The timing of auditory sensory deficits in Norrie disease has implications for therapeutic intervention

Norrie disease is caused by mutation of the NDP gene, presenting as congenital blindness followed by later onset of hearing loss. Protecting patients from hearing loss is critical for maintaining their quality of life. This study aimed to understand the onset of pathology in cochlear structure and function. By investigating patients and juvenile Ndp-mutant mice, we elucidated the sequence of onset of physiological changes (in auditory brainstem responses, distortion product otoacoustic emissions, endocochlear potential, blood-labyrinth barrier integrity) and determined the cellular, histological, and ultrastructural events leading to hearing loss. We found that cochlear vascular pathology occurs earlier than previously reported and precedes sensorineural hearing loss. The work defines a disease mechanism whereby early malformation of the cochlear microvasculature precedes loss of vessel integrity and decline of endocochlear potential, leading to hearing loss and hair cell death while sparing spiral ganglion cells. This provides essential information on events defining the optimal therapeutic window and indicates that early intervention is needed. In an era of advancing gene therapy and small-molecule technologies, this study establishes Ndp-mutant mice as a platform to test such interventions and has important implications for understanding the progression of hearing loss in Norrie disease.

C Arrow indicates loose fibrils in the space close to capillary D A cell adjacent to a capillary appears to be shrinking.
E Abnormal marginal cell. The cell has lost intense electron dense staining, and its baso-lateral infoldings, appearing to have become more rounded. The nucleus is more rounded in shape than its usual elongated morphology. Whole mount lateral wall preparations were imaged under consistent conditions and multiple fields of view from the apical and middle regions were analysed.
A Shows the analysis method: the extravascular region was delineated using the Endomucin channel; this outline was overlayed on the FITC channel and mean fluorescence intensity was measured in the extravascular region; N = 3 WT and 3 Ndp-KO mice, over 7 fields of view from each sample.

Animals
Norrie disease mice (Ndp tm1Wbrg ) generated in the Berger laboratory carried a loss of function Ndp mutation, in which exon 2 of the mouse Ndp gene is replaced by a neomycin resistance cassette to disrupt the function of the gene [1]. The Ndp tm1Wbrg mutation was maintained on a predominantly C57BL/6 genetic background following backcrossing of the 129 founder to the C57BL/6 inbred strain for multiple generations. Genotyping was carried out using forward primer GTATTGCATCCATATTTCTTGG and reverse primer CTCTCCATCCCCTGACAAGGA in a PCR protocol of 94°C for 5 minutes followed by 40 cycles of 94°C for 60 seconds, 62°C for 60 seconds and 72°C for 90 seconds. The reaction finished with a final step of 72°C for 5 minutes. This yielded a 528bp product from the wild type allele and/or a ~1628 bp product from the knockout allele.

Auditory Brainstem Response (ABR) recordings
Physiology tests were performed under a single session of anaesthesia, using 0.1 ml / 10 g bodyweight intra-peritoneal urethane (20% w/v in water). Brainstem auditory evoked potentials were measured using a previously described method (Ingham et al., 2019). In brief, anesthetised mice were placed on a heated blanket (Harvard Apparatus) inside a sound attenuating chamber (IAC Limited, Winchester, UK) and positioned to face a loudspeaker (FF1 magnetic speaker, Tucker Davis Technologies TDT, Alachua, FL, USA) at a distance of 10 cm. Subdermal needle electrodes [SD51-426-1 NeuroDart, Spes Medica, Italy] were inserted in the skin overlying the right and left bullae (ground and reference electrodes, respectively) and on the midline vertex (active electrode) and were connected to the appropriate input of a TDT RA4LI headstage (with RA4PA preamplifier). Under control of custom software, stimuli were generated, amplified and presented to the FF1 speaker via a TDT RZ6 multifunction processor. The same software and processor also gathered digitised evoked potential recordings from the electrodes via the TDT RA4LI/RA4PA. Acoustic stimuli (broad-band click transients of 10 μs duration, and tone pips of 6, 12, 18, 24, 30, and 42 kHz of 5 ms duration with a 1 ms onset and offset ramp) were presented to the mouse at levels ranging from 0-95 dB SPL (in 5 dB steps) at a rate of 42.3 stimuli per second. Evoked responses, bandpass filtered at 300 -3000 Hz, 20 ms in duration, were recorded following the onset of each individual stimulus presentation and were averaged in a digital signal buffer on the TDT RZ6 before being retrieved by the software for online display and saved for offline analyses.
ABR findings in mice aged 1 month were also confirmed using an independent ABR recording apparatus. ABR from anesthetised mice were obtained using subdermal needle electrodes (Rochester Medical), one inserted at the vertex, and one each behind the ipsilateral and contralateral pinnae. Electrode signals were low-pass filtered (7.5 kHz cut-off frequency) and recorded at 24 kHz sampling rate (TDT RA4LI, RA4PA and RX5). For analysis, ABR data were filtered using a bandpass filter (100-3000 Hz). Stimuli were tone pips (10 ms total duration with 1.5 ms rise/fall time; frequencies 5.7, 8, 11.3, 16, 22.6, 32 and 45.3 kHz) or clicks (50 μs duration), with intensities 0-80 dB SPL in 5 dB steps, delivered at a rate of 25/s. ABR thresholds were determined visually by estimating the lowest sound level at which at least two deflections in the ABR waveform were greater than the background variability in the waveforms.
Measurements of wave amplitudes were performed using custom Matlab software. The user selected a time window containing the wave of interest, and the software then detected maxima and minima of the ABR traces within that window. ABR wave I amplitudes were measured from the peak to the following trough.

Distortion Product Otoacoustic Emissions (DPOAE) recordings
After completion of the ABR measurements mice were laid on their right side and their left pinna was removed. A hollow perspex conical speculum was used as a couple between the ear canal and the DPOAE measurement probe; an Etymotic ER10B+ system (Etymotic Research Inc, Elk Grove Village, IL, USA) coupled to TDT MF1 loudspeakers via 5 cm tubes. Stimuli were generated and DPOAE responses recorded using the TDT RZ6 multifunction processor, under the control of TDT BioSigRZ software. Stimulus tones, f1 and f2 were generated on independent output channels. Frequencies for f2 were set to match some of the ABR tone-pip frequencies used (6, 12, 18, 24 and 30kHz). Frequencies for f1 tones were set such that f2 = 1.2 x f1. Sound pressure levels of the f2 stimulus ranged from -10 dB to 65 dB in 5 dB steps and were -10 dB relative to the SPL of the f1 component. ER10B+ microphone signals during stimulus presentation were recorded via the RZ6 processor. Online Fast Fourier Transformation of the microphone signal revealed a power spectrum that contains the f1 and f2 stimulus components, the main 2f1-f2 DPOAE component of interest and the measurement noise-floor. These spectra were exported for analysis.

Endocochlear Potential (EP) recordings
Following completion of DPOAE recording, a tracheal cannula was inserted, and the mouse placed in a custom-built head-holder. An Ag-AgCl pellet was inserted into the musculature of the neck to serve as a reference / ground electrode for the measurement. Muscle and tissue were surgically retracted using forceps to expose the wall of the auditory bulla. After opening the bulla, a small hole was made in the bone of the lateral wall of the basal turn of the cochlea and the tip of a 150 mM KCl-filled glass micropipette electrode was positioned to be in fluid contact with the spiral ligament tissue within the hole. A custom-built electrometer was zeroed, and the microelectrode was slowly advanced through the spiral ligament under manual control. The electrometer potential was monitored until a stable positive potential, the endocochlear potential (EP) was reached as the electrode entered the scala media [5][6][7].

Fixation and decalcification of cochleae
Auditory bullae were isolated, and the cochleae exposed and fixed by perfusion of 4% paraformaldehyde (PFA) injected directly via the round and oval windows. Fixation was continued by immersion of the cochlea in PFA for 90-120 mins followed by decalcification in 4% EDTA in PBS (w/v), pH 7.4, for 48 h.

3D imaging of cochlear vasculature
Fixed, decalcified cochleae were dehydrated in sequential stages of 25%, 50%, 75% and 100% methanol for 1 hour each. They were then incubated in 1.5% hydrogen peroxide (Sigma H1009) solution made up in methanol overnight at 4°C on a roller. The following day they were rehydrated in sequential solutions each lasting 1 hour: 75%, 50% and 25% methanol, PBS 0.2% Triton X-100.
Cochleae were permeabilised (0.2% Triton X-100, 20% DMSO in PBS) at 4°C for three days. The cochleae were then blocked (0.2% Triton X-100, 10% DMSO, 10% FBS, and 3% BSA in PBS) for 6 hours at room temperature on a roller. They were then incubated with a rat anti-endomucin antibody (Santa Cruz, sc-53941; at a dilution of 1:50 made up in a buffer of PBS, 0.2% Tween 20, 5% DMSO, 10% FBS, 3% BSA) for 4 days at 4°C on a roller. This was followed by six washes in PBS with 0.2% Tween 20, each lasting an hour at room temperature on a roller. The cochleae were then incubated with secondary antibody (goat anti-rat IgG Alexa Fluor 647; Life Technologies, A221247; with 0.2% Tween 20, 5% DMSO, 10% FBS, 3% BSA in PBS) for 3 days at 4°C on a roller. This was followed by six washes in PBS 0.2% with Tween 20, each lasting an hour at room temperature on a roller before a final step in the same solution overnight at 4°C on a roller. They were then dehydrated in sequential stages of 25%, 50%, 75% and 100% methanol for 1 hour each.

Morphometric analysis of stria vascularis
The stria vascularis was isolated from the apical region of fixed and decalcified cochleae. Samples were permeabilised and blocked (0.3% BSA, 0.5% Triton X-100 in PBS) for 30 minutes at room temperature. They were then incubated with GS-IB4 Alexa Fluor 594 at a dilution of 1:50 (ThermoFisher, I21413; 0.3% BSA, 0.1% Triton X-100, 1mM CaCl2 in PBS) for three days at 4°C. This was followed by six washes of 15 minutes each before mounting in Fluoroshield (Sigma, F6182), and samples were imaged using a spinning disk confocal microscope (Yokogawa, CSU22).
Image analysis was carried out using FIJI/ImageJ [8]. Vascular morphology was determined using GS-IB4 fluorescence. Vessels were delineated using the drawing tool and shape descriptors were recorded. Circularity (4π*area/perimeter^2) and solidity (area/convex) were calculated. Vessel diameter was determined at their middle section (i.e. approximately equal distance between two branching points) using the line tool.

Immunostaining of whole mount lateral wall and organ of Corti
The otic capsule was carefully removed from fixed, decalcified cochleae, and the lateral wall was separated from the organ of Corti with microscissors. Samples were blocked for 1 hour at room temperature (1% BSA, 0.5% Triton X-100, 5% FBS in PBS), stained with primary antibodies or phalloidin-A647 at appropriate dilutions (1% BSA, 0.5% Triton X-100, 5% FBS in PBS) followed by Image analysis for pericyte coverage was carried out using Image-J, based on a modification of a previously published protocol [9]. Z-stack images were acquired from the apical and middle regions of cochlear lateral wall wholemounts labelled for endomucin (endothelial cells) and desmin (pericytes) using a 40x objective and z-projections prepared. Single channel images were thresholded to delineate pericytes and the outlines of vessels were delineated using Image-J drawing tools. Pericyte coverage was calculated as the percentage of desmin-labelled area to the endomucin-labelled area within the total vessel outlines. Vessel coverage was also analysed after classification of vessels as either those with low endomucin and transversely wrapped pericytes (low Emcn), or those with high endomucin and longitudinally arranged pericytes (high Emcn).

Electron microscopy
The cochlear tissues were fixed by direct perfusion with 2.5% glutaraldehyde in 0.1 M cacodylate buffer, pH 7.35 with 3 mM CaCl2 via the round and oval windows with a hole punctured in the apex to aid immersion. The entire opened bulla was then immersed in that fixative and maintained for 24 h at 4°C. After decalcification with 4% EDTA (w/v) prepared in cacodylate buffer for 72 h, the bulla was trimmed and the entire, intact cochlea was post-fixed in 1% OsO4 in cacodylate buffer. The samples were partially dehydrated in an ethanol series to 70% ethanol, incubated overnight in uranyl acetate in 70% ethanol before completion of dehydration, and then embedding in plastic.
Sections of the entire cochlea were cut parallel to the modiolus and partial serial sections collected onto silicon wafers. The sections were stained with uranyl acetate and lead citrate. The silicon wafers were mounted on scanning electron microscope (SEM) support stubs using conductive silver paint and the samples were then coated with carbon by evaporation to provide a conducting layer ca. 7.5 nm thick. The samples were examined and imaged in a JEOL 6700F SEM using backscatter detection. This procedure enables examination in a single section of an entire cochlea -basal to apical turns and of all tissues -without the fields of view interrupted by the grid bars of the section supports used in transmission electron microscopy. The images collected have been reverse contrasted to provide "photographic positive" contrast for display.
The cross-sectional areas of all strial capillaries in a single section of the entire cochlea of every animal were measured by tracing around the circumference of the lumen using Image-J. The thickness of the stria, from the luminal surface to the interface between the basal aspect of the basal cell and the spiral ligament, at its mid-point between the Reissner's membrane and the spiral prominence was measured for all cochlear turns in a single section of the entire cochlea for every animal using Image-J.

Vascular permeability assay
Fluorescein isothiocyanate-conjugated bovine serum albumin (FITC-BSA; A9771, Sigma-Aldrich) was dissolved in PBS at 5% concentration (w/v). 50 µl or 100 µl of solution was injected into the tail vein of 1-or 2-month-old mice, respectively. After 5 hours, mice were sacrificed, cochleae isolated and the vasculature counter-stained with an anti-endomucin antibody for analysis, as described above.
Image analysis for extravascular FITC-BSA signal intensity was carried out using a custom Image-J macro. Stitched tile scan images or the stria vascularis were obtained from the apical and middle regions of cochlear lateral walls from FITC-BSA injected mice as described above. Fields of view of equivalent areas were analysed from WT and Ndp-KO mice. The extravascular regions were delineated using a combination of thresholding and manual tracing of vessels. This mask was then superimposed on the FITC channel of the image and the mean grey value of each region was measured and an area weighted mean was calculated.

qRT PCR analysis
Auditory bullae were isolated, the cochleae exposed and separated from the vestibule, immediately snap frozen on dry ice and stored at -80°C .Total RNA was extracted using a modified protocol combining TRIZOL and column based extraction [10]. cDNA was synthesized using the RevertAid  Image analysis was carried out using FIJI/ImageJ (Schindelin et al., 2012). Quantitative analysis of neurons was carried out on Tubb3 or Nf200 immunoreactivity images as described in the results.

Immunostaining of spiral ganglion neurons
The multipoint tool was used to count neuronal cell bodies and the drawing tool was used to trace the perimeter of cell bodies to measure neuronal size.

Statistical analysis
The number of mice (N) used for each experiment is stated in the legends. Error bars always represent standard deviation (SD). Data were analysed using statistical tests as appropriate for each data set. The type of statistical tests, significance levels (p-values) post hoc analysis are presented in the respective figure legends using either a two-tailed Student's t-test or one-or twoway ANOVA with respective post hoc tests, or as indicated for auditory function tests using Mann-Whitney Rank Sum Test. P value less than 0.05 was considered significant. Data, graphs and figures were organized, analysed, and assembled using GraphPad Prism 6, SigmaPlot, R (ggplot2), Adobe Illustrator and Inkscape (https://inkscape.org).

Study Approval
Informed consent was obtained from parents or guardians of patients prior to participation. The study was approved by the National Research Ethics Committee London Queens Square Hearing test results were examined from 6 patients with a diagnosis of Norrie disease and hearing loss. In two children, information about the newborn hearing screen was available, and they passed this test in both ears. In five out of the six patients, the first available audiogram was recorded at the age of 2-6 years and in one patient at the age of 35 years. The period over which audiograms were available in individual patients, varied from 1-30 years. The British Society of