Contents lists available at ScienceDirect

Neuroscience Letters

journal homepage: www.elsevier.com/locate/neulet

Surgery-free video-oculography in mouse models: enabling quantitative andshort-interval longitudinal assessment of vestibular function

Xiaojie Yanga,b, Shiyue Zhoua, Jiaojiao Wua, Qun Liaoa, Changquan Wangc, Minghua Liua,Lei Qua,d, Yuan Zhange, Cheng Chenge,f,g, Renjie Chaie,f,h,i, Kun Zhanga, Xiaojie Yuj,Pingbo Huangj,k,l, Lian Lium, Wei Xiongm, Shi Chenb, Fangyi Chena,⁎

a Department of Biomedical Engineering, Southern University of Science and Technology, Shenzhen, Guangdong 518055, Chinab Key Laboratory of Combinatorial Biosynthesis and Drug Discovery, Ministry of Education, School of Pharmaceutical Sciences, Zhongnan Hospital, Wuhan University,Wuhan, Hubei 430072, Chinac Institute of Microelectronics of Chinese Academy of Sciences, Beijing 100029, Chinad State Key Laboratory of Reliability and Intelligence of Electrical Equipment, Department of Biomedical Engineering, Hebei University of Technology, Tianjin 300130,Chinae Key Laboratory for Developmental Genes and Human Disease, Ministry of Education, Institute of Life Sciences, Southeast University, Nanjing, Jiangsu 210096, Chinaf Research Institute of Otolaryngology, Nanjing, Jiangsu 210008, Chinag Department of Otolaryngology-Head and Neck Surgery, Nanjing Drum Tower Hospital Affiliated to Nanjing University Medical School, Nanjing, Jiangsu 210008, Chinah Co-innovation Center of Neuroregeneration, Nantong University, Nantong, Jiangsu 226001, Chinai Jiangsu Province High-Tech Key Laboratory for Bio-Medical Research, Southeast University, Nanjing, Jiangsu 211189, ChinajDivision of Life Science, Hong Kong University of Science and Technology, Hong Kong, Chinak Department of Chemical and Biological Engineering, Hong Kong University of Science and Technology, Hong Kong, Chinal State Key Laboratory of Molecular Neuroscience, Hong Kong University of Science and Technology, Hong Kong, Chinam School of Life Sciences, IDG/McGovern Institute for Brain Research, Tsinghua University, Beijing 100084, China

A R T I C L E I N F O

Keywords:Vestibulo-ocular reflex (VOR)Behavioral assayGenetic screen3,3′-iminodipropionitrile (IDPN)OtotoxicityVestibular functionVideo-oculography (VOG)Mouse model

A B S T R A C T

Vestibulo-ocular reflex (VOR) responding to acceleration stimuli is originated from the vestibular apparatusesand thus widely used as an in vivo indicator of the vestibular function. We have developed a vestibular functiontesting (VFT) system that allows to evaluate VOR response with improved efficiency. The previously requiredsurgical procedure has been avoided by using a newly designed animal-immobility setup. The efficacy of our VFTsystem was demonstrated on the mice with vestibular abnormalities caused by either genetic mutations(Lhfpl5−/− or Cdh23−/−) or applied vestibulotoxicant (3,3′-iminodipropionitrile, IDPN). Daily longitudinalinspection of the VOR response in the IDPN-administered mice gives the first VOR-based daily-progressionprofile of the vestibular impairment. The capability of VOR in quantifying the severity of toxicant-inducedvestibular deficits has been also demonstrated. The acquired VOR-measurement results were validated againstthe corresponding behavioral-test results. Further validation against immunofluorescence microscopy was ap-plied to the VOR data obtained from the IDPN-administered mice. We conclude that the improved efficiency ofour surgery-free VFT system, firstly, enables the characterization of VOR temporal dynamics and quantificationof vestibular-impairment severity that may reveal useful information in toxicological and/or pharmaceuticalstudies; and, secondly, confers our system promising potential to serve as a high-throughput screener foridentifying genes and drugs that affect vestibular function.

1. Introduction

The vestibular system is responsible for the balance control and

motor function by sensing both linear and angular accelerations.Functional assessment of the vestibular system may reveal the extentand time-course characteristics of the vestibular impairment following

https://doi.org/10.1016/j.neulet.2018.12.036Received 6 November 2018; Received in revised form 18 December 2018; Accepted 26 December 2018

⁎ Corresponding author.E-mail addresses: yangxj3@sustc.edu.cn (X. Yang), zhousy@mail.sustc.edu.cn (S. Zhou), wujj1913@mails.jlu.edu.cn (J. Wu), qunliao_0902@163.com (Q. Liao),

wangchangquan@ime.ac.cn (C. Wang), liumh@sustc.edu.cn (M. Liu), qulei21@163.com (L. Qu), 230179765@seu.edu.cn (Y. Zhang),230159248@seu.edu.cn (C. Cheng), renjiec@seu.edu.cn (R. Chai), zhangk@mail.sustc.edu.cn (K. Zhang), xyuaq@ust.hk (X. Yu), bohuangp@ust.hk (P. Huang),liulian15@mails.tsinghua.edu.cn (L. Liu), wei_xiong@tsinghua.edu.cn (W. Xiong), shichen@whu.edu.cn (S. Chen), chenfy@sustc.edu.cn (F. Chen).

Neuroscience Letters 696 (2019) 212–218

Available online 28 December 20180304-3940/ © 2018 Published by Elsevier B.V.

T

genetic disorders or treatment of vestibulotoxicants and, therefore,provide information regarding the occurrence and type of the muta-tions, as well as the toxicity of the drugs. Such assessment is usuallyconducted on rodent models. Mice have been widely employed in ge-netic studies due to their abundant human-ortholog genes, includingthose ones related to the vestibular system [1,2]; while mice and ratsare both prevalent in toxicological studies due to their similarities investibular anatomy, function, and impairment/recovery process com-pared with humans [3,4].

At present, behavioral test, electrophysiological examination, andmorphological evaluation are commonly used methods for vestibularassessment in rodents. Both behavioral test and electrophysiologicalexamination are noninvasive approaches and thus suitable for long-itudinal studies. However, behavioral test typically relies on the jud-gement from observers, leading to subjective results [5]. Electro-physiological examination can provide objective and quantitative data,but is usually disturbed by the acoustic artifact generated from the vi-bration exciter when producing acceleration pulses [6]. Moreover, therequired time-consuming animal-anaesthesia procedure precludeselectrophysiological-examination methods from high-throughputscreening. Morphological evaluation based on histology and micro-scopy is able to estimate vestibular impairment/recovery by char-acterizing morphological alterations in vestibular apparatuses at thesubcellular level, but not suitable for longitudinal studies owing to itsrequirements of animal sacrifice and sample preparation [3]. Therefore,noninvasive techniques and instruments for objective, quantitative, andhigh-efficiency assessment of vestibular function are demanded.

Vestibulo-ocular reflex (VOR), as the primary mechanism for sta-bilizing the visual field when the head moves, functions by producingresponsive eye movement in the opposite direction to the head move-ment and thus cancelling the retinal slip of visual images. Such com-pensatory eye movement is originally triggered by the vestibular ap-paratuses in the presence of sensible acceleration [7]. Therefore, a fullyfunctional vestibular system is required for the elicitation of normal-appearing VOR. The VOR gain, defined as the ratio of magnitude [8] orangular velocity [9] between the response and stimulus, gives an ob-jective and quantitative measure of the vestibular function. Damagedvestibular function is indicated by a reduced VOR-gain level.

To date, three eye-tracking techniques, electro-oculography (EOG),the search-coil method, and video-oculography (VOG), have been em-ployed for the estimation of VOR response. Both EOG and the search-coil method can accurately specify the VOR-response level by trackingthe electric-potential dynamics resulted from eye movement. However,the cross-talk interference between the electrodes in the two eyes, aswell as the time-consuming procedures of electrode placement and re-petitive calibration, makes EOG less preferable in diminutive animals[10]. The search-coil method is challenged by not only its invasive andlaborious coil implantation but also the under evaluation of the VORgain because of the mechanical impediments to the eye movement,which can be attributed to the contact nature of coil and ocular dis-comfort caused by post-surgery scarring and inflammation [8]. VOGmeasures the VOR response by tracking the dynamics of pupil positionfrom the video record of the eye movement. The noncontact naturemakes VOG highly preferable in preclinical studies [1,11,12]. Specificrequirements of VOG for preclinical VOR measurement include im-mobilization of the animal head, constriction of the animal’s pupils, andadequate contrast between the pupil and the surrounding iris. Theserequirements can be met at least in mice, on which the head fixing andmiosis [2] have been achieved in previously reported studies. The pu-pils of pigmented mice are also sufficiently darker than iris [8,9].

Several VOG-based vestibular function testing (VFT) systems havebeen developed [2,8,9]. A typical design contains a motor-drivenplatform (i.e., turntable) exerting acceleration stimuli, a restrainer forimmobilizing the animal, video-detection facilities for recording the eyemovements, and a dark environment for testing in order to prevent theinspected eyes from the disturbance induced by visible light.

Illumination for the video recording is provided with infrared (IR) orUV light. The animal-immobility setup involved in these reported pre-clinical VOG-based VFT systems, however, typically employs head boltsthat need to be surgically mounted onto the skull. Animal anaesthesiaand the following recovery period, is still necessary.

In the present study, we have developed a portable binocular VOG-based VFT system designed for measuring horizontal VOR (hVOR) inmice. A newly designed mouse holder has been used as the animal-immobility setup that avoids any surgical procedures and thus enablesnoninvasive immobilization of the mouse. Using such a system, VORmeasurement was performed on the mice suffering from vestibulardeficits caused by either genetic mutations or acute exposure to avestibulotoxic nitrile, 3,3′-iminodipropionitrile (IDPN). For the firsttime, the progression of toxicity-induced vestibular impairment hasbeen quantitatively characterized through short-interval (˜24 h) long-itudinal VOR measurement and, moreover, the severity of such ves-tibular impairment induced under different toxicant doses was quan-tified. Gross behavioral test and immunofluorescence microscopy wereboth performed to verify the VOR-measurement results from the IDPN-administered mice. The results from both the mice with genetic muta-tions and the mice receiving IDPN treatment suggest that our VFTsystem is able to objectively and quantitatively assess vestibular func-tion by performing surgery-free VOR measurement with improved ef-ficiency. These advantages give our system promising potential in high-throughput screen for genes and drugs affecting vestibular function.

2. Materials and methods

2.1. VOG-based VFT system

A binocular VOG-based VFT system was developed for both deli-vering stimuli to and detecting VOR response from the mouse. A hor-izontally placed octagonal motion platform (diameter 300mm) wasemployed to carry the illumination and video-detection facilities, aswell as the mouse holder (Fig. 1 A and D). The motion platform wasdriven by a stepper motor (Melike®, China), providing stimuli of sinu-soidal counter rotation following designed rotation modes. The entiresystem was encased within a cubic chamber (420mm × 460mm ×395mm), which guaranteed not only the portability of the system butalso the required dark environment. The mouse holder was placed to-wards the centre of the platform and coupled with a 3D translationstage. As the mouse head was carefully aligned with this platformcentre (labelled on the motion platform), undesired linear accelerationwould be minimized. Illumination for the video recording was achievedby two near-infrared (NIR) light-emitting diode (LED) lamps (940 nm,Chundaxin®, China). By the two hot mirrors placed in front of mouseeyes, the NIR images of the individual mouse eyes could be reflected totheir relevant cameras (MI®, China). These cameras were both mountedon translation stages at 45° angles to the anteroposterior axis of themouse. Such arrangement would maximize the possibility that the opticaxis of each camera points towards the centre of the relevant mouseeye.

2.2. Animal-immobility setup

A noninvasive animal-immobility setup was developed and em-ployed in the presented VFT system. As shown in Fig. 1 B and C, twoclamps with curved shapes covered the dorsal surface of a mouse bodyat the locations of neck and back, respectively. When screwing theclamps down toward the mouse-holder pedestal, the mouse lying inbetween was restrained. The extent to which screws were tightened wasempirically chosen for each mouse with concerns about both its bodysize and tolerance. The pedestal and clamps were 3D printed using UV-curable resin. Additional stabilization of the mouse head was achievedby a bite-block and another inverted-U-shaped nose clamp. The heightsof the nose clamp and the steel bar were both strategically selected so

X. Yang et al. Neuroscience Letters 696 (2019) 212–218

213

that the loaded mouse head could be positioned ˜30° nose down. As aresult, the horizontal semicircular canals were maintained parallel tothe horizontal plane and exclusively evoked [13].

2.3. Animal preparation

All procedures for animal care and protocols for animal experimentsdescribed below were approved by the Animal Care and EthicsCommittee of Southern University of Science and Technology. The in-volved mice were 6-to-13-week-old ones including both genders. Forthe VOR measurement in mice with genetic mutations, C57 mice witheither targeted knockout of Lhfpl5 (also known as Tmhs) or point mu-tation of Cdh23 and CBA mice with targeted knockout of Lhfpl5, as wellas their heterozygous and wild-type (WT) controls, were involved in thestudy. For the toxicological study, C57 mice treated with IDPN wereemployed for gross behavioral test, VOR measurement, and the sub-sequent immunofluorescence microscopy.

2.4. Grouping, drug administration, and experiment design

2.4.1. Mice with genetic mutationsBecause our employed Lhfpl5-mutant mice were generated from

either C57 or CBA strains, the VOR response from these mice wereseparately measured. For each of the strains, VOR measurement wasperformed on 4 homozygous (Lhfpl5−/−), 4 heterozygous (Lhfpl5+/−),and 4 WT mice. Mice with mutation of Cdh23 were all generated fromthe C57 strain. We measured the VOR response in 4 Cdh23−/−, 4Cdh23+/−, and 4 WT mice.

2.4.2. IDPN-administered miceIn total, twenty-four C57 mice were involved in the toxicological

study and evenly divided into three groups. Acute administration (i.p.injection) of IDPN (TCI®, Japan) was applied to the mice from two of thegroups, with the doses of 4mL/kg (≈ 4000mg/kg) and 2mL/kg, re-spectively. Saline was given (4mL/kg) to the mice in the control group.

VOR measurement was performed both before and after the drug ex-posure. Results acquired before the treatment served as the backgroundreferences. After the treatment, one day was given to the mice for re-covery, followed by daily VOR measurement performed for 7 days.

2.5. Sinusoidal counter rotation

Pilocarpine was applied for miosis before the measurement. Afterplaced into the chamber, the mouse was allowed to rest for 3–5min,and then acclimatized with the rotation stimulus for 1–3min in dark-ness to avoid nervousness or anxiety [9]. To inspect the VOR responseunder various stimulus conditions, four rotation modes (Table 1) withtheir frequencies ranging from 0.25 Hz to 1 Hz, and angular amplitudesof either± 20° or± 60° (corresponding peak velocity ranging from31.4°/s to 125.6°/s) were selected. To collect adequate video data, ro-tation stimuli at the frequencies of 0.25 Hz, 0.5 Hz, and 1 Hz werecontinuously applied for no less than 120 s, 100 s, and 80 s, respec-tively. The duration of rotation stimulus might be extended when ar-tifacts (e.g., eyeblink) appeared in a considerable amount of videoframes. The mouse was checked for any signs of distress between trials.

2.6. Recording and monitoring of VOR response

Binocular VOR response of the mouse was video recorded by thecameras mounted on the motion platform. The video resolution was1280 ╳ 720 with a frame rate of 60 fps. Two tablet computers (MEIZU®,China) placed outside the cubic chamber were connected to the twocameras, respectively, via wireless local network, allowing control andvisual monitoring of the video recording. In addition, the immobiliza-tion of the mouse head could be also monitored by observing the canthiof the eye. Stationary canthi indicated sufficient immobilization of themouse head during the VOR measurement (Appendix).

2.7. Analysis and calibration of the eye movement

The VOR magnitudes were firstly calculated from the recorded eyemovement using customized software. The region of interest (ROI),which contains the pupil, in each frame was automatically selectedfollowing a template-matching method. The matching template wasindividually generated for each video file by manually delineating thepupil boundary in the first frame. The accurate location of the pupil inother frames was further determined by finalizing the pupil boundaryfrom the ROI using the starburst algorithm [14]. Subsequently, ellipse

Fig. 1. The VOG-based VFT system. (A) Overview of the system (The cubic chamber is not shown). (B) Overview of the animal-immobility setup. (C) Sideview of theanimal-immobility setup. (D) Layout of the components on the motion platform. TS: translation stage; BB: bite-block; Cl: clamp; NCl: nose clamp; Ca: camera; HM:hot mirrors; NL: near-infrared light-emitting diode lamps.

Table 1The frequencies and angular amplitudes of the four selected rotation modes.

Mode 1 Mode 2 Mode 3 Mode 4

Frequency (Hz) 0.25 0.50 1.00 0.25Angular amplitude (°) ± 20 ±20 ±20 ±60

X. Yang et al. Neuroscience Letters 696 (2019) 212–218

214

Fig. 2. A typical view (blue solid line) of thedynamics of horizontal eye positions within10 s under the stimuli of sinusoidal counterrotation following the rotation modes of (A)0.25 Hz,± 20°; (B) 0.50 Hz,± 20°; (C)1.00 Hz,± 20°; and (D) 0.25 Hz,± 60° andtheir curve fitting results (gray dashed line).The R2 values of the curve fittings were cal-culated as 0.9697, 0.8762, 0.9235, and 0.9300,respectively, from A to D.

Fig. 3. VOR gains (means and standard de-viations) measured from the homozygous(Cdh23−/− from C57 strain, Lhfpl5−/− fromC57 strain, and Lhfpl5−/− from CBA strain;n=4), heterozygous (Cdh23+/- from C57strain, Lhfpl5+/- from C57 strain, and Lhfpl5+/-

from CBA strain; n=4), and WT mice fromstrains of C57 (as the Cdh23 control and theLhfpl5 control; n=4) and CBA (as the Lhfpl5control; n=4) under the stimuli of sinusoidalcounter rotation following the rotation modesof (A) 0.25 Hz,± 20°, (B) 0.50 Hz,± 20°, (C)1.00 Hz,± 20°, and (D) 0.25 Hz,± 60°. The pvalues (*** means p < 0.001; ** means p <0.01; * means p < 0.05) indicating sig-nificance were calculated using t-test.

Fig. 4. Longitudinal dynamics of VOR gains(means and standard deviations) in all controland IDPN-administered mice under the sti-mulus modes of (A) 0.25 Hz,± 20°, (B)0.50 Hz,± 20°, (C) 1.00 Hz,± 20°, and (D)0.25 Hz,± 60°. IDPN was administered on Day0 (arrow head). No data were acquired on Day0 or Day 1 (gray area). Pre-administration dataserving as the background reference are pre-sented at Day P. Normalization with respect tothe corresponding control data was performedand thus all normalized data from the controlmice became one (presented as the blue dashedlines).

X. Yang et al. Neuroscience Letters 696 (2019) 212–218

215

fit was applied to define the center of the pupil. The plotted pupil-centre position along frames reflected the eye-movement track. As wefocused on hVOR response in this study, the horizontal component wasextracted from the eye movement. Using the previously proposed eye-rotation model [8], calibration converting the acquired translationaldistance to the eye-rotation angle was performed.

In general, the angle dynamics of the pupil in response to the sti-mulus of sinusoidal counter rotation also exhibits a sinusoidal pattern atthe stimulus frequency (Fig. 2). Ripples at higher frequencies, in-dicating ocular saccades, appeared at times in the signal [2,8]. How-ever, the effect of high-frequency ripples in VOR magnitude could beexcluded because we applied fast Fourier transform to the signal andselected the angular amplitude only at the stimulus frequency as theobtained VOR magnitude. The VOR gain were then calculated as theamplitude ratio between response and stimulus.

2.8. Gross behavioral test

Gross behavioral test was performed to all of the involved mice forverifying the VOR measurement results. We chose the characteristicventral bending in the tail-hanging-reflex test [5] and two types ofspontaneous behavioral disorders, head shakes and abnormal gait, asthree indicative features demonstrating the presence of vestibulardysfunction. For the tail-hanging-reflex test, the mouse was lifted by thetail and then slowly lowered onto the ground. Mice with normal ves-tibular function were expected to show “landing response” by ex-tending their forelimbs toward the ground, whereas the affected micewould ventrally bend their bodies instead. For observing the sponta-neous behavioral disorders, the mouse was placed individually in aclean polycarbonate cage with fresh bedding and video recorded for5min.

2.9. Immunofluorescence analysis

Immediately after the longitudinal VOR measurement, the utriclesof mice involved in the toxicological study were dissected and thenfixed (4% paraformaldehyde) for immunofluorescence analysis. Wechose utricles instead of semicircular canals because of the practical

difficulty in semicircular-canal sample preparation. This re-presentativity is validated by the consistence of vestibular apparatusesin their responsive morphological alterations to IDPN [4]. Followingthe standard protocol [15], the fixed and rinsed specimens were in-cubated in 0.5% Triton X-100, 10% donkey serum, and 1% bovineserum albumin (BSA) in phosphate buffer saline (PBS) for 1 h at roomtemperature. Then the incubation solution was updated with 0.1%Triton X-100, 5% donkey serum, and 1% BSA in PBS. The anti-myosinVII a antibody was applied for labelling the hair cells. The specimenswere then processed overnight at 4 °C, followed by the rinsing with PBS.Subsequently, the specimens were incubated in 0.1% Triton X-100 and1% BSA in PBS with Alexa Fluor® 488 donkey anti-rabbit IgG and AlexaFluor® 555 phalloidin for 1 h at room temperature. After the final rin-sing, the specimens were mounted on the glass slides and examinedunder a Nikon® A1R confocal microscope. 3D scans were performed toobtain maximum-intensity-projection images with both 20× and 100×objectives.

2.10. Statistics

We employed t-test for statistical analysis. The detected differencewas defined to be significant when p < 0.05. According to the ac-quired p values, three levels of significance (p < 0.05, p < 0.01, andp < 0.001) were rated and indicated in the results.

3. Results

3.1. VOR response in mice with genetic mutations

The VOR results from the mice with genetic mutations are shown inFig. 3. VOR data are presented as gain values, which are defined as theratio of amplitude between response and stimulus [8]. All homozygousmice from the two types of genetic mutations for both strains exhibitedsignificant reduction in the VOR gain compared with the control andheterozygous ones; while no significant difference was detected be-tween the control and heterozygous mice. The detected VOR deficiencyin homozygous mice was verified by their behavioral-test results. Incontrast, no behavioral abnormalities were observed from the hetero-zygous or WT mice, supporting their normal-appearing response ac-quired in the VOR measurement. These results are consistent with thereported function and characteristics of these genes [16,17] and,therefore, demonstrate the capability of our VFT system in screening forthe genetic mutants that affect vestibular function.

3.2. Longitudinal effect of IDPN on vestibular function

The time-course characteristics of the VOR gain were captured fromboth the control and IDPN-treated mice at either a low (2mL/kg) orhigh (4mL/kg) dose (Fig. 4). In order to minimize the disturbance in-troduced by day-to-day variation of VOR, all data were normalized withrespect to their corresponding control. The background reference datawere calculated by averaging the VOR gains measured from the IDPN-administered mice on two randomly selected days within 7 days beforethe administration. Since we defined the IDPN-administration day asDay 0, the absence of data then appeared on Day 0, due to the ad-ministration, and Day 1, due to the post-administration recovery.

As shown in Fig. 4, decrease in VOR gain after IDPN administration,indicating toxicity-induced vestibular impairment, was clearly detectedfrom both low- and high-dose groups under all stimulus modes. Theexhibited difference in the extent of VOR-gain decrease between thesetwo affected groups indicates successful quantification of vestibular-impairment severity. Following the sharp decrease appearing since 2–4days after the administration, the VOR gains from both groups had atrend of reaching plateau levels. The sharp-decrease window appearedearlier for the high-dose group than low-dose group, suggesting ashorter impairment latency at higher treatment doses. These results are

Fig. 5. VOR gains (means and standard deviations) in the IDPN-administeredand control mice on the 8th day after administration under different stimulusmodes that were provided by the applied sinusoidal counter rotation with (A)the same angular amplitude but multiple frequencies and (B) the same fre-quency but different angular amplitudes. The p values (*** means p < 0.001;** means p < 0.01; * means p < 0.05) indicating significance were calcu-lated using t-test.

X. Yang et al. Neuroscience Letters 696 (2019) 212–218

216

consistent with the persistency of the vestibular damage caused by ni-triles [4,5]. The findings obtained from gross behavioral test exhibitedgood match with and thus verified these longitudinal results. Beha-vioral abnormalities occurred in all 8 mice in the high-dose group and 2(out of 8) mice from the low-dose group. The symptoms appeared oneor two days earlier in the mice from the high-dose group than in theones from the low-dose group, which further verified the observed la-tency difference.

3.3. IDPN-induced decrease in VOR response following different rotationmodes

Fig. 5 shows the VOR gains acquired from Day 8. The VOR gains,quantitatively measuring the severity of vestibular impairment, ex-hibited significant difference between every two groups under all sti-mulus modes, except for the one between control and low-dose groupsunder the rotation with±60° amplitude. By comparing the p values,the stimulus with 1.00-Hz frequency and±20° amplitude was de-monstrated to possess the best differentiation capability of various

vestibular-impairment severities in our toxicological study.

3.4. IDPN-induced pathological alteration on vestibular sensory epithelia

Images acquired from immunofluorescence microscopy illustrateIDPN-induced pathological features on the utricular epithelia obtainedfrom the IDPN-administered mice (Fig. 6). Normal-appearing hairbundles were shown in both the control specimens (Fig. 6 E and M) andthe specimens from mice treated with low-dose IDPN but exhibited noVOR-gain decrease or behavioral abnormalities (denoted as mild; Fig. 6F and N). However, the density of the hair cells seemed to be slightlyreduced in the low-dose specimens (Fig. 6 B and J) than in the controlones (Fig. 6A and I). Mice treated with low-dose IDPN and subsequentlyexhibited decreasing VOR gain and behavioral abnormalities (denotedas severe), as well as the mice receiving high-dose IDPN administration,showed marked (Fig. 6 G and O) to nearly complete (Fig. 6 H and P) lossof hair bundles in their utricles. Furthermore, globular-shaped hair cellswere observed in both of these two types of specimens (Fig. 6 C and K,D and L); while the hair-cell density was significantly lower in the

Fig. 6. Immunofluorescence microscopy of the utricular epithelia dissected from the IDPN-administered and control mice. (A˜H) were acquired under 20╳ objective,the scale bar indicates 100 μm; (I˜P) were the characteristic regions acquired under 100╳ objective, the scale bar indicates 20 μm. In (A˜D) and (I˜L), hair cells werestained with anti-myosin VII a antibody; in (E˜H) and (M˜P), hair bundles were stained with phalloidin. The arrow heads in (P) indicate the residual hair bundles.

X. Yang et al. Neuroscience Letters 696 (2019) 212–218

217

specimens with high-dose treatment (Fig. 6 D and L). These findingssuggest the dose-dependent severity of vestibular impairment in theIDPN-treated mice and, therefore, verify the VOR-measurement results.

4. Discussion

The genes involved in this study, Lhfpl5 and Cdh23, both contributeto the integrity of stereocilia bundles, via which the hair cells in ves-tibular apparatuses can conduct the perception of acceleration. Lhfpl5encodes for the tetraspan membrane protein that regulates the forma-tion of tip links in the stereocilia bundles; while Cdh23 encodes forCadherin 23, a structural protein component of the stereocilia tip links[16,17]. Mutations in these two genes, therefore, lead to failed as-sembly of functional stereocilia and thus mechanically insensitive haircells. IDPN has been reported to degenerate the hair cells when beingapplied following an acute administration, although the molecularmechanism is yet to be fully uncovered [3]. As a result, impaired ves-tibular function and thus reduced VOR response are expected from themice either with genetic mutants or IDPN administration. We selectedthe utricular epithelium as the representative vestibular sensory epi-thelium for the immunofluorescence analysis, although VOR evokes theepithelium in the horizontal semicircular canals (cristae) instead. Suchrepresentativity is validated by the consistent responsive morphologicalalterations from all types of vestibular epithelia under IDPN adminis-tration [3,4].

We noted that the VOR gain of the mouse might vary along withtime during the measurement. Increase of the VOR gain occurred tomost of the mice after the pre-measurement acclimatization that wasusually performed for 1˜3min. This could be explained by the fact thatthe mouse stopped struggling and was obedient to the immobilization.In this study, we applied pre-treatment acclimatization to all of themice involved in VOR measurement and began to acquire data after themice became calm and steady. The acquired VOR gains, in this scenario,would reach a plateau level after the acclimatization-induced en-hancement. Therefore, listing pre-measurement acclimatization as aroutine procedure onto the protocol of VOR measurement is suggested.Since our VOR measurement was conducted within a dark environment,adaptation of the eye movement induced by visual inputs [18] was notexpected to happen.

One challenge in extending the application of our developed VFTsystem to more detailed and sophisticated functional analysis of thevestibular system is its limited types of stimulus paradigms at the cur-rent stage. Linear translation and eccentric rotation, as two examples ofadditional stimulus paradigms augmenting the capability of the systemin evoking more types of VOR responses [2], may need to be enabled inthe future. Further update of the system by integrating modules testingother vestibular-function indicators (e.g., optokinetic response) shouldbe also considered. The phase delay between VOR response and sti-mulus is also worth detecting and analyzing in future research.

5. Conclusions

In this study, a binocular VOG-based VFT system with a newly de-signed surgery-free animal-immobility setup has been developed, en-abling VOR measurement with improved efficiency in mouse models.The application of this VFT system in a short-interval longitudinal studyof the VOR response under vestibulotoxic conditions, for the first time,revealed the quantitative dynamics of VOR gain from the mice treatedwith IDPN. The detected extents and onsets of the VOR-gain decreaseprovided objective quantification of not only the severity but also theprogression of the toxicity-induced vestibular impairment. Moreover,the exhibited higher efficiency in VOR measurement gives our VFTsystem promising potential in serving as a high-throughput screener forboth genes and drugs affecting vestibular function.

Acknowledgements

This work was supported by the National Natural ScienceFoundation of China (Grant Nos. 81470701 and 81771882) and theFundamental Research (Discipline Layout) Foundation from ShenzhenCommittee of Science, Technology and Innovation (Grant No.JCYJ20170817111912585) to F. Chen; the PhD Start-Up Fund ofNatural Science Foundation of Guangdong Province (Grant No.2018A030310130) and the SUSTech Presidential PostdoctoralFellowship to X. Yang; Hong Kong Research Grants Council (Grant Nos.GRF16111616 and GRF16102417) to P. Huang. We thank Dr. WeitaoJiang for his valuable suggestions referring to both data analysis andmanuscript writing.

Appendix A. Supplementary data

Supplementary material related to this article can be found, in theonline version, at doi:https://doi.org/10.1016/j.neulet.2018.12.036.

References

[1] P.P. Hubner, R. Lim, A.M. Brichta, A.A. Migliaccio, Glycine receptor deficiency andits effect on the horizontal vestibulo-ocular reflex: A study on the SPD1J mouse, J.Assoc. Res. Otolaryngol. 14 (2013) 249–259, https://doi.org/10.1007/s10162-012-0368-6.

[2] P.A. Armstrong, S.J. Wood, N. Shimizu, K. Kuster, A. Perachio, T. Makishima,Preserved otolith organ function in caspase-3-deficient mice with impaired hor-izontal semicircular canal function, Exp. Brain Res. 233 (2015) 1825–1835, https://doi.org/10.1007/s00221-015-4254-4.

[3] J. Llorens, D. Dememes, Hair cell degeneration resulting from 3,3’-iminodipropio-nitrile toxicity in the rat vestibular epithelia, Hearing Res. 76 (1994) 78–86,https://doi.org/10.1016/0378-5955(94)90090-6.

[4] C. Soler-Martin, N. Diez-Padrisa, P. Boadas-Vaello, J. Llorens, Behavioral dis-turbances and hair cell loss in the inner ear following nitrile exposure in mice,guinea pigs, and frogs, Toxicol. Sci. 96 (2007) 123–132, https://doi.org/10.1093/toxsci/kfl186.

[5] S. Al Deeb, K. Al Moutaery, H.A. Khan, M. Tariq, Exacerbation of iminodipropio-nitrile-induced behavioral toxicity, oxidative stress, and vestibular hair cell de-generation by gentamicin in rats, Neurotoxicol. Teratol. 22 (2000) 213–220,https://doi.org/10.1016/s0892-0362(99)00075-6.

[6] G.C. Gaines, T.A. Jones, Effects of acute administration of ketorolac on mammalianvestibular sensory evoked potentials, J. Am. Assoc. Lab. Anim. 52 (2013) 57–62PMID: 28620284.

[7] H.G. MacDougall, S.T. Moore, Functional assessment of head-eye coordinationduring vehicle operation, Optom. Vis. Sci. 82 (2005) 706–715, https://doi.org/10.1097/01.opx.0000175623.86611.03.

[8] J.S. Stahl, A.M. van Alphen, C.I. De Zeeuw, A comparison of video and magneticsearch coil recordings of mouse eye movements, J. Neurosci. Meth. 99 (2000)101–110, https://doi.org/10.1016/s0165-0270(00)00218-1.

[9] M. de Jeu, C.I. de Zeeuw, Video-oculography in mice, J. Vis. Exp. 2012 (2012)e3971, https://doi.org/10.3791/3971.

[10] W. Swart, C. Scheffer, K. Schreve, A video-oculography based telemedicine systemfor automated nystagmus identification, Am. J. Respir. Med. 7 (2013), https://doi.org/10.1115/1.4024647 art.031002.

[11] Q. Yang, P. Sun, S. Chen, H.Z. Li, F. Chen, Behavioral methods for the functionalassessment of hair cells in zebrafish, Front. Med-Prc. 11 (2017) 178–190, https://doi.org/10.1007/s11684-017-0507-x.

[12] P. Sun, Y. Zhang, F. Zhao, J.-P. Wu, S.H. Pun, C. Peng, M. Du, M.I. Vai, D. Liu,F. Chen, An assay for systematically quantifying the vestibulo-ocular reflex to assessvestibular function in zebrafish larvae, Front. Cell. Neurosci. 12 (2018), https://doi.org/10.3389/fncel.2018.00257 art.257.

[13] D.R. Calabrese, T.E. Hullar, Planar relationships of the semicircular canals in twostrains of mice, J. Assoc. Res. Otolaryngol. 7 (2006) 151–159 https://doi.org/.

[14] D. Zoccolan, B.J. Graham, D.D. Cox, A self-calibrating, camera-based eye tracker forthe recording of rodent eye movements, Front Neurosci.-Switz. 4 (2010) 1–12,https://doi.org/10.3389/fnins.2010.00193.

[15] F. Rua, M. Buffard, L. Sedo-Cabezon, G. Hernandez-Mir, A. de la Torre, S. Saldana-Ruiz, C. Chabbert, J.M. Bayona, A. Messeguer, J. Llorens, Vestibulotoxic propertiesof potential metabolites of allylnitrile, Toxicol. Sci. 135 (2013) 182–192, https://doi.org/10.1093/toxsci/kft127.

[16] F. Di Palma, R.H. Holme, E.C. Bryda, I.A. Belyantseva, R. Pellegrino, B. Kachar,K.P. Steel, K. Noben-Trauth, Mutations in Cdh23, encoding a new type of cadherin,cause stereocilia disorganization in waltzer, the mouse model for Usher syndrometype 1D, Nat. Genet. 27 (2001) 103–107, https://doi.org/10.1038/83660.

[17] W. Xiong, N. Grillet, H.M. Elledge, T.F.J. Wagner, B. Zhao, K.R. Johnson,P. Kazmierczak, U. Muller, TMHS is an integral component of the mechan-otransduction machinery of cochlear hair cells, Cell 151 (2012) 1283–1295,https://doi.org/10.1016/j.cell.2012.10.041.

[18] P.P. Hubner, S.I. Khan, A.A. Migliaccio, Velocity-selective adaptation of the hor-izontal and cross-axis vestibulo-ocular reflex in the mouse, Exp. Brain Res. 232(2014) 3035–3046, https://doi.org/10.1007/s00221-014-3988-8.

X. Yang et al. Neuroscience Letters 696 (2019) 212–218

218