- Email: [email protected]
Evidence against a role of gap junctions in vestibular compensation
Neuroscience Letters 450 (2009) 97–101
Contents lists available at ScienceDirect
Neuroscience Letters journal homepage: www.elsevier.com/locate/neulet
Evidence against a role of gap junctions in vestibular compensation M. Beraneck a,∗ , A. Uno a , I. Vassias a , E. Idoux b , C. De Waele a , P.-P. Vidal a , N. Vibert c a Laboratoire de Neurobiologie des Réseaux Sensorimoteurs, CNRS UMR7060 – Université Paris Descartes, UFR Biomédicale des Saints-Pères, 45 rue des St-pères, Paris 75270, France b Dept. of Biomedical Engineering, Boston University, USA c Centre de Recherche sur la Cognition et l’Apprentissage, CNRS UMR6234 – Université de Poitiers, France
a r t i c l e
i n f o
Article history: Received 24 October 2008 Received in revised form 25 November 2008 Accepted 29 November 2008 Keywords: Vestibular compensation Gap junction In vitro
a b s t r a c t Vestibular compensation following unilateral labyrinthectomy is associated with modifications of the membrane and firing properties of central vestibular neurons. To determine whether gap junctions could be involved in this process, immunofluorescent detection of neuronal connexin 36 and astrocytic connexin 43 was performed in the medial vestibular nucleus (MVN) of rats. In non-lesioned animals, strong staining was observed with anti-connexin 43 antibodies, while moderate staining was obtained with the anti-connexin 36 antibody. However, the expression of either type of connexin was not modified following unilateral labyrinthectomy. These morphological observations were complemented by pharmacological tests performed during extracellular recordings of MVN neurons in guinea pig brainstem slices. In non-lesioned animals, the gap junction blocker carbenoxolone reversibly decreased or suppressed the spontaneous discharge of about 60% of MVN neurons. This reduction was often associated with a long-duration disruption of the regularity of spike discharge. Both effects were mimicked by several other gap junction blockers, but not by glycyrrhizic acid, an analog of carbenoxolone that does not block gap junctions but reproduces its non-specific effects, nor by the selective inhibitor of astrocytic connexin-based networks endothelin-1. Similar effects of carbenoxolone were obtained on the spontaneous activity of ipsilesional MVN neurons recorded in brainstem slices taken from labyrinthectomized animals. Altogether, these results suggest that neuronal gap junctions are involved in shaping the spontaneous activity of MVN neurons. However, unilateral labyrinthectomy does not affect the expression of gap junctions in vestibular nuclei nor their implication in the regulation of neuronal activity. © 2008 Elsevier Ireland Ltd. All rights reserved.
Gap junctions are probably one of the main pathways for communication between glial cells and in particular between astrocytes, and are known to be strongly involved in the regulation of neuronal physiology and excitability [13]. Although the presence of gap junctions in the vestibular nuclei (VN) has been demonstrated, their functional role in the neuronal networks that control stabilization of gaze and posture is still elusive. The neuronal connexin (Cx) 36 [6] and astrocytic Cx43 [23] are present in the VN as shown by immunohistochemical and/or in situ hybridization methods. The precise anatomical distribution of Cx43 within the vestibular complex has been assessed in young chickens [14], where gap junctions might play a role in establishing the excitability of medial VN neurons (MVNn) during development [18]. The present study aimed at revisiting the role of gap junctions in the VN. In particular, we searched for a potential role of gap junctions in the compensation process that follows labyrinthine
∗ Corresponding author. Tel.: +33 1 42 86 33 86. E-mail address: [email protected] (M. Beraneck). 0304-3940/$ – see front matter © 2008 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.neulet.2008.11.062
lesions. Immediately after unilateral labyrinthectomy (UL), the resting activity of most ipsilesional, deafferented vestibular neurons is drastically reduced, resulting in major postural and oculomotor disturbances [7]. In vertebrates, most of these postural and oculomotor deficits disappear during the following week. In vivo neuronal recordings indicate that this recovery is associated with the progressive restoration of a normal resting activity by the deafferented neurons; however, the molecular mechanisms involved remain largely unidentified [21]. In rat, the first week of vestibular compensation is associated with an astroglial and microglial reaction in the deafferented VN, which could contribute to the recovery of the resting activity of MVNn [4]. Because gap junctions are probably one of the main pathways for intercellular communication [15], they could be involved in a glial cells-mediated regulation of the activity of vestibular neurons [14]. To determine whether gap junctions were involved in vestibular compensation, we investigated the expression of Cx36 and Cx43 in the VN of control and UL animals with immunofluorescence techniques. On top of this structural information, the functional role of gap junction was studied using carbenoxolone (CBX), a popular gap junction blocker, both in control and UL animals. Since car-
98
M. Beraneck et al. / Neuroscience Letters 450 (2009) 97–101
benoxolone is known to have unspecific effects beyond its action on gap junctions, several others structurally related and unrelated gap junctions blockers were used [8,16,20,1] as well as glycyrrhizic acid (GZA), a structural analog of CBX that does not block gap junctions but mimics its unspecific effects [10]. Methods. All animal studies were carried out in accordance with the European Community Council directive of November 24th, 1986. Adult male pigmented Long-Evans rats (CERJ, Le Genest St. Isle, France) were used for immunohistochemical experiments. UL was performed under halothane anesthesia with the help of an operating microscope and using a retro-auricular approach as previously described (see for details [4]). The usual UL-induced statics deficits [7] were observed in all animals immediately following surgery and progressively disappeared within 3–4 days. The animals were maintained in normal housing conditions until their brain was removed. Eight animals were used, including two normal rats, two sham-operated animals and four labyrinthectomized rats killed either 1 day (UL1 rats, n = 2) or 7 days (UL7 rats, n = 2) after the lesion. The animals were anesthetized using chloral hydrate and their brains taken and frozen in powdered dry ice. Coronal cryostat sections (14 m thick) were fixed in absolute ethanol at −20 ◦ C for 10 min and processed for immunofluorescent staining of Cx43 or Cx36. Sections were incubated overnight at room temperature in phosphate buffered saline (PBS 0.1 M, pH 7.4, NaCl 9 g/l) with 5% normal goat serum and 0.3% Triton X-100. The antibodies were an anti-Cx43 rabbit polyclonal antibody (Zymed Laboratories 71-0700, San Francisco, USA) used at a 1/250 dilution, an anti-Cx43 mouse monoclonal antibody specific of a dephosphorylated form of Cx43 (Zymed Laboratories 13-8300) at 1/500 dilution, or an anti-Cx36 rabbit polyclonal antibody (Zymed Laboratories 51-6200) at 1/200 dilution. After PBS rinsing, Alexa 488-conjugated goat anti-rabbit immunoglobulins (Molecular Probes, Leiden, The Netherlands) or cyanine-3-conjugated goat anti-mouse immunoglobulins (Jackson ImmunoResearch Labs, West Grove, USA) were applied. Omission of the primary antibody resulted in a total absence of staining with the fluorochrome. Immunofluorescence intensity in the MVN area was measured on both sides of the brainstem using image-analysis software (Metamorph software, Roper Scientific, Trenton, USA) coupled with a CCD camera (CoolSNAP, Roper Scientific). All images were acquired using the same exposure time and illumination conditions and coded on a 16-bits digital scale (4096 arbitrary units). For each antibody, three to six non-overlapping images measuring 250 m × 186 m were taken on both side of four slices taken from the rostro-caudal extent of the MVN. The average fluorescence intensity was measured and corrected by subtraction of background intensity. Images acquisition and quantification were carried out blind to the status of the animal.
For double-immunolabeling, 25 m vibratome sections were obtained from three transcardially perfused (with ice-cold 4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.4) rats. A monoclonal anti-MAP 2 antibody (Sigma–Aldrich M4403, SaintQuentin-Fallavier, France) used at a 1/500 dilution was then added to the polyclonal anti-Cx43 antibody and immunohistochemistry performed as mentioned. Photographs of double-labeling were obtained with a confocal, laser-scanning microscope (Leica Microsystems, Wetzlar, Germany). Electrophysiological studies were carried out on adult, pigmented guinea pigs. UL were performed as previously described (see for detail [2]). As above, typical UL-induced deficits were observed in all lesioned animals. Coronal brainstem slices were prepared using standard techniques [22]. Extracellular recordings were performed in the MVN with 2 M NaCl-containing glass microelectrodes (2–8 M). The slices were superfused with normal artificial cerebro-spinal fluid (ACSF, control condition) or with a low Ca2+ (0.2 mM)/high Mg2+ (6.3 mM) ACSF to abolish chemical synaptic transmission [22]. The specificity of the effects of CBX was checked using various compounds, as summarized in Table 1. All drugs were obtained from Sigma (Sigma–Aldrich), except endothelin-1 from Neosystem (Strasbourg, France). CBX (100 M), glycyrrhizic acid (100 M) and endothelin-1 (0.5 or 1 M) were dissolved in ACSF prior to use, while 18␣- and 18-glycyrrhetinic acid (50 M), FFA (100 M) and spironolactone (1 M) were dissolved in ACSF containing 0.5% dimethylsulfoxide. We checked on four neurons that ACSF with 0.5% dimethylsulfoxide had no effect per se when applied for 20 min. Drugs were applied only on the neurons that displayed a stable firing rate (FR) for at least 3 min. Because in the low Ca2+ /high Mg2+ solution, very irregular or oscillatory FR were encountered, the FR was computed as the average number of spikes/s obtained over 30 s periods, and tested only when the mean activity had been stable for >3 min. As already observed [16], the kinetics of CBX (100 M) effects were slow and highly variable, and drugs were therefore all applied for 20 min durations. Drug perfusion was stopped if the neuron stopped firing before this maximal duration. Only the MVNn whose spontaneous firing rate was >3 spikes/s (∼80% of the neurons) were analyzed. To eliminate drug-independent variations, the FR was considered to have increased or decreased significantly when reversibly modified by ≥20%. Calculations of means ± standard deviations and statistical processing of all results were carried out using the Systat 8.0 software (SPSS Inc., USA). Statistical comparisons were achieved through non-parametric tests, i.e. Wilcoxon signed-rank tests for paired comparisons, Mann–Whitney U-tests to compare independent samples and chi-square tests to compare proportions of neurons. Friedman test was used to evaluate variation of immunochemistry labeling. Significance level was set at p = 0.05.
Table 1 Summary of pharmacological experiments. Compound
Effect on gap junctions
N (% inhibited) ACSF
Ca2+ –Mg2+ 16 (88%) 2 (100%) – – 6 (0%) –
Non-lesioned animal
Carbenoxolone (CBX) 18␣- and 18-glycyrrhetinic acid Saturating concentration of spironolactone Flufenamic acid (FFA) Glycyrrhizic acid (GZA) Endothelin-1
Gap junction inhibitor Gap junction inhibitors, structurally related Specific mineralocorticoid receptor antagonist Gap junction inhibitors, structurally unrelated Structural analog, no effect on gap junctions Astrocyte specific gap junction inhibitor
20(60%) 2 (100%) 15 (0%) (CBX:75%) 8 (100%) 5 (0%) (CBX:80%) 7 (0%) (CBX:100%)
UL
Carbenoxolone (CBX)
Gap junction inhibitor
47 (66%)
3 (100%)
All compounds were tested on normal, non-lesioned animals, except CBX that was also tested on UL animals. Compounds were tested in normal and uncoupling (low Ca2+ /high Mg2+ ) ACSF. The number of MVNn tested and between brackets the percentage of inhibited cells are reported. For the compounds that did not show any effect, the underlined number shows the percentage of cells inhibited by subsequently applied CBX. Asterisk signals the significant difference between normal ACSF and synaptic uncoupling conditions (compare Fig. 2A2 and B2).
M. Beraneck et al. / Neuroscience Letters 450 (2009) 97–101
99
Fig. 1. Expression of Cx before and after UL. Expression of Cx43 (A) and 36 (B) in the MVN of normal animals. (A2) Double-immunolabeling for Cx43 (in green) and MAP-2 (in red) in the MVN. After UL, no asymmetry was found between the ipsilesional and contralesional sides with either Cx43 (C) or Cx36 (D). (C1 and D1) Immunolabeling with the anti-Cx43 and anti-Cx36 antibody, respectively. Note that Cx36 expression in the MVN was clearly weaker than that in the molecular and granular layers of the cerebellar cortex (upper middle part of the image). (C2 and D2) Mean ± S.D. immunofluorescence intensity obtained for the MVN of control rats (in white) or on the ipsi (in gray) and contralesional side (in black) of sham-operated and previously labyrinthectomized animals. Scale bars in A = 10 m, B = 40 m, C and D = 400 m. For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.
Results. In control rats, all VN were strongly immunopositive for the antibodies directed against Cx43 (Fig. 1A1). Doublelabeling with the neuronal marker MAP-2 examined under confocal microscopy showed that the punctate immunoreactivity was outlining the soma and neuropil of vestibular neurons, and was also present around microvessels (Fig. 1A2). Labeling by the polyclonal anti-Cx43 antibody did not reveal any significant variation of the immunofluorescence intensity over the rostro-caudal extent of the MVN (Friedman ANOVA, p = 0.21). In UL1 and UL7 as well as shamoperated rats, staining obtained with the two antibodies against Cx43 revealed no significant asymmetry of labeling between the ipsilesional and contralesional VN (Fig. 1C1). Moreover, the average fluorescence intensity in the MVN was not significantly different between the four groups of rats (Fig. 1C2). Only moderate staining for the neuron-specific Cx36 was observed in the VN of normal rats (Fig. 1B) while the intensity was higher in the inferior olive as expected (571 ± 37 versus 186 ± 39 au, p < 0.001, not shown). Scattered immunoreactive puncta were distributed over the MVN (Fig. 1B) and in the other subdivisions of the vestibular complex. In UL1 and UL7 rats as well as in sham-operated animals, no asymmetry was found between the two sides of the brainstem (Fig. 1D1). No significant difference of CX36 expression was found between the four experimental groups of animals either (Fig. 1D2). Altogether, the immunofluorescence study did not reveal any change in the expression of Cx43 and Cx36 following UL. CBX was applied on 20 MVNn (mean FR: 17.3 ± 8.5 spikes/s) recorded in normal ACSF. 60% of these had their FR reduced or stopped by CBX (Fig. 2A1). The mean effect of CBX was a reduction
of the FR by ∼40% (Fig. 2A2). The onset latency of the action of CBX was ∼7 min and the peak of the inhibitory effect was reached after ∼16 min. Most effects of CBX were accompanied by a disruption of the regularity of the FR of the recorded neuron (see the evolution of the coefficient of variation (CV) in Fig. 2A1). Furthermore, recovery from the inhibitory effects of CBX was associated in 75% with low-frequency (≤0.07 Hz), high-amplitude oscillations of the FR (Fig. 2C). In about 1/3 of the cases, the inhibitory action of CBX was preceded by a transient excitation of the neuron (asterisk in Fig. 2A1). In a few instances, this transient excitation was actually superimposed on the main inhibitory effect of the drug (asterisk in Fig. 2C). Comparable effects of CBX were observed in the low Ca2+ /high Mg2+ ACSF (Fig. 2B). The proportion of affected cells was nevertheless higher (88%, p = 0.007, Table 1) and the mean inhibition stronger (∼60%). There was no difference in either the kinetics of the effect or the disruption of FR regularity between the two recording conditions. To assess the specificity of the effect of CBX we also tested some of its analogs, as summarized in Table 1. GZA, a structurally related analog of CBX devoid of any effect on gap junctions, did not inhibit MVNn while CBX reduced the spontaneous FR of 80% of the neurons on which both compounds were applied (Fig. 2C). All other tested compounds, whether structurally related to CBX (18␣- and 18glycyrrhetinic acids) or not (FFA), reproduced the inhibitory effect. Perfusion of endothelin-1 did not have any effect on the FR of MVNn, which were all inhibited by subsequently applied CBX (Table 1). This suggests that CBX primarily affects neuronal or neuron-related gap junctions. To check whether CBX could act by
100
M. Beraneck et al. / Neuroscience Letters 450 (2009) 97–101
Fig. 2. Pharmacological study of the role of gap junctions on the spontaneous discharge of MVNn. (A) Effect of CBX on the activity of a MVNn recorded in normal ACSF. (A1) The effects were accompanied by a disruption of the regularity of the cell firing rate, as shown by the CV values. Asterisks here and in panel (C) indicate a transient excitation induced by CBX before its main inhibitory effect. (A2) Summary showing the number and the proportion of neurons that were differentially affected by CBX. (B) Effect of CBX on the activity of a MVNn recorded in synaptic uncoupling conditions. (B1) CBX and its analog 18-glycyrrhetinic acid strongly reduced the firing rate. (B2) In synaptic uncoupling ACSF, the inhibitory effects were stronger than in normal medium. (C) Example of a neuron that did not respond to glycyrrhizic acid but was inhibited by CBX. The enlargements show that the recovery from the inhibitory effect of carbenoxolone was associated with low-frequency, high-amplitude oscillations of the neuron’s firing rate. This kind of oscillations (white arrows) was also observed in synaptic uncoupling conditions (B1, in presence of spironolactone) and in normal ACSF (data not shown). (D) Summary of the effects of CBX on MVNn recorded on the ipsilesional side of brainstem slices taken from previously labyrinthectomized animals. There is no significant difference compared to sham-operated or non-lesioned animals.
activating mineralocorticoid receptors, saturating concentrations of spironolactone were applied before CBX (Table 1). Spironolactone alone did not modify the FR of any neuron. In presence of spironolactone, CBX still reduced the FR of ∼2/3 of MVNn and thus did not act through mineralocorticoid receptors. Altogether, these results suggest that CBX inhibits MVNn by blocking gap junctions, and that gap junctions are involved in shaping the spontaneous activity of MVNn in slices. To assess the involvement of gap junctions in vestibular compensation, we searched whether CBX had different effects on slices taken from labyrinthectomized animals (Fig. 2D). UL were performed in 17 guinea pigs that were allowed to compensate for 1–30 days. Fifty spontaneously active ipsilesional MVNn were recorded in either normal or low Ca2+ /high Mg2+ ACSF (Table 1). Since the effects of CBX did not depend on the compensation time (Kruskal–Wallis ANOVA, p = 0.56), all data were pooled together for analysis. In accordance with previous data [16], the mean FR of all ipsilesional neurons recorded after 1–30 days of vestibular compensation was not significantly different from that of neurons obtained in normal animals. CBX reduced or suppressed the FR of 2/3 of ipsilesional MVNn by ∼45% in average (Fig. 2D). The propor-
tion of MVNn inhibited by CBX and the magnitude of their response were not different from those obtained in normal animals (respective p of 0.90 and 0.76). The latency of onset and peak of the action of CBX were not modified either (p = 0.80). As in normal animals, the inhibitory effects of CBX were accompanied by a disruption of the regularity of the FR, associated with low-frequency oscillations and/or preceded by small, transient excitations. In summary, we found no evidence of a differential effect of CBX on MVNn after UL, which suggests that gap junctions are not involved in the changes of the membrane and firing properties of MVNn that occur during vestibular compensation. Discussion. Because the effects of CBX were observed both in normal and synaptic uncoupling ACSF, they were due to direct effects of the compound on the recorded neurons or neighboring astrocytes. Despite the imperfect selectivity of CBX [16], a number of arguments strongly suggests that the inhibitory effects of CBX on MVNn resulted from the blockade of gap junctions present in the vestibular nuclei. First, the inhibitory and FR disrupting effects of CBX were reproduced by perfusion of 18␣- and 18glycyrrhetinic acids, but not of their inactive structural analog GZA, which has the same unspecific effects as CBX. In addition, the
M. Beraneck et al. / Neuroscience Letters 450 (2009) 97–101
inhibitory effects were reproduced by FFA, another blocker of gap junctions structurally unrelated to glycyrrhetinic acids and with different unspecific effects. Second, the effects of CBX and its analogs were reversible, which suggests that the drug did not damage the recorded cells. Finally, a comparable decrease in the resting discharge of neurons was associated in the inferior olive with the blockade of gap junctions by CBX [3]. Recently, Chepkova et al. [5] showed that CBX blocked NMDA glutamate receptors in the mouse hippocampus. Since NMDA receptors are present on most MVNn and were shown to be involved in the maintenance and regulation of their spontaneous discharge both in vivo and in vitro [9,17,19], they could mediate the inhibitory and FR disrupting action of CBX on MVNn. However, Chepkova et al. [5] mention that the effect of CBX on NMDA receptors was mimicked by GZA, in sharp contrast with our findings. In addition, Smith and Darlington [19] demonstrated that the inhibition of MVNn by NMDA receptors antagonists was weaker on the ipsilesional MVNn of previously labyrinthectomized animals than on control neurons, whereas the effect of CBX reported here was not modified by the lesion. As gap junctions help maintain and regulate the spontaneous activity of MVNn, we wondered if their contribution was modified following UL when the ipsilesional MVNn recover a normal spontaneous activity [21]. Since vestibular compensation is quite similar in all rodents and because of the lack of guinea pig-specific Cx antibodies, adult rats were used for the immunochemical studies. Despite the fact that the first week of vestibular compensation is associated in rats with a strong astroglial reaction [4], and that membrane properties of MVNn are modified during the first weeks following the lesion [2,21], no difference was found between the effects of gap junction blockers on slices taken from labyrinthectomized versus control animals. Consistent with the absence of functional change, we did not find any modification of the expression of Cx36 and Cx43 in the MVN up to 1 week after unilateral labyrinthectomy. This contrasts with data obtained in another model of sensory deafferentation [12]. The discrepancy could be that unlike the sciatic nerve section, labyrinthectomy does not induce an immediate degeneration of the sensory vestibular fibers coming from the lesioned labyrinth [11]. Altogether, no electrophysiological or immunohistochemical evidence was found that gap junctions contribute to the recovery of a normal resting discharge by deafferented central vestibular neurons during vestibular compensation. Acknowledgments Financial support: the French Ministère de la Recherche and Ministère des Affaires Etrangères, the Centre National d’Etudes Spatiales and the Osaka University Medical School. We thank Dr. Giaume and Koulakoff for their help and critical reading of earlier versions of this manuscript.
101
References [1] M. Andersson, F. Blomstrand, E. Hanse, Astrocytes play a critical role in transient heterosynaptic depression in the rat hippocampal CA1 region, J. Physiol. 585 (2007) 843–852. [2] M. Beraneck, E. Idoux, A. Uno, P.P. Vidal, L.E. Moore, N. Vibert, Unilateral labyrinthectomy modifies the membrane properties of contralesional vestibular neurons, J. Neurophysiol. 92 (2004) 1668–1684. [3] T.A. Blenkinsop, E.J. Lang, Block of inferior olive gap junctional coupling decreases Purkinje cell complex spike synchrony and rhythmicity, J. Neurosci. 26 (2006) 1739–1748. [4] A. Campos Torres, P.P. Vidal, C. de Waele, Evidence for a microglial reaction within the vestibular and cochlear nuclei following inner ear lesion in the rat, Neuroscience 92 (1999) 1475–1490. [5] A.N. Chepkova, O.A. Sergeeva, H.L. Haas, Carbenoxolone impairs LTP and blocks NMDA receptors in murine hippocampus, Neuropharmacology 55 (2008) 139–147. [6] D.F. Condorelli, N. Belluardo, A. Trovato-Salinaro, G. Mudo, Expression of Cx36 in mammalian neurons, Brain Res. Brain Res. Rev. 32 (2000) 72–85. [7] I.S. Curthoys, Vestibular compensation and substitution, Curr. Opin. Neurol. 13 (2000) 27–30. [8] J.S. Davidson, I.M. Baumgarten, E.H. Harley, Reversible inhibition of intercellular junctional communication by glycyrrhetinic acid, Biochem. Biophys. Res. Commun. 134 (1986) 29–36. [9] C. de Waele, N. Vibert, M. Baudrimont, P.P. Vidal, NMDA receptors contribute to the resting discharge of vestibular neurons in the normal and hemilabyrinthectomized guinea pig, Exp. Brain Res. 81 (1990) 125–133. [10] F.P. Elsen, E.J. Shields, M.T. Roe, R.J. Vandam, J.D. Kelty, Carbenoxolone induced depression of rhythmogenesis in the pre-Botzinger complex, BMC Neurosci. 9 (2008) 46. [11] R.R. Gacek, J.E. Schoonmaker, Morphologic changes in the vestibular nerves and nuclei after labyrinthectomy in the cat: a case for the neurotrophin hypothesis in vestibular compensation, Acta Otolaryngol. 117 (1997) 244–249. [12] W.E. Li, J.I. Nagy, Activation of fibres in rat sciatic nerve alters phosphorylation state of connexin-43 at astrocytic gap junctions in spinal cord: evidence for junction regulation by neuronal-glial interactions, Neuroscience 97 (2000) 113–123. [13] E.A. Newman, New roles for astrocytes: regulation of synaptic transmission, Trends Neurosci. 26 (2003) 536–542. [14] A. Popratiloff, S.M. Pollack, C. Giaume, K.D. Peusner, Differential expression of connexin 43 in the chick tangential vestibular nucleus, J. Neurosci. Res. 71 (2003) 617–628. [15] N. Rouach, E. Avignone, W. Meme, A. Koulakoff, L. Venance, F. Blomstrand, C. Giaume, Gap junctions and connexin expression in the normal and pathological central nervous system, Biol. Cell 94 (2002) 457–475. [16] R. Rozental, M. Srinivas, D.C. Spray, How to close a gap junction channel. Efficacies and potencies of uncoupling agents, Methods Mol. Biol. 154 (2001) 447–476. [17] M. Serafin, A. Khateb, C. de Waele, P.P. Vidal, M. Muhlethaler, Medial vestibular nucleus in the guinea-pig: NMDA-induced oscillations, Exp. Brain Res. 88 (1992) 187–192. [18] M. Shao, A. Gottesman-Davis, A. Popratiloff, K.D. Peusner, Dye coupling in developing vestibular nuclei, J. Neurosci. Res. 86 (2008) 832–844. [19] P.F. Smith, C.L. Darlington, Comparison of the effects of NMDA antagonists on medial vestibular nucleus neurons in brainstem slices from labyrinthine-intact and chronically labyrinthectomized guinea pigs, Brain Res. 590 (1992) 345–349. [20] M. Srinivas, D.C. Spray, Closure of gap junction channels by arylaminobenzoates, Mol. Pharmacol. 63 (2003) 1389–1397. [21] H. Straka, N. Vibert, P.P. Vidal, L.E. Moore, M.B. Dutia, Intrinsic membrane properties of vertebrate vestibular neurons: function, development and plasticity, Prog. Neurobiol. 76 (2005) 349–392. [22] N. Vibert, A. Bantikyan, A. Babalian, M. Serafin, M. Muhlethaler, P.P. Vidal, Postlesional plasticity in the central nervous system of the guinea- pig: a “topdown” adaptation process? Neuroscience 94 (1999) 1–5. [23] P.A. Wackym, P. Popper, L.A. Abelson, P.H. Ward, P.E. Micevych, Molecular biology of the vestibular system, Acta Otolaryngol. 481 (Suppl.) (1991) 141–149.
Contents lists available at ScienceDirect
Neuroscience Letters journal homepage: www.elsevier.com/locate/neulet
Evidence against a role of gap junctions in vestibular compensation M. Beraneck a,∗ , A. Uno a , I. Vassias a , E. Idoux b , C. De Waele a , P.-P. Vidal a , N. Vibert c a Laboratoire de Neurobiologie des Réseaux Sensorimoteurs, CNRS UMR7060 – Université Paris Descartes, UFR Biomédicale des Saints-Pères, 45 rue des St-pères, Paris 75270, France b Dept. of Biomedical Engineering, Boston University, USA c Centre de Recherche sur la Cognition et l’Apprentissage, CNRS UMR6234 – Université de Poitiers, France
a r t i c l e
i n f o
Article history: Received 24 October 2008 Received in revised form 25 November 2008 Accepted 29 November 2008 Keywords: Vestibular compensation Gap junction In vitro
a b s t r a c t Vestibular compensation following unilateral labyrinthectomy is associated with modifications of the membrane and firing properties of central vestibular neurons. To determine whether gap junctions could be involved in this process, immunofluorescent detection of neuronal connexin 36 and astrocytic connexin 43 was performed in the medial vestibular nucleus (MVN) of rats. In non-lesioned animals, strong staining was observed with anti-connexin 43 antibodies, while moderate staining was obtained with the anti-connexin 36 antibody. However, the expression of either type of connexin was not modified following unilateral labyrinthectomy. These morphological observations were complemented by pharmacological tests performed during extracellular recordings of MVN neurons in guinea pig brainstem slices. In non-lesioned animals, the gap junction blocker carbenoxolone reversibly decreased or suppressed the spontaneous discharge of about 60% of MVN neurons. This reduction was often associated with a long-duration disruption of the regularity of spike discharge. Both effects were mimicked by several other gap junction blockers, but not by glycyrrhizic acid, an analog of carbenoxolone that does not block gap junctions but reproduces its non-specific effects, nor by the selective inhibitor of astrocytic connexin-based networks endothelin-1. Similar effects of carbenoxolone were obtained on the spontaneous activity of ipsilesional MVN neurons recorded in brainstem slices taken from labyrinthectomized animals. Altogether, these results suggest that neuronal gap junctions are involved in shaping the spontaneous activity of MVN neurons. However, unilateral labyrinthectomy does not affect the expression of gap junctions in vestibular nuclei nor their implication in the regulation of neuronal activity. © 2008 Elsevier Ireland Ltd. All rights reserved.
Gap junctions are probably one of the main pathways for communication between glial cells and in particular between astrocytes, and are known to be strongly involved in the regulation of neuronal physiology and excitability [13]. Although the presence of gap junctions in the vestibular nuclei (VN) has been demonstrated, their functional role in the neuronal networks that control stabilization of gaze and posture is still elusive. The neuronal connexin (Cx) 36 [6] and astrocytic Cx43 [23] are present in the VN as shown by immunohistochemical and/or in situ hybridization methods. The precise anatomical distribution of Cx43 within the vestibular complex has been assessed in young chickens [14], where gap junctions might play a role in establishing the excitability of medial VN neurons (MVNn) during development [18]. The present study aimed at revisiting the role of gap junctions in the VN. In particular, we searched for a potential role of gap junctions in the compensation process that follows labyrinthine
∗ Corresponding author. Tel.: +33 1 42 86 33 86. E-mail address: [email protected] (M. Beraneck). 0304-3940/$ – see front matter © 2008 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.neulet.2008.11.062
lesions. Immediately after unilateral labyrinthectomy (UL), the resting activity of most ipsilesional, deafferented vestibular neurons is drastically reduced, resulting in major postural and oculomotor disturbances [7]. In vertebrates, most of these postural and oculomotor deficits disappear during the following week. In vivo neuronal recordings indicate that this recovery is associated with the progressive restoration of a normal resting activity by the deafferented neurons; however, the molecular mechanisms involved remain largely unidentified [21]. In rat, the first week of vestibular compensation is associated with an astroglial and microglial reaction in the deafferented VN, which could contribute to the recovery of the resting activity of MVNn [4]. Because gap junctions are probably one of the main pathways for intercellular communication [15], they could be involved in a glial cells-mediated regulation of the activity of vestibular neurons [14]. To determine whether gap junctions were involved in vestibular compensation, we investigated the expression of Cx36 and Cx43 in the VN of control and UL animals with immunofluorescence techniques. On top of this structural information, the functional role of gap junction was studied using carbenoxolone (CBX), a popular gap junction blocker, both in control and UL animals. Since car-
98
M. Beraneck et al. / Neuroscience Letters 450 (2009) 97–101
benoxolone is known to have unspecific effects beyond its action on gap junctions, several others structurally related and unrelated gap junctions blockers were used [8,16,20,1] as well as glycyrrhizic acid (GZA), a structural analog of CBX that does not block gap junctions but mimics its unspecific effects [10]. Methods. All animal studies were carried out in accordance with the European Community Council directive of November 24th, 1986. Adult male pigmented Long-Evans rats (CERJ, Le Genest St. Isle, France) were used for immunohistochemical experiments. UL was performed under halothane anesthesia with the help of an operating microscope and using a retro-auricular approach as previously described (see for details [4]). The usual UL-induced statics deficits [7] were observed in all animals immediately following surgery and progressively disappeared within 3–4 days. The animals were maintained in normal housing conditions until their brain was removed. Eight animals were used, including two normal rats, two sham-operated animals and four labyrinthectomized rats killed either 1 day (UL1 rats, n = 2) or 7 days (UL7 rats, n = 2) after the lesion. The animals were anesthetized using chloral hydrate and their brains taken and frozen in powdered dry ice. Coronal cryostat sections (14 m thick) were fixed in absolute ethanol at −20 ◦ C for 10 min and processed for immunofluorescent staining of Cx43 or Cx36. Sections were incubated overnight at room temperature in phosphate buffered saline (PBS 0.1 M, pH 7.4, NaCl 9 g/l) with 5% normal goat serum and 0.3% Triton X-100. The antibodies were an anti-Cx43 rabbit polyclonal antibody (Zymed Laboratories 71-0700, San Francisco, USA) used at a 1/250 dilution, an anti-Cx43 mouse monoclonal antibody specific of a dephosphorylated form of Cx43 (Zymed Laboratories 13-8300) at 1/500 dilution, or an anti-Cx36 rabbit polyclonal antibody (Zymed Laboratories 51-6200) at 1/200 dilution. After PBS rinsing, Alexa 488-conjugated goat anti-rabbit immunoglobulins (Molecular Probes, Leiden, The Netherlands) or cyanine-3-conjugated goat anti-mouse immunoglobulins (Jackson ImmunoResearch Labs, West Grove, USA) were applied. Omission of the primary antibody resulted in a total absence of staining with the fluorochrome. Immunofluorescence intensity in the MVN area was measured on both sides of the brainstem using image-analysis software (Metamorph software, Roper Scientific, Trenton, USA) coupled with a CCD camera (CoolSNAP, Roper Scientific). All images were acquired using the same exposure time and illumination conditions and coded on a 16-bits digital scale (4096 arbitrary units). For each antibody, three to six non-overlapping images measuring 250 m × 186 m were taken on both side of four slices taken from the rostro-caudal extent of the MVN. The average fluorescence intensity was measured and corrected by subtraction of background intensity. Images acquisition and quantification were carried out blind to the status of the animal.
For double-immunolabeling, 25 m vibratome sections were obtained from three transcardially perfused (with ice-cold 4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.4) rats. A monoclonal anti-MAP 2 antibody (Sigma–Aldrich M4403, SaintQuentin-Fallavier, France) used at a 1/500 dilution was then added to the polyclonal anti-Cx43 antibody and immunohistochemistry performed as mentioned. Photographs of double-labeling were obtained with a confocal, laser-scanning microscope (Leica Microsystems, Wetzlar, Germany). Electrophysiological studies were carried out on adult, pigmented guinea pigs. UL were performed as previously described (see for detail [2]). As above, typical UL-induced deficits were observed in all lesioned animals. Coronal brainstem slices were prepared using standard techniques [22]. Extracellular recordings were performed in the MVN with 2 M NaCl-containing glass microelectrodes (2–8 M). The slices were superfused with normal artificial cerebro-spinal fluid (ACSF, control condition) or with a low Ca2+ (0.2 mM)/high Mg2+ (6.3 mM) ACSF to abolish chemical synaptic transmission [22]. The specificity of the effects of CBX was checked using various compounds, as summarized in Table 1. All drugs were obtained from Sigma (Sigma–Aldrich), except endothelin-1 from Neosystem (Strasbourg, France). CBX (100 M), glycyrrhizic acid (100 M) and endothelin-1 (0.5 or 1 M) were dissolved in ACSF prior to use, while 18␣- and 18-glycyrrhetinic acid (50 M), FFA (100 M) and spironolactone (1 M) were dissolved in ACSF containing 0.5% dimethylsulfoxide. We checked on four neurons that ACSF with 0.5% dimethylsulfoxide had no effect per se when applied for 20 min. Drugs were applied only on the neurons that displayed a stable firing rate (FR) for at least 3 min. Because in the low Ca2+ /high Mg2+ solution, very irregular or oscillatory FR were encountered, the FR was computed as the average number of spikes/s obtained over 30 s periods, and tested only when the mean activity had been stable for >3 min. As already observed [16], the kinetics of CBX (100 M) effects were slow and highly variable, and drugs were therefore all applied for 20 min durations. Drug perfusion was stopped if the neuron stopped firing before this maximal duration. Only the MVNn whose spontaneous firing rate was >3 spikes/s (∼80% of the neurons) were analyzed. To eliminate drug-independent variations, the FR was considered to have increased or decreased significantly when reversibly modified by ≥20%. Calculations of means ± standard deviations and statistical processing of all results were carried out using the Systat 8.0 software (SPSS Inc., USA). Statistical comparisons were achieved through non-parametric tests, i.e. Wilcoxon signed-rank tests for paired comparisons, Mann–Whitney U-tests to compare independent samples and chi-square tests to compare proportions of neurons. Friedman test was used to evaluate variation of immunochemistry labeling. Significance level was set at p = 0.05.
Table 1 Summary of pharmacological experiments. Compound
Effect on gap junctions
N (% inhibited) ACSF
Ca2+ –Mg2+ 16 (88%) 2 (100%) – – 6 (0%) –
Non-lesioned animal
Carbenoxolone (CBX) 18␣- and 18-glycyrrhetinic acid Saturating concentration of spironolactone Flufenamic acid (FFA) Glycyrrhizic acid (GZA) Endothelin-1
Gap junction inhibitor Gap junction inhibitors, structurally related Specific mineralocorticoid receptor antagonist Gap junction inhibitors, structurally unrelated Structural analog, no effect on gap junctions Astrocyte specific gap junction inhibitor
20(60%) 2 (100%) 15 (0%) (CBX:75%) 8 (100%) 5 (0%) (CBX:80%) 7 (0%) (CBX:100%)
UL
Carbenoxolone (CBX)
Gap junction inhibitor
47 (66%)
3 (100%)
All compounds were tested on normal, non-lesioned animals, except CBX that was also tested on UL animals. Compounds were tested in normal and uncoupling (low Ca2+ /high Mg2+ ) ACSF. The number of MVNn tested and between brackets the percentage of inhibited cells are reported. For the compounds that did not show any effect, the underlined number shows the percentage of cells inhibited by subsequently applied CBX. Asterisk signals the significant difference between normal ACSF and synaptic uncoupling conditions (compare Fig. 2A2 and B2).
M. Beraneck et al. / Neuroscience Letters 450 (2009) 97–101
99
Fig. 1. Expression of Cx before and after UL. Expression of Cx43 (A) and 36 (B) in the MVN of normal animals. (A2) Double-immunolabeling for Cx43 (in green) and MAP-2 (in red) in the MVN. After UL, no asymmetry was found between the ipsilesional and contralesional sides with either Cx43 (C) or Cx36 (D). (C1 and D1) Immunolabeling with the anti-Cx43 and anti-Cx36 antibody, respectively. Note that Cx36 expression in the MVN was clearly weaker than that in the molecular and granular layers of the cerebellar cortex (upper middle part of the image). (C2 and D2) Mean ± S.D. immunofluorescence intensity obtained for the MVN of control rats (in white) or on the ipsi (in gray) and contralesional side (in black) of sham-operated and previously labyrinthectomized animals. Scale bars in A = 10 m, B = 40 m, C and D = 400 m. For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.
Results. In control rats, all VN were strongly immunopositive for the antibodies directed against Cx43 (Fig. 1A1). Doublelabeling with the neuronal marker MAP-2 examined under confocal microscopy showed that the punctate immunoreactivity was outlining the soma and neuropil of vestibular neurons, and was also present around microvessels (Fig. 1A2). Labeling by the polyclonal anti-Cx43 antibody did not reveal any significant variation of the immunofluorescence intensity over the rostro-caudal extent of the MVN (Friedman ANOVA, p = 0.21). In UL1 and UL7 as well as shamoperated rats, staining obtained with the two antibodies against Cx43 revealed no significant asymmetry of labeling between the ipsilesional and contralesional VN (Fig. 1C1). Moreover, the average fluorescence intensity in the MVN was not significantly different between the four groups of rats (Fig. 1C2). Only moderate staining for the neuron-specific Cx36 was observed in the VN of normal rats (Fig. 1B) while the intensity was higher in the inferior olive as expected (571 ± 37 versus 186 ± 39 au, p < 0.001, not shown). Scattered immunoreactive puncta were distributed over the MVN (Fig. 1B) and in the other subdivisions of the vestibular complex. In UL1 and UL7 rats as well as in sham-operated animals, no asymmetry was found between the two sides of the brainstem (Fig. 1D1). No significant difference of CX36 expression was found between the four experimental groups of animals either (Fig. 1D2). Altogether, the immunofluorescence study did not reveal any change in the expression of Cx43 and Cx36 following UL. CBX was applied on 20 MVNn (mean FR: 17.3 ± 8.5 spikes/s) recorded in normal ACSF. 60% of these had their FR reduced or stopped by CBX (Fig. 2A1). The mean effect of CBX was a reduction
of the FR by ∼40% (Fig. 2A2). The onset latency of the action of CBX was ∼7 min and the peak of the inhibitory effect was reached after ∼16 min. Most effects of CBX were accompanied by a disruption of the regularity of the FR of the recorded neuron (see the evolution of the coefficient of variation (CV) in Fig. 2A1). Furthermore, recovery from the inhibitory effects of CBX was associated in 75% with low-frequency (≤0.07 Hz), high-amplitude oscillations of the FR (Fig. 2C). In about 1/3 of the cases, the inhibitory action of CBX was preceded by a transient excitation of the neuron (asterisk in Fig. 2A1). In a few instances, this transient excitation was actually superimposed on the main inhibitory effect of the drug (asterisk in Fig. 2C). Comparable effects of CBX were observed in the low Ca2+ /high Mg2+ ACSF (Fig. 2B). The proportion of affected cells was nevertheless higher (88%, p = 0.007, Table 1) and the mean inhibition stronger (∼60%). There was no difference in either the kinetics of the effect or the disruption of FR regularity between the two recording conditions. To assess the specificity of the effect of CBX we also tested some of its analogs, as summarized in Table 1. GZA, a structurally related analog of CBX devoid of any effect on gap junctions, did not inhibit MVNn while CBX reduced the spontaneous FR of 80% of the neurons on which both compounds were applied (Fig. 2C). All other tested compounds, whether structurally related to CBX (18␣- and 18glycyrrhetinic acids) or not (FFA), reproduced the inhibitory effect. Perfusion of endothelin-1 did not have any effect on the FR of MVNn, which were all inhibited by subsequently applied CBX (Table 1). This suggests that CBX primarily affects neuronal or neuron-related gap junctions. To check whether CBX could act by
100
M. Beraneck et al. / Neuroscience Letters 450 (2009) 97–101
Fig. 2. Pharmacological study of the role of gap junctions on the spontaneous discharge of MVNn. (A) Effect of CBX on the activity of a MVNn recorded in normal ACSF. (A1) The effects were accompanied by a disruption of the regularity of the cell firing rate, as shown by the CV values. Asterisks here and in panel (C) indicate a transient excitation induced by CBX before its main inhibitory effect. (A2) Summary showing the number and the proportion of neurons that were differentially affected by CBX. (B) Effect of CBX on the activity of a MVNn recorded in synaptic uncoupling conditions. (B1) CBX and its analog 18-glycyrrhetinic acid strongly reduced the firing rate. (B2) In synaptic uncoupling ACSF, the inhibitory effects were stronger than in normal medium. (C) Example of a neuron that did not respond to glycyrrhizic acid but was inhibited by CBX. The enlargements show that the recovery from the inhibitory effect of carbenoxolone was associated with low-frequency, high-amplitude oscillations of the neuron’s firing rate. This kind of oscillations (white arrows) was also observed in synaptic uncoupling conditions (B1, in presence of spironolactone) and in normal ACSF (data not shown). (D) Summary of the effects of CBX on MVNn recorded on the ipsilesional side of brainstem slices taken from previously labyrinthectomized animals. There is no significant difference compared to sham-operated or non-lesioned animals.
activating mineralocorticoid receptors, saturating concentrations of spironolactone were applied before CBX (Table 1). Spironolactone alone did not modify the FR of any neuron. In presence of spironolactone, CBX still reduced the FR of ∼2/3 of MVNn and thus did not act through mineralocorticoid receptors. Altogether, these results suggest that CBX inhibits MVNn by blocking gap junctions, and that gap junctions are involved in shaping the spontaneous activity of MVNn in slices. To assess the involvement of gap junctions in vestibular compensation, we searched whether CBX had different effects on slices taken from labyrinthectomized animals (Fig. 2D). UL were performed in 17 guinea pigs that were allowed to compensate for 1–30 days. Fifty spontaneously active ipsilesional MVNn were recorded in either normal or low Ca2+ /high Mg2+ ACSF (Table 1). Since the effects of CBX did not depend on the compensation time (Kruskal–Wallis ANOVA, p = 0.56), all data were pooled together for analysis. In accordance with previous data [16], the mean FR of all ipsilesional neurons recorded after 1–30 days of vestibular compensation was not significantly different from that of neurons obtained in normal animals. CBX reduced or suppressed the FR of 2/3 of ipsilesional MVNn by ∼45% in average (Fig. 2D). The propor-
tion of MVNn inhibited by CBX and the magnitude of their response were not different from those obtained in normal animals (respective p of 0.90 and 0.76). The latency of onset and peak of the action of CBX were not modified either (p = 0.80). As in normal animals, the inhibitory effects of CBX were accompanied by a disruption of the regularity of the FR, associated with low-frequency oscillations and/or preceded by small, transient excitations. In summary, we found no evidence of a differential effect of CBX on MVNn after UL, which suggests that gap junctions are not involved in the changes of the membrane and firing properties of MVNn that occur during vestibular compensation. Discussion. Because the effects of CBX were observed both in normal and synaptic uncoupling ACSF, they were due to direct effects of the compound on the recorded neurons or neighboring astrocytes. Despite the imperfect selectivity of CBX [16], a number of arguments strongly suggests that the inhibitory effects of CBX on MVNn resulted from the blockade of gap junctions present in the vestibular nuclei. First, the inhibitory and FR disrupting effects of CBX were reproduced by perfusion of 18␣- and 18glycyrrhetinic acids, but not of their inactive structural analog GZA, which has the same unspecific effects as CBX. In addition, the
M. Beraneck et al. / Neuroscience Letters 450 (2009) 97–101
inhibitory effects were reproduced by FFA, another blocker of gap junctions structurally unrelated to glycyrrhetinic acids and with different unspecific effects. Second, the effects of CBX and its analogs were reversible, which suggests that the drug did not damage the recorded cells. Finally, a comparable decrease in the resting discharge of neurons was associated in the inferior olive with the blockade of gap junctions by CBX [3]. Recently, Chepkova et al. [5] showed that CBX blocked NMDA glutamate receptors in the mouse hippocampus. Since NMDA receptors are present on most MVNn and were shown to be involved in the maintenance and regulation of their spontaneous discharge both in vivo and in vitro [9,17,19], they could mediate the inhibitory and FR disrupting action of CBX on MVNn. However, Chepkova et al. [5] mention that the effect of CBX on NMDA receptors was mimicked by GZA, in sharp contrast with our findings. In addition, Smith and Darlington [19] demonstrated that the inhibition of MVNn by NMDA receptors antagonists was weaker on the ipsilesional MVNn of previously labyrinthectomized animals than on control neurons, whereas the effect of CBX reported here was not modified by the lesion. As gap junctions help maintain and regulate the spontaneous activity of MVNn, we wondered if their contribution was modified following UL when the ipsilesional MVNn recover a normal spontaneous activity [21]. Since vestibular compensation is quite similar in all rodents and because of the lack of guinea pig-specific Cx antibodies, adult rats were used for the immunochemical studies. Despite the fact that the first week of vestibular compensation is associated in rats with a strong astroglial reaction [4], and that membrane properties of MVNn are modified during the first weeks following the lesion [2,21], no difference was found between the effects of gap junction blockers on slices taken from labyrinthectomized versus control animals. Consistent with the absence of functional change, we did not find any modification of the expression of Cx36 and Cx43 in the MVN up to 1 week after unilateral labyrinthectomy. This contrasts with data obtained in another model of sensory deafferentation [12]. The discrepancy could be that unlike the sciatic nerve section, labyrinthectomy does not induce an immediate degeneration of the sensory vestibular fibers coming from the lesioned labyrinth [11]. Altogether, no electrophysiological or immunohistochemical evidence was found that gap junctions contribute to the recovery of a normal resting discharge by deafferented central vestibular neurons during vestibular compensation. Acknowledgments Financial support: the French Ministère de la Recherche and Ministère des Affaires Etrangères, the Centre National d’Etudes Spatiales and the Osaka University Medical School. We thank Dr. Giaume and Koulakoff for their help and critical reading of earlier versions of this manuscript.
101
References [1] M. Andersson, F. Blomstrand, E. Hanse, Astrocytes play a critical role in transient heterosynaptic depression in the rat hippocampal CA1 region, J. Physiol. 585 (2007) 843–852. [2] M. Beraneck, E. Idoux, A. Uno, P.P. Vidal, L.E. Moore, N. Vibert, Unilateral labyrinthectomy modifies the membrane properties of contralesional vestibular neurons, J. Neurophysiol. 92 (2004) 1668–1684. [3] T.A. Blenkinsop, E.J. Lang, Block of inferior olive gap junctional coupling decreases Purkinje cell complex spike synchrony and rhythmicity, J. Neurosci. 26 (2006) 1739–1748. [4] A. Campos Torres, P.P. Vidal, C. de Waele, Evidence for a microglial reaction within the vestibular and cochlear nuclei following inner ear lesion in the rat, Neuroscience 92 (1999) 1475–1490. [5] A.N. Chepkova, O.A. Sergeeva, H.L. Haas, Carbenoxolone impairs LTP and blocks NMDA receptors in murine hippocampus, Neuropharmacology 55 (2008) 139–147. [6] D.F. Condorelli, N. Belluardo, A. Trovato-Salinaro, G. Mudo, Expression of Cx36 in mammalian neurons, Brain Res. Brain Res. Rev. 32 (2000) 72–85. [7] I.S. Curthoys, Vestibular compensation and substitution, Curr. Opin. Neurol. 13 (2000) 27–30. [8] J.S. Davidson, I.M. Baumgarten, E.H. Harley, Reversible inhibition of intercellular junctional communication by glycyrrhetinic acid, Biochem. Biophys. Res. Commun. 134 (1986) 29–36. [9] C. de Waele, N. Vibert, M. Baudrimont, P.P. Vidal, NMDA receptors contribute to the resting discharge of vestibular neurons in the normal and hemilabyrinthectomized guinea pig, Exp. Brain Res. 81 (1990) 125–133. [10] F.P. Elsen, E.J. Shields, M.T. Roe, R.J. Vandam, J.D. Kelty, Carbenoxolone induced depression of rhythmogenesis in the pre-Botzinger complex, BMC Neurosci. 9 (2008) 46. [11] R.R. Gacek, J.E. Schoonmaker, Morphologic changes in the vestibular nerves and nuclei after labyrinthectomy in the cat: a case for the neurotrophin hypothesis in vestibular compensation, Acta Otolaryngol. 117 (1997) 244–249. [12] W.E. Li, J.I. Nagy, Activation of fibres in rat sciatic nerve alters phosphorylation state of connexin-43 at astrocytic gap junctions in spinal cord: evidence for junction regulation by neuronal-glial interactions, Neuroscience 97 (2000) 113–123. [13] E.A. Newman, New roles for astrocytes: regulation of synaptic transmission, Trends Neurosci. 26 (2003) 536–542. [14] A. Popratiloff, S.M. Pollack, C. Giaume, K.D. Peusner, Differential expression of connexin 43 in the chick tangential vestibular nucleus, J. Neurosci. Res. 71 (2003) 617–628. [15] N. Rouach, E. Avignone, W. Meme, A. Koulakoff, L. Venance, F. Blomstrand, C. Giaume, Gap junctions and connexin expression in the normal and pathological central nervous system, Biol. Cell 94 (2002) 457–475. [16] R. Rozental, M. Srinivas, D.C. Spray, How to close a gap junction channel. Efficacies and potencies of uncoupling agents, Methods Mol. Biol. 154 (2001) 447–476. [17] M. Serafin, A. Khateb, C. de Waele, P.P. Vidal, M. Muhlethaler, Medial vestibular nucleus in the guinea-pig: NMDA-induced oscillations, Exp. Brain Res. 88 (1992) 187–192. [18] M. Shao, A. Gottesman-Davis, A. Popratiloff, K.D. Peusner, Dye coupling in developing vestibular nuclei, J. Neurosci. Res. 86 (2008) 832–844. [19] P.F. Smith, C.L. Darlington, Comparison of the effects of NMDA antagonists on medial vestibular nucleus neurons in brainstem slices from labyrinthine-intact and chronically labyrinthectomized guinea pigs, Brain Res. 590 (1992) 345–349. [20] M. Srinivas, D.C. Spray, Closure of gap junction channels by arylaminobenzoates, Mol. Pharmacol. 63 (2003) 1389–1397. [21] H. Straka, N. Vibert, P.P. Vidal, L.E. Moore, M.B. Dutia, Intrinsic membrane properties of vertebrate vestibular neurons: function, development and plasticity, Prog. Neurobiol. 76 (2005) 349–392. [22] N. Vibert, A. Bantikyan, A. Babalian, M. Serafin, M. Muhlethaler, P.P. Vidal, Postlesional plasticity in the central nervous system of the guinea- pig: a “topdown” adaptation process? Neuroscience 94 (1999) 1–5. [23] P.A. Wackym, P. Popper, L.A. Abelson, P.H. Ward, P.E. Micevych, Molecular biology of the vestibular system, Acta Otolaryngol. 481 (Suppl.) (1991) 141–149.