Fast inhibitory synaptic transmission in the medial vestibular nucleus (MVN) is mediated by GABAA receptors (GABAARs) and glycine receptors (GlyRs). To assess their relative contribution to inhibition in the MVN, we recorded miniature inhibitory postsynaptic currents (mIPSCs) in physiologically characterized type A and type B MVN neurons. Transverse brain stem slices were prepared from mice (3–8 wk old), and whole cell patch-clamp recordings were obtained from visualized MVN neurons (CsCl internal; Vm = –70 mV; 23°C). In 81 MVN neurons, 69% received exclusively GABAAergic inputs, 6% exclusively glycinergic inputs, and 25% received both types of mIPSCs. The mean amplitude of GABAAR-mediated mIPSCs was smaller than those mediated by GlyRs (22.6 ± 1.8 vs. 35.3 ± 5.3 pA). The rise time and decay time constants of GABAAR- versus GlyR-mediated mIPSCs were slower (1.3 ± 0.1 vs. 0.9 ± 0.1 ms and 10.5 ± 0.3 vs. 4.7 ± 0.3 ms, respectively). Comparison of type A (n = 20) and type B (n = 32) neurons showed that type A neurons received almost exclusively GABAAergic inhibitory inputs, whereas type B neurons received GABAAergic inputs, glycinergic inputs, or both. Intracellular labeling in a subset of MVN neurons showed that morphology was not related to a MVN neuron's inhibitory profile (n = 15), or whether it was classified as type A or B (n = 29). Together, these findings indicate that both GABA and glycine contribute to inhibitory synaptic processing in MVN neurons, although GABA dominates and there is a difference in the distribution of GABAA and Gly receptors between type A and type B MVN neurons.
Medial vestibular nucleus (MVN) neurons are a major central target for afferents originating in the horizontal semicircular canal, and their projections play an important role in reflex control of eye and head movement (for review, see Wilson and Melvill Jones 1979). MVN neurons are tonically active, and their output is synaptically modulated by vestibular afferent input, reciprocal connections with contralateral vestibular nuclei, and inputs from cerebellar Purkinje cells (for review, see Barmack 2003; Smith and Curthoys 1989). These synaptic inputs are crucial for motor learning within the vestibuloocular reflex and the plasticity that occurs after peripheral vestibular lesions (Curthoys and Halmagyi 1995; Darlington and Smith 2000).
The contribution of fast inhibitory synaptic transmission in modulating MVN neuron output remains unclear as few studies have directly examined the properties of inhibitory amino acid receptors in any vestibular nucleus neurons, mammalian (Chun et al. 2003) or nonmammalian (Shao et al. 2003, 2004; Straka and Dieringer 1996, for review, see Straka et al. 2005). In mammals, it is known that activation of γ-aminobutyric acid receptors (GABAA receptors) is important in the commissural pathway that connects medial vestibular nuclei on either side of the brain stem (Furuya et al. 1991; Shimazu and Precht 1966) and in cerebello-vestibular circuitry (Babalian and Vidal 2000; Cox and Peusner 1990; du Lac and Lisberger 1992). Also application of the GABAA receptor agonist muscimol leads to a dose-dependent decrease in the resting discharge of MVN neurons (Johnston et al. 2001; Yamanaka et al. 2000). Furthermore, changes in GABAA receptor responsiveness have been implicated in the recovery of resting activity in ipsilesional MVN neurons after unilateral labyrinthectomy (Cameron and Dutia 1997; Johnston et al. 2001; Yamanaka et al. 2000).
Both GABA and glycine immunoreactive neurons, and GABAA and glycine receptors, have been identified in the MVN (Eleore et al. 2004, 2005; Horii et al. 2004; Spencer and Baker 1992; Takazawa et al. 2004; Vibert et al. 2000). Despite this, most electrophysiological data regarding inhibitory synaptic transmission in the MVN has focused on the effect of GABAA agonists on MVN neuron discharge (Johnston et al. 2001; Yamanaka et al. 2000). No studies have addressed the contribution of glycine acting at functional synaptic connections in modulating MVN output, although application of glycine has also been shown to decrease spontaneous discharge in MVN neurons (Lapeyre and De Waele 1995; Precht et al. 1973; Vibert et al. 2000).
Recordings in in vitro slice preparations have shown that MVN neurons remain tonically active and can be classified into two major types according to the properties of their action potential afterhyperpolarization (AHP) (Beraneck et al. 2003; Gallagher et al. 1985; Johnston et al. 1994; Serafin et al. 1991). Type A MVN neurons display a large-amplitude, monophasic AHP, whereas type B neurons show a smaller-amplitude, biphasic AHP. The varied action potential properties of MVN neurons, including AHP shape, are thought to be due to differences in intrinsic membrane properties (for review, Goldberg 2000). When combined with the effects of synaptic inputs, these intrinsic properties allow regulation of MVN discharge in response to a diverse range of head movements.
In the present study, we used whole cell recording techniques to examine GABAAergic and glycinergic inhibitory synaptic transmission in mouse MVN neurons. Using a combination of voltage- and current-clamp techniques, we tested whether GABAAergic and glycinergic transmission differed in the two major physiological classes of MVN neurons (types A and B). Eighty-one of 96 recorded neurons displayed resolvable mIPSCs, and of these neurons, most (69%) expressed exclusively GABAA receptors, 6% expressed exclusively glycine receptors, and 25% expressed both types of receptor.
Mice (C57/Bl6 strain, both sexes) aged 3–8 wks were used in this study. These ages were chosen since by the beginning of postnatal week three (P15) MVN neurons display intrinsic electrophysiological properties that allow identification of type A and type B neurons (Dutia and Johnston 1998). All procedures were in accordance with The University of Newcastle's animal care and ethics committee guidelines. Mice were deeply anesthetized with an intra-peritoneal injection of ketamine (100 mg/kg) and decapitated. The parietal, and part of the occipital bone were removed together with the cerebellum to expose the brain stem. To maintain tissue viability throughout this procedure, the brain and brain stem were constantly bathed in ice-cold sucrose-modified artificial cerebrospinal fluid (sACSF) containing (in mM) 236 sucrose, 25 NaHCO3, 11 glucose, 2.5 KCl, 1 NaH2PO4, 1 MgCl2, and 2.5 CaCl2 (Graham et al. 2003). This solution was continually gassed with Carbogen (95% O2-5% CO2) to achieve a final pH of 7.2–7.3. The brain stem (inferior colliculi to obex) was then isolated and removed from the surrounding bone. The excised brain stem was mounted on a styrofoam block, rostral end down, secured to the stage of a vibrating microtome (VT1000s, Leica Microsystems, Nuslock, Germany) using cyanoacrylate glue (Loctite 454, Loctite, Caringbah, Australia), and transferred to a cutting chamber filled with ice-cold sACSF. Slices (300 μm thick) were cut, and those containing the MVN (5–6 slices extending from hypoglossal nuclei to facial nerve) were transferred to an incubation chamber and allowed to equilibrate for ∼1 h before recording.
After incubation, slices were transferred to a small glass-bottom recording chamber and held in place by nylon threads fixed to a U-shaped flattened platinum wire. The chamber was continually perfused with ACSF (118 mM NaCl substituted for sucrose in sACSF) at room temperature (21–23°C). This temperature was selected to compare receptor kinetics with the large GABAA/glycine receptor literature. Furthermore a recent study by Takazawa et al. (2004) concluded that although spike parameters may be affected at room temperature, changes in the classification of MVN neurons is unlikely. Slices were viewed via a fixed-stage microscope (BX51WI, Olympus, Tokyo, Japan) at low power (×4) to identify the MVN. Near infra-red differential interference contrast (IR-DIC) optics and a ×40 water-immersion lens were used for visualizing individual MVN neurons. Positioning of the recording electrode was controlled using a motorized manipulator (Model MS-314, Marzhauser Wetzlar, Wetzlar-Steindorf, Germany).
Recording electrodes (2–4 MΩ resistance) were pulled from borosilicate glass tubing (1.5 mm OD, Harvard Apparatus, Kent, UK) on a Brown Flaming P-97 micropipette puller (Sutter Instruments, Novato, CA). For characterizing miniature inhibitory postsynaptic currents (mIPSCs), pipettes were filled with an internal electrode solution containing (in mM) 130 CsCl, 10 HEPES, 10 EGTA, 1 MgCl2, 2 MG2ATP, and 0.3 Na3GTP (pH adjusted to 7.3 with 1 M CsOH). To assess the effects of Cs-based internal solution on action potential (AP) properties, we compared APs with those recorded in a potassium-based internal solution containing (in mM) 135 KMeSO4, 8 NaCl, 10 HEPES, 0.1 EGTA, 2 Mg2ATP, and 0.3 Na3GTP (pH adjusted to 7.3 with 1 M KOH). For mIPSCs, whole cell recordings were amplified and filtered (2 kHz) using an Axopatch 1D amplifier (Axon Instruments, Foster City, CA). Signals were captured (10 K samples/s) using an Apple Macintosh G4 computer, an ITC-16i Analogue/Digital converter (Instrutech, Long Island, NY) and Axograph v4.8 data-acquisition software (Axon Instruments).
Whole cell voltage clamp was used to record mIPSCs. After forming a high-resistance seal (>1 GΩ) on the soma of an IR-DIC visualized MVN neuron, the whole cell recording configuration was established by applying brief suction to the pipette tip. Series and input resistance were calculated from the response to a 10-mV hyperpolarizing voltage step from a holding potential of –70 mV. Input resistance was monitored throughout each experiment by presenting this pulse every 10 s. Data were rejected if the series resistance changed by >20% during the course of an experiment. Changes in input resistance, on the other hand, were not used as a criterion for excluding neurons because this parameter could be affected by receptor blockade. Series resistance and membrane capacitance were uncompensated. In voltage-clamp, using CsCl internal, GABAA- and glycine-mediated mIPSCs were pharmacologically isolated by adding 6-cyano-7-nitroquinoxaline-2-3-dione (CNQX, 10 μM) and tetrodotoxin (TTX; 1 μM) to block AMPA-kainate type glutamate receptors and voltage-activated sodium channels, respectively. These blockers were added after characterization of AP properties. mIPSCs recorded under these conditions represent combinations of GABAA receptor (GABAAR)- and glycine receptor (GlyR)-mediated mIPSCs. The GABAAR antagonist bicuculline (10 μM) was then added to the bath. If all mIPSCs were abolished after the addition of bicuculline, the previously recorded mIPSCs were classified as being exclusively GABAAergic. mIPSCs that remained after addition of bicuculline were classed as glycinergic, and this was confirmed by the abolition of all activity using the GlyR antagonist, strychnine (1 μM). In a limited number of neurons, the order of inhibitory antagonist application was reversed (strychnine prior to bicuculline). Altering the conditions in this way did not change GABAA or glycine receptor kinetics. All mIPSC experiments were carried out at a holding potential of –70 mV to block N-methyl-d-aspartate (NMDA) glutamate receptors. At least 3 min of data was used for analysis of the various types of mIPSC. To pharmacologically isolate GABAA or glycine receptor-mediated mIPSCs required between 11 and 16 min in total. In cases where more than one cell was recorded per slice, ≥30 min was allowed to ensure drug washout. TTX was obtained from Alomone Laboratories (Jerusalem, Israel). All other drugs were obtained from Sigma Chemicals (St Louis, MO).
mIPSCs recorded in type A and B MVN neurons
One of the major aims of this study was to examine the contribution of GABAAergic and glycinergic transmission in type A and B MVN neurons. The use of a CsCl internal solution in the recording pipette is a well-accepted method for recording mIPSCs (Graham et al. 2003; Lim et al. 2003; Singer et al. 1998) and allows direct comparison of kinetics with previous studies of GABAA/glycine receptors. However, use of a CsCl internal solution presents difficulties when recording and analyzing action potential properties as cesium ions block potassium channels, which are responsible for features of the AP including the AHP (see Hille 2001). To study mIPSC and action potential properties in the same neuron, we adapted the recently described method of Hefti and Smith (2003) to analyze AP characteristics in recordings obtained with a CsCl internal solution (Fig. 1). According to this technique, APs recorded immediately after switching from cell-attached to the whole cell recording configuration (breakthrough; arrow in Fig. 1B) can be used for analysis of AP characteristics. During this period, the CsCl-based internal solution has not fully dialyzed the cell, and therefore intrinsic membrane conductances have not been dramatically altered. In our hands, it was possible to classify a majority of MVN neurons based on their AP shape (described in the following text), 1–5 s after breakthrough. This allowed classification of both AP properties and mIPSC characteristics.
Recordings obtained in a typical experiment involved first establishing a gigaohm seal on a visualized MVN neuron in voltage-clamp mode (see Fig. 1). The amplifier was then switched to current-clamp mode. In some cases, occasional extracellular APs were observed under these conditions (Fig. 1A1). In later experiments, this activity was allowed time (2–5 min) to become more regular and served as an indicator for achieving successful whole cell current-clamp configuration. In current clamp, the tonic AP discharge, a characteristic of MVN neurons, was now observed (Fig. 1A2). Typically, APs recorded immediately after breakthrough were characterized by a robust overshoot (more positive than +20 mV) and a deep AHP (more negative than –60 mV). As might be expected, cesium diffusing from the pipette into the neuron, gradually increased AP discharge rate and attenuated spike height (Fig. 1B). After ∼6–30 s, the amplifier was switched to voltage clamp and the holding potential set to –70 mV for subsequent characterization of mIPSCs (Fig. 1A3). These mIPSCs were then pharmacologically dissected into GABAAergic or glycinergic components for subsequent analysis.
mIPSC analysis Both classes of mIPSCs were analyzed off-line using a semi-automated, sliding template protocol within the Axograph analysis package (Clements and Bekkers 1997). A detection criterion was calculated based on the optimum scaling factor and the quality of the fit. Events are detected when the criterion crosses a threshold level. The algorithm also automatically compensates for changes in recording noise. Amplitudes of at least three times the noise SD (3 σ) were accepted. mIPSCs detected by the template were individually assessed and accepted for analysis based on two criteria: mIPSCs did not overlap and records displayed a stable baseline (2.5 ms) prior to the rising phase and after the decay phase of the mIPSC. Accepted mIPSCs were aligned at their onset and averaged. Peak amplitude, rise time (calculated over 10–90% of peak amplitude), and decay time constant (calculated over 20–80% of the decay phase) were calculated within the Axograph analysis software. The decay phase of both GABAAR- and GlyR-mediated mIPSCs were best fit by a single decay time constant.
QUANTITATIVE ANALYSIS OF AP PROPERTIES.
In previous in vitro current-clamp studies, MVN neurons have been classified, as either types A and B (Johnston et al. 1994) or A, B, and C (Serafin et al. 1991) using qualitative features of their AP shape. Type A neurons have a monotonically rising AHP, whereas type B neurons have a biphasic AHP. Figure 2 (top and middle) illustrates the features of APs from type A and B MVN neurons. It has been suggested that a small group of neurons displaying intermediate characteristics can be classified as type C. In our study, only neurons that could be clearly identified as either type A or B were analyzed because those displaying intermediate characteristics may have been contaminated by cesium-based internal solution (see results). The qualitative features were further analyzed using quantitative measures (Beraneck et al. 2003) to confirm our initial classification. These quantitative methods are outlined in Fig. 2 (bottom). For each neuron, ≥10 individual action potentials collected following breakthrough were overlaid at their onset (defined as 5% of peak) and averaged to determine AP shape. The first derivative (dV/dt) of the AHP was then calculated and plotted. As Beraneck et al. (2003) have shown, the first derivative's shape is markedly different in type A and B neurons. Throughout the AHP, dV/dt of type A neurons remains positive but decreases toward zero. For type B neurons, dV/dt shows a transient negativity. This difference was independent of internal solution (CsCl vs. KMeSO4—see Fig. 5) and was present at all discharge frequencies (≤60 Hz) reported here. Because only clearly defined type A and B neurons were parsed in the initial selection, further quantitative measures, such as convexity/concavity proved unnecessary. Similar to Beraneck et al. (2003), it should be noted, that ΔdV/dt values in this study are reported as absolute differences between the maximum and minimum deflections of the AHP dV/dt.
In our initial unbiased sample (n = 29), the proportion of neurons classified as type A and B were comparable to previous in vitro studies, accounting for ∼30 and 70%, respectively (Johnston et al. 1994; Serafin et al. 1991). To provide an adequate sample of type A neurons for analysis, we deliberately biased our sample toward these neurons during latter experiments.
In addition to characterizing the MVN neuron, we also measured the following AP properties. Spike amplitude was defined as the maximum peak voltage (mV) taken from baseline (2.5 ms before the AP peak). AHP amplitude was defined as the absolute difference between the minimum of the voltage trace and baseline. Spike width was defined as the width of the action potential at 50% of peak amplitude. Spike rise time was defined as the time from 10 to 90% of spike amplitude. In addition, discharge rate was also measured and, like the previous properties, was restricted to the first two seconds after breakthrough before significant changes in spike characteristics were observed.
In some experiments, 0.5% Neurobiotin (Vector Laboratories, Burlingame, CA) was included in the recording electrode solution. Neurobiotin was allowed to passively diffuse into the cell during the recording session. Subsequently, slices were fixed in 4% paraformaldehyde overnight (4°C). Neurobiotin labeling was assessed using an avidin-biotinylated peroxidase procedure (ABC Kit Elite, Vector Laboratories, Burlingame, CA) with diaminobenzidine (DAB) as the chromogen. Cobalt and nickel intensification was also used in the reaction (Adams 1981). Following visualization, slices were mounted on poly-l-lysine coated slides, air-dried, and counterstained with 1% Cresyl violet. They were then passed through an ascending series of ethanols (70%, 95%, 2 × 100%), cleared in xylene, and cover-slipped in Permount (Fisher Scientific).
Labeled neurons were examined under bright field illumination, and the location of their cell bodies within the MVN was drawn on an outline of the brain stem slice using a drawing tube attached to a microscope (Zeiss Axioskop). Because the shape of the brain stem changes significantly throughout the rostrocaudal extent of the MVN, a set of four templates (≈300 μm apart) from The Mouse Brain (Paxinos and Franklin 2001) were used to map the locations of labeled MVN neurons (see Fig. 7). Using the seventh cranial nerve as a landmark to determine the most rostral section, subsequent slices were sequentially assigned to the four, rostral to caudal templates. To correct for any distortion due to processing, the outline of the cut slice was scaled to match the template. The scaling factors used were then applied to the labeled neuron's dorsoventral and mediolateral position within the slice to calculate its normalized location within the template.
For analysis of neuron morphology, each cell was drawn at high magnification (×1,000). Soma area was taken as the area enclosed by a smooth line that closely followed the soma's border but excluded emerging dendritic processes (Callister et al. 1996). Soma area and maximal diameter were calculated using an image-analysis package (Bioquant Image Analysis, Nashville, TN). In all cases, an attempt was made to follow the axon trajectory within the slice. Invariably, axons were followed to one of the cut surfaces and therefore could not be traced to their target. For similar reasons, the complete dendritic branching pattern could not be determined.
ANOVA or t-test were used for comparisons between variables used to assess mIPSC characteristics and action potential properties. All data are presented as means ± SE. When analyzing the frequency distribution of GABA-only, glycine-only, and mixed mIPSCs in type A and B neurons, we used a Williams' corrected 3 x 2 G-Test of independence (Sokal and Rohlf 1981). Statistical significance was set at P < 0.05.
To investigate the contribution of GABAAR- and GlyR-mediated inhibitory inputs on MVN neurons, we made whole cell patch-clamp recordings in both current- and voltage-clamp configurations from mouse brain stem slices. Recordings were made from 96 MVN neurons in the presence of CNQX (10 μM) and TTX (1 μM). In 81 of these neurons, we could record clearly resolvable mIPSCs at a holding potential of –70 mV. We examined the contribution of GABAA- and glycine-mediated synaptic transmission in neurons that were classified as either type A or B (n = 52). A subset of MVN neurons were also labeled with 0.5% Neurobiotin to determine whether there were morphological differences between MVN neurons based on their inhibitory profiles (n = 15) or action potential properties (n = 29).
Inhibitory profiles of MVN neurons
To determine whether mIPSCs were mediated by GABAA, glycine, or both types of inhibitory receptor, we added bicuculline (10 μM), a selective antagonist for GABAARs, to the bath. In some neurons, the addition of bicuculline abolished all synaptic activity. Such neurons were classified as expressing exclusively GABAARs (Fig. 3A). In other neurons, the addition of bicuculline had no effect on mIPSC frequency, and the subsequent addition of strychnine (1 μM), a selective antagonist of GlyRs, abolished all mIPSCs. Such neurons were classified as expressing GlyRs exclusively (Fig. 3B). Finally, in some neurons addition of bicuculline decreased the frequency of mIPSCs, indicating that some, but not all, mIPSCs were GABAAR-mediated (Fig. 3C). Subsequent addition of strychnine abolished the remaining synaptic activity indicating that the mIPSCs were GlyR-mediated. Such neurons were classified as expressing both GABAA and glycine receptors. Using this approach, we could establish an inhibitory profile for each recorded MVN neuron; cells were classified as receiving either exclusively GABAAergic, exclusively glycinergic, or a mixture of GABAAergic and glycinergic mIPSCs (hereafter termed mixed). From the sample of 81 MVN neurons receiving resolvable mIPSCs, 69% (56/81) received inhibitory inputs mediated exclusively by GABAARs (Fig. 3A). Six percent (5/81) received inhibitory inputs mediated exclusively by GlyRs (Fig. 3B) and 25% (20/81) received mixed inhibitory inputs (Fig. 3C). Whether GABAAR and GlyRs are clustered underneath the same release site and activated by the co-release of both amino acid transmitters (Jonas et al. 1998) was not addressed in this study.
The properties of GABAAR- and GlyR-mediated mIPSCs recorded from MVN neurons are compared in Table 1 and Fig. 4. Neither input nor series resistances were significantly different in MVN neurons used in our analysis of GABAA- and GlyR-mediated mIPSC properties. Thus any differences in mIPSC properties are not attributable to differences in intrinsic membrane properties or recording conditions. GABAAR- and GlyR-mediated mIPSCs showed significant differences in several mIPSC properties. GABAAR-mediated mIPSC amplitudes were smaller than those for GlyR-mediated mIPSCs (22.6 ± 1.8 vs. 35.3 ± 5.3 pA). The kinetics of the two receptors differed significantly, with GABAAR-mediated mIPSCs having slower rise (1.3 ± 0.1 vs. 0.9 ± 0.1 ms) and decay times (10.5 ± 0.3 vs. 4.7 ± 0.3). The frequency of each type of mIPSC varied widely between neurons (0.2 –9.0 and 0.15 –2.5 Hz for GABAAR- and GlyR-mediated events, respectively); however, mean rates were not significantly different for the two types of events (1.1 ± 0.2 vs. 1.0 ± 0.5).
Quantitative classification of MVN neurons based on AHP properties
A principal aim of this study was to compare the contribution of GABAA- and glycine-receptor-mediated inhibition in the two major types of MVN neuron observed in in vitro recordings. Our approach was to use a CsCl-based internal solution in our recording pipette and to determine MVN type, using methods described by Beraneck and co-workers (2003), and then assess mIPSC characteristics in the same neuron. These experiments depend on characterizing action potential AHP features before they have been markedly altered by the accumulation of cesium ions in the recorded neuron. Thus we performed a series of experiments where we compared the effect of Cs ions on maximum ΔdV/dt values during a 7-s period after achieving the whole cell recording configuration with either a KMeSO4- or CsCl-filled pipette. These data are compared in Fig. 5. Using a KMeSO4 internal (n = 5), the maximum ΔdV/dt values are not altered within 7 s of breakthrough. In contrast, in recordings made with CsCl-filled pipettes (n = 52), ΔdV/dt values decrease exponentially toward zero during the same initial 7-s period. However, ΔdV/dt values recorded within the first 2 s (epochs 1 and 2) are similar to those recorded in KMeSO4 internal (left of dashed lines in Fig. 5B). Using this approach, we were able to reliably distinguish type A and B MVN neurons recorded with a CsCl internal as long as the analysis was restricted to the first 2 s after breakthrough. This time is significantly shorter than the 30–45 s reported in Hefti and Smith (2003), where CsCl-based internal solution was used to classify cortical neurons based on the number of spikes fired, a characteristic much less sensitive to the effects of internal cesium.
Based on the preceding criteria, 20 MVN neurons were classified as type A. They displayed a single-component AHP, an A-like rectification (see Beraneck et al. 2003; Hille 2001 for description of A-like current), and a positive dV/dt (Fig. 2A). Thirty-two MVN neurons were classed as type B neurons as they displayed a biphasic AHP and dV/dt showed a transient negativity (Fig. 2B). Eight neurons could not be confidently classified as either type A or B based on the criteria described above. Other studies (Beraneck et al. 2003; Gallagher et al. 1985; Serafin et al. 1991) have also described MVN neurons that do not fit clearly into either the type A or B class and are sometimes referred to as type C (Beraneck et al. 2003; Serafin et al. 1991). As described in the preceding text, neurons that could not be confidently classified as either type A or type B were not included in this study of MVN neuron firing properties.
We also measured several AP properties in type A and B MVN neurons, and these are shown in Table 2. The action potential width for type A and B neurons was not statistically different (2.2 ± 0.2 vs. 2.1 ± 0.1 ms, respectively), and neither was spike amplitude (63.7 ± 2.7 vs. 71.0 ± 2.4 mV). In contrast, AHP amplitude (16.9 ± 1.3 vs. 13.8 ± 0.8 mV; P < 0.05) differed significantly in type A versus B MVN neurons, respectively. The discharge rate of type A neurons was also significantly higher than for type B neurons (29.7 ± 4.0 vs. 14.3 ± 1.3 Hz; P < 0.001).
Inhibitory profiles of type A and B MVN neurons
Because MVN neurons can be classified according to the shape of their AHP and the type of inhibitory inputs they receive, we determined whether there was a differential distribution of inhibitory inputs onto type A and B MVN neurons. In 52 MVN neurons that could be confidently assigned as type A or B (see preceding text), we were able to determine their inhibitory profiles (Fig. 6). Type A MVN neurons were dominated by exclusively GABAAergic mIPSCs (n = 15 of 20) with only a single neuron showing both GABAAergic and glycinergic (mixed) inhibition. The remaining type A neurons (n = 4) had no detectable mIPSCs. Type B neurons also displayed exclusively GABAAergic mIPSCs (n = 15); however, significantly more neurons received glycinergic input, some receiving exclusively glycinergic mIPSCs (n = 2) as well as mixed inputs (n = 10). The remaining type B neurons (n = 5) had no detectable mIPSCs. The difference in inhibitory profiles of type A and B neurons was statistically significant (GSTAT = 7.28, df = 2; P < 0.05).
Morphological characteristics of MVN neurons
A subset of MVN neurons were labeled with 0.5% Neurobiotin to determine whether there were location, and/or morphological differences between MVN neurons based on either action potential properties, or inhibitory profiles (Table 3, and Fig. 7 ). In 29 neurons, 7 were classified as type A and 22 were classified as type B (Table 3). Soma area and maximal soma diameter were similar in type A and B MVN neurons (179.2 ± 27.8 vs. 162.1 ± 15.9 μm2 and 21.4 ± 2.2 vs. 18.6 ± 1.6 μm, respectively). Similarly, no differences were observed in the number of primary dendrites (4.0 ± 0.5 vs. 3.9 ± 0.3). The morphological features of MVN neurons were also compared in cells with different inhibitory profiles. A sample of 15 neurons was classified in this manner. Of these neurons, 10 received exclusively GABAAergic inhibitory inputs, 1 received exclusively glycinergic inputs, and 4 received mixed inhibitory inputs. Soma area, maximum soma diameter, and number of primary dendrites were similar in neurons receiving exclusively GABAAergic, exclusively glycinergic, or both types of mIPSCs (Table 3). Taken together, these data suggest that neuron morphology is not associated with either its discharge characteristics or the type of inhibitory inputs it receives. Morphological characteristics were also not helpful distinguishing MVN neuron discharge properties in rat (Sekirnjak and du Lac 2002).
Inhibitory inputs to MVN neurons
Prior to this report, only one study has recorded mIPSCs and characterized the nature of spontaneous inhibitory synaptic transmission on MVN neurons in a mammal. These experiments used young rats (P13-17) and showed that spontaneous IPSCs were mediated exclusively by GABAARs (Chun et al. 2003). Because eye opening in rats and mice occurs at ∼P12-13 (Henneberger et al. 2005; Zhang et al. 2005) and MVN neurons in both species display electrophysiological properties similar to those in adult animals after P15 (Murphy and Du Lac 2001), it is possible that the lack of glycine receptors reflects an earlier developmental synaptic profile of MVN neurons (for review, see Straka et al. 2005). Inhibitory postsynaptic events have been recorded from embryonic (E16) and neonatal (1-day-old hatchlings) second-order vestibular neurons in the chick tangential nucleus (Shao et al. 2003, 2004). Although a nonmammalian species, these studies showed the presence of both receptors and a significant developmental shift in the contribution of GABAAR-mediated (74 to 44%) and GlyR-mediated (26 to 56%) spontaneous IPSCs. Our data in mouse brain stem slices show that for MVN neurons displaying resolvable mIPSCs, almost all (94% = 69% GABAAergic only +25% Mixed) received GABAAR-mediated inhibitory inputs. In contrast, only 31% receive GlyR-mediated inhibitory inputs, (6% glycinergic only and 25% mixed).
Our data are consistent with numerous studies that have provided evidence for GABAARs and/or GlyR expression in the postsynaptic membranes of MVN neurons in a number of species. Physiological studies have been largely based on the affects of applied agonists or antagonists on the spontaneous discharge rate of MVN neurons. (Dutia et al. 1992; Lapeyre and De Waele 1995; Precht et al. 1973; Vibert et al. 2000). Anatomical studies have also provided evidence for the presence of GABAARs and GlyRs in the MVN (for review, see de Waele et al. 1995). More recently, exhaustive analyses using in situ hybridization (Pirker et al. 2000) together with immunofluorescence (Eleore et al. 2004, 2005; Horii et al. 2004) have described the presence of various GABAAR and GlyR subunits in rat MVN neurons. However, these studies only reveal the presence or absence of a particular receptor, and they do not distinguish between receptors located under a functional synapse or those located extra-synaptically. Our study shows GABAA- mediated synaptic inhibition dominates in the mouse MVN, although the contribution of GlyRs is significant.
GABAAR-mediated mIPSCs in mouse MVN neurons displayed smaller amplitudes and slower kinetics (rise times and decay times) than GlyR-mediated mIPSCs. These results are similar to those reported for other rodent CNS regions where GABAA- and GlyR-mediated mIPSCs have been compared (Donato and Nistri 2000; Graham et al. 2003; Nabekura et al. 2004; O'Brien and Berger 2001; O'Brien et al. 2004; Russier et al. 2002). Furthermore, the kinetics for GABAA- and GlyR-mediated mIPSCs are similar to those recorded in other brain stem nuclei under similar conditions (Callister et al. 1999; O'Brien and Berger 2001). When combined with recent in situ hybridization and protein expression studies, some insight into the subunit composition of the relevant receptors in mouse MVN can be gained. The available data in rat suggest MVN GABAARs are composed of α1, β2 /3, and γ2 subunits, whereas GlyRs are composed of α1 and β subunits (Eleore et al. 2004, 2005; O'Brien and Berger 2001). As both GABAARs and GlyRs in mice have similar kinetics to those in rats, we would predict similar subunit compositions in the mouse MVN (O'Brien and Berger 2001; Singer et al. 1998).
Although our results show that both GABAARs and GlyRs are clustered under release sites and contribute to fast inhibitory synaptic transmission in mouse MVN neurons, the exact origin of these inhibitory inputs is unclear. Previous studies have demonstrated that GABAA-mediated inhibitory inputs onto MVN neurons come from at least two sources: the cerebellum, via GABA-releasing Purkinje cells (Babalian and Vidal 2000; Cox and Peusner 1990; du Lac and Lisberger 1992), and from contralateral MVN neurons, via commissural connections between the MVN on either side of the brain stem (Furuya et al. 1991; Shimazu and Precht 1966). The origin of glycinergic inhibitory inputs to MVN neurons is less clear. Some authors suggest that GlyR-mediated inhibitory inputs arise from contralateral MVN neurons and project across the midline via the commissural pathway (Precht et al. 1973), whereas others suggest that these inputs arise from ipsilateral local circuit MVN interneurons (Furuya et al. 1991; Shimazu and Precht 1966).
Classification of MVN neurons
Previous studies have classified MVN neurons into types A, B, or C (Beraneck et al. 2003; Gallagher et al. 1985; Johnston et al. 1994; Serafin et al. 1991) based on the shape of the action potential's AHP. It is accepted, however, that MVN neurons probably form a continuum between canonical type A and B neurons (Beraneck et al. 2003; du Lac and Lisberger 1995). In the present study, we classified neurons as either type A or B. Early studies (Johnston et al. 1994; Serafin et al. 1991) report broader spike width, and slower rise times for type A neurons. In our study no significant differences were observed in AP characteristics such as spike width, and rise time, for type A and B MVN neurons. This could be attributed to our recordings using patch electrodes with a CsCl-based internal solution at room temperature, Indeed, increased background discharge rates are consistent with the known inhibitory effect of cesium on potassium channels responsible for action potential repolarization (Hille 2001). A more recent study (Beraneck et al. 2003), however, concludes that spike width and rise time are not sufficient to distinguish the two types of MVN neurons.
Inhibitory drive in type A and B MVN neurons
The major goal of this study was to compare the nature of fast inhibitory synaptic drive in type A and B MVN neurons. We used a CsCl-based internal solution and recorded spontaneous action potentials immediately after (Hefti and Smith 2003) achieving the whole cell current-clamp configuration (see Figs. 1 and 5). This approach allowed classification of neurons based on their AP properties (see Fig. 2), and subsequent voltage-clamp recording of low noise mIPSCs at near resting membrane potential (−70 mV). Moreover, recordings made under these conditions can be readily compared with the large rodent literature on GABAA- and GlyR-mediated mIPSC properties (see Table 1) in other CNS regions (Graham et al. 2003; Legendre 2001; O'Brien et al. 2004).
Our results show that GABAAR-mediated inhibitory drive is directed to both type A and B MVN neurons. In contrast, GlyR-mediated inhibitory drive is confined almost exclusively to type B neurons. Our data are based on the observation that glycinergic mIPSCs are absent in type A neurons during the 3 min after bicuculline block. This, however, may not reflect a total lack of GlyRs. For instance, small glycinergic events may be obscured at room temperature due to their low frequency and reduced kinetics. However, because all neurons, regardless of type, were exposed to the same conditions, this strongly argues that the results reflect a fundamental difference in the inhibitory profiles of type A and B neurons. It is also possible that glycinergic receptors are located on distal dendrites of type A neurons and thus in some cases, were removed during tissue processing. Again this still represents a significant asymmetry in glycinergic synaptic transmission between type A and type B neurons.
The functional significance of these findings are open to interpretation. Given the differences in amplitude and time course of GABAA and GlyR-mediated mIPSCs (Table 1), the characteristics of the hyperpolarization produced by activation of either receptor would be markedly different. For example, the longer decay time of GABAAR-mediated input would provide prolonged hyperpolarization, shunting of excitatory events, and decreased MVN neuron discharge to both type A and B neurons. In contrast, activation of GlyRs, available almost exclusively to type B neurons, would allow additional brief blockade of MVN neuron discharge.
To understand the implication of our results, precise information regarding the in vivo function of type A and B MVN neurons is needed. Previously, in vivo MVN neurons have been classified according to their responses to ipsilateral head rotation (type I and type II, and in some cases type III) (Duensing and Schaefer 1958; Precht and Shimazu 1965; Shimazu and Precht 1965). Type I neurons have been further subdivided into tonic and kinetic, based on the regularity of their background discharge pattern (Shimazu and Precht 1965). Indirect attempts to consolidate in vivo and in vitro schemes (I and II vs. A and B) have been made and suggest that type A MVN neurons (in vitro) correspond with type I tonic neurons (in vivo), and type B neurons correspond with type I kinetic neurons (Babalian et al. 1997; Vidal et al. 1996). Thus it is possible that type A (type I tonic) neurons are modulated almost exclusively by GABAAR-mediated inputs, whereas type B (type I kinetic) neurons may be modulated by exclusively GABAAR-mediated inputs, exclusively GlyR-mediated inputs, or a combination of both. The differential distribution of inhibitory inputs onto type A and B MVN neurons observed in this study would therefore confer distinct physiological functions for tonic and kinetic neurons recorded in vivo.
A recent study has implied an alternative relationship. It is reported that neurons displaying properties similar to type A are “interneurons” (as suggested by Takazawa et al. 2004) and by inference, equivalent to in vivo type II, and not type I, neurons. If correct, this suggests that type II neurons receive predominately GABAAergic inputs. In addition, Takazawa et al., showed that glutamatergic neurons always display type B characteristics whereas GABAergic neurons were more heterogenous. Therefore another possibility is that glutaminergic neurons receive glycinergic input whereas GABAergic neurons may not. The resolution of these issues awaits further study.
This study represents the first electrophysiological investigation of inhibitory synaptic input onto physiologically classified type A and B MVN neurons. Our data demonstrate a clear differential distribution of inhibitory inputs onto type A and B MVN neurons. Type A neurons receive almost exclusively GABAAergic inhibitory inputs, whereas type B neurons can receive a significant additional glycinergic input. Although these data provide potential evidence for discrete roles for GABAAergic and glycinergic synaptic transmission in type A and B MVN neurons, it is not possible at this stage to confidently state what this means for MVN neurons in vivo. The technically challenging in vivo patch-clamp recording technique as recently applied to spinal and cortical neurons (Brecht et al. 2004; Graham et al. 2004), could be applied to MVN neurons to address this issue. Using this approach, it would be possible to examine subthreshold events associated with a neuron's response to both natural stimulation (e.g., head rotation) and current injection. Such experiments would therefore allow the precise role of inhibitory synaptic transmission in functioning vestibular-mediated pathways to be determined.
Support for this work was provided by the National Health and Medical Research Council of Australia, The Garnett Passe and Rodney Williams Foundation, and Hunter Medical Research Institute.
Thanks to M. Walsh and S. Hilton for assistance with the processing of labeled neurons and production of Fig. 7 and to E. Lamont for proofreading the manuscript.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2006 by the American Physiological Society