ms (0.06
nA/s), 3,400 ms (0.09 nA/s), 1,800 ms (0.17 nA/s), 600 ms (0.5 nA/s), and 200
ms (1.5 nA/s), respectively. Because the whole stimulus was 5,000 ms long,
there was no plateau after the slowest ramp, which was used, as stated in the
preceding text, to assess the cell's firing threshold
(Fig. 2D). We computed
for each ramp the rate of increase of the instantaneous firing rate of the
cell (kIF in spikes ·
s–1 · nA–1)
during the depolarizing, ramp-like portion of the current injection, i.e.,
over the time taken to reach the plateau of current. This gives an indication
of the sensitivity of the cell to current injections. We also measured in each
case the difference between the firing rate reached at the end of the
depolarizing current injection and the final, stable discharge rate reached at
the end of the plateau (overshoot in spikes/s). This parameter gives an
indication of the nonlinear, dynamic properties of neurons. To assess how the
level of polarization of MVNn influenced their responses, the whole sequence
of ramp stimulations was repeated while the neuron was at its resting membrane
potential.
Injection of sinusoidal currents
A third series of stimuli consisted of current sine waves applied for 5,000
ms at various frequencies ranging from 0.2 to 50 Hz
(Du Lac and Lisberger 1995b;
see for details Ris et al.
2001b). The amplitude of the stimulus was adjusted at the 0.2-Hz
frequency to keep the membrane potential variation 10 mV peak to peak.
Typically, the first series of sinusoidal currents was delivered while the
cell was at its resting membrane potential and spontaneously fired action
potentials (Fig. 3). For each
frequency of stimulation inferior or equal to about one-third of the neuron's
resting discharge, the modulation of the instantaneous firing rate of MVNn was
fitted with a sine wave that was then used to calculate the amplitude and the
phase of the IF modulation (IF, Fig.
3). Ris et al.
(2001b) have shown that in
this condition, the IF modulation of MVNn was linear. When the frequency of
stimulation passed a third of the neuron's firing rate, the amplitude of the
IF modulation of MVNn was calculated in an empirical way as the difference
between the minimum and maximum IF reached by the neuron during the
stimulation. No phase measurements were obtained in this situation. Using this
method, we could evaluate IF from 0.2 Hz to a maximum stimulus
frequency that varied from cell to cell according to its resting discharge and
the sensitivity of its discharge to current injection but could reach 50 Hz in
some cases. The underlying mean membrane potential excursion
(V) was computed for each stimulus frequency using a
Mathematica script, which performed a Fourier analysis of the total membrane
potential response. The magnitude of the Fourier component corresponding to
the stimulation frequency was taken as the potential response. This procedure
was only valid when the components due to the shape and frequency of the
action potentials were not overlapping those of the stimulation frequency.
This requirement was true for frequencies <1 Hz
(Fig. 3). IF and
I were used to evaluate at 0.4 Hz the cell sensitivity to
current injection by dividing IF by the amplitude of the injected
current (IF/I in spikes ·
s–1 · nA–1).
The sensitivity of the firing rate of the cell to variations of the mean
membrane potential IF/V was quantified in spikes
· s–1 ·
nA–1. We calculated the "active"
impedance Z of the cell as the amplitude of the membrane potential
change obtained for the 0.4-Hz stimulus divided by the amplitude of the
injected current (V/I in M).
|
When possible, a similar series of sinusoidal stimuli was given while the cell was maintained at a depolarized membrane potential by a steady-state current injection of 0.15–0.25 nA to assess how the level of discharge of MVNn modified their responses.
Some of the cells were also submitted to the same series of sinusoidal
current injections while they were maintained at 10–20 mV below their
threshold for discharge, so that no spike was evoked by the stimulation. The
amplitude of the membrane potential change (Vh) was
computed for each frequency using a Mathematica script, and the response to
the 0.4-Hz stimulus was used to evaluate the impedance Zh
of the cell maintained under a steady-state hyperpolarization
(Zh = Vh/I in
M).
As reported by Ris et al.
(2001b), the amplitude of the
modulation of the membrane potential or instantaneous firing rate of MVNn by
sinusoidal currents displayed resonant properties. For each MVNn, the response
increased with increasing stimulation frequency to reach a maximum at what was
defined as the peak frequency of resonance. Then the modulation
progressively dropped to lower levels. The "amplitude" of the
resonance was defined as the ratio between the maximum amplitude of the firing
rate modulation at the peak frequency of resonance and the amplitude obtained
at the lowest frequency we used, namely 0.2 Hz. The amplitude of the resonance
was measured in the same way for the membrane potential when the neurons were
hyperpolarized to suppress action potentials.
Statistical analysis
Calculations of means SD and further processing of all results were carried out using the Systat 8.0 software (SPSS, Chicago, IL) on a PC-compatible computer. For each parameter, normality of the distributions was assessed using one sample Kolmogorov-Smirnov tests, with significance set at P ≤ 0.05. Statistical comparisons between numerical values were achieved through either parametric (if the distribution of the parameter was normal for all the samples involved and each sample included ≥15 values) or otherwise nonparametric tests, with the threshold for significance set at P ≤ 0.05. Type B +LTS neurons were pooled together with the other B neurons for analysis. ANOVA or the nonparametric Kruskal-Wallis ANOVA was first performed to search for significant differences between the mean values obtained for type A and B neurons in control slices and in slices taken from labyrinthectomized animals (which defined 4 categories of neurons). Two-by-two comparisons among the four cell groups were then performed using Student's t-test or the nonparametric Mann-Whitney U tests. Type C neurons were excluded from the analysis except for comparisons performed using t-test or Mann-Whitney U tests between the whole sample of neurons obtained on control slices and the whole sample of neurons recorded on slices taken from labyrinthectomized animals. Paired parametric (ANOVA followed by paired t-test) or nonparametric tests (Friedman ANOVA followed by Wilcoxon signed-rank tests) were used to compare for each cell type the responses evoked by ramps of different slopes. They were also used to determine how the responses to ramps and sinusoidal currents were modified according to the level of steady-state polarization of the cell (2 levels for the ramps and 3 levels for the sinusoidal currents).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
1 mo before (compensation times ranged from 23 to 43
days). All mean values are presented with their SD. Categorization of the MVNn using quantitative criteria
The quantitative criteria used for the classification of MVNn into type A
and B neurons were set from the sample of 89 control MVNn, which included 32
MVNn that were recorded by Ris et al.
(2001b) and 57 MVNn recorded
afterward. Because there was no significant difference between the two sets of
result, data from the two samples were pooled together.
We have defined in the methods section the three parameters we viewed as the most pertinent to categorize MVNn into type A and B neurons. They were the presence and strength of an A-like rectification during the inter-spike interval, the presence and strength of a double AHP, and the convexity of the voltage trace during the inter-spike interval. The qualitative classification of MVNn into type A and B MVNn previously used has shown that these two categories of neurons had also significantly different spike widths and AHP amplitudes. Because of that, we checked whether one or both of these parameters could be unambiguously used to categorize MVNn. When considering the whole sample of MVNn, the width of the spike ranged from 0.70 to 2.20 ms around a mean of 1.15 ± 0.23 ms, while the size of the AHP ranged from 6.3 to 29.2 mV around a mean of 16.77 ± 4.50 mV (Table 1). Both distributions were normal, and no sign for the existence of two distinct groups of neurons was obtained when considering only these parameters. Hence, neither the width of the spike nor the size of the AHP could be used to unambiguously categorize control MVNn.
|
Thirty-four of the 89 control MVNn (38%) displayed no measurable A-like rectification on the average spike profile obtained during spontaneous firing. For the 55 other cells, the strength of the A-like rectification ranged from 0.04 to 2.56 V/s around a mean of 0.59 ± 0.53 V/s. Double AHP was present during spontaneous firing in 32 of the control MVNn (36%), and its strength ranged from 0.1 to 5.4 V/s around a mean value of 1.06 ± 1.11 V/s. The 57 other MVNn had no double AHP visible during spontaneous discharge. The maximum convexity of the voltage trace during the inter-spike interval, which was added as a pertinent parameter for the reasons described in METHODS, ranged from 0 to 2.50 mV around a mean value of 0.80 ± 0.63 mV (n = 89).
These three parameters were used to plot the three-dimensional graph shown
on Fig. 4A. On this
plot, most of the MVNn are clearly distributed along two separate,
perpendicular planes, one defined by the presence of a double AHP (double AHP
vs. convexity plane), the other one by the presence of an A-like rectification
(A-like rectification vs. convexity plane). In other words, the presence of a
large A-like rectification and the presence of a double AHP appeared mutually
exclusive, with the exception of three MVNn that clearly displayed both
(
on Fig. 4A).
This means that these two parameters can be used to define two distinct groups
of MVNn as shown on the two-dimensional plot of
Fig. 4B. On this plot,
the neurons that display a double AHP (corresponding to type B MVNn) are
aligned along or close to the vertical axis because most of them display no or
a small A-like rectification. In contrast, the MVNn that display a large
A-like rectification (corresponding to type A MVNn) are aligned along the
horizontal axis because most of them (except the 3 cells marked as 
on
Fig. 4B) display no
double AHP.
|
While MVNn can clearly be categorized into two groups by this method, there
is still a sizeable proportion of MVNn that display "intermediate"
properties (see Fig.
4A, graph). This is in accordance with the idea that
there is a continuum of neurons with intermediate properties between type A
and B MVNn, put forward by Du Lac and Lisberger
(1995a). The
"intermediate" MVNn, which have both no double AHP and no large
A-like rectification, are grouped at or near the origin of the two-dimensional
graph or are aligned along the "no double AHP – no A-like
rectification" axis on the three-dimensional graph. According to the
graphs of Fig. 4, the
intermediate MVNn do not form a distinct group of neurons; this would
correspond to the former type C neurons. There was therefore no objective
reason to keep these MVNn as a separate category. Therefore as already
suggested by Johnston et al.
(1994), MVNn are best
categorized by defining only two groups of neurons corresponding to the type A
and B MVNn. Because there is a continuum of neurons between these two cell
types, the limit between the two groups has to be set somewhat arbitrarily. As
shown on Fig. 4B, we
decided to fixate the threshold of what could be considered as a
"large" A-like rectification at 0.15 V/s. Indeed, several MVNn
displaying a clear double AHP during spontaneous discharge were also endowed
with a small A-like rectification, whose strength was inferior to 0.15 V/s.
Furthermore, most of the intermediate MVNn, which had both no double AHP and
no or a small A-like rectification, displayed the large convexity typical of
type B MVNn (Fig.
4A).
Altogether, as shown on Fig. 4B, most MVNn were categorized as either type A or B MVNn according to the following criteria (see also Table 1): 1) the 44 MVNn displaying either no A-like rectification (n = 32) or an A-like rectification with an amplitude <0.15 V/s (n = 12) were classified as type B MVNn. Twenty-nine of them (66%) had a clear double AHP visible during spontaneous firing (Fig. 5A). Altogether, the mean convexity of type B MVNn reached 1.07 ± 0.66 mV, the mean strength of their double AHP was 0.73 ± 1.07 V/s, and the mean strength of their A-like rectification was 0.02 ± 0.04 V/s. 2) The 42 MVNn displaying an A-like rectification stronger than 0.15 V/s, and no double AHP were classified as type A MVNn. Their mean convexity was 0.56 ± 0.48 mV and was lower than for type B MVNn (P < 0.001). The mean strength of the A-like rectification reached 0.72 ± 0.54 V/s. And 3) the three MVNn that stood out of the two main axes on the two-dimensional graph displayed both a double AHP and an A-like rectification >0.15 V/s. They could not be unambiguously categorized as either type A or type B MVNn and were therefore considered as the only true type C MVNn.
|
According to this classification scheme, there were 42 type A neurons (47.1%), 44 type B neurons (49.5%) including 4 B+LTS neurons, and 3 type C neurons (3.4%) among the MVNn recorded in control slices.
Altogether, the five main parametric differences that characterize type A versus type B MVNn can be summarized as follows.
First, typical type A MVNn display a strong A-like rectification, a high-amplitude AHP and broad spikes; in contrast, they have low convexity values and no double-component AHP. Even when the analysis is restricted to the 42 type A MVNn defined in the preceding text, there are significant correlations between the AHP amplitude, strength of the A-like rectification, and convexity values. Type A MVNn with the strongest A-like rectification are those that display the smallest convexity values (r = –0.43, P = 0.004) and the largest AHPs (r = 0.33, P = 0.046), which results in a significant, negative relationship between the amplitude of the AHP and the convexity (r = –0.39, P = 0.017). But none of these parameters is significantly correlated with the width of spikes when only the type A MVNn were considered.
Second, typical type B MVNn display no or only a small A-like rectification, a small amplitude AHP and thin spikes; in contrast, they have high convexity values and often display a double component AHP. When the analysis is restricted to the 44 type B MVNn defined in the preceding text, only two significant correlations persist between these five parameters. As for type A MVNn, the strength of the A-like rectification is negatively correlated with the convexity value (r = –0.52, P < 0.001). Besides, the neurons displaying strong double-component AHPs are those with the thinner spikes (r = –0.49, P = 0.001). Within type B neurons, there was no correlation between the amplitude of the AHP, the strength of the double AHP and the convexity of the voltage trace during the inter-spike interval.
Membrane and response properties of the MVNn recorded in control slices
The control data that confirm what has already been reported
(Ris et al. 2001b) will be
only briefly summarized. Only new results will be presented in detail.
BASIC MEMBRANE AND FIRING PROPERTIES OF MVNN. As when
they were classified using qualitative criteria, the type A neurons displayed
a single deep AHP (19.6 ± 3.8 mV) and a wider action potential than
type B neurons (P < 0.001). The type B neurons had narrower action
potentials and were endowed with a significantly smaller AHP (14.0 ±
3.4 mV, P < 0.001). Whereas 91% of type B neurons displayed
subthreshold plateau potentials (Serafin
et al. 1991a), only plateau potentials of much shorter duration
(P = 0.004) could be triggered in 38% of type A neurons
(Table 1). There was no
difference between the spontaneous firing rate of type A and B MVNn recorded
at their resting membrane potential, but the regularity of the discharge of
type A MVNn (assessed by the CV) was significantly greater than for type B
MVNn (P = 0.01). The membrane resistance of hyperpolarized MVNn was
similar for both types of MVNn (Table
1).
RESPONSES TO RAMP-LIKE CURRENTS. Of the five ramps applied to
each cell, the 600-ms (slope of 0.5 nA/s) and 200-ms ramps (slope of 1.5 nA/s)
gave the most significant results and were taken as the main indices of the
response of MVNn to ramp-like currents (Figs.
6 and
7). As described by Ris et al.
(2001b), type B MVNn were more
responsive to ramps than type A MVNn. The mean overshoot (see
METHODS) was larger for type B than for type A MVNn for both the
600-ms (5.1 ± 4.7 vs. 1.8 ± 2.1 spikes/s, P = 0.036)
and 200-ms ramps (8.4 ± 5.3 vs. 2.4 ± 2.2 spikes/s, P =
0.002) delivered from the resting membrane potential. When ramps were
delivered from a hyperpolarized level (Fig.
6, A and B), this difference persisted for the
200-ms ramp (13. 3 ± 9.1 vs. 6.8 ± 5.6 spikes/s, P =
0.006) but appeared only as a trend for the 600-ms one (5.8 ± 4.5 vs.
3.5 ± 3.1 spikes/s, P = 0.08). In contrast, the rate of
increase of the instantaneous firing rate kIF over the ramp-like
portion of the current injection (see METHODS) was not
significantly different between type A and B MVNn for any of the ramps we
tested. The respective kIF obtained for the 600- and 200-ms ramps
delivered from the resting membrane potential were 123.2 ± 38.0 and
126.2 ± 32.9 spikes · s–1 ·
nA–1 for type B MVNn versus 112.2 ± 33.8
and 115.3 ± 34.5 spikes · s–1
· nA–1 for type A MVNn (P >
0.05 in both cases).
|
|
Within each cell type, there was no relationship between the overshoot and the sensitivity of MVNn to current injection given by kIF for the 600-ms ramps. A significant positive relationship appeared between the overshoot and the rate of increase of the firing rate kIF only when all MVNn were pooled together. The coefficients of correlation reached 0.33 (P = 0.048) for the 600-ms ramps delivered from a hyperpolarized level and 0.51 (P = 0.012) for those delivered from rest.
Whatever the level of polarization of the cells, the mean kIF and overshoot of both types of neurons significantly increased when the slope of the ramps went from 0.06 to 1.5 nA/s. Compared with the ramps delivered from a hyperpolarized level, the mean overshoots and kIF of the ramps delivered from the resting membrane potential tended to be smaller for all slopes.
MEMBRANE POTENTIAL RESPONSES TO SINUSOIDAL CURRENTS DELIVERED DURING
STEADY-STATE HYPERPOLARIZATION IN THE ABSENCE OF ACTION POTENTIALS. Both
types of MVNn recorded on control slices responded to sinusoidal current
injections in a similar way. The membrane potential modulation
Vh displayed a sizeable resonance at a median peak
frequency of 1 Hz for type A MVNn and 0.7 Hz for type B MVNn
(Table 2,
Fig. 8A1). In
accordance with this slight resonance, the membrane potential response of both
types of MVNn displayed a small phase lead re the injected current at the
lowest frequencies of stimulation, which decreased to zero and became a phase
lag at higher frequencies (Fig.
8A2). These results demonstrate that the membrane does
not behave in a purely passive way at these moderately hyperpolarized
levels.
|
|
RESPONSES TO SINUSOIDAL CURRENTS DELIVERED AT THE RESTING MEMBRANE
POTENTIAL. In terms of instantaneous firing rate, type B MVNn were more
sensitive to sinusoidal current injection at low frequency than type A MVNn
(Fig. 9, A1 and
B1). The greater sensitivity of type B MVNn was
associated with a trend for the sensitivity of their discharge to membrane
potential variations (IF/V) to be higher than for type
A MVNn (P = 0.10 at 0.4 Hz, Table
3). Interestingly, there was a strong trend for the type A MVNn to
have a higher frequency of resonance than type B MVNn. Indeed, the peak of the
resonance was reached at a median frequency of 8 Hz for type A MVNn versus 4
Hz for type B MVNn (P = 0.069,
Fig. 9, A1 and
B1). No significant difference was found between the
active impedance of type A and B MVNn
(Table 3), which suggests that
the difference between type A and B MVNn obtained by Ris et al.
(2001b) on a smaller sample of
neurons might have been linked to a sampling bias.
|
|
The active impedance Z (V/I) at
0.4 Hz of MVNn recorded in control slices was for both cell types much lower
than the impedance Zh obtained in the absence of action
potentials (P < 0.001 when pooling all MVNn together, using the
paired Wilcoxon signed-rank test). The relative amplitude of the resonance of
the firing rate modulation (Fig.
8B1) was not significantly different from the relative
amplitude of the resonance of the membrane potential modulation induced by
sinusoidal currents in the absence of action potentials (during
hyperpolarization) because of the large dispersion of the values. However, the
peak frequency of the resonance (Fig.
8B1) was strongly increased compared with what was
observed in the absence of spikes (P = 0.004 when pooling all MVNn
together, Table 3).
Consistent with the increased peak frequency of the resonance, both type A and B MVNn displayed a slightly greater phase lead at low frequency compared with the phase values obtained during steady-state hyperpolarization (Fig. 8, A2 and B2). On the other hand, the phase lag obtained at high frequency was greater than the one displayed by the potential modulation during steady-state hyperpolarization.
RESPONSES TO SINUSOIDAL CURRENTS DELIVERED DURING STEADY-STATE DEPOLARIZATION. The difference between the sensitivity of type A and B MVNn observed at rest disappeared when the neurons were maintained under a steady-state depolarization (P = 0.62 at 0.4 Hz). The peak of the resonance was reached at a median frequency of 12 Hz for type A versus 8 Hz for type B MVNn (Fig. 9, A2 and B2), but the statistical trend for type A MVNn to have a higher frequency of resonance disappeared (P = 0.29) because of the large dispersion of the values (Table 3).
Compared with the values obtained at rest, the active impedance of type A
and B MVNn maintained under steady-state depolarization was lower. The mean
Z value at 0.4 Hz was 47.9 ± 36.4 M versus 58.5
± 33.9 for currents delivered at rest (P = 0.001 pooling all
MVNn together, Wilcoxon signed-rank test). In accordance with this decrease,
the sensitivity of the instantaneous firing rate of both types of MVNn to
sinusoidal current injection decreased with depolarization
(Table 3,
Fig. 8C1). This
decrease was significant when considering the 23 MVNn that were submitted to
sinusoidal currents both at rest and during depolarization (P = 0.001
at 0.4 Hz, Wilcoxon signed-rank test).
The amplitude of the resonance increased compared with rest for type A MVNn (Fig. 9A, P = 0.047) but was not modified for type B MVNn (Table 3). The peak frequency of the resonance was increased compared with rest for both cell types. Indeed, the median peak frequency of modulation of the firing rate reached 12 versus 8 Hz at rest for type A MVNn (P = 0.005) and 8 versus 4 Hz at rest for type A MVNn (P = 0.005, Fig. 9). In accordance with this increase of the peak frequency of resonance, the mean phase function of the depolarized MVNn was shifted by a few degrees toward smaller phase lags at intermediate and high frequencies compared with rest (Fig. 8, B2 and C2).
Membrane and response properties of MVNn recorded in slices taken from guinea pigs 1 mo after UL
CLASSIFICATION OF THE MVNN RECORDED IN SLICES TAKEN FROM LABYRINTHECTOMIZED GUINEA PIGS. The 78 MVNn recorded on the ipsilesional side of slices taken from lesioned animals were characterized as type A, B, or C neurons according to the quantitative criteria developed in control slices. The graph showing the distribution of ipsilesional MVNn obtained when using the three parameters used for the classification is shown on Fig. 10.
|
First, 35 of the 78 ipsilesional MVNn displayed either no A-like rectification (n = 26) or an A-like rectification <0.15 V/s (n = 9) and were classified as type B MVNn. Only 16 of them (46%), instead of 66% in control slices, had a clear double AHP during spontaneous firing, which ranged from 0.05 to 2.20 V/s around a mean of 0.64 ± 0.68 volts/s (see Fig. 5, A and B). Altogether, the mean strength of the double AHP of the type B MVNn recorded on the deafferented side was 0.30 ± 0.56 V/s and was significantly decreased compared with the type B MVNn recorded on control slices (P = 0.013). In contrast, long-term deafferentation did not modify the convexity, or strength of the small A-like rectification, displayed by type B MVNn (Table 1).
Second, the 42 MVNn displaying an A-like rectification >0.15 V/s and no double AHP were classified as type A MVNn. Their mean convexity was 0.34 ± 0.50 mV and was significantly reduced compared with control slices (P = 0.004). The mean strength of the A-like rectification of ipsilesional type A MVNn was not significantly modified compared with control slices (Table 1).
Third, only 1 of the 78 ipsilesional MVNn (1.3%) displayed both a double AHP and an A-like rectification >0.15 V/s, and was therefore categorized as a type C MVNn.
The proportion of the different types of neurons found in the MVN was slightly modified compared with control slices, but this difference was not significant (compare Figs. 4 and 10). The proportion of type A neurons increased to 53.8% instead of 47.1% in intact animals, whereas the proportion of type B neurons decreased from 49.5 to 44.9%. Among the type B cells, the proportion of B+LTS MVNn tended to increase (6 of 35, i.e., 17 instead of 9%). As in control slices, the spikes of type A MVN were significantly wider than those of B neurons (P < 0.001), and the AHP of type A neurons significantly higher (P < 0.001).
BASIC MEMBRANE AND FIRING PROPERTIES OF MVNN. Compared with the neurons recorded in control slices, the mean resting membrane potential of all types of ipsilesional MVNn was increased by 5 to 10 mV (Fig. 5B, Table 1). The mean potential of type A neurons shifted from –56.8 ± 7.3 to –51.8 ± 5.5 mV (P = 0.002), whereas the mean potential of type B neurons increased from –60.8 ± 9.4 to –51.0 ± 3.8 mV (P < 0.001). This depolarization of the mean resting membrane potential was accompanied by a similar increase in the firing threshold of the cells (Table 1). The other main change compared with control neurons was a significant increase of the amplitude of the AHP displayed by type B neurons, which reached a mean value of 16.1 ± 2.6 versus 14.0 ± 3.4 mV (P = 0.007, Fig. 5C). As already mentioned in the preceding text, this was concomitant to a decrease in the proportion of type B neurons displaying a double AHP (Fig. 5B). The increase in the amplitude of the AHP was associated with a significant increase of the regularity of the spontaneous discharge of type B MVNn assessed by their CV (P = 0.008, Table 1). As a consequence, the difference in the regularity of the spontaneous discharge observed between type A and B MVNn in control slices disappeared.
The increase in the amplitude of the AHP of the ipsilesional type B MVNn was not a consequence of the depolarization of their mean resting membrane potential or of the increase of their spontaneous discharge rate. Indeed, for control as well as ipsilesional B MVNn, the AHP was significantly smaller when the neurons had more depolarized resting potentials and higher spontaneous firing rates. In other words, there was a negative correlation between the amplitude of the AHP and the level of depolarization of type B MVNn (Fig. 5D). Furthermore, the amplitude of the AHP of ipsilesional type A neurons following the deafferentation was not significantly modified (P = 0.46, Table 1), despite the fact they were also depolarized.
Surprisingly, there was only a trend for the spontaneous firing rate of the
whole sample of ipsilesional MVNn taken at their resting membrane potential to
increase compared with control slices
(Table 1, P = 0.10).
The discharge rate of type B MVNn increased by 30% (P = 0.015),
but there was no significant modification of the firing rate of type A MVNn.
Neither the proportion of type A (33%) and type B MVNn (79%) exhibiting
subthreshold plateau potentials nor the duration of these plateau potentials
was different from control (Table
1).
Because of the general depolarization and increase of firing threshold of the ipsilesional MVNn, current steps were generally delivered at less negative membrane potentials than in control slices. Indeed, the level of steady-state hyperpolarization used was set relative to the firing threshold of each cell. Despite this, the deafferented type A MVNn displayed a higher than normal input resistance (Rm) value (P = 0.046). The same trend was visible for type B neurons (Table 1), so that the membrane resistance of the MVNn recorded on the deafferented side increased by 23% compared with the MVNn recorded on control slices (P = 0.018, Table 1).
Responses to ramp-like currents
The main effect of previous deafferentation on the response of MVNn to 600- and 200-ms ramps was a strong increase of the overshoot displayed by both types of neurons (Figs. 6 and 7A). For the ramps delivered from a hyperpolarized level, the mean overshoot of MVNn was almost multiplied by two for the 600-ms ramps (P = 0.007) and increased by 43% for the 200-ms ramps (P = 0.037, Fig. 7A). The overshoot of type B MVNn stayed significantly bigger than the one of type A MVNn (P = 0.03 for the 200-ms ramps). Similar results were obtained for the overshoot of the ramps delivered from rest except that the significant difference between type A and B MVNn observed in control slices did not persist following long-term deafferentation. Indeed, the increase of the overshoot was almost restricted to type A MVNn.
For the ramps delivered from a hyperpolarized level, there was no
significant variation of the rate of increase of the firing rate of MVNn
during the ramps (kIF) after long-term deafferentation for
either type A or B MVNn (Figs.
6 and
7B). However, a
significant increase of kIF was observed for the ramps
delivered from the resting membrane potential, while MVNn were generally
spontaneously active. When considering the whole sample of MVNn,
kIF was increased by 20% for both the 600- and 200-ms
ramps (P = 0.05 in both cases,
Fig. 7B). This
increase tended to be stronger for type B than for type A MVNn.
As
