One hundred forty-five male C57BL/6J mice were purchased from Charles River, China. PV-Cre and loxP-flanked ErbB4 (Floxed-ErbB4) mice have been described previously (Wen et al., 2010; Yi et al., 2020) and backcrossed with C57BL/6J for over 10 generations.

PV-Cre mice were crossed with floxed-ErbB4 mice to generate PV-Cre; floxed-ErbB4^+/+ (ErbB4 KO) mice, in which ErbB4 was ablated in PV neurons (Yi et al., 2020). Forty-eight PV-Cre or ErbB4 KO mice were used for behavior analysis, most of which (~70%) were male.

Less than 5 mice were housed per cage at 22°C–24°C and 55%–80% humidity, on a 12:12 h light/dark cycle, with water and food available. The Animal Care and Use Committee of Huazhong University of Science and Technology approved all experiments.

UL has been previously described (Campos-Torres et al., 2005; Chen et al., 2019; Simon et al., 2020; Magyar et al., 2022) in detail. Briefly, after adult male mice (8–10 weeks) were anesthetized with sodium pentobarbital (50 mg/kg, i.p.), an incision was made behind the ear, and the tympanic membrane, malleus, incus, and stapes were surgically removed. A crooked syringe needle tip was inserted into the oval window for mechanical damage and then rinsed with 100% ethanol for chemical demolition of the vestibule. The space created by labyrinthectomy was filled with a hemostatic sponge. For the sham group, the tympanic membrane, malleus, incus, and stapes were removed without damaging the vestibule.

Only the adult male mice (8–10 weeks) were used in behavioral tests. C57BL/6J mice were acclimatized for 7 days after purchase. All mice were handled in the testing room for at least 30 min per day for 3 days before taking behavioral measurements. Tests of static compensation include scoring for posture asymmetry, spontaneous nystagmus, and head tilt. One day prior to UL and one to 5 days following UL, the tests were performed at the same time of day. Blinded assessors scored the static symptoms.

Posture deficits were scored as previously described (Chen et al., 2019): 10 points for spontaneous barrel rolling; 9 points for barrel rolling elicited by a light touch or puff of air; 8 points for the recumbent position on the deafferented side without leg support; 7-6 points for some ipsilateral leg support; 5 points for moving with bilateral leg support; 4-3 points for occasional postural asymmetry; 2-1 points for barely perceptible postural asymmetry.

Electronystagmography was recorded by inserting electrodes into the nasal and lateral orbital margins of the two eyes. The frequency of the fast-phase pullback of the cornea-retinal potential was used to evaluate spontaneous nystagmus (Pietkiewicz et al., 2012).

Head tilt was measured based on the angle between the line from the center of the sacrum to the center of the first thoracic vertebra and the line from the tip of the nose to the center of the parietal.

The medial vestibular nuclei were isolated from both sides after the mice were anesthetized. Tissues were homogenized in ice-cold RIPA buffer (Beyotime Biotechnology) supplemented with proteinase inhibitors and subjected to centrifugation at 12,000 rpm for 15 min at 4°C to collect the supernatant. SDS-PAGE separated proteins were transferred to PVDF membranes. After being blocked with tris-buffered saline (TBS) containing 0.1% Tween-20, and 5% nonfat powdered milk for 1 h at room temperature, PVDF membranes were incubated with primary antibodies overnight at 4°C with antibody dilution buffer (Elabscience, Cat. No. E-IR-R125). After washing, the PVDF membrane was incubated with horseradish peroxidase (HRP)-labeled secondary antibodies (AS014, ABclonal, 1:10000) in TBS and 0.1% Tween-20 for 1 h at room temperature. Signal revelation was performed using an enhanced chemiluminescence substrate (RM00021, ABclonal) detection system. Band intensities were scanned using MicroChemi 4.2 (DNR Bio-imaging Systems, Israel), and quantified using ImageJ (National Institutes of Health, United States). Primary antibodies used were: rabbit anti-α-tubulin (AC003, ABclonal, 1:4000), rabbit anti-parvalbumin (A2791, ABclonal, 1:800), and rabbit anti-ErbB4 (A10853, ABclonal, 1:800), and rabbit anti-GABA (A2052, Sigma, 1:50).

Mice were anesthesia with sodium pentobarbital (50 mg/kg, i.p.) and perfused transcardially with 0.1 mol/L phosphate-buffered saline (PBS) and 4% paraformaldehyde (PFA) in PBS. After post-fixed at 4°C overnight with PFA and gradient dehydration with 15% and 30% sucrose solution, brains were frozen in OCT medium (Tissue-Tek, Sakura, Japan), and sliced at 35 μm using a Leica cryostat (Thermo Scientific, HM550). After washing with PBS five times, slices were mounted to a slide with anti-florescence decay sealant containing DAPI (Solarbio, China). Fluorescent signals were imaged with Olympus Fluoview FV1000. Five films were analyzed per mouse, and each group contained 3 mice.

Adult mice were anesthesia with sodium pentobarbital (50 mg/kg, i.p.) and head-fixed in a stereotaxic device (RWD life science; 68025). Viruses were injected into the unilateral medial vestibular nucleus coordinates relative to bregma: anteroposterior, −6.05 mm; mediolateral, ±0.75 mm; dorsoventral, −4.3 mm using a glass pipette (Cetin et al., 2006) (300 nL per mouse, 50 nL/min). After injection, the glass pipette was left in place for 15 min and removed slowly. The skin was sutured and sterilized with iodophors. The titers of AAV-EF1α-DIO-GCaMP6m-WPRE-hGH-polyA (Brain VTA, PT-0283), AAV-EF1α-DIO-hM3Dq-EGFP-WPRE (Obio Technology, Shanghai, H15959), AAV-EF1α-hM4Di-EGFP-WPRE (Obio Technology, Shanghai, H15963), AAV-retro-hSyn-DIO-hM3Dq-EGFP (GeneChem technologies, AAV0071), AAV-retro-CMV-DIO-EGFP (ViGene Biosciences, AV200001) were 10¹² genome copies per mL. Mice were injected with clozapine-N-oxide [CNO, dissolved in saline 3 mg/kg, i.p.; 10 μM, 1 μL per mouse, guide cannula, (coordinates relative to bregma: anteroposterior, −6.05 mm; mediolateral, ±0.75 mm; dorsoventral, −4.3 mm)] or vehicle to active hM3Dq or hM4Di, guide cannula was implanted 100 μm above the MVN (coordinates relative to bregma: anteroposterior, −6.05 mm; mediolateral, 0.75 mm; dorsoventral, −4.2 mm), 30 min before the behavioral tests.

The procedures have been described previously (Chen et al., 2021). To record contralateral MVN PV neuron’s activity after UL, PV-Cre mice were injected with viruses (AAV-EF1α-DIO-GCaMP6m-WPRE-hGH-polyA) in contralateral MVN. Three weeks later, a gradient-index (GRIN) lens (0.5 mm in diameter and 5.7 mm in length, Fsphotonics Technology, Shanghai, China) was placed above MVN with 100 μm. The GRIN lens was fixed on the skull with dental cement. After a day of recovery, mice underwent UL and recorded 0 h, 1 h, 3 h, 6 h in contralateral MVN. During recording, the distance of the microscope to the GRIN lens was adjusted by a micro-manipulator until the field of view was in focus. Images were collected at 30 frames per second using the MiniScope V2.0 software.

Images were processed using MATLAB R2020b (Mathworks, Natick, MA) with customized code (https://github.com/thinkertech333/analysisforminiscope, Thinkertech, Nanjing, China), and Cellsort 4.0 software. After motion correction (set to 10) and denoising (set to 100), the region of interest was manually selected according to the fluorescence intensity. The Ca²⁺ signals in the 0 h were considered a baseline and the average Ca²⁺ signal in this period was used as a reference (F₀) to normalize the fluorescence signal (ΔF/F). The formula is as the following:

F_signal is the real-time Ca²⁺ signal intensity of the PV neurons of interest, and F₀ is the average Ca²⁺ signal intensity in the baseline period.

Data were analyzed by paired or unpaired t-tests, one-way ANOVA followed by Tukey’s test, or two-way ANOVA followed by the Bonferroni test, using GraphPad Prism 8.00 (GraphPad Software) software. Data are expressed as the mean ± SEM, and statistical significance was considered when p < 0.05.

To investigate the role of PV in bilateral MVN after UL, we first evaluated the expression of PV protein by western blotting. Compared to the sham group, PV expression was increased in the contralateral rather than ipsilateral MVN at 6 h after UL, (Figures 1A,B), one-way ANOVA, [F (4, 15) = 3.231, p = 0.0423], returning to baseline at 1 day after UL.

To further validate whether enhanced protein expression is associated with PV neurons activity, we observed the calcium signals of contralateral MVN PV neurons. Interestingly, we discovered that the majority of PV neurons (83.3%) (Figure 1C) displayed enhanced activity in 3–6 h after UL [Figures 1D,E, One-way ANOVA, F (3, 76) = 11.05, p < 0.0001]. In comparison, the other (16.7%) maintained a stable activity throughout the procedure (data not shown). Together, these results suggest that the expression of PV protein may be selectively increased in the contralateral MVN and accompanied by increased PV neuron activity.

To investigate whether ErbB4 is involved in vestibular compensation, we first detected the alterations in ErbB4 protein expression in MVN following UL. Compared with the control group, the expression of ErbB4 protein significantly increased in the contralateral MVN at 3–24 h after UL, while there was no significant change in the ipsilateral MVN [Figures 2A–D, one-way ANOVA, F (7, 21) = 8.604, p < 0.0001 for contralateral]. We next examined the mice that specifically knock out ErbB4 in PV neurons (PV-Cre; ErbB4-floxed^+/+, ErbB4 KO) (Figures 2E,F), thereby attenuating the activation of PV neurons in both sides of MVN after UL. ErbB4 KO mice exhibited faster vestibular compensation, [Figures 2G–I, postural asymmetry, F (1, 6) = 6.919, p = 0.039; spontaneous nystagmus, F (1, 6) = 39.94, p = 0.0007; head tilt, F (1, 6) = 21.2, p = 0.0037], suggesting ErbB4 in MVN PV neurons is detrimental to the vestibular compensation.

To confirm whether the hyperactivation of PV neurons in contralateral MVN is detrimental to the vestibular compensation, we employed a chemogenetic technique to specifically inhibit the activity of PV neurons in the contralateral MVN after UL. PV-Cre mice were injected with AAV-EF1α-hM4Di-EGFP-WPRE virus in the contralateral MVN (Figures 3A–E), and wild-type mice were used as control. Twenty days later, these mice were subjected to UL followed by CNO i.p. injection for five days (once the mice had undergone UL, CNO was given immediately). Behavior tests were performed 30 min after CNO administration. For the control mice, vestibular symptoms were gradually alleviated, including postural asymmetry, spontaneous nystagmus, and head tilt; interestingly, in PV-Cre mice, all these behavior symptoms were alleviated faster than wild-type mice [Figures 3F–H, postural asymmetry, F (1, 6) = 13.97, p = 0.0096; spontaneous nystagmus, F (1, 6) = 11.18, p = 0.0156; head tilt, F (1, 6) = 20.93, p = 0.0038], suggesting inhibition of contralateral MVN PV neurons effectively accelerated vestibular compensation.

Previously we have demonstrated that reducing the activity of PV neurons in the contralateral MVN is beneficial to vestibular compensation, it needs to be further investigated whether vestibular compensation could also be promoted by enhancing the activity of ipsilateral MVN PV neurons to achieve a new balance between bilateral MVN PV neurons. By injecting AAV-EF1α-DIO-hM3Dq-EGFP-WPRE virus into the ipsilateral MVN of PV-Cre mice, and intraperitoneal injecting CNO (Figures 4A–D), we found that for the wild-type control mice, spontaneous vestibular compensation could be observed 1–5 days after UL, however, activation of PV neurons in the ipsilateral MVN could also further promote vestibular compensation [Figures 4E,G, postural asymmetry, F (1, 6) = 26.05, p = 0.0022; spontaneous nystagmus, F (1, 6) = 35.75, p = 0.001; head tilt, F (1, 6) = 11.08, p = 0.0158]. Together, these results suggest that a new balance of activity in bilateral MVN PV neurons contributes to the amelioration of static deficits after UL.

Vestibular commissures connections are essential in vestibular compensation and are mediated predominantly by inhibitory GABAergic neurons (Gliddon et al., 2005; Malinvaud et al., 2010). To verify whether PV neurons project from one side of MVN to the other, we injected AAV-retro-CMV-DIO-EGFP virus into one side of MVN (Figure 5A). Twenty-eight days later, we found PV neuron projections from the other side of MVN (Figure 5B). Next, we counted the number of PV neurons in bilateral MVN, the injection side accounted for 70.4% of the total, while the contralateral accounted for 29.6% (Figure 5C), suggesting considerable PV neurons in MVN projecting to the opposite side MVN, though we could not exclude the possibility of PV neurons may also innervate other neurons in the same side of MVN. These results suggest PV neurons may influence vestibular compensation through commissural inhibition (Figure 5D).

The vestibular commissural inhibitory system is essential for vestibular compensation (Bergquist et al., 2008; Malinvaud et al., 2010). Modulation of the vestibular commissures inhibitory system facilitates vestibular compensation by enhancing ipsilateral to contralateral GABAergic neuron projections (Chen et al., 2019); however, the cell type of these neurons are unknown. We wondered whether ipsilateral MVN PV neurons could contribute to vestibular compensation through this way. The AAV-retro-hSyn-DIO-hM3Dq-EGFP virus was injected into the contralateral MVN, then a guide cannula was implanted into the ipsilateral MVN on day 10. This allowed for the local injection of CNO into one side of MVN and selective activation of only PV neurons that were projecting from ipsilateral to contralateral MVN (Figures 6A–C). Behavioral analysis revealed that activation of these PV neurons promoted vestibular compensation [Figures 6D–F, postural asymmetry, F (1, 6) = 53.8, p = 0.0003; spontaneous nystagmus, F (1, 6) = 17.5, p = 0.0058; head tilt, F (1, 6) = 10.7, p = 0.017]. These results indicate that PV neurons in the ipsilateral MVN regulate the activity of neurons in the contralateral MVN through commissural inhibition between both sides of MVN, and activation of these PV neurons can facilitate vestibular compensation, which may be brought about by the restoration of activity balance in both sides of MVN (Figure 7).