In previous studies, neurons in the medial vestibular nucleus (MVN) were classified mainly into 2 types according to their intrinsic membrane properties in in vitro slice preparations. However, it has not been determined whether the classified neurons are excitatory or inhibitory ones. In the present study, to clarify the relationship between the chemical and electrophysiological properties of MVN neurons, we explored mRNAs of cellular markers for GABAergic (glutamic acid decarboxylase 65, 67, and neuronal GABA transporter), glutamatergic (vesicular glutamate transporter 1 and 2), glycinergic (glycine transporter 2), and cholinergic neurons (choline acetyltransferase and vesicular acetylcholine transporter) expressed in electrophysiologically characterized MVN neurons in rat brain stem slice preparations. For this purpose, we combined whole cell patch-clamp recording analysis with single-cell reverse transcription–polymerase chain reaction (RT-PCR) analysis. We examined the membrane properties such as afterhyperpolarization (AHP), firing pattern, and response to hyperpolarizing current pulse to classify MVN neurons. From the single-cell RT-PCR analysis, we found that GABAergic neurons consisted of heterogeneous populations with different membrane properties. Comparison of the membrane properties of GABAergic neurons with those of other neurons revealed that AHPs without slow components and a firing pattern with delayed spike generation (late spiking) were preferential properties of GABAergic neurons. On the other hand, most glutamatergic neurons formed a homogeneous subclass of neurons exhibiting AHPs with slow components, repetitive firings with constant interspike intervals (continuous spiking), and time-dependent inward rectification in response to hyperpolarizing current pulses. We also found a small number of cholinergic neurons with various membrane properties. These findings clarify the electrophysiological properties of excitatory and inhibitory neurons in the MVN, and the information about the preferential membrane properties may be useful for identifying GABAergic and glutamatergic MVN neurons electrophysiologically.
The vestibular nuclei (VN) are centers for processing of neural signals related to gaze and posture (Goldberg and Fernandez 1984; Precht 1979). It has been shown that the vestibular neurons display a variety of firing responses to head rotation and tilt (Fuchs and Kimm 1975; Melvill Jones and Milsum 1971; Scudder and Fuchs 1992; Shimazu and Precht 1965; Shinoda and Yoshida 1974). The diversity of discharge patterns may permit a precise temporal response to a wide range of head movements. The discharge patterns of VN neurons have been attributed to the inputs from the primary vestibular afferents (Ezure et al. 1978; Fernandez and Goldberg 1971; Keller 1976; Louie and Kimm 1976; for review, see Goldberg 2000), the local and commissural vestibular circuits (Precht and Shimazu 1965; Shimazu and Precht 1966), and the cerebellar inputs (Baker et al. 1972; Lisberger et al. 1994; Precht et al. 1976; Stahl and Simpson 1995; Zhang et al.1995). However, evidence has been accumulated supporting the idea that various ionic conductances in heterogeneous VN neurons also play a critical role in their discharge patterns (Ris et al. 2001; Sekirnjak and du Lac 2002; Smith et al. 2002).
Neurons exhibiting different membrane properties in the medial vestibular nucleus (MVN) were demonstrated in previous studies performed using intracellular recording in in vitro slice preparations (du Lac and Lisberger 1995; Gallagher et al. 1985; Johnston et al. 1994; Serafin et al. 1991a). According to their intrinsic membrane properties, MVN neurons were classified into types A, B, and intermediate type (type C) (Serafin et al. 1991a). The type A MVN neurons exhibited action potentials with long durations followed by a single, large afterhyperpolarization (AHP). They also exhibited a transient rectification ascribed to A-like currents. On the other hand, the type B MVN neurons exhibited action potentials with short durations followed by double AHPs with an early fast and a delayed slow component. Some of them displayed prolonged plateau potentials or low-threshold calcium spikes (LTS). These intrinsic properties were also observed in isolated in vitro whole brain preparations (Babalian et al. 1997; Vidal et al. 1996).
Although MVN neurons were characterized based on their intrinsic membrane properties, it has not been clarified whether the classified neurons are excitatory or inhibitory. Immunocytochemical, in situ hybridization, and electrophysiological studies have demonstrated that there are a variety of neurons with several neurotransmitters in the MVN (for review, see de Waele et al. 1995). Excitatory vestibuloocular MVN neurons are considered to be glutamatergic and/or aspartergic (Demêmes and Raymond 1982; Kevetter and Hoffman 1991), whereas inhibitory ones are mainly considered glycinergic (Spencer et al. 1989). Inhibitory commissural connections linking both sides of the MVN are borne by GABAergic as well as glycineric neurons (Furuya et al. 1992; Holstein et al. 1999; Precht et al. 1973). Furthermore, cholinergic as well as glutamatergic projections to the cerebellum have been revealed (Barmack et al. 1992). Therefore further information about neurotransmitters contained in each neuron would advance this basic understanding of the functional roles of the electrophysiologically classified MVN neurons.
As in other parts of the CNS, the relationships between the chemical and electrophysiological properties of neurons were previously investigated by performing intracellular recordings in slice preparations combined with immunocytochemical staining (Kaneko et al. 1995; Kawaguchi 1993; Kawaguchi and Kubota 1996). However, it is usually difficult to use several kinds of antibodies simultaneously to detect multiple proteins regulating the metabolism of neurotransmitters following electrophysiological analysis of the tested neurons. To overcome this problem, we attempted to detect mRNAs for synthesizing enzymes, neuronal transporters, and vesicle transporters related to neurotransmitters using the whole cell patch-clamp technique combined with reverse transcription–polymerase chain reaction (RT-PCR) analysis. Using this technique, we investigated the relationship between intrinsic membrane properties and 4 kinds of neurotransmitters: glutamate, GABA, glycine, and acetylcholine, in MVN neurons.
Slices of the brain stem were prepared from young Wistar rats (16–21 postnatal days old). The procedures on animals followed the guide for animal experimentation approved by the Animal Research Committee of Gunma University Graduate School of Medicine. The animal was decapitated under deep anesthesia with isoflurane. After decapitation, the brain was quickly removed and submerged in ice-cold sucrose solution containing (in mM): 234 sucrose, 2.5 KCl, 1.25 NaH2PO4, 10 MgSO4, 0.5 CaCl2, 26 NaHCO3, and 11 glucose, bubbled with 95% O2 and 5% CO2 for more than 5 min. The brain block extending from the inferior colliculus to the obex was isolated and submerged again in the ice-cold sucrose solution for more than 5 min. The rostral side of the block was fixed to the stage of a Microslicer (DTK-1000, Dosaka EM, Kyoto, Japan) with cyanoacrylate, and frontal slices of 250-μm thickness were cut. The location of the MVN was defined on the basis of the rat brain atlas (Paxinos and Watson 1998). The slices containing the rostral and caudal edge of the MVN were discarded because it was difficult to determine the boundary of the MVN there. They were incubated in standard Ringer solution at room temperature for more than 1 h before recording. The standard Ringer solution contained (in mM): 125 NaCl, 2.5 KCl, 2 CaCl2, 1 MgCl2, 26 NaHCO3, 1.25 NaH2PO4, and 25 glucose, and was bubbled continuously with 95% O2-5% CO2 (pH 7.4). After incubation, the slice was placed in a conventional submersion-type recording chamber on an upright microscope (Axioskop FS, Zeiss, Oberkochen, Germany) and continuously superfused with standard Ringer solution (2 ml/min of flow rate) using a peristaltic pump (Minipuls 3, Gilson, Villiers, France).
Whole cell patch-clamp recordings
MVN neurons visualized with Nomarski optics were arbitrarily selected in the dorsal region of the MVN for recording. The distance between the selected neurons was far enough not to overlap the distal dendrites for morphological analysis (2–3 neurons from one side of the MVN). Whole cell patch-clamp recordings were made under the visual control of patch pipettes. Patch pipettes were prepared from borosilicate glass capillaries (GC150TF-15, Clark Electromedical Instruments, Pangbourne, UK) with a micropipette puller (P97, Sutter Instrument, Novato, CA). The pipette was filled with an internal solution containing (in mM): 120 K-methylsulfate, 5 KCl, 0.2 EGTA, 2 MgATP, 0.3 NaGTP, 10 HEPES, and 0.1 spermine, pH-adjusted to 7.3 with KOH. In the present study, we used a potassium methylsulfate–based internal solution because potassium gluconate–based internal solution has been reported to eliminate calcium-dependent potassium conductance (Velumian et al. 1997). To stain the recorded neurons, biocytin (5 mg/ml, Sigma-RBI, St. Louis, MO) was dissolved in the solution before recordings. The osmolarity of the internal solution was 280–290 mOsm/l. The liquid junction potential between the pipette solution and the standard Ringer solution, which was measured according to the method reported by Neher (1992), was about −5 mV, and thus the membrane potentials were corrected by −5 mV. The resistance of electrodes was 4.0–7.0 MΩ in the bath solution, and the series resistance during recording was 10–30 MΩ. Whole cell current-clamp recording was performed using an EPC-8 patch-clamp amplifier (HEKA, Darmstadt, Germany). We performed the recording mainly at room temperature (20–24°C), but we also collected data at a bath temperature of 30–32°C controlled with a solution in-line heater (SH-27A, Warner Instruments, Hamden, CT). Voltage signals were filtered at 3 kHz and digitized at 10–100 kHz. Neurons with membrane potentials more negative than −50 mV immediately after patch membrane rupture and exhibiting overshooting action potentials were used for further analysis. The membrane properties of MVN neurons were examined by application of depolarizing and hyperpolarizing current pulses (400 ms in duration) routinely from 2 different levels of holding potential (−65 to −50 mV, and −85 to −75 mV) maintained with constant current injections. Data were acquired using a pClamp6 or pClamp8 system (Axon Instruments, Foster City, CA).
To visualize neurons that were injected with biocytin, the patch pipette was carefully detached from the cell after recording. Slices were fixed with 4% paraformaldehyde in phosphate buffer (0.05 M, pH 7.4) for 2–3 days at 4°C. The slices were rinsed in phosphate-buffered saline (PBS, 0.06 M, pH 7.4) and incubated in PBS/methanol solution (PBS:methanol = 1:1) containing 0.6% H2O2 for 30 min. After rinsing again in PBS solution containing 0.1% Triton-X100, the slices were incubated in the solution containing 1% avidin–biotin peroxidase complex (Vector Laboratories, Burlingame, CA) for 2 h. The slices were rinsed in PBS and Tris-buffered saline (TBS, 0.05 M, pH 7.6), and then incubated in TBS containing 0.01% diaminobenzidine tetrahydrochloride (DAB), 1% nickel ammonium sulfate, and 0.0003% H2O2 for 15–20 min. All of the procedures were performed at room temperature. The slices were mounted on silane-coated slides, counterstained with cresyl violet, dehydrated, and then coverslipped. The visualized neurons with intact somata and proximal dendrites were drawn using a camera lucida attached to a light microscope (BX41, Olympus, Tokyo, Japan).
Single-cell RT-PCR analysis
Electrophysiological properties of MVN neurons were recorded using the whole cell patch-clamp technique described above. The recordings were performed at room temperature. In our preliminary study, when a patch pipette was inserted into slices (about 100 μm from the surface) and withdrawn without aspiration, the PCR reaction described below generated no detectable products that we explored. This was the case in all of the 8 trials. This strongly suggested that as far as an insertion of a patch pipette was restricted to near the surface of slices, there was no contamination from nonrecorded cells. Therefore we selected MVN neurons located near the surface of slices [<100 μm (1–3 layers of cells)]. Patch pipettes were filled with the autoclaved internal solution containing (in mM): 140 K-methylsulfate, 0.2 EGTA, 2 MgCl2, 10 HEPES, and 0.1 spermine, pH-adjusted to 7.3 with KOH. The resistance of the patch electrodes was 2.0–4.0 MΩ in the bath solution and the liquid junction potential was −5 mV. The electrode holder and glass pipettes were autoclaved and kept clean before use. After recording, the contents of the recorded cell were aspirated into the patch pipette. Monitoring the seal resistance, we carefully performed aspiration so as not to detach the pipette from the cell. The nuclei of target cells were not aspirated into the patch pipettes to avoid contamination of genomic DNA. The aspiration was terminated before breaking the giga-seal and the pipette was carefully detached from the cell and withdrawn from the slice. The contents of the cell were expelled into a 0.5-ml tube containing 2 μl of a solution containing dNTPs (2.5 mM), random hexamer (25 μM), oligo-dT15 (2.5 μM), and RNase-free water. After expelling the contents of the patch pipette, 0.25 μl of dithiothreitol (0.4 M), 0.25 μl of RNasin (40 U/μl), and 0.25 μl of Sensiscript reverse transcriptase (Qiagen, Hilden, Germany) were added. The mixture was incubated at 37°C overnight.
In this study, we explored mRNAs of glutamic acid decarboxylase 65, 67 (GAD65 and GAD67) and neuronal GABA transporter (GAT-1) for GABAergic neurons (Cherubini and Conti 2001; Erlander et al. 1991; Yasumi et al. 1997), vesicular glutamate transporter 1 and 2 (VGluT-1 and VGluT-2) for glutamatergic neurons (Aihara et al. 2000; Ni et al. 1994), choline acetyltransferase (ChAT) and vesicular acetylcholine transporter (VAChT) for cholinergic neurons (Gasnier 2000; Usdin et al. 1995), and glycine transporter 2 (GLYT2) for glycinergic neurons (Liu et al. 1993; Poyatos et al. 1997). The PCR primers for GAD65 (GenBank accession number: M72422), GAD67 (M76177), GAT-1 (M33003), VGluT-1 (U07609), VGluT-2 (AF271235), VAChT (U09211), and GLYT-2 (L21672) were designed by using commercially available primer analysis software (OMIGA 2.0, Accelrys, San Diego, CA), and those for ChAT were designed by referring to the sequences described previously (Brice et al. 1989; Yan and Surmeier 1996). For the GLYT-2 primers, 2 primer sets that recognized different positions were made to detect GLYT-2 effectively. The primers were synthesized at Invitrogen (Carlsbad, CA). The primer sets used in the present study are listed in Table 1.
The amplification of cDNA fragments by the PCR method was performed as described previously (Cauli et al. 1997). Briefly, 70 μl of a solution containing the PCR buffer, Taq DNA polymerase (2.5 U, Qiagen) and water was added to the mixture and heated at 95°C for 5 min. Then, 20 μl of a solution containing the primer mixtures (a total of 18 primers, 5 pmol each) was added. The thermal cycling of 95°C for 30 s and 55°C for 1 min for 20 cycles was performed with a thermal cycler (Takara PCR Thermal Cycler 480, Takara Bio, Ohtsu, Japan). An aliquot (0.2 μl) of this reaction was added to the 2nd PCR reaction mixture (10 μl reaction volume) containing 50 mM KCl, 20 mM Tris-HCl (pH 8.3), 1.5 mM MgCl2, 50 μM dNTPs Taq DNA polymerase (5 U/μl, Sigma), and a primer pair (0.1 μM each). The thermal cycling of 94°C for 15 s, 55°C for 30 s, and 72°C for 20 s for 35 cycles was performed with a thermal cycler (GeneAmp PCR system 9700, Applied Biosystems, Foster City, CA). The 2nd PCR products were separated on 1.5% agarose gels and stained with ethidium bromide. All procedures were performed wearing sterile gloves.
Positive controls for primer efficiency (Fig. 1A) were checked using cDNA of the brain stem of adult rats obtained as described previously (Tsuzuki et al. 2001). To verify the specificity of the amplification, the 2nd PCR product was subjected to restriction enzyme analysis. Two different restriction enzymes cutting the PCR product once at different sites were used separately for each PCR product (Table 1). The mixture containing 1.5 μl of an appropriate buffer, 0.3 μl of restriction enzyme (final concentration = 5–10 U/μl), 11.2 μl of H2O, and 2 μl of PCR product was incubated at an appropriate temperature for 2 h. The digested fragments were separated on a 3% agarose gel and stained with ethidium bromide. Positive controls to check the restriction enzyme efficiency were performed using amplified cDNAs obtained from the brain stem (Fig. 1, B–J).
All primers except for those of VAChT were designed not to amplify genomic DNA being referred to the genomic map of Rattus norvegicus (http://www.ncbi.nlm.nih.gov/mapview/map_search.cgi?taxid=10116). Although VAChT primers amplified genomic DNA prepared from a rat tail (data not shown), no amplified product was detected from contents of single cells without adding reverse transcriptase. This confirmed that the nuclei of target cells were not aspirated into the patch pipette, and that all signals detected in the present study were obtained from the amplified cDNAs that were reverse-transcribed from mRNAs.
When at least one of cellular markers for GABAergic neurons was detected in an MVN neuron, we classified the neuron as GABAergic (see Table 2). Similarly, we classified neurons as either glutamatergic, cholinergic, or glycinergic when one of the cellular markers for each neuron type was detected. However, when markers for more than 2 different neuron types, such as GABAergic and glutamatergic neurons, were simultaneously detected in a single neuron, we refrained from classifying the neuron (“Others” in Table 2).
Off-line analysis was performed with Axograph (Axon Instruments) and Origin (Microcal Software, Northampton, MA) software. The properties of AHP, firing patterns, and hyperpolarizing responses were analyzed from voltage responses to injected current pulses. The shapes of action potentials were analyzed from the recordings in which one action potential was evoked during injection of a current pulse at holding potential of −85 to −75 mV. Although the current pulse of 40 or 80 pA usually evoked one action potential, the pulse evoked multiple action potentials (<4) in some neurons. In such a case, the first action potential was analyzed. The shapes of voltage responses to hyperpolarizing current pulses were analyzed after holding the membrane potential at −65 to −50 mV. The time to peak of AHP was estimated as the time from the peak of the action potential to the peak of AHP. The amplitude of AHP was estimated as the difference between the threshold and the most negative membrane potential (defined as the peak of AHP). The threshold of the action potential was determined as the intersection between linear fits to the depolarizing phase preceding the action potential (from 10 to 5 ms preceding the action potential peak) and the rising phase of the action potential (from 1 ms preceding the action potential to the peak). Afterdepolarization (ADP, see Fig. 2), a transient depolarization after a fast AHP, was defined according to the following criteria: 1) within 5 ms from the peak of action potential, the first derivative of a voltage trace was greater than the SD of the baseline for 100 ms before injection of current pulses, and 2) the amplitude of depolarization was larger than 1 mV, which was measured from the membrane potential at the time when the first derivative was greater than the SD of the baseline to the peak of the depolarization. The half-width of the action potential was defined as the spike width of half-amplitude from the threshold. The adaptation ratio, which was defined as the ratio of the last interspike interval (ISI) to the first ISI, was analyzed from a spike train with 6–8 action potentials in 400-ms duration. The duration of the decay after the termination of a current pulse was defined as the duration from the time of the termination of the pulse to the time when the voltage reached the baseline. All values presented are the means ± SD. Statistical significance was examined using Student's t-test (unpaired data) or a one-way ANOVA with a post hoc Scheffé test.
Because we performed recordings of MVN neurons using different experimental procedures and animals from those used in previous studies (du Lac and Lisberger 1995; Gallagher et al. 1985; Johnston et al. 1994; Serafin et al. 1991a), we first investigated the intrinsic membrane properties of MVN neurons. The analysis was focused not only on the properties of AHPs, but on firing patterns to depolarizing current pulses and responses to hyperpolarizing current pulses. Whole cell recordings without single-cell RT-PCR analysis were made from 135 MVN neurons at room temperature. MVN neurons exhibited repetitive firings without injection of current pulses when the holding membrane potential was set to more positive than −50 mV.
Spike afterhyperpolarization of MVN neurons
Typical examples of AHPs observed in this study are shown in Fig. 2A. Figure 2A1 shows an AHP that was characterized by a hyperpolarization (arrow) followed by a monotonic depolarization. The time to peak of the AHP was 5.1 ms. Figure 2A2 shows an AHP exhibiting an early fast (arrowhead) and a delayed slow (closed arrow) component. The 2 components of the AHP were clearly visible because of the intercalation of a transient depolarization (ADP, open arrow). The times to peak of the early and the delayed components were 0.9 and 22.1 ms, respectively. Of 135 neurons recorded, 54 neurons showed ADP (6.0 ± 3.7 mV, ranging from 1.1 to 16.5 mV). Figure 2A3 shows an AHP that exhibited a slow time to peak (arrow, 20.8 ms). During the hyperpolarizing phase, an obvious deflection (arrowhead) was observed, suggesting that the AHP contained a fast component as well as a slow one. These characteristics of AHPs may be similar to those demonstrated in previous studies (Gallagher et al. 1985; Johnston et al. 1994; Serafin et al. 1991a), in which the AHPs were qualitatively classified into 2 types: a single AHP and a double AHP. In the present study, we measured the time to peak of the AHP from the peak of the action potential (Fig. 2B), and classified AHPs according to whether they had a slow component. The distribution of the time to peak of a slow component was obtained from the delayed slow components of AHPs that were intercalated by ADPs (n = 54, black bars in Fig. 2B), because their slow components were easily discernable. The shortest time to peak of the slow components was 12.6 ms. Therefore an AHP whose time to peak was longer than 12.6 ms (gray bars, n = 58) was considered to have a slow component, whereas the others (white bars, n = 23) were considered not to have one. Thus in the present study, we classified AHPs into 3 types: AHPs without a slow component [AHP(s−)], AHPs with a slow component [AHP(s+)], and AHPs with both a slow component and ADP [AHP(s+) with ADP].
Previous studies indicated that the width of the action potential and the amplitude of the peak of the AHP were different between neurons with a single AHP and those with a double AHP (Johnston et al. 1994; Serafin et al. 1991a). Therefore we compared the spike width (Fig. 2C) and the amplitude of the AHP (Fig. 2D) among the 3 types of AHP classified in the present study. The half-width of the action potential in neurons exhibiting AHP(s+) with ADP (black bar, 0.52 ± 0.13 ms) was significantly narrower than that in neurons exhibiting AHP(s−) (white bar, 0.74 ± 0.37 ms, P < 0.05, ANOVA post hoc test) and in neurons exhibiting AHP(s+) (gray bar, 0.66 ± 0.18 ms, P < 0.05). The amplitudes of AHPs of neurons exhibiting AHP(s−) (white bar), AHP(s+) (gray bar), and AHP(s+) with ADP (black bar) were 28.8 ± 7.8, 29.9 ± 5.2, and 25.7 ± 4.9 mV, respectively. A significant difference was observed only between neurons exhibiting AHP(s+) with and without ADP (P < 0.05, ANOVA post hoc test).
Firing patterns of MVN neurons
When depolarizing current pulses were applied to neurons of which membrane potentials were held at −65 to −50 mV, most MVN neurons exhibited repetitive firings with relatively constant ISIs (data not shown). The difference in the firing pattern was clear when depolarizing current pulses were applied to neurons of which membrane potentials were held at −85 to −75 mV (Figs. 3 and 4). Figure 3A shows a firing pattern of an MVN neuron. This neuron exhibited repetitive firings with relatively constant interspike intervals when the depolarizing current pulses were applied to a neuron that was held at −79 mV. Analysis of the adaptation ratio (see methods) revealed that most of these neurons had values near 1.0 (Fig. 3B), indicating that they exhibited weak or no spike-frequency adaptation. These neurons were designated as continuous-spiking neurons (n = 76). In many continuous-spiking neurons, the membrane potentials returned to the baseline within 100 ms after the termination of depolarizing current pulses (see Fig. 3D). In more than one third of them (30/76), however, the decay time to the baseline was prolonged for longer than 100 ms. Figure 3C shows a neuron exhibiting a prolonged depolarization, which is clearly illustrated in the faster sweep records (Fig. 3C2). Action potentials often overrode the depolarization. Figure 3D shows a histogram illustrating the duration of the decay after the termination of current pulses. Of 30 continuous-spiking neurons exhibiting prolonged depolarization, 9 showed depolarization that lasted for longer than 300 ms (white bar in Fig. 3D).
Examples of other firing patterns observed in this study are shown in Fig. 4. Figure 4A shows a firing pattern with a delay in the generation of the first spike. The delayed spike generation was caused by a transient hyperpolarization that occurred after the onset of the membrane depolarization (arrow). Neurons exhibiting delayed spike generation (n = 41) were designated as late-spiking neurons. Figure 4B shows a firing pattern exhibiting a cluster of more than 2 spikes (transient burst) at a threshold for spike generation. The transient burst was often followed by solitary action potentials (bottom trace). Neurons exhibiting a transient burst were designated as burst-spiking neurons (n = 7). Burst-spiking neurons may correspond to the LTS neurons identified in previous studies (Babalian et al. 1997; Serafin et al. 1991a). Figure 4C shows a firing pattern exhibiting the termination of repetitive firings during the current pulse. The termination of spike generation has been reported to be attributable to strong spike-frequency adaptation (Locke and Nerbonne 1997; Wang and McKinnon 1995). Therefore neurons exhibiting termination of repetitive firings during the current pulse were designated as neurons with strong spike-frequency adaptation (n = 11). Because burst-spiking neurons and neurons with strong spike-frequency adaptation were not observed frequently, they may be minor components in the MVN.
The mean input resistances of neurons exhibiting continuous-spiking, late-spiking, burst-spiking neurons, and neurons with strong spike-frequency adaptation were 341.1 ± 151.8, 424.7 ± 205.8, 529.9 ± 211.7, and 322.5 ± 132.4 MΩ, respectively. There was no significant difference among the subtypes (ANOVA post hoc test).
Response to hyperpolarizing current pulse
In response to hyperpolarizing current pulses, 3 types of voltage responses were observed (Fig. 5). Figure 5A exemplifies a voltage response that shows an almost linear relationship between the amount of injected current and the amplitude of the response (Fig. 5A2). Of 135 MVN neurons analyzed, 18 exhibited this response. On the other hand, a majority of MVN neurons (117/135) exhibited voltage responses with a decrease in input resistance at hyperpolarized levels (Fig. 5, B and C). The voltage responses with a decrease in input resistance are ascribed to ionic currents giving rise to inward rectification such as inward rectifier potassium (IRK) currents (Hagiwara and Takahashi 1974; Standen and Stanfield 1978) and hyperpolarization-activated currents (Ih, If, or Iq) (DiFrancesco and Ojeda 1980; Halliwell and Adams 1982; Mayer and Westbrook 1983; Yanagihara and Irisawa 1980). The neuron for which the data are shown in Fig. 5B exhibited a voltage response characterized by “voltage sag,” a fast hyperpolarization (open circle in Fig. 5B1, 2) followed by a slow depolarization that reached a plateau level (filled circle in Fig. 5B1, 2). This type of voltage response was observed in 81 MVN neurons. Such a voltage response as “voltage sag” was presumably attributable to activation of Ih conductances (Sekirnjak and du Lac 2002). The neuron for which the data are shown in Fig. 5C did not exhibit an obvious voltage sag but did show a hyperpolarization accompanied by a decrease in input resistance (Fig. 5C2). This voltage response was presumably attributable to activation of IRK conductances and was observed in 36 MVN neurons.
Classification of membrane properties of MVN neurons
The intrinsic membrane properties of MVN neurons recorded in the present study are summarized in Fig. 6A. Most continuous-spiking neurons exhibited AHP with a slow component (70/76 of continuous-spiking neurons, black and gray bars). In particular, the neurons exhibiting AHP(s+) with ADP (black bars) were predominant (42/70 of neurons exhibiting AHP with a slow component). In response to hyperpolarizing current pulses, more than 70% of continuous-spiking neurons (55/76) exhibited time-dependent inward rectification. In late-spiking neurons, about 40% (16 of 41) of the neurons exhibited AHP(s−) (white bars). The ratio of neurons exhibiting AHP(s−) was high in late-spiking neurons in contrast to neurons exhibiting other firing patterns. It is notable that only a few late-spiking neurons exhibited AHP(s+) with ADP (4/25 of neurons exhibiting AHP with a slow component). The distribution of the 3 types of responses to hyperpolarizing current pulses was almost even in late-spiking neurons. All burst-spiking neurons exhibited AHP(s+) with ADP. Time-independent inward rectification was predominantly observed in burst-spiking neurons. In neurons with strong spike-frequency adaptation, AHP(s+) and time-dependent inward rectification were predominant.
Although the data presented above were obtained from recordings performed at room temperature, in most previous in vitro studies (du Lac and Lisberger 1995; Gallagher et al. 1985; Johnston et al. 1994; Serafin et al. 1991a), the recordings were performed at a temperature above 30°C. Therefore to investigate the distribution of MVN neurons recorded at a temperature above 30°C, we collected the data at a bath temperature of 30–32°C. Figure 6B shows the distribution of MVN neurons recorded at a temperature of 30–32°C (n = 209). The shortest time to peak of the slow component of AHP(s+) with ADP was measured to be 11.2 ms. Thus AHPs whose time to peak were shorter than 11.2 ms were regarded as AHP(s−), whereas the others were regarded as AHP(s+). Although the number of neurons in each subclass was different between the 2 different recording conditions, the overall distribution of neurons at 30–32°C appeared to be similar to that of neurons at room temperature. We analyzed the half-width of action potential and the amplitude of AHP as shown in Fig. 2, C and D. The half-width of action potential in neurons exhibiting AHP(s+) with ADP (0.35 ± 0.10 ms, n = 61) was significantly narrower than that in neurons exhibiting AHP(s−) (0.46 ± 0.17 ms, n = 35) and in neurons exhibiting AHP(s+) (0.41 ± 0.13 ms, n = 113) (P < 0.01, ANOVA post hoc test). On the other hand, the amplitude of AHP in neurons exhibiting AHP(s+) with ADP (25.1 ± 4.5 mV) was significantly smaller than that in neurons exhibiting AHP(s−) (28.8 ± 5.0 mV) and in neurons exhibiting AHP(s+) (29.1 ± 5.0 mV) (P < 0.01, ANOVA post hoc test). Furthermore, the half-width of action potential in neurons exhibiting AHP(s−) recorded at 30–32°C was significantly narrower than that in neurons exhibiting AHP(s−) recorded at room temperature (P < 0.05). Similarly, action potentials in neurons exhibiting AHP(s+) and AHP(s+) with ADP recorded at 30–32°C were significantly narrower than those in neurons recorded at room temperature (P < 0.05). The amplitude of AHP in neurons exhibiting AHP(s+) recorded at 30–32°C was significantly larger than that in neurons recorded exhibiting AHP(s+) at room temperature (P < 0.05). However, it was not significantly different between 30–32°C and room temperature in neurons exhibiting AHP(s−) and those exhibiting AHP(s+) with ADP. The mean amplitude of ADP at 30–32°C was 4.4 ± 3.1 mV (ranging from 1.3 to 13.4 mV, n = 61), and was significantly smaller than that at room temperature (P < 0.05, t-test).
In the present study, AHP(s−) and AHP(s+) were classified according to the shortest time to peak of the slow component of AHP(s+) with ADP (12.6 ms at room temperature and 11.2 ms at 30–32°C). Recordings of the membrane properties from the same neurons during change in temperatures revealed that neurons whose time to peak was longer than 12.6 ms at room temperature (22.5 ± 3.9 ms, n = 8) showed the time to peak longer than 11.2 ms at 30–32°C (18.8 ± 6.3 ms). Also, neurons whose the time to peak were shorter than 12.6 ms at room temperature (5.6 ± 3.5 ms, n = 2) showed the time to peak shorter than 11.2 ms at 30–32°C (3.6 ± 1.6 ms). Firing patterns and responses to hyperpolarizing current pulses did not change at the different temperatures. The amplitude of ADPs decreased as the temperature increased (n = 4). Although the decrease was prominent in ADPs with a large amplitude (n = 2, 6.0 → 2.5 mV and 7.0 → 5.1 mV), ADPs with a small amplitude decreased only slightly (n = 2, 2.3 → 2.2 mV and 1.8 → 1.7 mV). Therefore the significant difference in the amplitude of ADPs between at room temperature and at 30–32°C as described above was attributed mainly to a decrease in ADPs with a large amplitude. This indicated that raising the temperature to 30–32°C did not abolish ADPs. Taken together, the classification of MVN neurons is unlikely to be affected by the difference in the temperature.
To observe the inward rectification responses to hyperpolarizing current pulses clearly (see following text), we performed the recordings using the intracellular solution containing 0.1 mM spermine, which is one of the intracellular factors responsible for intrinsic gating and rectification of IRK channels (Fakler et al. 1994; Ficker et al. 1994; Lopatin et al. 1994). However, spermine could block calcium-activated potassium conductances (Drouin and Hermann 1994). Thus we attempted to record the membrane properties using the intracellular solution without spermine. We obtained MVN neurons exhibiting AHP(s+) (n = 6) and those exhibiting AHP(s+) with ADP (n = 13). The amplitude of AHP and the half-width of action potential were 24.5 ± 3.2 mV (from 20.7 to 29.2) and 0.78 ± 0.17 ms (from 0.60 to 1.05) in AHP(s+), and 23.5 ± 5.4 mV (from 13.5 to 31.6) and 0.64 ± 0.14 ms (from 0.47 to 0.95) in AHP(s+) with ADP. The amplitude of ADP was 4.7 ± 2.2 mV (from 1.6 to 8.4 mV). All values obtained without spermine were in the range of the values obtained with 0.1 mM spermine. Furthermore, the adaptation ratio of continuous-spiking neurons (n = 13) was 1.09 ± 0.21. All these results were comparable to those obtained using intracellular solution containing 0.1 mM spermine. Thus the effect of intracellular spermine at a concentration of 0.1 mM on the profiles of action potential was negligible.
Morphological characteristics of MVN neurons
Among 135 neurons investigated electrophysiologically, 50 were successfully stained with biocytin and subjected to morphological classification. Although most dendrites as well as axons were truncated in the 250-μm-thick slices, several morphological characteristics were disclosed. The MVN neurons labeled with biocytin were roughly separated into 5 groups (Fig. 7). As shown in Fig. 7, A–C, 3 groups of neurons were characterized by a wide mediolateral extension of dendrites. The groups were separated into multipolar-type neurons (group A, n = 19, Fig. 7A), bipolar-type neurons (group B, n = 5, Fig. 7B), and neurons with fusiform somata tending to the dorsoventral direction (group C, n = 6, Fig. 7C). In the fourth group, the dendrites extended dorsoventrally (group D, n = 6, Fig. 7D). The fifth had a local dendritic extension (group E, n = 14, Fig. 7E). Approximate cell size was estimated by the measurement of the area of cell soma. The averaged areas in groups A–E were 132.0 ± 55.4, 107.5 ± 25.9, 153.3 ± 44.9, 133.2 ± 47.7, and 100.1 ± 40.3 μm2, respectively. There was no significant difference among the groups (ANOVA post hoc test). However, the percentage of neurons whose somata were smaller than 100 μm2 was larger in group E (50%, 7/14) than in other groups. In group C, there was no neuron with soma smaller than 100 μm2.
The relationships between the morphological characteristics and the intrinsic membrane properties are shown in Fig. 7, F–H. Although AHP with a slow component was predominant in all of the 5 morphological subgroups, AHP(s+) with ADP was predominant in group A (Fig. 7F). AHP(s−) was observed mainly in neurons belonging to group E. As for firing patterns (Fig. 7G), most of the neurons in groups A and C exhibited continuous firing, whereas those in groups B and E exhibited late-spiking properties predominantly. Neurons in group D were divided evenly into continuous- and late-spiking neurons. Among the continuous-spiking neurons (n = 26), a prolonged depolarization was observed in 8 of 15 neurons in group A, 4 of 5 in group C, 0 of 2 in group D, and 1 of 4 in group E. In response to hyperpolarizing current pulses (Fig. 7H), all of the groups except for group E predominantly exhibited time-dependent inward rectification. In group E, neurons exhibiting time-independent inward rectification were predominant.
Detection of mRNAs expressed in MVN neurons
To clarify the relationship between the electrophysiological and chemical properties of MVN neurons, we analyzed the mRNAs of functional proteins regulating the metabolism of transmitters at the single-cell level after the whole cell recordings. The intrinsic membrane properties were determined based on the criteria described above. Figure 8 shows typical examples of the results obtained from electrophysiological and RT-PCR analysis. In Fig. 8A, the neuron exhibited the properties of late-spiking (a), AHP(s+) (b), and time-dependent inward rectification (c). Single-cell RT-PCR analysis revealed that this neuron expressed GAD-65, GAD-67, and GAT-1 mRNAs (d). The specificity of the amplified fragments of GAD-65, GAD-67, and GAT-1 was confirmed by restriction enzyme analysis (e, f, and g). This finding identified this neuron as a GABAergic neuron. On the other hand, the neuron shown in Fig. 8B exhibited the properties of continuous-spiking (a), AHP(s+) (b), and time-dependent inward rectification (c). Single-cell RT-PCR analysis revealed that this neuron expressed VGluT-2 mRNA (d). The specificity was confirmed by restriction-enzyme analysis (e). This finding identified this neuron as a glutamatergic neuron.
Of 146 MVN neurons analyzed, the mRNAs of neurotransmitter-related proteins were detected in 122 neurons. Among them, 60 neurons were identified as either GABAergic (n = 36), glutamatergic (n = 18), cholinergic (n = 5), or glycinergic (n = 1) (Table 2). In the remaining 62 neurons, more than 2 cellular markers for different kinds of neurotransmitters (for example, both GAD-65 and VGluT-2) were detected in a single neuron (“Others” in Table 2). Previous immunocytochemical studies demonstrated the codistribution of at least 2 different neurotransmitters in the CNS (Kosaka et al. 1988; Manns et al. 2001; Sandler and Smith 1991). Indeed, recent physiological studies have demonstrated that cotransmission with different kinds of neurotransmitters occurs at central synapses (GABA and glycine: Jonas et al. 1998; O'Brien and Berger 1999; GABA and glutamate: Gutiérrez 2000; Walker et al. 2001). Therefore MVN neurons belonging to “Others” in Table 2 might contain multiple neurotransmitters. However, we did not further analyze the relationship between intrinsic membrane properties and neurotransmitters in MVN neurons belonging to “Others.”
As shown in Table 2, about 70% (25/36) of neurons identified as GABAergic neurons expressed more than 2 cellular markers, whereas more than half of neurons identified as glutamatergic neurons (10/18) expressed VGluT-2 mRNA alone. Only 5 neurons were identified as cholinergic neurons. In the majority of neurons expressing ChAT and/or VAChT, mRNAs of the cellular markers for GABAergic and/or glutamatergic neurons were also detected (n = 38, “Others” in Table 2). The number of MVN neurons expressing GLYT-2 mRNA (n = 5) was small, and only one neuron was identified as glycinergic.
The relationships between the neurotransmitters and the intrinsic membrane properties in 60 identified neurons are summarized in Fig. 9. Figure 9A shows the relationship between the neurotransmitters and the characteristics of AHPs: AHP(s−) (white bars), AHP(s+) (gray bars), and AHP(s+) with ADP (black bars). Although all types of AHP were observed in both GABAergic and glutamatergic neurons, the distribution of neurons exhibiting each type of AHP was different between them. In GABAergic neurons, 16 out of 36 neurons exhibited AHP(s−), whereas in glutamatergic neurons (n = 18), only one exhibited AHP(s−). Among neurons exhibiting AHPs with slow components, the ratio of neurons exhibiting AHP(s+) was large in GABAergic neurons (12/20), whereas the majority of glutamatergic neurons exhibited AHP(s+) with ADP (12/17). Cholinergic neurons also exhibited all types of AHPs, and the distribution was similar to that in GABAergic neurons. Figure 9B shows the relationship between the neurotransmitters and the firing patterns. More than half of GABAergic neurons (21/36) exhibited the late-spiking property (black bar), and the rest exhibited the other 3 firing patterns. Strong spike-frequency adaptation was observed in GABAergic neurons (dark gray bar). Such a variety of firing patterns was observed only in GABAergic neurons. A majority of glutamatergic neurons (16/18) were continuous-spiking neurons (white bar). It was notable that no glutamatergic neuron showed the late-spiking property. Like GABAergic neurons, some cholinergic neurons showed the late-spiking property. Figure 9C shows the relationship between the neurotransmitters and the hyperpolarizing responses. Time-dependent inward rectification (white bars) was predominant in the 3 types of neurons, except for a glycinergic neuron. Both GABAergic and glutamatergic neurons exhibited all types of hyperpolarizing responses, but the proportions of neurons exhibiting time-independent inward rectification or no rectification were higher in GABAergic neurons than in glutamatergic neurons.
Focusing on neurons identified as GABAergic and glutamatergic neurons, we again made histograms illustrating the distributions of the neurons exhibiting the different intrinsic membrane properties (Fig. 9D). A comparison of the distributions between GABAergic and glutamatergic neurons revealed that GABAergic neurons consisted of heterogeneous populations that showed different intrinsic membrane properties, whereas most of glutamatergic neurons belonged to a single subclass such as continuous-spiking neurons with AHP(s+) (with ADP) and time-dependent inward rectification. Although continuous-spiking neurons included both GABAergic and glutamatergic neurons, the decay times after termination of the depolarizing current pulses were different between these 2 types of neurons (Fig. 9E). A substantial number of continuous-spiking neurons identified as glutamatergic (7/16) exhibited a prolonged depolarization longer than 100 ms, whereas only a few GABAergic continuous-spiking neurons (2/10) did so. Therefore most continuous-spiking neurons exhibiting a prolonged depolarization are considered likely to be glutamatergic neurons.
In the present study, we classified MVN neurons according to the following membrane properties: 1) characteristics of AHPs: AHP with a slow component, AHP with a slow component and ADP, and AHP without a slow component; 2) firing patterns: continuous-spiking, late-spiking, burst-spiking, and strong spike frequency adaptation; 3) hyperpolarization response: time-dependent inward rectification, time-independent inward rectification, and no rectification. Single-cell RT-PCR analysis revealed that GABAergic neurons consisted of heterogeneous populations with different membrane properties. On the other hand, most glutamatergic neurons formed a homogeneous subclass of neurons exhibiting AHPs with slow components, continuous-spiking, and time-dependent inward rectification.
Characterization of AHP
The previous study demonstrated that AHPs are divided into the 3 types: a single large AHP (type A neuron), a double AHP with an early fast and a delayed slow component (type B neuron), and the intermediate characteristics (type C neuron) (Serafin et al. 1991a). The classification was carried out mainly from the shapes of action potentials occurring spontaneously. Although we did not examine spontaneous firings, du Lac and Lisberger (1995) showed spontaneous firings of the chick MVN neurons at higher than 30°C and reported the similar profiles of AHPs to ours. du Lac and Lisberger (1995) also demonstrated that an increase in firing rate reduced the amplitude of a slow AHP. Therefore it was difficult to discern the slow AHP in neurons exhibiting spontaneous firings at high frequency. Furthermore, they pointed out that hyperpolarization of membrane potential by injection of DC currents occasionally unmasked a slow AHP in neurons that lacked the slow AHP at their resting membrane potential. Therefore to clarify the profiles of AHPs precisely in this study, we analyzed AHPs after single spikes elicited by the application of depolarizing current pulses to neurons held at −85 to −75 mV.
To classify the characteristics of AHPs, Johnson et al. (1994) analyzed AHPs from action potential shapes that were averaged from successive spontaneous action potentials (spike-shape averaging). Using this method, the authors found that all MVN neurons had either a single or a double AHP (Johnston et al. 1994). In the present study, we classified AHPs according to whether an AHP had a slow component, presumably derived from apamine-sensitive calcium-activated potassium currents (IAHP or SK-currents: de Waele et al. 1993; Johnston et al. 1994; Serafin et al. 1991b; Smith et al. 2002). Although AHPs were classified on the basis of analysis of the time to peak of AHPs in this study, AHPs with a slow component [AHP(s+) and AHP(s+) with ADP] and without a slow component [AHP(s−)] may be similar to a double and a single AHP in the previous studies, respectively. Dutia and Johnston (1998) showed that immature type A cells of the MVN in mice exhibited broad action potentials with small single AHPs. The times to peak of the single AHPs of immature type A cells appeared to be longer than those of adult type A cells. Therefore MVN neurons exhibiting slow AHPs might contain neurons with a single immature AHP. However, we did not find neurons exhibiting broad action potentials such as those reported by Dutia and Johnston (1998). Furthermore, in the rat MVN, the mature properties of action potentials are mostly established at the beginning of the 3rd postnatal week (Murphy and du Lac 2001). These findings suggest that the properties of AHPs of MVN neurons obtained from young rats may be similar to those in adult rats, although some MVN neurons might still have been in the course of development in the rats used in the present study. Although quantitative analysis for classification of AHPs using the derivative of the averaged spike profile has recently been reported (Beraneck et al. 2003), the measurement of the time to peak of AHP may provide simple quantitative criteria for the classification of AHPs.
In this study, AHPs with slow components were further subdivided into those with and without ADP. The appearance of the ADP between a fast and a slow AHP may suggest that the depolarization occurs simply because of a time lag between a fast and a slow AHP. Alternatively, the ADP may be caused by active conductances such as calcium-dependent nonselective currents (Haj-Dahmane and Andrade 1997). In any case, the ADP is an important property for identification of glutamatergic neurons, as described below.
Classification of MVN neurons
The classification of MVN neurons into types A, B, and C based on the profiles of action potentials was initially proposed in brain stem slice of guinea pig (Serafin et al. 1991a). Similar classification has been applied to in vitro studies (Babalian et al. 1997; Beraneck et al. 2003; Him and Dutia 2001; Johnston et al. 1994; Ris et al. 2001), and several functional characteristics of the classified neurons have been investigated (Babalian et al. 1997; Beraneck et al. 2003; Him and Dutia 2001; Ris et al. 2001). In view of the classification into types A, B, and C, late-spiking neurons exhibiting AHP(s−) may be typical type A cells, whereas continuous- and burst-spiking neurons exhibiting AHP(s+) and AHP(s+) with ADP may correspond to type B cells. Other neuron groups classified in the present study may be included collectively in type C neurons. Thus neurons obtained in the present study could also be classified according to the previous classification. However, our classification of MVN neurons on the basis of firing pattern and response to hyperpolarizing current pulse in addition to AHP profiles would be useful for clarifying the relationship between the membrane electrophysiological properties and information about transmitters adopted by each neuron.
Expression of mRNAs of cellular markers for neurotransmitters in MVN neurons
The previous in situ hybridization and immunocytochemical studies demonstrated that VGluT-1 and VGluT-2 were differently distributed in the brain: intense VGluT-1 signals were detected in the telencephalic regions, whereas intense VGluT-2 signals were detected in the diencephalic and lower brain stem regions (Fremeau et al. 2001; Herzog et al. 2001; Hisano et al. 2000; Kaneko et al. 2002; Takamori et al. 2001). The present study showed that more than half of MVN neurons identified as glutamatergic expressed VGluT-2 mRNA alone (Table 2), in accord with the previous anatomical studies. Some MVN neurons, however, expressed VGluT-1 mRNA alone or both VGluT-1 and VGluT-2 mRNAs. A recent in situ hybridization study demonstrated that VGluT-1 mRNA signals were observed in the MVN, but the signals were localized only in the ventral magnocellular part of the MVN (Hisano et al. 2002). Because we obtained recordings mainly from neurons in the dorsal parvocellular part in the present study, our results indicate that some neurons in the parvocellular part also express VGluT-1 mRNA, although its expression may be very low.
An immunohistochemical study using ChAT antibodies revealed that cholinergic neurons, some of which projected to the vestibular cerebellum, were widely distributed in the MVN (Barmack et al. 1992). In the present study, we detected ChAT and/or VAChT in 44 out of 122 neurons tested. Interestingly, most of the neurons (89%, 39/44) coexpressed mRNAs for GAD, VGluT, and/or GLYT-2. This finding suggests the possibility of coexistence of ACh and other neurotransmitters in MVN neurons.
Only a few MVN neurons expressed GLYT-2 mRNA, in contrast to the abundance of neurons expressing the markers of other neurotransmitters. GLYT-2 has been thought to be specifically associated with inhibitory glycinergic transmission (Liu et al. 1993; Poyatos et al. 1997), and to be localized in glycine-immunopositive neurons (Poyatos et al. 1997). An immunohistochemical study using anti-glycine antibody has demonstrated that glycine-immunopositive neurons are distributed mainly in the ventral part of the MVN (Rampon et al. 1996). This regional difference in the distribution of glycinergic neurons may have led to the detection of only a small number of neurons expressing GLYT-2 in the present study.
Characteristics of GABAergic and glutamatergic neurons
Single-cell RT-PCR analysis combined with the whole cell recordings revealed that GABAergic neurons in the MVN consisted of a heterogeneous population of neurons with respect to the intrinsic membrane properties (Fig. 9D). Comparison of the intrinsic membrane properties of GABAergic neurons with those of other neurons revealed that the AHP(s−) was a predominant property in GABAergic neurons (Fig. 9A). Similarly, late-spiking neurons were predominant in GABAergic neurons (Fig. 9B). The hyperpolarizing responses of time-independent inward rectification and no rectification were mainly observed in GABAergic neurons (Fig. 9C). These specific intrinsic membrane properties may be useful for identification of GABAergic neurons in the MVN electrophysiologically. In particular, all neurons exhibiting the properties of late-spiking, AHP(s−), and time-independent inward rectification simultaneously were GABAergic. Morphologically, neurons exhibiting AHP(s−), late-spiking, and time-independent inward rectification belonged mainly to group E, in which neurons exhibited local dendritic extensions (Fig. 7). These findings strongly suggest that the neurons in group E are GABAergic interneurons in the MVN.
On the other hand, most glutamatergic neurons formed a homogeneous subclass of neurons exhibiting AHP(s+) (with ADP), continuous-spiking, and time-dependent inward rectification (Fig. 9D). Of 12 neurons exhibiting the continuous-spiking property together with both AHP(s+) with ADP and time-dependent inward rectification, 10 neurons were glutamatergic (the other 2 were GABAergic and cholinergic neurons), indicating that continuous-spiking neurons exhibiting both AHP(s+) with ADP and time-dependent inward rectification are predominant in glutamatergic neurons. Furthermore, continuous-spiking neurons exhibiting prolonged depolarization continuing longer than 100 ms after termination of the current pulses were predominant in glutamatergic neurons (7/9, Fig. 9E). Based on these findings, neurons exhibiting AHP(s+) with ADP, the continuous-spiking property, time-dependent inward rectification, and prolonged depolarization are considered to be glutamatergic neurons. Morphologically, neurons exhibiting AHP(s+) with ADP, the continuous-spiking property, and time-dependent inward rectification belonged mainly to group A, in which neurons were multipolar in shape and exhibited a wide mediolateral extension of dendrites (Fig. 7). MVN neurons extending their dendrites widely in a mediolateral direction were recognized in a previous Golgi study (Hauglie-Hanssen 1968). Furthermore, intra-axonal injections of HRP into physiologically identified type I MVN neurons in cats revealed that the type I neurons exhibited wide mediolateral extensions of dendrites as well as rostrocaudal extensions (Ohgaki et al. 1988). Thus some of the glutamatergic neurons described above may be type I neurons that presumably project to the abducens nucleus and the spinal cord. Because continuous-spiking neurons exhibited a weak or no spike frequency adaptation (Fig. 3B), most of the glutamatergic neurons in the MVN may relay signals with high fidelity.
Previous in vivo studies have demonstrated that MVN neurons can be classified into type I, which exhibit an increase in discharge during horizontal head rotation to the ipsilateral direction, and type II, which show enhanced discharge during the contralateral rotation (Duensing and Schafter 1958; Gernandt 1949; Shimazu and Precht 1965). The type I MVN neurons are further classified into tonic and kinetic neurons according to their spontaneous firing patterns and responses to horizontal head rotation (Shimazu and Precht 1965). Although it is not clear which types of neurons classified in the present study correspond to those classified in the previous in vivo studies, a previous study using isolated in vitro whole brain preparations obtained from guinea pigs suggested putative relationships between the cell types identified in vivo and in vitro (Babalian et al. 1997; Vidal et al. 1996). An analysis of the regularity of spontaneous firings suggested that the type A and type B MVN neurons classified in vitro corresponded to tonic and kinetic neurons, respectively (Babalian et al. 1997). If this relationship could be applied to our results obtained with rat slices, tonic neurons may include GABAergic neurons because late-spiking neurons, which may belong to type A neurons, are predominant in GABAergic neurons. On the other hand, kinetic neurons may include both GABAergic and glutamatergic neurons. Babalian et al. (1997) also demonstrated that short-latency EPSPs were evoked in all types of MVN neurons (types A, B, and C) in response to stimulation of the ipsilateral vestibular nerve, suggesting that 2nd-order vestibular neurons belong to all types of MVN neurons classified in in vitro studies. Furthermore, all types of MVN neurons received inhibitory as well as excitatory inputs from the contralateral side (Babalian et al. 1997). Because our classification of MVN neurons according to their intrinsic membrane properties was different from the classification of cell types by Serafin et al. (1991a), it is not clear which types of MVN neurons classified in the present study are 2nd-order neurons. Further studies will be needed to clarify this issue.
Of 76 continuous-spiking neurons, 30 exhibited a prolonged depolarization after termination of the current pulses (Fig. 3, C and D), which is similar to the plateau potential demonstrated in the previous slice experiments (Him and Dutia 2001; Serafin et al. 1991a,b). The plateau potentials were induced in response to current injection in an isolated in vitro whole brain preparation of guinea pigs and could be induced by ipsilateral vestibular stimulation (Babalian et al. 1997; Vidal et al. 1996). Serafin et al. (1991b) demonstrated that type B MVN neurons exhibited subthreshold plateau potentials caused by persistent sodium conductances. It has been suggested that the rostral part of MVN as well as the prepositus hypoglossi nucleus is involved in the integrative process from the head velocity signals encoded by the primary vestibular afferents to position signals of the eye and head (Godaux et al. 1993). Therefore the plateau potential might be a substrate for the velocity-position and/or velocity-storage integrator (Babalian et al. 1997), although a recent in vivo study on neurons in an oculomotor neural integrator of the goldfish demonstrated that persistent changes in firing rate, which were correlated with changes in eye position, were caused by changes in the rate or amplitude of synaptic inputs (Aksay et al. 2001).
This study was supported by Grants-in-Aid from the Ministry of Education, Culture, Sports, Culture, Science and Technology of Japan (15016021). Y. Saito also appreciates a grant from the Brain Science Foundation.
The authors thank Dr. Yasuo Kawaguchi for technical suggestions, Dr. Tadashi Isa for continuous encouragement, Y. Yoshida for technical assistance, and T. Horikoshi, H. Matsuoka, and S. Takakusaki for partial participation in the experiments.
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 © 2004 by the American Physiological Society