Labyrinthine nerve-evoked monosynaptic excitatory postsynaptic potentials (EPSPs) in second-order vestibular neurons (2°VN) sum with disynaptic inhibitory postsynaptic potentials (IPSPs) that originate from the thickest afferent fibers of the same nerve branch and are mediated by neurons in the ipsilateral vestibular nucleus. Pharmacological properties of the inhibition and the interaction with the afferent excitation were studied by recording monosynaptic responses of phasic and tonic 2°VN in an isolated frog brain after electrical stimulation of individual semicircular canal nerves. Specific transmitter antagonists revealed glycine and GABAA receptor-mediated IPSPs with a disynaptic onset only in phasic but not in tonic 2°VN. Compared with GABAergic IPSPs, glycinergic responses in phasic 2°VN have larger amplitudes and a longer duration and reduce early and late components of the afferent nerve-evoked subthreshold activation and spike discharge. The difference in profile of the disynaptic glycinergic and GABAergic inhibition is compatible with the larger number of glycinergic as opposed to GABAergic terminal-like structures on 2°VN. The increase in monosynaptic excitation after a block of the disynaptic inhibition in phasic 2°VN is in part mediated by a N-methyl-d-aspartate receptor-activated component. Although inhibitory inputs were superimposed on monosynaptic EPSPs in tonic 2°VN as well, the much longer latency of these IPSPs excludes a control by short-latency inhibitory feed-forward side-loops as observed in phasic 2°VN. The differential synaptic organization of the inhibitory control of labyrinthine afferent signals in phasic and tonic 2°VN is consistent with the different intrinsic signal processing modes of the two neuronal types and suggests a co-adaptation of intrinsic membrane properties and emerging network properties.
Central vestibular neurons play a key role in the sensory-motor transformation of body motion-related multi-sensory signals. Head motion is decomposed by semicircular canal and macula organs into different spatial vector and dynamic components and mediated as neuronal discharge by a large number of vestibular nerve afferent fibers to second-order vestibular neurons (2°VN) (see Straka and Dieringer 2004). The presence of physiologically different types of hair cells, vestibular nerve afferents and 2°VN suggests that labyrinthine signals are processed in frequency-tuned channels (for review, see Goldberg 2000; Straka et al. 2005). The relative proportions of 2°VN with high and low response dynamics vary between different species and are likely related to locomotor styles and/or dynamics (Straka et al. 2005). Although vertebrate 2°VN form separate subpopulations, the vast majority of these neurons receive monosynaptic inputs, although in different proportions, from physiologically diverse vestibular nerve afferents, i.e., thick, irregular-, and thin, regular-firing afferents (Goldberg et al. 1987; Straka and Dieringer 2000; Straka et al. 2004; see Goldberg 2000).
Vestibular nerve afferent-evoked monosynaptic excitatory postsynaptic potentials (EPSPs) in most 2°VN are superimposed by inhibitory inputs as shown in monkey and frog (Goldberg et al. 1987; Straka and Dieringer 1996, 2000; Straka et al. 1997). In frog, these superimposed inhibitory postsynaptic potentials (IPSPs) have disynaptic onset latencies and are mediated by glycinergic and GABAergic neurons in the ipsilateral vestibular nuclei (Straka and Dieringer 1996). Moreover, these inhibitory neurons are activated by the thickest vestibular nerve afferent fibers (Straka and Dieringer 2000) and allow a control of the magnitude of the monosynaptic afferent excitatory inputs. This morpho-physiological organization is identical to the suggested neuronal circuit of the side-loop model of Minor and Goldberg (1991).
This model was developed to explain in monkeys the absence of changes in the performance of the vestibuloocular reflex during a block of irregularly firing thick vestibular nerve afferent fibers (Angelaki and Perachio 1993; Chen-Huang et al. 1997; Minor and Goldberg 1991). However, the direct experimental verification in frogs suggests that the functional circuitry, on which the model is based, is a general feature of the vertebrate vestibulomotor network. Because most 2°VN receive monosynaptic inputs from irregularly firing vestibular nerve afferent fibers (Boyle et al. 1992; Goldberg et al. 1987; Highstein et al. 1987; Straka et al. 2004), the inhibitory feed-forward circuitry could control the dynamics of the responses in 2°VN evoked from the same fiber types and in particular the timing and extent of the afferent-evoked spike discharge. Consequently, the activation of disynaptic IPSPs would render the monosynaptic excitation more transient. However, it is unknown if the inhibitory side-loop exerts its effect on all functional subtypes of 2°VN, independent of their intrinsic response properties (Straka et al. 2005).
The present study is based on earlier results showing that electrical stimulation of vestibular nerve afferent fibers evoke disynaptic GABAergic as well as glycinergic IPSPs that sum on monosynaptic EPSPs in most but not all central vestibular neurons (Straka and Dieringer 1996, 2000). Because only few vestibular neurons co-localize these two transmitters (Reichenberger et al. 1997), the presence of a GABAergic as well as a glycinergic IPSP in most 2°VN suggests that these inhibitory inputs are mediated by separate neuronal subpopulations. However, it is unknown if the two pharmacological types of IPSPs have the same kinetics and how this inhibition affects the synaptically activated discharge. Furthermore, in earlier studies, all recorded frog 2°VN were treated as a single homogeneous population (Straka and Dieringer 1996, 2000). Frog 2°VN have been recently differentiated into two distinct functional subtypes based on differences in intrinsic membrane and discharge properties (Beraneck et al. 2007; Straka et al. 2004). This latter distinction opens the possibility that a disynaptic inhibition might only be present in phasic 2°VN which form the majority of vestibular neurons (Straka et al. 2004). In these neurons, the disynaptic inhibitory input would then reinforce the highly dynamic membrane properties (Beraneck et al. 2007). Alternatively, both types are targets for ipsilateral vestibular inhibitory interneurons because tonic and phasic 2°VN receive inputs from the thickest vestibular nerve afferents (Straka et al. 2004) that also activate the disynaptic inhibition.
The present study uses an isolated brain of adult frogs to investigate the properties of the ipsilateral semicircular canal nerve afferent-evoked GABAergic and glycinergic inhibition and their respective roles in the control of the monosynaptic EPSPs and spike discharge in phasic and tonic 2°VN. Preliminary results have been published in abstract form (Biesdorf and Straka 2004).
Glutamate, glycine, and GABA immunocytochemistry of terminal-like structures in the vestibular nuclei was studied in five adult grass frogs (Rana temporaria). The characterization of the antibodies and the immunostaining with positive and negative controls have been performed and described along with methodological details in earlier studies (Reichenberger and Dieringer 1994; Reichenberger et al. 1997). The presence of glutamate, glycine, and GABA in terminal-like structures surrounding central vestibular neurons was quantitatively analyzed on three consecutive 0.5-μm-thick transverse sections at particular rostrocaudal levels of the frog hindbrain. This was repeated on a total of 28 levels that were separated by 80 μm and spanned the whole rostrocaudal extent of the vestibular nuclear complex, including the medial (MVN), superior (SVN), lateral (LVN), and descending vestibular nuclei (DVN).
Electrophysiology and pharmacology
In vitro experiments were performed on the isolated brains of 15 adult grass frogs (R. temporaria) and complied with the “Principles of Animal Care, ” Publication No. 86-23, revised 1985 by the National Institutes of Health. As described earlier (Straka and Dieringer 1993), animals were deeply anesthetized with 0.1% 3-aminobenzoic acid ethyl ester (MS-222) and perfused transcardially with iced Ringer solution [(in mM) 75 NaCl, 25 NaHCO3, 2 CaCl2, 2 KCl, 0.5 MgCl2, and 11 glucose; pH 7.4]. Thereafter, the skull and the bony labyrinth were opened by a ventral approach. After dissecting the three semicircular canals on each side, the brain was removed from the skull with all labyrinthine end organs attached to the VIIIth nerve. Subsequently, the brain was submerged in iced Ringer and the dura, labyrinthine end organs and choroid plexus covering the IVth ventricle were removed. In all experiments, the cerebellum and forebrain were disconnected. Brains were used ≤4 days after their isolation and were stored overnight at 6°C in continuously oxygenated Ringer solution (carbogen: 95% O2-5% CO2) with a pH of 7.4 ± 0.1. For the experiments, the brain stem was glued with cyanoacrylate to a plastic mesh which was fixed with insect pins to the silicone elastomer (Sylgard) floor of a chamber (volume: 2.4 ml) and continuously perfused with oxygenated Ringer solution at a rate of 1.3–2.1 ml/min. The temperature of the Ringer solution in the chamber was electronically controlled and maintained at 14 ± 0.1°C. Technical aspects of the isolation and the maintenance of the preparation have been published earlier (Luksch et al. 1996; Straka and Dieringer 1993).
Neuronal activity in the vestibular nuclei was evoked by electrical stimulation of individual ipsilateral vestibular nerve branches (Straka et al. 1997). For practical purposes and to constrain the study to one set of vestibular endorgans, only the three semicircular canal nerves were stimulated separately with single constant current pulses (0.2 ms; 1–15 μA) applied across suction electrodes (diameter: 120–150 μm). Pulses were produced by a stimulus isolation unit (WPI A 360) at a rate of 0.5 Hz. Glass microelectrodes for extra- and intracellular recordings were made with a horizontal puller (P-87 Brown/Flaming). Electrodes for extracellular field potential recordings were beveled (30°, 20 μm tip diameter) and filled with 2 M sodium chloride (∼1 MΩ). Electrodes for intracellular recordings were filled with a mixture of 2 M potassium acetate and 3 M potassium chloride (10:1) which gave a final resistance of 80–100 MΩ. Neuronal recordings were made in bridge mode (SEC-05L; npi electronic GmbH, Tamm).
At the beginning of each experiment, pre- (N0) and postsynaptic field potential components (N1) evoked by separate stimulation of the three semicircular canal nerve branches were recorded at a reference recording site to optimize the position of the stimulus electrodes and to determine the stimulus threshold (T) for each nerve branch (Straka et al. 1997). This reference recording site was located 0.4 mm caudal to the VIIIth nerve root at a depth of 0.4 mm below the top of the brain stem. Neurons were recorded in all vestibular nuclei except the most medial parts of the medial vestibular nucleus. Vestibular neurons were identified as second order by their monosynaptic EPSPs and were classified as either phasic or tonic based on their discharge pattern evoked by the injection of long, positive current steps (Beraneck et al. 2007; Straka et al. 1997, 2004). Only neurons with a resting membrane potential more than −58 mV and spike amplitudes >60 mV were included in this study. As reported earlier, no differences were encountered in the parameters related to intrinsic membrane properties or synaptically activated responses of neurons recorded on subsequent days after the isolation of the brain stem (Luksch et al. 1996; Straka and Dieringer 2004).
Bath application of the glycine receptor antagonist strychnine hydrochloride (1 μM; Sigma) or of the GABAA receptor antagonist bicuculline methochloride (1–8 μM; Sigma) was used to unmask inhibitory components superimposed on afferent-evoked responses in 2°VN. Both substances in the used chemical form have been shown to be specific for the block of the respective neurotransmitter-evoked responses, e.g., in chicken tangential vestibular neurons (Shao et al. 2003, 2004). Bath application of 7-chloro kynurenic acid (7-Cl KYNA; Sigma; 10 μM) was used to reveal the N-methyl-d-aspartate (NMDA) receptor-mediated component of the afferent nerve-evoked EPSPs. Antagonist-related changes of the evoked responses occurred after ∼5 min and reached steady state after 10–15 min. After 30–40 min, the washout of the different antagonists was usually complete.
Synaptic potentials were digitized at 20 kHz and analyzed from averages of 20–30 single sweeps after electronic subtraction of the extracellular field potential recorded in the vicinity. Statistical differences of the analyzed parameters were obtained using the Mann-Whitney U test (unpaired parameters) or the Wilcoxon signed-rank nonparametric test (paired parameters; Prism, Graphpad Software). Averaged results were expressed as means ± SD. Graphical presentations were made with the aid of commercially available computer software (Origin, Microcal Software; Corel Draw, Corel).
Immunohistochemistry of amino acids in terminal-like structures on vestibular neurons
Glutamate, glycine, and GABA immunoreactive puncta were abundant throughout the vestibular nuclei although in differential quantity for the three amino acids (Fig. 1, A–C). Immunoreactive puncta of all three amino acids were observed in close contact to cell bodies and stem dendrites and likely reflect synaptic structures that often outlined dendrites and cell bodies (green and red arrows in Fig. 1, A2–C2). In general, glutamatergic terminal-like structures were located more on proximal dendrites (green arrows in Fig. 1A2), whereas glycinergic and GABAergic terminal-like structures were more densely distributed around the cell bodies (red arrows in Fig. 1, B2 and C2). However, GABAergic structures seemed to be less numerous than glycinergic puncta independent of the location within the vestibular nuclei. The numbers of immunoreactive, terminal-like structures was quantified on 0.5-μm-thick serial sections through all vestibular nuclei for cell bodies with a visible nucleus (e.g., * in Fig. 1, A2–C2). Central vestibular neurons (n = 145) were more frequently contacted by glutamate (n = 1,632)- than by glycine (n = 1,375)- or GABA-immunoreactive terminal-like structures (n = 866). The significantly higher number of glycinergic compared with GABAergic terminal-like structures (P ≤ 0.0001) on individual vestibular neurons suggests that glycinergic inputs are more pronounced than GABAergic inputs, assuming that other pre- and postsynaptic parameters of these inhibitory synapses are otherwise similar.
Unmasking of glycinergic and GABAergic components in semicircular canal nerve-evoked field potentials
Field potentials evoked by separate electrical stimulation of the ipsilateral horizontal (HC), anterior (AC) and posterior vertical canal (PC) nerves at an intensity of 4 × T were recorded either in the caudal part (i.e., DVN/LVN; Fig. 1, D–F) or in the rostral part (SVN/MVN; not shown) of the vestibular nuclear complex. Independent of the recording site and the activated nerve branch, the evoked responses consisted of a short, transient pre- (N0) and a longer postsynaptic (N1) component that usually lasted up to ∼30 ms (e.g., Fig. 1D1). The N0-component reflects the activity of vestibular nerve afferent fibers and the N1-component with a monosynaptic onset (3.2 ± 0.8 ms, n = 12) the activation of 2°VN (see Straka et al. 1997).
Disynaptic inhibitory components that overlapped with the monosynaptic excitatory responses were unmasked following bath application of the glycine receptor antagonist strychnine (1 μM; Fig. 1, D1–F1) as well as the GABAA receptor antagonist bicuculline (1 μM; Fig. 1, D3–F3). Strychnine increased the peak amplitudes of the control field potentials, gave rise to spike-like events (red arrows), and considerably delayed the return to the baseline (compare black and red traces in Fig. 1, D1–F1). Bicuculline also increased the peak amplitudes and triggered spike-like events (red arrows) but left the longer-latency components of the field potentials largely unaffected (Fig. 1, D3–F3).
Electronic subtraction of the traces recorded before and in the presence of strychnine or bicuculline revealed the magnitude and profile of the unmasked glycinergic (red areas in Fig. 1, D2–F2) and GABAergic inhibition (red areas in Fig. 1, D4–F4), respectively. The onset of the glycinergic (5.8 ± 1.8 ms, n = 12) and GABAergic component (5.9 ± 1.6 ms, n = 12) was delayed with respect to the monosynaptic N1-component and complies with a disynaptic latency following stimulation of semicircular canal nerve branches (see Straka et al. 1997). Although the peak amplitudes of both inhibitory components were similar, the profile of the unmasked glycinergic component had a considerably longer time course compared with that of the GABAergic component (Figs. 1, D, 2 and 4, to F, 2 and 4, and 2A, 1 and 2). The area of the unmasked glycinergic component within the first 50 ms was more than twice as large as the GABAergic component (P ≤ 0.0001) and suggests that EPSPs in individual 2°VN might be differently controlled by the two pharmacological types of inhibitory inputs.
Combined application of the NMDA receptor antagonist 7-Cl KYNA and strychnine or bicuculline was used to reveal a potential contribution of the NMDA receptor-mediated component to the increase of the field potentials in the presence of either one of the two inhibitory transmitter antagonists. Application of 7-Cl KYNA (10 μM) and strychnine (Fig. 1, D1–F1) or bicuculline (Fig. 1, D3–F3) reduced the peak amplitudes and facilitated the return to the baseline (green traces) of the field potentials that were increased beforehand by the two inhibitory transmitter antagonists (red traces). The relative effect of 7-Cl KYNA was similar for HC and AC, somewhat less for PC nerve-evoked field potentials, and generally larger in the presence of strychnine than in the presence of bicuculline (Fig. 1, D, 2 and 4, to F, 2 and 4). In addition, the spike-like events in the presence of strychnine or bicuculline (red arrows in Fig. 1, D, 1 and 3, to F, 1 and 3) were blocked completely. Electronic subtraction of the traces before and in the presence of 7-Cl KYNA (green areas in Fig. 1, D, 2 and 4, to F, 2 and 4) revealed a monosynaptic onset of the NMDA component (3.4 ± 1.7 ms; n = 12). These results indicate that a considerable proportion of the field potential in the presence of the inhibitory transmitter antagonists is mediated by NMDA receptors. This component is larger when compared with that in the absence of GABAA or glycine receptor blockers (Straka et al. 1996). Even though the interaction between the two different inhibitory and the excitatory component, respectively, is likely nonlinear, the larger unmasked NMDA component in the presence of strychnine compared with bicuculline (Fig. 1, D, 2 and 4, to F, 2 and 4) coincides with the longer time course of the glycinergic inhibitory component.
Characterization of frog phasic and tonic second-order semicircular canal neurons
The effect of strychnine and bicuculline on the monosynaptic EPSPs and the profile of the unmasked inhibitory components were studied in 60 intracellularly recorded 2°VN. Based on the differences of the responses to intracellular injection of long positive current pulses, all neurons were characterized as either phasic or tonic as in earlier studies (Beraneck et al. 2007; Straka et al. 2004). The response of most neurons (n = 54; 90%) consisted of a short burst of one to three spikes within the first 50 ms and the absence of a subsequent continuous discharge (Figs. 2B1 and 3A), characteristic for phasic frog 2°VN. In contrast, a continuous discharge (Fig. 4A) with a frequency that increased with current amplitude following the injection of a long positive current pulse, typical for tonic 2°VN was encountered in few neurons (n = 6; 10%). As in earlier studies, the presence, as opposed to the lack of a continuous discharge during positive current steps unequivocally distinguished tonic from phasic 2°VN (Beraneck et al. 2007; Straka et al. 2004). The difference in the discharge pattern evoked by the current step was paralleled by a significant difference (P ≤ 0.001; Mann Whitney U test) in the input resistance between phasic (12.6 ± 4.3 MΩ; n = 54) and tonic 2°VN (23.7 ± 5.4 MΩ; n = 6). In contrast, the resting membrane potential of the two neuronal types was similar (phasic 2°VN: −67 ± 5 mV, n = 54; tonic 2°VN: −68 ± 7 mV, n = 6).
All recorded neurons were identified as 2°VN by the presence of a monosynaptic EPSP (Fig. 2, B2 and C, 1 and 2) after electrical stimulation of a particular ipsilateral semicircular canal nerve branch (Straka et al. 1997). The majority of 2°VN (88%) received the monosynaptic afferent input from only one of the three semicircular canals that originated in similar proportions from the HC (n = 14), the AC (n = 19), or the PC nerve (n = 20). In the remaining seven neurons (12%), a monosynaptic response was evoked by stimulation of two or all three ipsilateral canal nerves. The monosynaptic onset latency (3.4 ± 1.5 ms; n = 70) was similar for EPSPs evoked from the different semicircular canals.
Differential effects of strychnine and bicuculline on monosynaptic EPSPs in phasic 2°VN
The activation of a disynaptic inhibitory component superimposed on the monosynaptic EPSP was usually not detected without pharmacological block of glycine or GABAA receptors. Occasionally, however, the membrane potential of phasic 2°VN (n = 4) could be depolarized with DC currents. This caused a small reduction of the EPSP peak amplitude but more importantly a change in the shape due to a faster decay (Fig. 2B2). The shortening of the decay time is compatible with the activation of a delayed IPSP at the EPSP peak (↓ in Fig. 2B2 and inset) that increased in amplitude with depolarization. The latency for this component (6.3 ± 1.2 ms; n = 4) is compatible with a disynaptic onset after electrical stimulation of the ipsilateral semicircular canal nerve. Because the cerebellum was removed in all experiments, this inhibition is likely mediated by inhibitory neurons in the ipsilateral vestibular nuclei as described earlier (Straka and Dieringer 1996).
The effect of the inhibitory responses on subsequent monosynaptic EPSPs in phasic 2°VN was revealed by stimulation of a semicircular canal nerve with two electrical pulses at different interstimulus intervals (Fig. 2, C and D). EPSPs evoked at various intervals after a preceding EPSP had smaller peak amplitudes (see 1–5 in Fig. 2C1). More precisely, the amplitude of the second EPSP depended on the interval between the two stimuli and was reduced by 35% if the two stimuli were separated by 25 ms (Fig. 2D). With larger interstimulus intervals, the reduction became gradually smaller and EPSPs evoked ∼200 ms after the first had the same amplitude. This reduction in amplitude of the second EPSP was significantly less pronounced in the presence of 1 μM strychnine (1 and 2 in Fig. 2C2) or 1 μM bicuculline. Application of strychnine increased the amplitudes of both the first and the second EPSPs (Fig. 2C2); however, the second EPSP was disproportionately larger relative to the first. Accordingly, the marked reduction in amplitude of the second EPSP in controls (35 ± 4%; n = 10) was only 11 ± 4% (n = 5) in the presence of strychnine and 22 ± 3% (n = 5) in the presence of bicuculline (see amplitudes of EPSPs separated by 25 ms in Fig. 2D). Combined application of strychnine and bicuculline was attempted in few 2°VN (n = 4) but regularly caused a DC depolarization >20 mV that did not allow a study of the combined effects of both blockers. This depolarization is likely a consequence of increased continuous excitatory drive due to the functional neuronal circuitry in the isolated frog brain. A contribution of a persistent Na-conductance to this depolarization is unlikely because the membrane properties of phasic 2°VN are dominated by powerful voltage-dependent K conductances that effectively counteract the possible activation of a persistent Na conductance (Beraneck et al. 2007). The increase in amplitude of the second EPSP in the presence of either one of the two inhibitory transmitter blockers corroborates the activation of an inhibition by the first stimulus pulse that caused a reduction of the amplitude of the second EPSP in controls. However, the different time course of the effects of strychnine and bicuculline (Fig. 2D) suggest that glycinergic IPSPs are larger and last longer than GABAergic IPSPs.
The profiles of pharmacologically unmasked glycinergic and GABAergic inhibitory components were studied in 36 phasic 2°VN, including five neurons where the effect of strychnine as well as bicuculline was tested by sequential application of the two blockers in the same 2°VN (see Fig. 3B, 1 and 2). To facilitate a comparison of the amplitudes and profiles of disynaptic inhibitory components in phasic 2°VN, semicircular canal nerves were stimulated with a similar current pulse intensity (2.3 × T ± 0.3; n = 36). Application of either strychnine or bicuculline increased the amplitude of the monosynaptic EPSPs in all phasic 2°VN (e.g., Fig. 3B, 1 and 2). Electronic subtraction of the traces before and in the presence of the inhibitory transmitter blockers indicated that the unmasked glycinergic and GABAergic component have similar disynaptic onset latencies (∼6.2 ms) and times to peak (∼8.1 ms; Table 1). However, the unmasked glycinergic component had a larger peak amplitude and a longer duration than the GABAergic component as revealed by the sequential application of strychnine and bicuculline in the same phasic 2°VN (compare ░ in Fig. 3B,1 and 2). Calculation of the area of the unmasked inhibitory components in all recorded phasic 2°VN indicated that this is a general feature of the glycinergic inhibition (Table 1). The larger area after normalization (Table 1) confirms the notion that a longer time course rather than a larger peak amplitude is the cause of the larger contribution of the glycinergic compared with the GABAergic component. Thus the extent of the ipsilateral disynaptic inhibition in phasic 2°VN depends on the pharmacological profile of the mediating population of interneurons, given the fact that only a small population of vestibular neurons has glycine and GABA co-localized (Reichenberger et al. 1997). In the five neurons where strychnine and bicuculline were applied sequentially, both antagonists revealed a superimposed disynaptic IPSP. This suggests that any given phasic 2°VN receives a glycinergic as well as a GABAergic inhibitory component.
Application of 7-Cl KYNA (10 μM) in the presence of either strychnine or bicuculline reduced the increased EPSPs slightly below control values, respectively (Fig. 3B, 1 and 2). This is compatible with the results of semicircular canal nerve-evoked field potentials (Fig. 1, D–F) and indicates that the increase in EPSP amplitudes provoked by inhibitory transmitter antagonists is caused to a considerable extent by a NMDA receptor-mediated component. However, at present it is unclear how much of the increase due to the block of the inhibition is caused by a concomitant increase of the non-NMDA (α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid; AMPA) receptor-mediated response, which comprises the major component of vestibular nerve afferent-evoked EPSPs (Straka and Dieringer 1996). Because the proportion of the AMPA and NMDA response depends on the afferent fiber spectrum that activates the two components (Straka et al. 1996), it is at present not possible to precisely determine how much of each component is affected by the disynaptic inhibition.
Effects of strychnine and bicuculline on monosynaptic EPSPs in tonic 2°VN
The presence of superimposed inhibitory components on monosynaptic EPSPs in tonic 2°VN (Fig. 4, B and C) after electrical stimulation (at 2.3 × T) of the AC (n = 2), HC (n = 1), and PC (n = 3) nerves was studied after application of strychnine (n = 3) or bicuculline (n = 3). The limited number of recorded neurons is due to their relatively smaller percentage in the total population of 2°VN (∼20%) (Beraneck et al. 2007; Straka et al. 2004) as well as to the limited period of stable recordings necessary for the pharmacological experiments. Nonetheless an inhibitory response superimposed on monosynaptic EPSPs was encountered in tonic 2°VN. In a few neurons, the inhibitory component became visible after depolarization of the membrane potential (Fig. 4B).The small notch following the peak of the monosynaptic EPSP at a resting membrane potential of −73 mV (Fig. 4B, ↓), was facilitated at −69 mV and truncated the monosynaptic EPSP, compatible with a delayed inhibition. The onset of this component (21.3 ± 3.4 ms; n = 3), however, was much longer than that of the inhibitory components in phasic 2°VN.
Application of 1 μM strychnine (Fig. 4C) or bicuculline increased the EPSP of tonic 2°VN and unmasked inhibitory components with amplitudes of 0.8–3 mV. No difference was encountered between the amplitude and time course of the unmasked glycinergic and GABAergic IPSP. The onset of the inhibitory component (Fig. 4C, ░) had an average latency of 23.2 ± 4.1 ms (n = 6) and was similar for the two pharmacological types of IPSPs. At higher stimulus intensities (>4 × T; n = 4), the number of synaptically-evoked action potentials superimposed on the longer-latency components of the EPSPs increased in the presence of both strychnine and bicuculline from approximately two to three to six spikes in different neurons (not shown). Unfortunately, the small sample size of tonic 2°VN (n = 6) did not allow a detailed analysis of possible differences of glycinergic and GABAergic IPSP profiles and corresponding changes in spike discharge pattern and timing. Nonetheless, the long latency of the inhibitory responses excludes a disynaptic origin of this component in tonic 2°VN.
Differential control of the spike discharge in phasic 2°VN by strychnine and bicuculline
Electrical stimulation of semicircular canal nerves at intensities >2.5 × T usually triggered action potentials on top of the monosynaptic EPSPs in phasic 2°VN (Fig. 5, A, 1 and 4, to C, 1 and 4). The average number of evoked spikes per stimulus pulse (spikes/pulse) increased with current intensity, saturated at stimulus intensities of 4–4.5 × T and was limited to approximately two spikes in most phasic 2°VN (mean: 2.06 spikes/pulse; Table 2). With increasing stimulus intensities the onset latency of the first spike as well as the first interspike interval became gradually shorter and less variable.
Application of 1 μM strychnine increased in all phasic 2°VN the average number of evoked spikes (Fig. 5, A2 and B2) to reach a maximum of more than 3 spikes/pulse at saturation (Table 2). Application of 1 μM bicuculline also increased the average number of evoked spikes (Fig. 5C2); however, the increase was less pronounced (∼2.4 spikes/pulse) compared with that provoked by strychnine (Table 2) and only occurred in ∼70% of the phasic 2°VN (10 of 14 neurons). This differential effect of strychnine and bicuculline complies with the different time courses of the glycinergic and GABAergic IPSPs at subthreshold stimulus intensities (Fig. 3B, 1 and 2).
In contrast to the differential increase of the average spike number per stimulus pulse by strychnine and bicuculline, both antagonists equally affected the timing of the first spike (Table 2). The latency of this spike as well as the first interspike interval in controls (∼6 and ∼5.5 ms, respectively) decreased to a similar extent in the presence of either one of the two blockers (Table 2). In addition, both transmitter antagonists reduced the variability of the latency of the first spike and the interval of the first two spikes (Table 2), leading to a higher synchronization of the synaptically evoked discharge.
The increase in average spike numbers per stimulus pulse in phasic 2°VN in the presence of the glycine or GABAA receptor antagonist was reversed beyond control values after application of 10 μM 7-Cl KYNA to the bath (Fig. 5, A3 to C3). At lower stimulus intensities, only the first spike was activated in the presence of the NMDA receptor antagonist (Fig. 5A3), whereas the average number of spikes decreased below 2 spikes/pulse at saturation intensities of >4 × T (Fig. 5B3; Table 2). In addition, the onset latency of the first spike and the duration of the first interspike interval as well the variability of the latter increased significantly after application of 7-Cl KYNA (Table 2). This is compatible with the idea that the effects of the inhibitory transmitter blockers were mediated in part by an afferent nerve-evoked NMDA component.
Semicircular canal nerve-evoked monosynaptic EPSPs in phasic but not in tonic frog 2°VN are superimposed by disynaptic glycinergic and GABAergic IPSPs that differ in their response profiles. The longer duration of the glycinergic compared with the GABAergic IPSPs reduced the amplitude of the monosynaptic EPSPs and the number of evoked spikes for a longer period. These differences in the inhibitory response profiles are in agreement with the overall larger number of glycinergic as opposed to GABAergic terminal-like structures on vestibular neurons. Both pharmacological types of IPSPs control the magnitude and timing of the monosynaptic excitation in phasic 2°VN thereby promoting the processing of transient signals in this vestibular neuronal subtype.
Differential control of afferent excitation in phasic and tonic 2°VN by ipsilateral inhibitory circuits
The activation of semicircular canal nerve-evoked disynaptic IPSPs superimposed on the labyrinthine nerve-evoked monosynaptic EPSPs in phasic but not tonic 2°VN suggests that the afferent excitation in the two types of vestibular neurons is differently controlled by ipsilateral inhibitory circuits. Because the cerebellum was removed in all experiments, the feed-forward inhibition in phasic 2°VN (Fig. 6) is likely mediated by interneurons in the vestibular nuclei on the same side (Straka and Dieringer 1996). Recurrent collaterals of inhibitory vestibular neurons such as those reported in squirrel monkey (McCrea et al. 1987) could further contribute to the disynaptic inhibition. However, because a disynaptic inhibition is present even in the absence of action potentials, the inhibitory inputs must originate from neurons other than the recorded one. The delayed onset of the inhibition with respect to the monosynaptic EPSP, promotes the processing of transient inputs by limiting the duration of the afferent activation, especially at more depolarized membrane potentials (see Fig. 2B2). In conjunction with the specific intrinsic membrane properties of phasic 2°VN that make these neurons particularly suitable for signal detection (Beraneck et al. 2007; Straka et al. 2004), the transformation of higher frequency vestibular nerve afferent inputs is enhanced. Thus the effect of the disynaptic inhibition on temporal aspects of evoked afferent synaptic responses in phasic 2°VN acts synergistically with their intrinsic membrane properties.
The absence of an inhibition with a disynaptic onset in frog tonic 2°VN suggests that the vestibular circuit that mediates a feed-forward inhibition to phasic 2°VN is specific and does not relay its signals onto all 2°VN. However, the presence of an ipsilateral inhibition with longer latency in tonic 2°VN indicates that IPSPs are not generally absent in these neurons, compatible with the presence of glycinergic and GABAergic terminal-like structures on virtually every frog central vestibular neuron (Fig. 1, B and C) (see Reichenberger et al. 1997). Even though both tonic and phasic 2°VN receive monosynaptic EPSPs from the thickest vestibular nerve afferents (Straka et al. 2004), the disynaptic inhibition that originates from the same fiber spectrum (Straka and Dieringer 2000) is present only in phasic 2°VN. This suggests that the control of afferent inputs from the thickest vestibular nerve fibers by a feed-forward inhibition (Fig. 6) is not a general feature of vestibular signal processing. Rather the differential organization of the inhibition selectively occurs in conjunction with the intrinsic membrane properties and synaptic response dynamics of phasic and tonic frog 2°VN and promotes the idea of parallel pathways for high and low dynamic vestibular signal transformation (Beraneck et al. 2007; Straka et al. 2004).
Differential profiles of glycinergic and GABAergic inhibitory responses
Disynaptic IPSPs in frog phasic 2° semicircular neurons are glycinergic as well as GABAergic (Fig. 6) (Straka and Dieringer 2000; Straka et al. 1997). The latter IPSPs are mediated only by GABAA but not by GABAB receptors (Straka and Dieringer 1996). Glycinergic IPSPs differ from GABAergic IPSPs by larger peak amplitudes and longer duration. The larger peak amplitude might be explained by an activation of more glycinergic synapses, compatible with larger numbers of glycinergic than GABAergic terminal-like structures on frog vestibular neurons (Fig. 1, B and C).
The longer duration of glycinergic IPSPs could be due to slower channel kinetics, additional activation of glycinergic synapses at longer latency, more distal synaptic termination sites, or a particular interaction with the excitatory inputs. Differences in ion channel activation kinetics or location of synapses are unlikely because the rise times of the glycinergic and GABAergic IPSPs (Straka et al. 1997) and the largely somatic location of the glycinergic and GABAergic terminal-like structures are similar (Reichenberger et al. 1997). More likely, an additional recruitment of glycinergic synapses at longer latency activated by higher-order inhibitory interneurons could cause a longer duration. This is also compatible with the generally larger number of glycinergic terminal structures on frog 2°VN. Alternatively, the longer duration of the glycinergic IPSPs might result from the larger peak amplitudes that more effectively reduce the monosynaptic EPSPs through the nonlinear interaction with the NMDA receptor-mediated component (see following text). A larger IPSP would cause a disproportionately larger reduction of the NMDA receptor component due to a nonproportional increase of the voltage-dependent Mg2+ block (Nowak et al. 1984). This would reduce a NMDA receptor-activated Ca2+ influx to a larger extent, and a block of the glycinergic IPSP would therefore unmask an inhibitory component with a longer duration.
Because only few frog vestibular neurons co-localize glycine and GABA (Reichenberger et al. 1997), separate subsets of inhibitory interneurons with different functional roles at multiple synaptic levels might be recruited following activation of ipsilateral semicircular canal afferents. Future experiments in the isolated whole brain employing more natural stimuli will reveal possible differences in the glycinergic and GABAergic feed-forward side-loops for the control of dynamic response parameters in 2°VN.
Disynaptic inhibition of different glutamate receptor-mediated components
The increase of the monosynaptic glutamatergic EPSP by glycinergic or GABAergic blockers in phasic 2°VN diminished after adding a NMDA blocker (Fig. 3). This indicates that the disynaptic inhibition exerts its effect in part by controlling the activation of the NMDA receptor-mediated component compatible with earlier results (Straka et al. 1996). This NMDA component is evoked only by the thickest vestibular nerve afferent fibers, whereas AMPA receptors are activated predominantly, although not exclusively, by thinner fibers (Straka and Dieringer 1996; Straka et al. 1996). Thus the monosynaptic NMDA component and the disynaptic inhibition originate from the same afferent fiber spectrum, i.e., the thickest vestibular afferent fibers (Straka and Dieringer 1996, 2000), compatible with the feed-forward side-loop model of Minor and Goldberg (1991).
The disynaptic inhibition when activated would reduce the magnitude of the NMDA receptor component in phasic 2°VN largely by two interrelated mechanisms. First, the evoked IPSPs decrease the input resistance by opening Cl− channels that cause a shunt of all subsequent inputs (Fig. 2, C and D) and at depolarized membrane potentials, truncate the underlying monosynaptic EPSP (see Fig. 2B, inset). Second, the hyperpolarization due to the inhibition disfacilitates a NMDA receptor activation by reinforcing the voltage-dependent Mg2+ block thereby minimizing long-lasting depolarizations and active dendritic events. Thus a reduction of the NMDA component by ipsilateral glycinergic and GABAergic circuits together with the activation of voltage-dependent K+ conductances (Beraneck et al. 2007) results in a limitation of long-lasting labyrinthine afferent activity in frog phasic 2°VN and particularly promotes transient inputs.
A concomitant decrease of the afferent nerve-evoked AMPA component by the disynaptic inhibition in phasic 2°VN is likely, although its magnitude might be smaller due to the absence of voltage-dependent activation requirements of this glutamate-receptor subtype. To compare the relative proportions of NMDA and AMPA receptor-mediated excitation, controlled by the disynaptic glycinergic and GABAergic feed-forward inhibition, the time courses and respective contribution of both components to the monosynaptic EPSP need to be determined. However, any attempt to study a differential control of the two glutamate receptor components by the inhibitory side-loop is constrained by the fact that the relative contribution of NMDA and AMPA components to the monosynaptic EPSP in frog 2°VN varies and depends on the activation of the presynaptic afferent fiber spectrum (Fig. 6) (Straka and Dieringer 1996, 2000; Straka et al. 1996).
The possibility that the disynaptic inhibition controls the afferent excitation in 2°VN through an interaction not only with the NMDA but also with the AMPA receptor-mediated component is not compromising the interpretation of the present results. According to the side-loop model of Minor and Goldberg (1991), the disynaptic feed-forward inhibition cancels the monosynaptic excitatory inputs from the same spectrum of thick afferent fibers. Assuming that this connectivity is in fact implemented in the frog vestibular circuitry, it is not surprising that the disynaptic inhibition would control both AMPA and NMDA components because the thickest vestibular nerve afferent fibers monosynaptically activate both receptor subtypes in 2°VN (Fig. 6) (Straka et al. 1996). However, the NMDA component considerably exceeds that of the AMPA component activated by these afferent fiber types (Straka et al. 1996). Future experiments on frog brain stem slice preparations with the VIIIth nerve attached and the local inhibitory network intact that allow reversible, repetitive application of AMPA, NMDA, glycine, and GABA receptor antagonists will reveal the precise cellular control mechanism of the ipsilateral disynaptic inhibition.
Common network circuitry underlying vertebrate central vestibular information processing
Based on intrinsic membrane properties and response dynamics, frog tonic and phasic 2°VN are functionally equivalent to rodent type A and type B MVN neurons (Bagnall et al. 2007; Beraneck et al. 2003, 2007; Johnston et al. 1994; Saito and Ozawa 2007; Sekirnjak and du Lac 2006; Takazawa et al. 2004; see Straka et al. 2005). Supported by neuronal modeling (Av-Ron and Vidal 1999; Quadroni and Knöpfel 1994), vestibular signal processing in separate frequency-tuned neuronal channels seems to be a common mechanism. Consequently, intrinsic membrane and emerging network properties of the different neuronal subtypes in mammals might also be co-adapted as in frogs (Beraneck et al. 2007). A differential insertion of vestibular neuronal subtypes into vestibular networks is substantiated by the differences in discharge regularity of type A and B MVN neurons in the guinea pig whole brain (Babalian et al. 1997) or of cat tonic and “kinetic” vestibular neurons (Shimazu and Precht 1965). The irregular resting activity of type B MVN neurons in the isolated whole brain (Babalian et al. 1997) as opposed to their regular discharge in brain slices is likely caused by considerable random spontaneous synaptic activity generated by inhibitory network circuitry. In fact, mouse type B MVN neurons (functionally equivalent to frog phasic 2°VN) receive more spontaneous glycinergic and GABAergic inputs from ipsilateral vestibular interneurons than type A MVN neurons (Camp et al. 2006). This predominance of IPSPs in one vestibular cell type complies with the activation of disynaptic IPSPs in only ∼50% of vestibular neurons after electrical stimulation of vestibular nerve afferents in monkey (Goldberg et al. 1987). In mammals, a feed-forward inhibition that exerts its effect on type B MVN neurons might originate from the specific subset of GABAergic neurons in the parvocellular part of the MVN that differ in their physiology from vestibuloocular projection neurons (Bagnall et al. 2007; Gittis and du Lac 2006; Sekirnjak and du Lac 2006). The presence of a set of specific conductances in type B MVN neurons (Beraneck et al. 2003; Eugene et al. 2007; Johnston et al. 1994) render these neurons particularly sensitive to inhibitory synaptic inputs from, e.g., interneuronal connections. A modifiable gain of the inhibitory side-loop through inhibitory or excitatory inputs would allow short- and long-term adaptive changes for an independent and graded control of responses in type B MVN neurons suitable for a continuous fine tuning of the dynamics as suggested by the feed-forward side-loop model of Minor and Goldberg (1991).
Thanks to Drs. M. Beraneck and P. P. Vidal for valuable comments on an earlier version of this manuscript and to Dr. L. E. Moore for critically reading the manuscript. The work was supported by funding from the Centre Nationale des Etudes Spatiales (CNES).
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 © 2008 by the American Physiological Society