Animals

Sprague–Dawley (SD) rats (at P1, 6, 8, 9, 12, 14, 57 and 60), Pvalbtm1.1(cre)Aibs (PV-Cre) mice (Jackson Laboratory, JAX 008069) (at P1, 8, 42, and 60), and C57BL/6 J mice (at P1, 10, 60) of either sex were used (Supplementary Table 1). All animal procedures were conducted according to the Guide for the Care and Use of Laboratory Animals (National Institutes of Health) and approved by the University of Hong Kong Committee on the Use of Live Animals in Teaching and Research. The day of birth of each pup was recorded as P1 and the litters were reared with a lactating female. All animals were housed in the Centre for Comparative Medicine Research under a 12-h light/dark cycle with ad libitum access to food and water. Effort was made to keep the animals used at a minimum and to decrease animal suffering.

Navigational behavior: dead reckoning test

The ability of rats and mice to return home via a direct path after foraging for food in an open field was assessed in the dead reckoning test [19, 20]. Rodents (both rats and mice) underwent behavioral acclimatization and baseline-cue training 1 week before conducting the dead reckoning test at P60. The test arena was a round table (205 cm in diameter for rats; 120 cm for mice) with 8 holes (11.5 cm in diameter each for rats; 5 cm for mice) arranged at equal distance around the perimeter of the table. These holes were designed to allow a plexiglass box (the “home base”) to be affixed underneath. The table was located 35 cm above the floor in a test room with multiple visual cues, including a ceiling-to-floor black curtain around the arena, computer boxes, a chair, and a refrigerator, which remained during the entire experimental period. Activities of the rodents on the table were recorded using a video camera with infrared taping abilities (HI-8 Sony) positioned above the centre of the round table.

Rodents were fasted for 12 h prior to training and test sessions and brought into the experimental room 1 h before the test began. Training differed slightly between rats and mice. For acclimatization, the rat was first placed on the table to forage for food pellets (1 g Supreme Mini-Treats pellets, Bio-Serv). Then, rats were subjected to baseline-cue training in which the rat was permitted to leave its home base (placed in the same location throughout the training session) through the hole to search for 4 randomly placed food pellets on the table within a 30 min time limit. Rats spontaneously carried each food pellet directly back to the home base. As mice would eat the pellet on the table, supervised training was required [19]. Only 1 pellet (45 mg Dustless Precision Pellets, Bio-Serv) was placed randomly on the table in each trial. When the mouse started to eat the pellet, the trainer would ring a bell and remove the pellet. Another pellet was placed on the table only after the mouse returned to its home base. Mice learned to carry the pellet back to the home base for consumption after 1 week of training. Up to 4 training sessions were given daily for all rodents. After each trial, the surface of table was rinsed completely with 75% ethanol so as to minimize olfactory cues. A trial was complete if the rodent (1) left the home base to search for food on the table, (2) found the food pellet, and (3) returned to the home base with the pellet. Rodents were considered ready for dead reckoning test when they completed 5 consecutive trials.

The dead reckoning test consisted of light probe test, dark probe test and new home location probe test. During each test session, 1 food pellet was placed randomly within a circular area (diameter 30 cm for rats, 20 cm for mice) around the middle of the table. The animal used both self-movement cues and environmental cues in the light probe test during which the lights in the room were kept on.

The dark probe test was conducted in the dark together with a ceiling-to-floor black curtain completely surrounding the table so as to remove visual cues. Rodents navigated mainly using self-movement cues. The dark and light test probes were carried out on alternate days. Four test sessions were conducted per day for each rodent. At least 8 successful test sessions in both light and dark probe tests were recorded and analyzed for each rodent.

In the new home location test, animals were released from a home base that was diametrically opposite to the original home base used for light and dark probe tests. Animals were subjected to 2 trials per day for 2 consecutive days. Completing the new location test required the self-movement cues generated during the outward-bound path of the animals since this test introduced a conflict between the allothetic cues acquired during baseline-cue training.

The paths of rodents were analyzed offline using TopScan (CleverSys Inc, USA) by experimenters blinded to the pre-treatment of the animals. Parameters measured include the duration of searching time and returning time (s), the heading angle (deg), and the mistakes undertaken during returning (number of errors).

Chemogenetic inhibition of PV-expressing interneurons during navigational behavior

Clozapine-N-oxide (CNO, Cayman, 16882) was dissolved in physiological saline to a working concentration of 0.1 mg/ml. To chemogenetically inhibit PV-expressing interneurons in the VN, CNO (1 mg/kg) was intraperitoneally administrated 30 min prior to dead reckoning test to mice that had been pre-transfected with virus carrying expression vector for hM4DGi (AAV5-hSyn-DIO-hM4DGi-mCherry) and received baseline-cue training as described above. To minimize the confounding effect of CNO metabolite, control animals were administered with CNO (1 mg/kg) as comparison.

Preparation and implantation of bicuculline-loaded Elvax slice

Biopolymer Elvax 40P (Dupont, 1 g) was dissolved in 10 ml of dichloromethane (10%, w/v, Merck) at room temperature (22–24 °C). Saline (lyophilized with Ficol, Sigma) or GABAAR antagonist bicuculline (BIC, 20 mM, Tocris, 0109) was mixed with 200 µl of dichloromethane and then added to the dissolved Elvax. The mixture was vortexed before frozen in liquid nitrogen and then stored at − 80 °C overnight. The Elvax block was kept at − 20 °C for 2 weeks before sectioning into 200 µm thick slices. To confine the direction of drug release from the Elvax slice only from one surface and not the other, one side of the Elvax slice was coated with polycaprolactone (PCL) (Sigma) using a spray gun [21]. With the use of 3H-BIC-loaded Elvax slices (1 mm × 1 mm, 200 µm thickness, 20 mM 3H-BIC), we have demonstrated PCL coating on both sides retained 95% of the loaded drug within the slice even after immersing in PBS for 1 week, while uncoated Elvax slices released the majority of the drug in the same time period [6].

Under isoflurane anesthesia (5%, 250 cm3/min, Zoetis), the fourth ventricle of P1 rats or mice was surgically exposed. A piece of Elvax slice (200 µm thick, 1 mm × 1 mm for rats, 0.8 mm × 0.8 mm for mice) loaded with saline or BIC was inserted into the fourth ventricle and placed on the surface of the bilateral VN such that the PCL-coated side of the Elvax slice faced the cerebellar primordium. Particular care was taken during this implantation not to damage the surrounding brain tissues. The skin incision was subsequently sutured. Pups were allowed to recover inside an animal intensive care unit (Thermocare) set to 35 °C before returning to their mothers. Their pups were given neomycin ointment, oral meloxicam and subcutaneous buprenorphine injection for 7 days post-operation. These pups showed positive health status and were kept until they were subjected to different behavioral, electrophysiological or immunohistochemical experiments. Animals were only included for analysis when at the end of the experiment both the Elvax slice was still in situ on the surface of the bilateral VN and the posterior cerebellum was intact.

Viral vectors

Viral vectors were injected to express ChR2 or designer receptor exclusively activated by designer drugs (DREADD) in PV-expressing neurons. Recombinant adeno-associated viruses (AAVs) carrying Cre-activated expression vectors encoding either optogenetic activator ChR2 (AAV5-hSyn-ChR2-eYFP, Penn Vector Core) or the inhibitory DREADD hM4DGi (AAV5-hSyn-DIO-hM4DGi-mCherry, Addgene) were purchased. Viral vectors were stored in aliquots at − 80 °C until use.

Stereotaxic surgery and injection

For experiments employing both retrograde tracing and optogenetics, P42 PV-Cre mice pre-implanted at P1 with either saline- or BIC-loaded Elvax slice were anesthetized with isoflurane (5% induction, 2% maintenance, 250 cm3/min) and mounted in a stereotaxic apparatus (Stoelting Instruments). A 5 µl microsyringe (Hamilton) fitted with a glass capillary (tip diameter 15–20 µm) was slowly advanced through an opening on the skull dorsoventrally to the specific brain area. For retrograde tracing, red retrobeads (Lumafluor; 0.3 µl) was injected into the bilateral SGN of P57 rats (AP: − 10.5; ML: ± 0.6; DV: − 5.4 mm) or cholera toxin subunit B (CTB, Invitrogen; 0.3 µl) to the right SGN of mice (AP: − 5.3; ML: 0.25; DV: − 3.5 mm). AAV5-hSyn-ChR2-eYFP virus (Penn Vector Core; 0.3 µl) was delivered into the right MVN of mice (AP: − 6.2; ML: 0.5; DV: − 3.5 mm). Injections were made under the control of a micropump (World Precision Instruments) at 100 nl/min. Leakage from the injection site was minimized by slowly withdrawing the syringe after a delay of at least 10 min after each injection. The viral titers were > 2 × 1013 virus particles/ml and full transgene expression occurred after 2 weeks. The skin incision was subsequently sutured. Animals remained in a temperature-controlled intensive care unit (Thermocare) until fully recovered from anesthesia. These animals were given neomycin ointment, subcutaneous buprenorphine injection and oral meloxicam for 7 days post-operation. The animals were allowed to recover for 1–2 weeks before the in vitro electrophysiological experiments.

For chemogenetic experiments, P42 PV-Cre mice were anesthetized with isoflurane (5% induction, 2% maintenance, 250 cm3/min). AAV5-hSyn-DIO-hM4DGi-mCherry was stereotaxically injected into the MVN on both sides of the brainstem as described above. The mice were allowed to recover for 2 weeks before behavioral experiments. After the dead reckoning test, immunohistochemical verification of mCherry in PV-expressing neurons within the MVN was done.

Immunohistochemistry

Immunostaining was performed on brain sections from rats and mice. Rats pre-implanted at P1 with either saline- or BIC-loaded Elvax slice were sacrificed at P6, P9 or P12 with overdose of pentobarbital sodium (Dorminal, AlfaMedic International). Similarly, adult mice for co-staining of small conductance calcium-activated potassium (SK) channels and GABAAR were sacrificed. After transcardiac perfusion with 0.9% sodium chloride solution followed by ice-cold 4% paraformaldehyde in 0.1 M phosphate-buffered saline (PBS), the brain was carefully removed and post-fixed at room temperature for 4 h. The brain was washed with PBS and then transferred to ethanol:saline solution for 45 min and stored overnight in 70% ethanol, followed by dehydration through ethanol series (80, 90, 95%, and absolute ethanol for 30 min each). After dehydration, the brain was placed in fresh xylene twice for 10 min each and then equilibrated with wax in a vacuum oven overnight. After embedding in wax, coronal brainstem sections (7 µm) were obtained from the wax block using semi-automated microtome (HM340E, Microm International) and mounted on glass slides.

Before staining, the slices were deparaffinized in fresh xylene 3 times for 3 min each, following rehydration (100, 95, 70, 50, and 30% ethanol for 3 min each). After rinsing with PBS, antigen retrieval was conducted by bathing the specimen slides in boiled sodium citrate buffer (10 mM sodium citrate, 0.05% Tween 20, pH 6.0) for 20 min in a pre-heated steamer. The buffer dish with slides was moved to room temperature and allowed to cool for 20 min before slides were removed and rinsed twice with PBS.

Double staining of perineuronal nets (PNN) and PV in rat brain sections was performed with the incubation of biotinylated Wisteria floribunda agglutinin (1:200, WFA, Sigma, L-1516) and antibody against PV (1:1000; Abcam, ab11427). For comparing expression of SK channels with GABAAR in mice brain sections, antibodies against SK channels (KCNN2, 1:500; Alomone Labs, APC-028), GABAAR (1:500; Millipore, MAB341) and VGluT1 (1:500; Synaptic Systems, 135 304) were used. Slices were incubated in PBS with 3% normal goat serum and 0.3% Triton X-100 for immunofluorescent staining. Signals were visualized with TRITC-conjugated streptavidin or Alexa Fluor 488-conjugated secondary antibody against rabbit IgG. The slides were then washed and mounted. Images were taken using a fluorescent microscope (Olympus) with the same settings for each group. Further offline processing (Adobe Photoshop) was limited to brightness and contrast adjustments which were applied equally across the entire image and also applied equally to controls. The number of MVN cells immuno-positive to PV, WFA or both was counted from coronal slices, using ImageJ (NIH). The results from the same group were pooled for statistical analyses. Cell counting was conducted by individuals blinded to the experimental group.

Brain tissue preparation for electrophysiology

Rats or mice of the desired age were decapitated under isoflurane anesthesia (5%, 250 cm3/min). For animals ≤ P14, the brains were immediately removed to ice-cold artificial cerebrospinal fluid (ACSF) composed of (in mM) 120 NaCl, 2 KCl, 2.5 CaCl2, 1.2 MgCl2, 1.2 KH2PO4, 26 NaHCO3, and 11 glucose, saturated with 95% O2 and 5% CO2 (adjusted to pH 7.3, 290 mOsm). Coronal brainstem slices containing VN (300 µm thick) were prepared on a vibratome (Campden Instruments, MA752) and then transferred to ACSF at 33 °C for 1 h [22]. The slices were then kept at room temperature with sustained aeration before recording.

For P60 animals, the brains upon removal from the skull were put into ice-cold N-Methyl-D-glucamine (NMDG; Sigma, M2004)-containing ACSF [23], composed of (in mM): 92 NMDG, 2.5 KCl, 1.25 NaH2PO4, 30 NaHCO3, 20 HEPES (Sigma; H4034), 25 glucose, 2 thiourea (Sigma; T7875), 5 Na-ascorbate (Sigma; A7631), 3 Na-pyruvate (Sigma; P2256), 0.5 CaCl2, and 10 MgSO4 × 7H2O (adjusted to pH 7.3, 290 mOsm). Coronal brainstem slices (250 μm) were cut on a vibratome. The slices were transferred to NMDG-ACSF at 33 °C for 10 min, and then placed in HEPES-ACSF consisting of (in mM): 92 NaCl, 2.5 KCl, 1.25 NaH2PO4, 30 NaHCO3, 20 HEPES, 25 glucose, 2 thiourea, 5 Na-ascorbate, 3 Na-pyruvate, 2 CaCl2, and 2 MgSO4 × 7H2O, at room temperature until recording.

Slices were transferred to a recording chamber mounted under an upright microscope equipped with infrared-differential interference contrast optics (BX51WI, Olympus) and a CCD camera. During recording, slices were perfused with ACSF containing (in mM) 119 NaCl, 2.5 KCl, 2 CaCl2, 1.25 NaH2PO4, 24 NaHCO3, 12.5 glucose, and 2 MgSO4 × 7H2O warmed to 32 °C using an inline heater (Warner Instrument Company) and aerated by bubbling with 95% O2 and 5% CO2 at a rate of 1.5 ml/min.

Whole-cell patch-clamp recording

MVN neurons identified visually by location, size and fluorescence were recorded using borosilicate glass pipettes of external diameter 1.2 mm/internal diameter 0.69 mm (Harvard Apparatus) pulled from Flaming/Brown micropipette puller (Model P-97, Sutter Instrument) and filled with high chloride internal solution (in mM): 140 CsCl, 10 HEPES, 1 EGTA, 2 MgCl2, 2 Na2ATP, and 1 Na2GTP (adjusted to pH 7.2, 290 mOsm). The advancement of the pipette was manually operated through the micromanipulator (Sutter Instrument).

Signals were amplified using MultiClamp700A (Axon Instruments) and acquired through a 16-bit data acquisition system (DIGIDATA 1322A; Axon Instruments). During whole-cell patch-clamp recording, membrane potentials were corrected for the liquid junction potential (10 mV), and the change of series resistance was sustained within 15%. Only recordings with series resistance smaller than 15 MΩ were included for subsequent analysis. Cell recording was discarded if the leaking currents went beyond 200 pA. The signals of the recording were digitized at 10 kHz and filtered at 3 kHz by the Multiclamp 700A amplifier, Digidata 1322A analog/digital interface board and pCLAMP 10.2 software (Axon Instruments). Data were captured by Clampex 10.2/Multiclamp Commander 1 (Axon Instruments) package.

Neuronal electrical properties/Spontaneous postsynaptic currents

sEPSCs of MVN neurons were recorded at − 70 mV with perfusion of ACSF containing BIC (10 μM) and antagonist of glycine receptor strychnine (Sigma; 1 μM) to remove inhibitory currents. Alternatively, sIPSCs were recorded at − 70 mV and isolated with bath application of AMPAR antagonist CNQX (Tocris; 10 μM) and NMDAR antagonist D-APV (Tocris; 50 μM).

SGN-projecting neurons in the MVN with red retrobeads were visualized under the fluorescence microscope. To activate vestibular afferent input, a bipolar tungsten electrode (225 µm tip separation, Microprobes) was placed in at the medial border between the lateral and medial vestibular nuclei to stimulate primary vestibular afferents. The stimulating electrode was connected to a current isolator (Grass Instrument). For experiments assessing synaptic plasticity, pulses of 0.1 ms duration were delivered at 0.05 Hz with stimulation intensity adjusted for every neuron such that the response was approximately midway between the minimal current and the saturated current (100–200 μA). Theta burst stimulation (TBS) consisting of 4 trains of 10 bursts delivered at 5 Hz with each burst containing 4 pulses at 100 Hz was employed to induce long-term plastic changes at GABAergic synapses. Neurons with a decrease in ePSCGABA amplitude > 20% were categorized as exhibiting LTDGABA. GABAAR-mediated current was recorded in the presence of strychinine (1 μM). BIC was added to the bath after LTD measurements to confirm the GABAAR-mediated identity of the recorded current. Blue laser light (Lumen Dynamics) was used to opto-activate PV-expressing MVN interneurons, transduced with ChR2, through wide-field blue light illumination (473 nm, 1 min) via an optic fiber lens directed at the MVN region. To limit variability of photo-induced activation due to differences in viral vector expression, recordings were only made from slices in which ChR2-eYFP expression was visible at 4X magnification.

For recording of voltage responses, cells were injected with depolarizing current steps from 0 to 800 pA with incremental steps of 20 pA (each sweep lasted 1000 ms duration). The rheobase was defined as the minimal current that could elicit an action potential. Neurons which only responded with one action potential within the stimulation period were classified as single-spiking neurons, while those with more than 1 spike were classified as multiple-spiking neurons. Resting membrane potential was recorded in current clamp mode within the first minute of membrane rupture.

Statistical analysis

No statistical methods were used to predetermine sample sizes, but our sample sizes were based on prior literature and best practices in the field. Data are presented as mean ± SEM. Graphpad Prism 8 was used to plot data and perform statistical analyses. Differences were considered statistically significance for P < 0.05. Data were subjected to D’Agostino-Pearson omnibus test for normal distribution and Levene’s test for equality of variances. Two-way ANOVA was used for Figs. 2A2,3, B2,3, 3F, and 5B. Paired student T-tests were used for data in Fig. 4C3, D2. For Figs. 4A2, B2, 6F1,2, G1,2, one-way ANOVA with Holm-Sidak’s multiple comparisons test was used for data with normal distribution. In cases when the data were not normally distributed, Dunn’s multiple comparisons test was used.