Reappearance of activity in the vestibular neurones of labyrinthectomized guinea-pigs is not delayed by cycloheximide

Abstract

1. In mammals, unilateral labyrinthectomy induces an immediate depression of the resting discharges in the neurones of the ipsilateral vestibular nuclei. Later on, a spontaneous restoration of this activity occurs. The aim of the present study was to test the possibility that protein synthesis could be involved in the start of this process in the guinea-pig.

2. Cycloheximide (CHX), a protein synthesis inhibitor, was injected intramuscularly 1 h before (30 mg kg⁻¹) and 5 h after (15 mg kg⁻¹) labyrinthectomy.

3. In a first group of animals, CHX was found to induce an inhibition of protein synthesis at levels ranging from 71 to 93 % for 9 h after labyrinthectomy.

4. In a second group of alert animals, we studied single unit activity of second-order vestibular neurones. It was found that, in the 12–16 h post-labyrinthectomy period, at a time when restoration began in guinea-pigs not treated with CHX, the discharges in the labyrinthectomized group treated with CHX were not different from those observed in a previous study in labyrinthectomized animals not treated with CHX.

5. We conclude that protein synthesis is not required for the start of restoration of activity in the vestibular neurones deprived of their ipsilateral labyrinthine input.

In all classes of vertebrates, unilateral labyrinthectomy causes immediate and dramatic disorders in the control of posture and gaze which, however, mostly recover during a process known as vestibular compensation (for review, see Smith & Curthoys, 1989; Dieringer, 1995). These behavioural changes are accompanied by important modifications in the activity of the ipsilateral vestibular nuclei. The resting discharge of the vestibular neurones is seriously depressed just after labyrinthectomy and then recovers spontaneously (Precht et al. 1966; McCabe & Ryu, 1969; Xerri et al. 1983; Pompeiano et al. 1984; Ried et al. 1984; Smith & Curthoys, 1988; Newlands & Perachio, 1990; Ris et al. 1995). In the guinea-pig, the resting activity in the ipsilateral vestibular nuclei, which remains depressed over a period of at least 10 h, starts to reappear from the twelfth hour after the lesion and is normal again 1 week later (Ris et al. 1997).

At present, the mechanism underlying this restoration of activity in the vestibular neurones deprived of their ipsilateral labyrinthine input remains largely unknown. Among the proposed hypotheses, some do not involve de novo protein synthesis, but others do (see Smith & Curthoys, 1989; Darlington et al. 1991). The spontaneous activity of normal vestibular neurones is due both to some pacemaker activity and to the synaptic drive. It could be that the pacemaker activity of the partially deafferented neurones would increase either as a result of synthesis of a type of transmembrane channel not expressed beforehand (protein synthesis-dependent mechanism) or as a consequence of phosphorylation of pre-existing channels (protein synthesis-independent mechanism). According to the ‘denervation supersensitivity hypothesis’, the vestibular neurones would become more sensitive to the excitatory neuromediators that act on them in the control state, either by an increase in the number of receptors (protein synthesis-dependent mechanism), or by a modification of the properties of pre-existing receptors (protein synthesis-independent mechanism). More recently, it has been proposed that the vestibular neurones would become less sensitive to inhibitory neuromediators as a consequence of a decrease in the number of related receptors (protein synthesis-independent mechanism; Cameron & Dutia, 1997). It has also been hypothesized that the immediate silencing of the ipsilateral cells after unilateral labyrinthectomy is induced by an elevation of Ca²⁺ levels in these cells, as a result of high-frequency injury discharges in the lesioned vestibular afferent axons (Darlington & Smith, 1996). In this model the recovery of resting discharge would occur as a result of the gradual restoration of a normal level of intracellular Ca²⁺ (protein synthesis-independent mechanism). Another possibility is that a presynaptic change would occur in the synaptic boutons of the remaining inputs, which would result in more powerful stimulation from these intact inputs. Here again, mechanisms dependent or independent on de novo protein synthesis could play a role. Finally, the reappearance of spontaneous activity in the partially deafferented vestibular neurones could result from a concerted reorganization of activity involving several regions of the nervous system. Such a distributed process could also involve modifications in the efficacy of synapses outside the ipsilateral vestibular nuclei and could, therefore, be dependent or independent on de novo protein synthesis.

In several other types of brain plasticity, it has been demonstrated that the short-term and long-term changes could be dependent on distinct molecular mechanisms (for review, see Bailey et al. 1996; Carew, 1996). Therefore, when one investigates the potential role of protein synthesis in a type of brain plasticity, it is important to define the target period of the study. In the model studied here, the earliest effect was observed 12 h after the lesion. The aim of the present work was to study the role of protein synthesis in this relatively early response of the brain to a lesion of one labyrinth.

Interestingly, it has recently been demonstrated that, within the first 10 h after a labyrinthectomy, several proteins are expressed in the vestibular nuclei as a consequence of a unilateral labyrinthectomy: Fos and Jun in the neurones (Cirelli et al. 1996; Darlington et al. 1996), S100 protein in astrocytes (Rickmann et al. 1995). These proteins are thus expressed at times when they could play a functional role in the start of restoration of spontaneous activity in the partially deafferented vestibular neurones.

Thus, here we have investigated whether protein synthesis is required for the start of restoration of the resting activity in the vestibular neurones deprived of their labyrinthine input. For this purpose, cycloheximide (CHX), a protein synthesis inhibitor (Wettstein et al. 1964; Sisler & Siegel, 1967), was injected intramuscularly according to a routine blocking protein synthesis for 9 h after labyrinthectomy. It was tested whether the first sign of recovery of the vestibular neuronal activity, which was known to occur 12 h after the lesion, was sensitive to the presence of CHX.

METHODS

Experimental studies used pigmented guinea-pigs obtained from an authorized supplier (Charles River, Saint Aubin les Elboeuf, France).

Protein synthesis studies

The aim of this first series of studies was to establish the time course of protein synthesis inhibition provoked by CHX injected intramuscularly twice: the first time at a dosage of 30 mg kg⁻¹ and the second time 6 h later at a dosage of 15 mg kg⁻¹. The level of protein synthesis was assessed by measuring [³⁵S]methionine radioactivity associated with the acid insoluble fraction of brain tissue containing the vestibular nuclei (that is, the part of the brain stem formed by medulla oblongata and pons). Thirty-six guinea-pigs weighing 350–450 g were used. These animals were paired, one receiving a CHX solution and the other receiving equivalent volumes of saline solution. The level of protein synthesis inhibition in each CHX-treated animal was obtained by dividing the radioactivity of the [³⁵S]methionine incorporated into the brainstem proteins by that observed in the brainstem of the control paired animal. Six animals were used for each time point (3 for CHX and 3 for saline).

CHX, obtained from Sigma, was dissolved in saline buffered at pH 7.4 with 0.1 m Hepes. Stock solutions of CHX were prepared at a concentration of 12 mg ml⁻¹. [³⁵S]Methionine ([³⁵S]Met) was obtained from ICN Inc. as an aqueous solution containing 85 %[³⁵S]Met, 15 %[³⁵S]cysteine and 10 mm 2-mercaptoethanol. The original preparation (10 mCi ml⁻¹) was diluted in saline to a concentration of 200 μCi ml⁻¹.

One hour prior to killing, drug-treated animals and controls received 45 μCi (100 g)⁻¹[³⁵S]Met, which was given by injection through a catheter inserted under halothane anaesthesia into the left jugular vein at least 3 h before injection. Volumes of injected isotope were approximately 1 ml. Guinea-pigs were killed by an overdose of pentobarbitone (Nembutal, Abbott) injected through the intraveneous catheter. Death occurred in less than 30 s. Necropsy of the brainstem was performed in approximately 10 min. Incorporation of [³⁵S]Met into TCA-precipitable proteins was measured by the method described by Nguyen & Atwood (1990), modified as follows. Each brainstem was homogenized in 700 μl of ice-cold 0.1 m Hepes and transferred to a capped centrifuge tube. Tubes were spun in a tabletop centrifuge (Biofuge A, Eraeus) at 800 g for 10 min at 4°C, and 50 μl of the supernatant was transferred to a separate microfuge tube for subsequent quantitative protein assay. The protein concentration was estimated by the bicinchoninic acid (BCA) protein microassay (Smith et al. 1985) using bovine serum albumin (BSA) as a standard. To 300 μl of the remaining supernatant were added 100 μl BSA (1 mg ml⁻¹) and 500 μl ice-cold 20 % (w/v) TCA. The mixture was vortexed and spun at 800 g for 10 min at 4°C. The supernatant was discarded and the remaining pellet washed with 1 ml ice-cold 10 % TCA. Each tube was again centrifuged at 800 g for 10 min. This 10 % TCA wash-spin cycle was repeated twice for each tube. The supernatant was discarded.

After the washes, 1 ml of 1 m KOH was added to each centrifuge tube to dissolve the pellet. An 800 μl sample of the mixture was transferred to individual scintillation vials containing 20 ml aqueous counting scintillant (Insta-Gel, Amersham Corp.) and 100 μl glacial acetic acid. Each vial was counted for 10 min in a scintillation counter (Tri-carb, 1600 CA, Packard). Counts were evaluated in the ¹⁴C channel, since the maximum emitted energy from ³⁵S is almost identical to that of ¹⁴C.

The radioactivity found in each vial reflected the amount of TCA-precipitable protein synthesized in one brainstem. All counts were expressed as counts per minute per microgram of protein and adjusted for background and decay. The percentage inhibition of [³⁵S]Met incorporation by CHX was calculated by comparing measurements from the CHX-exposed brainstem with those performed in the absence of inhibitor in the brainstem of the paired animal.

Neuronal activity studies

General design of the experiments

Data about the effect of CHX on the resting discharge of the vestibular neurones in intact and unilaterally labyrinthectomized guinea-pigs were obtained from twelve pigmented animals weighing between 425 and 625 g. All experiments conformed to the recommendations of the Guide for the Care and Use of Laboratory Animals (Department of Health, Education and Welfare Publication, NIH85–23, 1985).

The study comprised four steps. (1) During a first operation (preliminary operation), all guinea-pigs were prepared for chronic recording of extracellular spikes from the neurones in the rostral part of the left vestibular nuclear complex. (2) One week after the preliminary operation, the twelve guinea-pigs received two intramuscular injections of CHX: the first one at a dosage of 30 mg kg⁻¹ and the second one, 6 h later, at a dosage of 15 mg kg⁻¹. (3) Thirty minutes after the first CHX injection, six of the twelve animals were anaesthetized for 1 h with halothane. During the anaesthesia, a left labyrinthectomy was performed 1 h after CHX injection (labyrinthectomized, CHX-treated group). The six remaining guinea-pigs, representing the intact, CHX-treated group, did not undergo surgery. (4) Thirteen hours after the first CHX injection, neuronal activity was recorded in the alert animal during a 4 hour session (i.e. 12 h following the lesion in the labyrinthectomized CHX-treated group). Control values in animals not treated with CHX were established in two previous studies (Ris et al. 1995, 1997), during a 4 h recording session beginning 12 h after labyrinthectomy (labyrinthectomized group) and during a 4 h recording session in a group of six intact individuals (control group).

The surgical procedures used in the preliminary operation and in the labyrinthectomy, the methods utilized for recording neuronal activities and for localizing the analysed neurones histologically, as well as the care taken to avoid any discomfort to the animals were exactly the same as those used and described in detail in our previous work (Ris et al. 1997). They will be only briefly described here.

Surgical procedure

In the preliminary operation, a head holder was fixed to the skull, a craniotomy was performed over the cerebellum, the dura mater was removed and a dental cement chamber was constructed around the hole. Between the recording sessions, the surface of the cerebellum was protected with a siliconized rubber sheet and the chamber sealed with bone wax. The animals were also prepared for recording eye movements by the scleral search coil technique.

One week later, a global left labyrinthectomy was performed in six guinea-pigs. The semicircular canals, utricule and saccule were reached via a retroauricular approach and extirpated.

In order to enable stimulation of the vestibular nerve, two stimulating electrodes consisting of two Teflon-coated silver wires denuded at their tips (ball electrodes) were placed in each intact guinea-pig at the end of the preliminary operation, and just after the lesion in each labyrinthectomized animal. The first electrode was placed on the round window. The second one was placed in front of the horizontal and anterior semicircular ampullae in the intact guinea-pigs and over the hole drilled in the horizontal and anterior semicircular ampullae in the lesioned animals.

Recording of neuronal activities

Each experimental session began by attaching the head of the alert animal to a holding bar located in the centre of a turntable and placed so that the plane defined by the two horizontal semicircular canals was coincident with the earth horizontal plane (Curthoys et al. 1975). The cranial opening was cleaned with sterile saline and antibiotics. Local anaesthetic (lidocaine, 2 %) was also used to irrigate the cement chamber to prevent any pain.

Before the start of the neuronal activity recording session, the vestibular nuclear complex was localized electrophysiologically. A glass microelectrode (1–5 MΩ impedance), attached to a micromanipulator, was lowered vertically through the cranial opening in the direction of the left vestibular nuclei. The N1 and N2 waves of the field potential, evoked by stimulation of the vestibular nerve, were used to map out the location of the rostral part of the vestibular complex.

To record neuronal activity in the vestibular nuclei, a glass micropipette was lowered into the region where the vestibular field potential was previously localized. It is important to realize that, in labyrinthectomized animals, many vestibular neurones were not spontaneously active and thus could only be detected by applying a search stimulus. The left vestibular nuclei were thus explored during stimulation of the left vestibular nerve. The intensity of the square pulses used as a search stimulus were between 2 and 3 times the threshold intensity required to obtain an N1 potential. The small amplitude of the field potential obtained with such pulses did not obscure evoked single action potentials. Once a unit was recruited, its threshold was determined to be the shock level where the neurone followed about 50 % of the pulses. The latent period of the neurone was then measured with a stimulation of twice the unit threshold. The resting activity of each studied neurone was obtained by measuring the mean firing rate of the spikes occurring in 1 min.

RESULTS

Cycloheximide-induced inhibition of [³⁵S]Met incorporation

We determined the extent of the inhibition exerted by CHX on [³⁵S]Met incorporation into neural tissue by injecting CHX intramuscularly into guinea-pigs and extracting a part of their brainstem (medulla oblongata and pons) at different time points (Fig. 1). Two injections of CHX were performed. Following the first one at a dosage of 30 mg kg⁻¹, incorporation was maximally suppressed 1 h post-CHX (93 ± 2 %, mean ±s.d.). After 5 h, the level of inhibition receded to 76 %. The second injection at a dosage of 15 mg kg⁻¹, performed 6 h after the first one, induced a reinforcement of inhibition. A new peak of suppression was observed 1 h after this second injection (90 ± 3 %). Afterwards, inhibition diminished again, but remained greater than 71 % up to 10 h after the first injection. Hence, our double injection routine inhibited brain protein synthesis at a level varying between 93 and 71 % for a duration of 9 h.

Figure 1. Time course of protein synthesis inhibition by CHX.

CHX was injected intramuscularly twice. The first injection at a dosage of 30 mg kg⁻¹ was followed 6 h later by a second one at a dosage of 15 mg kg⁻¹. Inhibition was measured from counts of incorporated [³⁵S]Met activity in the part of the brainstem containing the vestibular nuclei (pons and medulla oblongata). The dosage of CHX that we used maintained an inhibition of protein synthesis at levels ranging from 71 to 93 % (see horizontal dashed lines) for 9 h. Inhibition was maximal 1 h after the first injection. Six animals were assayed for each time point (3 for CHX and 3 for saline control). Each point and the associated error bar represents the mean and the standard deviation of three measurements performed on three pairs of animals.

General characteristics of the studied neurones

Because we wanted to compare the behaviour of neuronal populations recorded in different animal groups, the studied neuronal populations had to be as similar as possible. We therefore defined criteria to select the neurones included in this work as follows. (1) The analysed neurones had to be recorded in the anterior part of the vestibular complex, and (2) had to be activated by electrical stimulation of the ipsilateral vestibular nerve at monosynaptic latencies.

The explored parts of the vestibular nuclear complex were: the superior vestibular nucleus (S), the rostral pole of the parvocellular part of the medial vestibular nucleus (Mp), the magnocellular part of the medial vestibular nucleus (Mm), the lateral vestibular nucleus (L) and the rostral pole of the descending vestibular nucleus (D) (Fig. 2, left column). The sampling of the neurones on the basis of their activation by electric shocks applied on the vestibular nerve had the major advantage of giving silent neurones (especially after labyrinthectomy) the same probability of being picked up as neurones active at rest. The focusing on the second-order vestibular neurones (monosynaptically recruited) had the advantage of allowing the analysis of a more homogeneous population of neurones. A neurone was considered as monosynaptically recruited when it was activated at a latency between 0.85 and 1.15 ms (see Ris et al. 1995). The number of recorded neurones fulfilling our two criteria was 211 in the labyrinthectomized CHX-treated group (the present study) and 222 in the labyrinthectomized group (in a previous study, Ris et al. 1997). Figure 2 (middle and right columns) shows that a possible anatomical bias was avoided. The distributions of the locations of the studied neurones were similar in the two groups.

Figure 2. Location of the studied neurones in the labyrinthectomized group and in the labyrinthectomized CHX-treated group.

Each point represents a neurone which was recruited at a monosynaptic latency (second-order vestibular neurones) by an electric shock applied on the vestibular nerve and whose resting activity was recorded 12–16 h after a left labyrinthectomy. The plane of the illustrated sections is tilted 40 deg posteriorly with respect to the vertical stereotaxic plane as defined by Rapisarda & Bacchelli (1977). The reference point is the centre of the abducens nucleus, which has a diameter of only 0.25 mm in the guinea-pig. Sections in the first, second, third, fourth and fifth rows are respectively in planes 600 μm anterior, 360 μm anterior, at the level, 440 μm posterior and 800 μm posterior with respect to the defined reference point. S, superior vestibular nucleus; Mp, parvocellular part of the medial vestibular nucleus; L, lateral vestibular nucleus; Mm, magnocellular part of the medial vestibular nucleus; X, group X; D, descending vestibular nucleus; 4th v, fourth ventricle; I, interstitial nucleus; C, cochlear nucleus; 5d, descending root of the trigeminal (5th) nerve; 7n, facial nerve; bc, brachium conjunctivalis; 8n, vestibular nerve; 7g, genu of the facial nerve; 6, abducens nucleus; sa, stria acustica; PH, prepositus hypoglossi nucleus; rb, restiform body.

Effect of CHX on the restoration of neuronal activity in unilaterally labyrinthectomized animals

The resting activities of the second-order vestibular neurones recorded 12–16 h after lesioning in the labyrinthectomized CHX-treated group on the one hand and in the labyrinthectomized group on the other were compared using two approaches.

In the first approach, the mean resting rate was calculated animal by animal and the mean of these individual resting rate means (grand mean, n= 6 animals) was calculated. The importance of this procedure comes from the fact that the individual resting rate means varied little from one animal to another, whichever group was studied. We previously established (Ris et al. 1995, 1997) that the resting rate grand mean, which fell as low as 7.1 ± 4.2 spikes s⁻¹ (mean ±s.d.) just after the labyrinthectomy, began to increase in the 12–16 h post-lesion period, reaching a level of 16.3 ± 3.9 spikes s⁻¹ (Fig. 3A). A double intramuscular injection of CHX did not affect this recovery. Between 12 and 16 h after a unilateral labyrinthectomy, the resting rate grand mean of the vestibular neurones in the animals treated with CHX reached a level (15.3 ± 2.6 spikes s⁻¹) which was not statistically different from that observed in the absence of CHX (Student's t test, P > 0.05) (compare Fig. 3B with Fig. 3A).

Figure 3. Effect of CHX on restoration of activity in the second-order vestibular neurones after labyrinthectomy.

The resting activity of the left second-order vestibular neurones was recorded 12–16 h after a left labyrinthectomy in a group of six alert guinea-pigs not treated with CHX (A and C) and in a group of six alert animals that had received two intramuscular injections of CHX (the first at a dosage of 30 mg kg⁻¹, 1 h before labyrinthectomy, and the second at a dosage of 15 mg kg⁻¹, 6 h after the first injection; B and D). The resting discharge of each electrically recruited neurone was measured. A and B, for each animal, the mean resting rate of the studied neurones was calculated. The mean of these six individual means was then calculated and referred to as the resting rate grand mean (n= 6). C and D, histogram of the resting discharges of the pooled neurones from each group of six animals (15.7 ± 20.2 spikes s⁻¹, n= 222 in C; 15.1 ± 18.3 spikes s⁻¹, n= 211 in D). Silent neurones (▪) and spontaneously active neurones () are considered separately.

In a second approach, the distributions of the resting discharges of the neurones recorded in either group were compared. In this approach, the data obtained in each group of six animals were pooled. The distribution of the resting discharges observed 12–16 h after a unilateral labyrinthectomy was established in a previous study (Ris et al. 1997) (Fig. 3C). In the labyrinthectomized group not treated with CHX, the mean (±s.d.) resting rate was 15.7 ± 20.2 spikes s⁻¹, the percentage of silent units was 41 % and, among the spontaneously active neurones, the most frequently observed resting rates ranged between 10 and 20 spikes s⁻¹ (Fig. 3C). Our CHX administration routine did not affect this distribution (Fig. 3D). In the labyrinthectomized CHX-treated group, the mean (±s.d.) resting rate was 15.1 ± 18.3 spikes s⁻¹, the percentage of silent units was 38 % and, here too, among the spontaneously active neurones, the most frequently observed resting rates ranged between 10 and 20 spikes s⁻¹. The mean resting rates measured in the two groups were not statistically different (Wilcoxon rank sum test, P= 0.88).

Effect of CHX on the neuronal activity in intact animals

It was checked whether CHX injections per se could have affected the second-order vestibular neurone resting activity. Indeed, a direct excitatory effect of the drug could have masked the impact of the protein synthesis inhibition on the restoration of the resting discharge in the deafferented neurones.

The values of resting activities of the vestibular neurones in control animals were established in a previous study (Ris et al. 1995, 1997). Our CHX administration routine did not affect these values. From 13 to 17 h after the first CHX injection (Fig. 4B), the resting rate grand mean of the recorded neurones was 33.7 ± 4.2 spikes s⁻¹ (n= 6 animals). This value was not statistically different from that observed in control animals (35.8 ± 6.0 spikes s⁻¹; n= 6 animals; Student's t test, P > 0.05) (Fig. 4A).

Figure 4. Effect of CHX on resting activity of second-order vestibular neurones in intact animals.

The resting activity of second-order vestibular neurones was measured in a group of six intact alert guinea-pigs not treated with CHX (A and C) and in a group of six intact alert animals that had received two intramuscular injections of CHX (the first at a dosage of 30 mg kg⁻¹, and the second at a dosage of 15 mg kg⁻¹, 6 h later). In the CHX-treated group, neuronal activity was recorded 13–17 h after the first CHX injection, as in the labyrinthectomized CHX-treated animals (see Fig. 3B). A and B, resting rate grand mean (n= 6 animals). C and D, histogram of the resting discharges of the pooled neurones from six animals (36.0 ± 21.2 spikes s⁻¹, n= 103 in C; 33.9 ± 20.3 spikes s⁻¹, n= 132 in D). There were no silent units in either group.

The distribution of the resting discharges of the pooled neurones (Fig. 4D) was similar to that observed in control animals (Fig. 4C). There were no silent units either in control animals or in intact, CHX-treated animals. The mean resting rate of the pooled neurones in the CHX-treated guinea-pigs, which was 33.9 ± 20.3 spikes s⁻¹ (n= 132 neurones), was not statistically different from that observed in the control guinea-pigs (36.0 ± 21.2 spikes s⁻¹; n= 103 neurones; Wilcoxon rank sum test, P= 0.65).

Table 1 shows that the neurones of the two groups were recorded in the same regions of the vestibular nuclear complex.

Table 1.

Location of the monosynaptically recruited vestibular neurones in control animals and in animals treated with CHX

Group	S	Mp	Mm	L	D
Control	19	55	22	5	2
Intact CHX-treated	7	31	56	12	27

S, superior vestibular nucleus; Mp, parvocellular part of the medial vestibular nucleus; Mm, magnocellular part of the medial vestibular nucleus; L, lateral vestibular nucleus; D, descending vestibular nucleus. Values are numbers of neurones.

DISCUSSION

The major result of this study was that inhibition of protein synthesis by CHX did not delay the start of the spontaneous restoration of the resting discharge in the vestibular neurones deprived of their labyrinthine input.

The experimental protocol

When faced with a negative result, it is important to examine critically the experimental design.

There is always the possibility that residual protein synthesis may be responsible for the restoration of neuronal activity in the vestibular neurones. We think that this is unlikely in this case. Indeed, the intensity and duration of the induced inhibition (between 93 and 71 % for 9 h) contrasted with the complete lack of effect of our CHX administration routine (compare Fig. 3D with Fig. 3C).

Since the major result of this study was obtained by comparison of the behaviour of two neuronal populations recorded from two distinct groups of animals (labyrinthectomized group and labyrinthectomized CHX-treated group), it was important that the neurones were recorded in the same areas in the two groups. The avoidance of a possible anatomical bias is attested by the fact that all the analysed neurones were recorded in the rostral part of the nuclear vestibular complex (a brain tissue volume of about 3 mm³) (Fig. 2). It is further attested by comparison of the precise locations of the recorded neurones (Fig. 2). Moreover, we previously demonstrated that the tendency for the second-order vestibular neurones to retrieve spontaneously their normal level of resting activity was similar in the different nuclei of the rostral part of the vestibular nuclear complex (Ris et al. 1995, 1997). Hence, small differences in the distribution of the recorded neurones between the two different groups studied could not be a significant bias.

Protein synthesis is not involved in the start of neuronal recovery

Our results demonstrate that de novo protein synthesis is not required for the start of the reappearance of the spontaneous activity. This implies that the different proteins whose expression has been reported to increase in the ipsilateral vestibular neurones within the first 10 h after labyrinthectomy do not play a role in this early phenomenon. However, as far as the proteins studied so far are concerned, it is not so surprising.

After a unilateral labyrinthectomy, the immunoreactivity of S100 protein increases on the ipsilateral side in astrocytes of the lateral vestibular nucleus only (Rickmann et al. 1995). This does not concur with the fact that, 12 h after the lesion, the reappearance of activity has started in the medial vestibular nucleus as well as in the lateral vestibular nucleus. In the medial vestibular nucleus, the resting discharge increases from 8.9 ± 20.9 spikes s⁻¹ 1–5 h after the lesion to 18.7 ± 21.1 spikes s⁻¹ 12–16 h after the lesion, while in the lateral vestibular nucleus this resting discharge increases from 5.9 ± 10.7 spikes s⁻¹ 1–5 h after the lesion to 21.4 ± 26.5 spikes s⁻¹ 12–16 h after the lesion (L. Ris, B. Capon, C. de Waele, P. P. Vidal & E. Godaux, unpublished data).

Krox-24, which is expressed around the tenth hour after lesioning, cannot be involved because it is present only in the mid-vestibular nuclear complex and remains absent in the rostral part (Darlington et al. 1996).

Around 10 h after a labyrinthectomy, Jun is expressed in all the subdivisions of the vestibular nuclear complex. However, the fact that this change is bilateral contrasts with the fact that the restoration of neuronal activity only concerns the ipsilateral vestibular neurones.

Fos has been reported to appear in the vestibular nuclei after a labyrinthectomy. Both in the rat (Cirelli et al. 1996) and in the guinea-pig (Darlington et al. 1996) the effect is bilateral. Moreover no expression of Fos was found 3 h and 6 h after lesioning in the lateral vestibular nucleus by Cirelli et al. (1996), whereas Darlington et al. (1996) reported only a very transient expression in this nucleus. The argument used in the case of S100 protein seems thus to be valid here too.

Protein synthesis and neuronal recovery at later times

The results of the present work rule out the possibility of a role for de novo protein synthesis in the start of the restoration of activity. However, the mechanisms that initiate recovery of neuronal activity in the partially deafferented vestibular nuclei might be distinct from those which complete it or maintain it.

In future work, it will thus be important to investigate a potential role for de novo protein synthesis at later times following labyrinthectomy. However, a long-duration inhibition of protein synthesis in the vestibular nuclei could not be achieved by a systemic injection of CHX. Indeed any further injection of CHX after our double injection was found to be lethal. Therefore, to study the role of protein synthesis in the days following labyrinthectomy, a procedure of local delivery of CHX by an osmotic pump into the vestibular nuclei will be necessary.

What is the mechanism of the start of neuronal recovery ?

If neuronal recovery results from a change in the intrinsic excitability of the vestibular neurones, as suggested by the recent data obtained in vitro by Cameron & Dutia (1997), several mechanisms independent of protein synthesis could play a role. A phosphorylation of pre-existing proteins in the vestibular neurones could be of prime importance (Sansom et al. 1997). The start the of restoration of activity could be due to a downregulation in the number of GABA_A receptors, which would diminish the part of the inhibitory influences among the remaining inputs (Cameron & Dutia, 1997). It is also possible, according to Darlington & Smith (1996), that the reappearance of activity could result from restoration of a normal level of Ca²⁺ in the vestibular neurones after it had been increased as a consequence of high-frequency injury discharges in the lesioned afferent fibres. Finally, another possible mechanism is that partial deafferentation of the vestibular neurones would induce in these neurones the insertion of an intracellular pool of appropriate channels into the plasma membrane (Jacob et al. 1986; Margiolta et al. 1987).

If restoration of activity does not result from a modification of the vestibular neurones themselves but, as first suggested by Llinas & Walton (1979), from a re-organization of the neuronal network of which the vestibular neurones are elements, modifications of synaptic efficacities by mechanisms not involving de novo protein synthesis could play a role: either in synapses between extralabyrinthine fibres and vestibular neurones or in synapses away from the deafferented vestibular neurones.