Background
Vestibular compensation, i.e. functional recovery after a unilateral lesion of the peripheral vestibular system, is a good model to study the cellular and molecular mechanisms involved in the neuroplasticity of the adult central nervous system. Unilateral loss of vestibular inputs to the brainstem vestibular nuclei (VN) is caused by lesioning the peripheral sensory receptors (chemical or mechanical unilateral labyrinthectomy, UL) or unilaterally transecting the vestibular nerve (UVN). These procedures produce a complex vestibular syndrome made of static and dynamic signs. The static signs include postural deficits (increased support surface, head tilt) and ocular motor deficits (spontaneous nystagmus) that are compensated within a few days or weeks, depending on the species and on the type of vestibular deafferentation. Recent findings in the cat indicated that vestibular deficits resulting from a functional deafferentation are compensated faster (2 weeks after mechanical UL) than those induced by an anatomical deafferentation (about 2 months after UVN) [
1]. On the contrary, the dynamic signs (vestibulo-ocular reflex asymmetry, postural instability and equilibrium) are compensated much less completely or over a longer time [
2,
3].
Several hypotheses have been proposed to explain the vestibular compensation process [
2,
4]. At least part of the behavioral recovery process has been attributed to plasticity of neuronal and non-neuronal elements in the deafferented VN. In line, recent findings have demonstrated reactive neurogenesis in the adult cat [
5,
6] and a strong microglial and astroglial reaction [
5,
7‐
9] after unilateral deafferentation in the deafferented VN. Campos-Torres et al. [
9] suggested that the glial reaction may be due to modifications of the environment of VN neurons and most probably to the central inflammatory tissue reaction consecutive to UL. Despite this inflammatory reaction, the central vestibular neurons did not show any sign of apoptosis after the permanent loss of their peripheral inputs [
8]. Glial reaction in the vestibular nuclei may be a beneficial mechanism. It could both protect the injured tissue from long-lasting inflammation and promote the survival of deafferented vestibular neurons [
9] or of newly generated neurons [
6]. Finally, the inflammatory reaction could help in the vestibular compensation process.
Upon insult or infection, the brain inflammatory response generally involves the production of proinflammatory cytokines such as tumor necrosis factor α (TNFα), interleukins (IL1β, IL-6...), interferon gamma (IFN-γ), and various reactive oxygen and nitrogen species [
10]. To date, the cellular inflammatory mediators produced by central vestibular glial cells and/or neurons after peripheral vestibular loss have not been characterized.
Among the inflammatory mediators, TNFα is a prominent effector of the immune response in the brain. It is constitutively expressed by astrocytes, microglia, and neurons in the normal brain [
11] and upregulated following various manifestations of brain injury, including neurodegenerative disorders (Alzheimer's and Parkinson's diseases, multiple sclerosis), brain trauma, and ischemic injury (for review, see [
12]). It has been suggested TNFα may thus have a pathogenic role in these disorders. Nevertheless, TNFα has been shown to exert both neurotoxic and neuroprotective effects in experimental models. TNFα can mediate cell death [
13,
14] but, conversely, TNFα is thought to enhance neuroprotective processes in models of acute neurodegeneration (ischemia, excitotoxicity, axotomy) and chronic (Alzheimer's disease) neurodegeneration [
15‐
20].
Increased levels of TNFα protect cells against damage from cytotoxic insults in part via the induction of cytoprotective genes, including Manganese Superoxide Dismutase (MnSOD) [
21‐
23]. The antioxidant enzyme MnSOD is critical for neuronal survival in various paradigms of stress-induced brain injury [
23‐
26]. Xu et al. [
27] demonstrated that the presence of an intronic nuclear factor κB (NFκB) site in the MnSOD gene was essential for the induction of MnSOD by TNFα. NFκB is a nuclear transcription factor that has a protective role in many tissues in response to stress, inflammation, and immunity. NFκB protects neurons from toxic insults and promotes neuronal growth and synaptic plasticity [
24,
28‐
30]. Widera et al. [
31] demonstrated that NFκB plays a crucial role in the proliferation of neural stem cells induced by TNFα.
The present study aimed to determine whether a chemical UL (transtympanic injection of sodium arsanilate) inflicts an inflammatory response in the vestibular nuclei. This reaction could be expressed by an increased production of proinflammatory cytokines such as TNFα and could induce a compensatory change in the expression of the neuroprotective factors NFκB and MnSOD during the course of the vestibular recovery. This was achieved using immunohistochemistry, notably by quantifying modulation of TNFα, NFκB, and MnSOD labeling in the four main ipsilateral and contralateral VN (medial, superior, lateral, and inferior). The immunohistochemical data were recorded in groups of rats killed 4 h, 8 h, 1 day, 3 days, 7 days, and 15 days after chemical UL. In addition, functional recovery after chemical UL was tested in another group of rats.
TNFα, NFκB, and MnSOD labeling was also quantified in the ipsilateral and contralateral medial (MVN) and inferior (IVN) nuclei in groups of rats submitted to mechanical UL and sacrificed 8 h, 1, and 7 days after the lesion. Effects of mechanical and chemical UL were compared to ensure that arsanilate-induced changes in the expression of inflammatory and neuroprotective factors were not due to the toxicity of arsanilate on brainstem structures.
Methods
Animals care and housing
The experiments were performed on male adult Long Evans rats (300-350 g) obtained from R. Janvier (Le Genest-St-Isle, France). Animals were maintained on a 12 h light/dark cycle at a constant temperature (22°C) with ad libitum access to food and water. The experiments were performed in accordance with the requirements of the French "Ministère de l'Agriculture et de la Pêche" Décret no. 87-848, October 19, 1987 and to the European Communities Council Directive of November 24, 1986 (86/609/EEC). Every attempt was made to minimize both the number and suffering of animals used in these experiments.
Vestibular lesions
Chemical UL was produced by a unilateral transtympanic injection of the toxin, sodium arsanilate. Sodium arsanilate (Sigma, St Louis, MO) was dissolved in 0.3 M sodium carbonate at the concentration of 300 mg/ml and titrated to pH 7.5.
The rats were anesthetized with an intramuscular injection of a solution of ketamine (Ketamine 1000, Virbac, Carros, France, 62.5 mg/kg) and medetomidine (Domitor, Orion Pharma, Espoo, Finland, 0.4 mg/kg). Guided through a dissecting microscope, a 26-gauge needle of injection was advanced through the tympanic membrane, and 100 μl of sodium arsanilate was infused in the left middle-ear. After the injection needle was withdrawn, the external ear canal was packed with sterile hemostatic sponge to prevent leakage of the toxin away from the injection site. When the surgery was completed, the effects of medetomidine were reversed by a subcutaneous injection of the specific antagonist atipamezole (0.33 mg/kg) (Pfizer, Paris, France). Using this protocol, the effects of anesthesia were rapidly reversed and, a few minutes later, the UL rats were able to stand and thereafter to behave nearly normally.
The mechanical UL was performed using a retroauricular approach. The entire length of the ventral edge of the left meatus was destroyed. The tympanic membrane, malleus, incus, and the stapes were removed to expose the pterygopalatine. This artery was coagulated using an electrocoagulator and the oval window was opened. Finally, the vestibule was drilled and aspirated using a suction tube. A postoperative i.m Terramycin (Pfizer, Paris, France) injection was given to prevent infection.
Histological preparation
Thirty rats were subjected to a chemical UL. They received unilateral transtympanic injection of sodium arsanilate on the left side and were killed at six recovery times after vestibular lesion: 4 h (n = 5), 8 h (n = 5), 1 day (n = 5), 3 days (n = 5), 7 days (n = 5), 15 days (n = 5). Sham-operated rats (n = 24, four rats per group) received a unilateral transtympanic injection of 0.9% sterile saline and were killed at the same recovery time as lesioned animals.
Six rats were submitted to mechanical UL on the left side. Based on data recorded in the chemical UL rats, they were killed at three recovery times: 8 h (n = 2), 1 day (n = 2), 7 days (n = 2).
Rats were deeply anesthetized with sodium pentobarbital (100 mg/kg IP) and perfused intracardiacally with 100 ml of 0.9% saline followed by 400 ml of 4% paraformaldehyde (Sigma, St Louis, MO) in 0.1 M phosphate buffer (PB), pH 7.4. The brains were rapidly removed and postfixed in a 4% paraformaldehyde solution for 24 h, cryoprotected by successive transfers into increasing concentrations (10%, 20%, and 30%) of sucrose solution in 0.1 M PB for 72 h at 4°C. Brains were then rapidly frozen on powdered dry ice (CO2 gas) and kept at -80°C until use. They were sectioned coronally at 30 μm on a cryostat, and the sections were stored into stock solution (0.1 M PB (pH 7.4), 30% (v/v) glycerol and 30% (v/v) ethylene glycol) at -20°C until immunohistochemistry experiments.
Immunohistochemistry (IHC)
Free-floating sections were rinsed three times with 0.1 M phosphate buffer saline (PBS) for 5 min, preincubated in a PBS 0.1 M solution containing 0.25% Triton X-100 and 10% bovine serum albumin (BSA) for 1 h at room temperature. Thereafter, sections were incubated under continuous agitation 24 h at 4°C in either a goat polyclonal TNFα antibody (1:200), a rabbit polyclonal NFκB antibody (1:200), or a rabbit monoclonal MnSOD antibody (1:200), diluted in a PBS 0.1 M solution containing 0.25% Triton X-100 and 2% BSA. The antibodies for TNFα and NFκB were obtained from Santa Cruz Biotechnology (Santa Cruz, CA), anti-MnSOD from Epitomic (Burlingame, CA).
On the following day, sections were washed 3 × 5 min in PBS 0.1 M, BSA 2% and 3 × 5 min in PBS 0.1 M, BSA 5%. Then, they were incubated in 1:200 biotinylated secondary antibody diluted in PBS 0.1 M, 2% BSA for 1 h at room temperature. Sections then underwent another round of washing with PBS 0.1 M, BSA 2% (3 × 5 min) and with PBS 0.1 M, BSA 5% (3 × 5 min) and were incubated in biotinylated horseradish peroxidase avidin (Vector Laboratories, Burlingame, CA) for 1 h at room temperature. Finally, for visualization, sections were reacted using 0.2 mg/ml diaminobenzidine and 0.03% hydrogen peroxide in PBS 0.1 M. After being washed with PB, the sections were mounted on microscope slides and allowed to dry overnight. The following day, sections were dehydrated in ascending series of alcohol concentrations (70%, 80%, 90%, and 100%), cleared with xylene, and coverslipped with DPX mounting media.
Immunolabeling for TNFα, NFκB, and MnSOD was examined 4 h, 8 h, 1, 3, 7 and 15 days after chemical UL, and 8 h, 1 and 7 days after mechanical UL. The specificity of the immunolabeling was validated by omission of the primary antibody. No labeling was observed in these control conditions.
Data quantification of TNFα, NFκB, and MnSOD immunoreactivity in the vestibular nuclei
The vestibular nuclei were identified according to Paxinos and Watson's stereotaxic atlas of the rat [
32]. The four VN (medial, superior, lateral, and inferior; hereinafter MVN, SVN, LVN, and IVN respectively) were easily differentiated without counterstaining. Data quantification in the MVN, SVN, LVN, and IVN was done in serial sections collected from the rostral (-9.96 mm) to the caudal (-11.76 mm) part of the brainstem. Stained cells were counted in the MVN (27 sections per rat), SVN (24 sections), LVN (15 sections), and IVN (24 sections). The variable number of sections depended on the rostrocaudal extent of the nuclei [
5]. The immunoreactivity quantification was performed on both sides.
The number of cells expressing TNFα, NFκB, or MnSOD was quantified via computer-assisted image analysis using a DMLB microscope (Leica Microsystems, Wetzlar, Germany) equipped with a DXM 1200 Nikon high-resolution digital camera (1024-1024 pixels; Nikon, Tokyo, Japan) interfaced to a PC computer employing image software for capturing and processing the images (Lucia G, Nikon, Champigny-sur-Marne, France). Quantification of the labeled cells was performed via a grey-level method by adjusting a threshold brightness value, and only cells labeled with a grey value above this threshold were taken into account. Reproducibility was assessed by comparing the same data analyzed independently by two researchers blind to the animal groups. Specific labeling was quantified in each section as the number of labeled cells and was automatically computed and expressed as the mean (± SEM) for each side, each rat, and each subgroup of rats [
5]. To eliminate quantification problems due to potentially asymmetric slides, data were collected from symmetrical slides only. The absence of structural asymmetry in the slides was verified on slides stained with cresyl violet.
Behavioral investigations
Behavioral investigations were performed in another group of rats (n = 6) to assess static and dynamic postural deficits after arsanilate lesion and the time course of the functional recovery.
Static postural deficits and recovery were evaluated by measuring the surface delimited by the four legs of the rat standing erect at rest, without walking. Support surface is a very sensitive parameter used for evaluating vestibular lesion-induced static posture deficit and recovery. To quantify the support surface, the rats were placed in a box equipped with a transparent bottom and filmed from under this box. A scale drawn on the bottom served to take measurements of the four paws location. Four repeated measurements were taken for each rat before chemical UL, and at regular intervals during the course of recovery until complete recovery (60 days). The support surface was measured using an image analysis system (Canvas 9TM, Deneba Software, Miami, FL). An average was calculated for each postlesion time. Data recorded after vestibular lesions were compared to the mean prelesion values using individual references, that is, each animal acted as its own control.
Dynamic postural deficits and recovery were evaluated in the same group of rats (n = 6) using two dynamic reflex tests, the air-righting reflex and the landing reflex. The air-righting reflex was used to test the rat's ability to right itself in the air, while the landing reflex was used to test the response of the rat to a vertical linear acceleration. The tests were considered to evaluate the level of the vestibular deficit and of the recovery process following the vestibular lesion.
During the air-righting test, the rats were held at 50 cm above a cushion, in supine position. The experimenter removed his/her hands as quickly and simultaneously as possible. In a normal rat, just after the drop, the vestibular system detects a change in linear acceleration that triggers the head turning, which in turn induces the sequence of body repositioning. After vestibular lesion, the righting response is totally abolished and it recovers with time through the vestibular compensation process. A score of "0" was assigned when the air-righting reflex was complete, "2" when the animal failed to return itself and fell on its back, and "1" when it showed a partial reaction only.
During the landing reflex, the rats were held by the base of the tail and lowered toward the ground. Animals with intact vestibular function flex their neck and extend their forelimbs as they approach the surface, whereas rats with vestibular lesion at an uncompensated stage show the right response only when their forepaws or vibrissae touch the ground. A score of "0" was assigned when the animals responded properly, and "2" if ground contact was required.
The scores for the air-righting reflex and the landing reflex were summed, providing a global score of dynamic vestibular function and recovery. A zero score indicated a normal vestibular function as observed before the vestibular lesion, or a complete recovery after UL. Scores ranging from 1 to 4 pointed to vestibular dysfunction and more or less complete vestibular compensation. Three repeated measurements were taken for statokinetic tests for each rat before chemical UL and at each postoperative time. A mean was calculated for each postlesion time and compared to the mean prelesion values. As for the support surface evaluation, each animal acted as its own control.
For some experimental animals, double-blind quantification was performed for the dynamic reflex tests, which did not show any dependence on the experimentator when he/she was trained to perform the tests.
Statistical analysis
TNFα, NFκB, and MnSOD immunolabeling in the VN observed after a chemical UL was analyzed using repeated-measures analysis of variance (ANOVAs) with lesion (sham vs arsanilate-lesioned rats) and postoperative recovery period (4, 8 h, and 1, 3, 7, 15 days) as between-factors, and with the nuclei (MVN, SVN, LVN and IVN) as within factors. When a significant difference was revealed, Student's t-tests were used for further two by two comparisons between sham and lesioned animals at the six postoperative times. The levels of significance were set at p < 0.01 (TNFα and NFκB) and p < 0.0083 (MnSOD) according to the number of multiple comparisons (ie, Bonferroni's principle).
TNFα, NFκB, and MnSOD immunolabeling in the MVN and IVN after mechanical and chemical UL was analyzed using repeated-measures ANOVA with type of lesion (mechanical vs chemical UL) and postoperative time (8 h, 1 and 7 days) as between-factors, and with the vestibular nuclei (MVN and IVN) as within factors. Unpaired t-tests were used for comparisons between mechanical and chemical UL animals at the three postoperative times. The levels of significance were set at p < 0.016 (TNFα, NFκB, and MnSOD) according to the Bonferroni's principle.
Support surface values at each postoperative time were compared with preoperative surface values for each rat by a paired Student's t-test. The global score of the rats in the air-righting and landing reflex tests recorded at each postoperative time was compared to the preoperative performance of each rat by a paired Student's t-test. The levels of significance were set at p < 0.0038 (support surface and global score) according to the number of multiple comparaisons (Bonferroni's principle).
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
ML(*) is the main investigator of this work in charge of the study design, analysis and interpretation of results, and writing. CM and LBD both participated in immunohistological investigations and analysis of the data. ML(**) participated mainly in the writing of the report. All authors read and approved the final manuscript.