Characterization of the Sheep Round Window Membrane

Abstract

Intratympanic injection is a clinically used approach to locally deliver therapeutic molecules to the inner ear. Drug diffusion, at least in part, is presumed to occur through the round window membrane (RWM), one of the two openings to the inner ear. Previous studies in human temporal bones have identified a three-layered structure of the RWM with a thickness of 70–100 μm. This is considerably thicker than the RWM in rodents, which are mostly used to model RWM permeability and assess drug uptake. The sheep has been suggested as a large animal model for inner ear research given the similarities in structure and frequency range for hearing. Here, we report the structure of the sheep RWM. The RWM is anchored within the round window niche (average vertical diameter of 2.1 ± 0.3 mm and horizontal diameter of 2.3 ± 0.4 mm) and has a curvature that leans towards the scala tympani. The centre of the RWM is the thinnest (55–71 μm), with increasing thickness towards the edges (< 171 μm), where the RWM forms tight attachments to the surrounding bony niche. The layered RWM structure, including an outer epithelial layer, middle connective tissue and inner epithelial layer, was identified with cellular features such as wavy fibre bundles, melanocytes and blood vessels. An attached “meshwork structure” which extends over the cochlear aqueduct was seen, as in humans. The striking anatomical similarities between sheep and human RWM suggest that sheep may be evaluated as a more appropriate system to predict RWM permeability and drug delivery in humans than rodent models.

INTRODUCTION

Hearing loss affects 466 million people globally; a number that is expected to grow to 900 million by 2050, with the vast majority due to cochlear disorders (World Health Organization 2017). One of the challenges in developing new treatments for hearing loss is the effective delivery of therapeutic compounds to the inner ear. The cochlea is protected by the blood-labyrinth barrier which effectively limits drug delivery to the inner ear after systemic administration (Hao and Li 2019). In addition, off-target side effects can be associated with systemic drug administration (Plontke and Salt 2018; Salt and Plontke 2009). The intratympanic route of drug administration therefore has attracted considerable interest for clinical practice (El Kechai et al. 2015; Liu et al. 2013). For such local administration, the drug formulation is passed directly into the middle ear space through an anaesthetised tympanic membrane. The middle ear cavity then serves as a “reservoir” for drugs to diffuse into the inner ear presumably via the following two openings: the round and the oval windows. The round window has a thin membrane of epithelial and connective tissues (the round window membrane, RWM). The RWM faces the middle ear externally and is bathed by the perilymph in scala tympani (ST) internally. The oval window connects with the perilymph of the scala vestibuli (SV) and is covered by the footplate of the stapes which rests on the endosteum of the inner ear, suggesting the oval window can also contribute to drug diffusion from the middle ear to the cochlea (Zdilla et al. 2018). Drug entry into the perilymph of the cochlea is presumed to occur predominantly via RWM into ST, with subsequent drug distribution to SV by local radial communication pathways between the ST and SV while distribution through the helicotrema at the cochlear apex seems limited (Salt et al. 2012; Salt and Plontke 2009). There is also a report on simultaneous drug entry via the bony otic capsule in guinea pigs (Mikulec et al. 2009). Although the oval window is also a likely point of entry for drugs in small animal models (King et al. 2017; Salt et al. 2016) and could be useful to gain access to the vestibular system, there is less evidence of the characteristics of drug penetration by this route. The amount of drug that can be delivered via intratympanic injection is limited by the absorption across the RWM and the molecular size of the compound (El Kechai et al. 2015; Liu et al. 2013). Ways to improve the permeability of the RWM, for example by using microbubbles (Shih et al. 2013) or pharmacological reagents (Creber et al. 2019), have been proposed, but further research is needed to understand the movement of therapeutic compounds across the RWM.

Part of the approach to developing effective drug transfer systems is to fully understand the cellular and anatomical properties of the RWM and its permeability under normal and diseased conditions. While data on the gross structure of the human RWM are available, most of the information on its permeability is derived from work on small rodents and guinea pigs, which do not approximate the size and thickness of the human RWM. For example, the average thickness of the human RWM is 70 μm (Goycoolea and Lundman 1997), whereas in small rodents, it is very thin (< 10 μm) (Nordang et al. 2003). As comparisons, the RWM in cats is 25–70 μm thick (Goycoolea et al. 1987) and in Rhesus monkey it is 40–60 μm on average (Goycoolea et al. 1988). In this regard, the sheep is a potential large animal model for translational studies of the human cochlea. Its longevity (10–12 years of life) allows modelling of age-related chronic conditions in humans, and the size similarities of sheep organs with those of humans provide a platform for testing novel medical devices easily adaptable for human use. Sheep hear over a similar sound frequency range (100 Hz–30 kHz) as humans (20 Hz–20 kHz) (Heffner and Heffner 1983), and there are similarities between the anatomy of the middle and inner ear structures (Gurr et al. 2011; Mantokoudis et al. 2016; Seibel et al. 2006). In addition, middle ear ossicular pressure and intracochlear sound pressure (Péus et al. 2017), histology of the cochlea (Soares and Lavinsky 2011) and cochlear geometry such as length, number of turns and the volume (Schnabl et al. 2012) are similar in sheep and humans.

The present study aims to comprehensively characterize the RWM of the sheep cochlea and compare it with published data on the human RWM. The fixed adult sheep cochlea was imaged using microCT to provide 3D maps of the RWM and its connection with other inner ear structures in combination with histological assessment.

METHOD AND MATERIALS

Sheep Temporal Bones

All sheep tissue samples used in this study were obtained from sheep cadavers from other research projects at the University of Auckland, or from the local abattoir. For sheep samples obtained from other research, fresh sheep temporal bones were obtained from pregnant 4–6-year-old female Romney/Suffolk sheep immediately after euthanasia using sodium pentobarbitone (Pentobarb 300 iv; Chemstock, Christchurch, New Zealand) as previously described (van den Heuij et al. 2016). Temporal bones obtained from the abattoir were kept on ice after animal death (2–6 h), followed by fixation in 4 % paraformaldehyde (PFA). Temporal bones obtained from another research group at the University of Auckland were processed immediately after euthanasia and placed in a fixative within 2 h of death. These tissue samples were used for microCT and some histological analyses. The study was approved by the University of Auckland Animal Ethics Committee under the Code of Ethical Conduct for Animals in Research and the New Zealand Animal Welfare Act.

Tissue Preparation and Examination of the Round Window Niche

A total of 19 fresh sheep temporal bones were processed in this study. After exposing the ventral surface of the skull, the tympanic bullae were cut open to expose the medial surface of the middle ear and to allow direct access to the round window niche. Temporal bones were fixed by immersion in 4 % PFA in 0.1 M phosphate buffer (PB, pH 7.4) for 48 h at room temperature. After washing with 0.1 M phosphate-buffered saline (PBS, pH 7.4), bone and soft tissues surrounding the inner ear were carefully removed with dissection tools and electric drills, excising the entire intact cochlea and part of the temporal bone. The cochlear surface was washed with PBS, and bright-field images of the round window niche were taken using Leica MZ12.5 stereomicroscope (Leica Biosystems, Germany). For visualization by Optical Coherence Tomography (OCT), the RWM was covered with a lubricating eye gel (GenTeal® Gel; Alcon, Switzerland) to fluid-couple the RWM with the probe of the OCT. Micron IV imaging system (Phoenix Research Labs; CA, USA) was used to obtain OCT images focusing on the RWM. Cochleae were either stored in PBS at 4 °C or immediately processed for decalcification.

Histological Analysis

Fixed cochleae were decalcified for 6–8 weeks in 8 % w/v ethylenediaminetetraacetic acid (EDTA) in 0.1 M PB (pH 7.4) at room temperature. Once the bone surrounding the cochlea was soft enough for microdissection, decalcified cochleae were dehydrated by sequentially increasing concentrations of ethanol (70–100 %, 10 min each) and with 100 % xylene, and then embedded in paraffin wax following standard protocols (Canene-Adams 2013). Tissue sections (10–20 μm) were cut using a microtome (Leica Jung RM2035, Germany) in either horizontal or vertical orientation, dried for one day at 60 °C, deparaffinized and rehydrated with xylene and sequentially decreasing concentration of ethanol and water (2 × 5 min each) before staining and mounting on glass slides for histological examination. The following staining protocols were used: Haematoxylin & Eosin (H & E) to stain cellular structure (Haematoxylin; 0.5 % Haematoxylin, 0.1 % sodium iodide, 1 % methanol and 2 % acetic acid in 30 % glycerol; and eosin; 1 % eosin Y, 1 % aqueous phloxine and 0.4 % glacial acetic acid in 95 % ethanol) (Ellis 2011); Toluidine Blue (TB, 0.1 % TB in 1 % NaCl, pH 2.0–2.5) (Toluidine Blue Staining Protocol for Mast Cells 2011); Verhoeff’s Van Gieson (VVG) for collagen and elastin fibres (Percival and Radi 2016) and Alcian Blue (AB, 1 % w/v in 3 % acetic acid, pH 2.5 followed by 0.1 % Nuclear Fast Red with 5 % aluminium sulphate). Bright field and fluorescence images taken by Leica CTR MIC upright microscope (Leica) were processed by ImageJ (Rasband, W.S., ImageJ, U.S. National Institutes of Health, Bethesda, Maryland, USA; https://imagej.nih.gov/ij/, 1997–2018) and Adobe Photoshop CC 2018 (Adobe Inc., USA).

MicroCT Image Acquisition

A total of five adult sheep temporal bones, fixed and fully decalcified or fixed and trimmed without decalcification, were used for X-ray microtomography (microCT) analysis. Temporal bone tissues surrounding the round window niche and cochlear capsule were carefully trimmed using number 15 surgical blades and fine forceps under the microscope. Tissue samples were then stained with 1 % w/v osmium tetroxide (ProSciTech Pty Ltd., Kirwan, Australia) in 0.1 PB for 2 days at room temperature. MicroCT data were acquired using a Bruker Skyscan 1272 (Bruker, Kontich, Belgium) at the Auckland Bioengineering Institute MicroCT facility. Scanning parameters of the Bruker SkyScan 1272 were optimized for each individual sample to achieve the best signal-to-noise ratio with sufficient contrast and pixel resolution in addition to a reasonable scan time, resulting in variability in scanning parameters (Table 1, samples 1 and 2 for whole cochlear microCT and samples 3–5 for RWM + round window niche samples). Images were reconstructed using InstaRecon (InstaRecon, Illinois, USA) and subsequently visualized in 3D using CTVox and analysed using CTAnal (Bruker, Kontich, Belgium).

Table 1.

MicroCT scanning parameters

Sheep sample	Filter metal and thickness (mm)	Current (μA)	Voltage (kV)	Exposure time (ms)	Resolution step (°)	Total rotation (°)	Pixel resolution (μm)
1	0.5 Al	100	100	950	0.3	360	7.4
2	0.5 Al	105	94	1275	0.28	360	8.0
3	0.5 Al	100	100	5000	0.5	360	3.5
4	0.5 Al	150	65	5900	0.15	180	1.5
5	0.25 Al	130	70	2700	0.15	180	1.0

Image Analyses and Segmentation

Image reconstruction parameters were optimized to have minimal ring artefact and best contrast around the soft tissue including the RWM by NRecon (MicroPhotonics Inc., USA). Reconstructed dataset was pseudo-coloured for visualization of the RWM where appropriate. DataViewer (Bruker, Belgium) was used to orientate the RWM and select a volume of interest. A new dataset of the reconstructed and re-sliced image stacks in the sagittal plane was loaded into SkyScan CTAnal (Bruker, Belgium), and selected regions of interest (ROI) were analysed. The image showing the beginning of the RWM from the lateral margin was set as the top image and the bottom was set at the point where the RWM was no longer visible. Each ROI was manually traced around the RWM, the surrounding bone (round window niche) and the apical, basal, medial and lateral margins. This was performed on every fifth image out of approximately 2000 images per dataset. ROI datasets were then loaded into the CTAnal, and all images were manipulated by a range of custom processing tasks, to remove the background noise. Each processed dataset of images was then reconstructed into three-dimensional models using CTVox (Bruker, Belgium).

Thickness Measurements

A custom “RWM thickness measurement” protocol was designed to allow for measurement of membrane thickness at each point (pixel) along the length of the RWM from the microCT dataset. This protocol comprises a custom-designed workflow (=macro) in ImageJ. Briefly, sagittal plane slice images of the RWM from microCT were converted to binary images and any branches on the image were pruned to obtain a single elongated object. The principle of this algorithm is based on the use of a Euclidean distance map which measures the distance of a pixel from the medial line of the segmented RWM along its whole length and converts the pixel distance into an intensity value. In parallel, the binary image of the RWM was skeletonized in order to obtain the medial line (equidistance from all pixel points) of the RWM. The Euclidean distance map was superimposed to the skeletonized image creating a Euclidean thickness values along the central line to the edge. Because the Euclidean thickness output here represents the distance from the edge of the object to the central midline (i.e. half the thickness of the object), this value was then multiplied by two to calculate the thickness of the whole RWM in conversion; 1 pixel = 1 μm. This custom ImageJ algorithm allowed us to quantify RWM thickness along all pixel points, but also calculate the length of the RWM and create a histogram of all the thickness (intensity) values, including the minimum, maximum and frequency of a specific thickness. The macro for this algorithm can be found in Fig. 9. Descriptive statistics of RWM measurements from image analyses were performed using Microsoft Excel.

Fig. 9.

Custom ImageJ script for RMW thickness quantification. (a) A custom-made “RWM thickness measurement” algorithm for accurate measurement of thickness at every point (pixel) along the length of the RWM. Created with Dr. Anekal, a senior technologist at the Biomedical Imaging Research Unit at the University of Auckland. (b–e) Example of consecutive workflow using the “RWM thickness measurement” algorithm to measure RWM thickness. (f) Histogram of measurement of thickness at every point (pixel) along the length of the RWM along with the maximum thickness measured

Guinea Pig RWM Histology

Although not a major part of this study, some general features of the RWM of the guinea pig cochlea were also examined for comparison. An adult guinea pig auditory bulla was obtained from another study fully approved by the University of Auckland Animal Ethics Committee. Male adult guinea pigs (6+ weeks old) were anaesthetised using an intraperitoneal injection of ketamine (75 mg/kg) and xylazine (10 mg/kg) and euthanised using an intraperitoneal injection of pentobarbitone (90 mg/kg). Extracted bullae were immersion fixed in Karnovsky’s fixative (3 % paraformaldehyde (PFA) and 0.5 % glutaraldehyde in 0.1 M phosphate buffer (PB)) for seven days at 4 °C and subsequently processed for histological analysis.

RESULTS

Gross Anatomy of the Sheep Round Window Niche

The round window sits within a recess in the petrous portion of the temporal bone, called the round window niche (Fig. 1). For the purpose of the RWM characterization, we defined two axes of the round window relative to the body axes as inferior-superior and anterior-posterior (Fig. 1a). Relative to the surrounding structures, the superior border is closest to the oval window, the anterior border of the round window is the bony overhang of the cochlear capsule, while the posterior aspect is part of the floor of the tympanum and is proximal with intricate vestibular system (Fig. 1a, insets a′ and a″). The sheep round window niche is generally circular in shape, with some variation in size. The semi-translucent RWM is visible within the round window niche. Interestingly, 2 out of 19 sheep exhibited an irregular round window niche, with either full obstruction of the round window niche or an additional round window (Fig. 1d, e). Sheep round window niche diameters measured across two orientations were comparable to published values for the human round window (Schnabl et al. 2012); the sheep round window had an average diameter from the inferior to superior orientation of 2.14 ± 0.3 mm and anterior to posterior orientation diameter of 2.3 ± 0.4 mm (n = 18) (Table 2). This is comparable to variations in size and shape of the round window that has been described in humans at 1.81 and 2.05 mm², respectively (Schnabl et al. 2012).

Fig. 1.

Shape and size of the sheep round window niche. (a) Schematic drawing of the sheep skull and ear canal showing RWM orientation relative to body axes. (a′) Schematic view of the sheep middle and inner ear when viewed through the ear canal. (a″) Schematic of the sheep middle and inner ear when viewed from the inferior/ventral aspect. (b, c) Representative images of the sheep round window niche. (d, f) Examples of abnormally shaped round window niche. Bright field photographs were taken from fixed sheep cochleae showing variable shape of the RWM; (d) with no identifiable round window niche; (e) additional opening to the cochlea (asterisk). (f) Small round window niche. Panels d–f have identical pixel scales. A, anterior; P, posterior; I, inferior; P, posterior definition of the RWM used in the subsequent figures of this article. Scale bars, 1 mm, identical for b′ and c

Table 2.

Comparison of sheep and human round window membrane niche

Diameter	Sheep (n = 19) (mm)	Humana (mm)
Inferior-superior diameter	2.14 ± 0.29	1.81 ± 0.08
Anterior-posterior diameter	2.30 ± 0.37	2.05 ± 0.09

^aHuman data according to Schnabl et al. (2012)

The Structure of the Sheep Round Window Membrane

The sheep cochlea was studied histologically and with microCT. For histology, 10 adult decalcified sheep cochleae were sectioned in two different orientations to compare two cross-sectional views. We first identified the RWM with histological examination of tissue sections taken across the two axes (anterior–posterior axis (A–P) (Fig. 2a–c) and the inferior–superior (I–S) axis (Fig. 2d–f)). We examined the margins where the RWM attaches to the surrounding bone (Fig. 2b, c, e and f) and demonstrated at least four distinct shapes of RWM attachment at the anterior (Fig. 2b), posterior (Fig. 2c), inferior (Fig. 2e) and superior (Fig. 2f) margins. The anterior top margin of the RWM fits into the groove formed in the bone while the posterior attachment forms over the triangular extrusion of the bone. The superior and inferior margins of the RWM both had a fan-like shape, but the inferior margin appears to be broader. These fan-shaped margins continued around the RWM. Intense H&E staining was observed at the boundary of the RWM and the bone, and at the inner ear side of the RWM.

Fig. 2.

Cross-sectional view of the sheep round window niche. Decalcified adult sheep RWMs were sectioned along the anterior-posterior (a–c) or inferior-superior (d–f) orientations, stained with H&E and visualized using bright field microscope. Black arrowheads indicate RWM. SM, scala media; ST, scala tympani; SV, scala vestibuli; SC, semicircular canal; ME, middle ear; CA, cochlear aqueduct. Scale bars; 1 mm for a, d; 200 μm for b, c, e and f

Previous studies have reported that the human RWM consists of two layers of epithelial cells and an intervening connective tissue layer (Goycoolea and Lundman 1997). The three-layered arrangement of the RWM was conserved in sheep, including an outer epithelium, a fibrous tissue layer and an inner epithelium (Fig. 3a, b). Thus, there was a uniform monolayer of epithelial cells on both the outer and inner surfaces of the RWM with a core of connective tissue in between (Fig. 3b). The flat outer epithelial cells were continuous with the mucosal lining on the middle ear side of the round window niche (Fig. 3c, d, arrowheads). Likewise, the inner epithelium of the RWM constituted a monolayer of cells which was continuous with the mesothelial lining of the inner ear (Fig. 3c, d, arrowheads). The three-layered structure of the sheep RWM more closely resembles those reported for human (Goycoolea and Lundman 1997). This is distinctively different from the RWM of smaller animals like guinea pig, which are thin (approx. 15 μm) and lack clearly defined three-layered structure (Fig. 4). Pigmented cells with a dark-brown appearance were present in the fibrous connective tissue layer (Fig. 3e), but were more frequently observed near the posterior attachment of the RWM. The fibrous tissue layers displayed bundles of collagen and elastic fibres stained with Van Gieson (VG) and Verhoeff’s Van Gieson (VVG) (Fig. 3f, g). The region immediately adjacent to outer and inner epithelial cells stained more intensely with VG and VVG, indicating elastin content close to the epithelial layers and an abundance of collagen spanning the middle connective tissue layer (Fig. 3f, g). Small vasculature, morphologically resembling capillaries were identified in the connective tissue immediately adjacent to the outer epithelial layer of the RWM, but not the inner epithelium (Fig. 3h). There is no apparent vasculature or pigmented melanocytes in the guinea pig RWM (Fig. 4).

Fig. 3.

Histological examination of the outer and inner epithelium and fibrous layer of the sheep RWM. Adult sheep RWM was decalcified, sectioned and stained with Alcian blue (a, c, d), H&E (b, e, h), Van Gieson (f) or Verhoeff Van Gieson (g) stains. Examination at high magnification reveals the following three distinct RWM layers: outer epithelium (OE), inner epithelium (IE) and fibrous layer (FL) (b). Outer and inner epithelium were in continuum with the middle ear mucosa and inner ear mesothelium, respectively, at both top and bottom attachements (c and d, black arrowheads). (e) Around the lower attachment, a number of pigmented cells (melanocytes) were observed (red arrowheads). (f and g) More intense VG and VVG staining was observed in the inner and outer epithelium. (h) Blood vessels were observed only in OE region on the middle ear side of the RWM (asterisks)

Fig. 4.

Comparison of guinea pig RWM and sheep RWM. (a–d) Adult guinea pig RWM sectioned in equivalent orientation as A-P cross-section of sheep RWM in this study (e), stained with toluidine blue. The RWM attachment margins show similar fan-like shape on anterior (b) and V-shaped attachment in posterior (c) margins. (d) High-magnification image of guinea pig RWM does not show evident three-layered structure compared to sheep RWM visualized using the same magnification (e). n = 3 guinea pig cochleae were examined for consistency

Location and Curvature of Sheep RWM Relative to Surrounding Structures

For microCT, five whole adult sheep cochleae were decalcified and stained with contrast-enhancing agent of 1 % osmium tetroxide for visualization, two were whole intact cochlea while three were micro-dissected to only have the RWM and surrounding niche. Whole cochlear microCT 3D reconstructed images were then pseudo-coloured based on the signal intensity so that bony tissue with high X-ray absorption is shown in orange, while less dense tissue is shown in green (Figs. 5 and 6). The 3D reconstructed model was then resliced to produce cross-sectional views in any plane and could be aligned to histological sections in similar orientations. When a cross-section is taken through the RWM from anterior to posterior axis (Fig. 5a, b), the RWM opens into the ST and the anterior edge aligns with the spiral ligament (SL) such that these appear to be in close proximity to the anterior region of the RWM (Fig. 5c). When a cross-section is taken through the inferior–superior axis (Fig. 5d), the superior aspect of the RWM closer to the oval window sits adjacent to the spiral ligament (Fig. 5e). Therefore, the anterior-superior quarter of the RWM is physically very close to the spiral ligament and basilar membrane of the basal turn. In the opposing side, the posterior-inferior quadrant of the sheep RWM opens directly to the scala tympani. Adjacent to this in the posterior-inferior region, the cochlear aqueduct (CA) is located near the RWM and is connected to the RWM by a fine connective tissue network, described as a “meshwork structure” (Fig. 6d). At the superior side towards the oval window, the RWM was situated above the semi-circular canal (Fig. 6i). At the mid-point along the inferior-superior axes, the RWM was situated approximately 1.3 mm inward from the bone of the round window niche (Fig. 6f). In the human, a long thin bone with a finger-like projection conformation named the fustis is found at the centre of the round window niche. The fustis has been used as a landmark to indicate the entrance of the round window niche as it is a constant structure (Luers et al. 2018; Marchioni et al. 2015; Marchioni et al. 2016). While we did not observe a bony structure equivalent to the fustis, a prominent small bone triangular in shape was identified on the posterior margin of the RWM (Fig. 6h, arrowhead).

Fig. 5.

MicroCT imaging of the sheep cochlea. Decalcified sheep cochlea was stained with OsO₄ and visualized using microCT to construct 3D model (a, d). (b, c) Cross-sectional view of reconstructed sheep cochlea taken through anterior-posterior axis as indicated in (a), pseudo-coloured to show tissue density: orange—high density, blue—low density. (e) Cross-sectional view of reconstructed sheep cochlea taken at inferior-superior axis as shown in (d). SM, scala media; ST, scala tympani; SV, scala vestibuli; OW, oval window; RW, round window. Scale bars; 1 mm

Fig. 6.

Anatomical features around RWM at different locations within the cochlea. (a) Decalcified sheep cochlea was stained with OsO₄ and visualized using microCT. (b and c) Decalcified adult sheep RWMs were microtome sectioned along anterior-posterior axis (b) or inferior-superior axis (c), stained with H&E and visualized using bright field microscope. (d–m) Cross-sectional view of reconstructed sheep cochlea taken through anterior-posterior axes (d–i) or inferior-superior axes (j–m), pseudo-coloured to show density: orange—high density, blue—low density. Red arrowheads indicate RWM. SM, scala media; ST, scala tympani; SV, scala vestibuli; SC, semicircular canal; ME, middle ear; CA, cochlear aqueduct. Arrowhead; a small triangular bone resembling the fustis in humans, albeit significantly smaller than expected from human

Next, we examined the shape and thickness of the RWM at high resolution (Fig. 7). These were then resliced along the inferior-superior axis (Fig. 7c–g) and the anterior-posterior axis (Fig. 7h–j, k and l) at different locations of the RWM. The cross-sectional examination of the RWM revealed three key features. First, the thickness of the RWM was not uniform. Instead, the membrane was much thicker at the edges (Fig. 7c, g, h, l). Second, the RWM had a characteristic kink in the middle (Fig. 7, arrowheads). Finally, as previously seen in the histological images (Fig. 2), there were clear differences in the shape of the areas where the RWM attaches to the bone. To visualize and map the location and variation in RWM curvature and thickness, the RWMs were segmented and reconstructed in 3D (Fig. 8). From the segmented images of the RWM, the thickness of the membrane was quantified at 10 different locations along inferior-superior axis by a custom designed algorithm using ImageJ macro which measures shortest distance across each point of the RWM (Fig. 9). Quantification was applied to three separate 3D reconstructions of the RWM (n = 3 sheep) and the mean ± SD summarized (Table 3). The location within the RWM is expressed as the relative distance in both the inferior-superior and anterior-posterior axes. The ten different locations in each axis were labelled from the inferior or anterior edge (0.1) to the superior or posterior edge (1.0) of the RWM (Table 3). Similar to human, the sheep RWM has a predominantly convex curvature towards the middle ear cavity (Fig. 8a–d, red line on far right). The average thickness of the anterior and posterior peripheral regions of the RWM was 137 and 126 μm, respectively, and this was similar to the average thickness of the inferior and superior peripheral regions (134 and 123 μm, respectively). The thinnest point of the membrane (56–74 μm was at approximately 90 % distance from the inferior boundary and 70 % from the anterior boundary (Fig. 8e and f). The thickness of the central portion of the RWM ranged between 55 and 71 μm and was skewed towards the superior half of the RWM (Table 3).

Fig. 7.

MicroCT analysis of the sheep round window membrane. (a) A representative photograph of the sheep cochlea showing the RWM and the orientation of images (blue and green lines) presented in c–l. (b) Pseudo-coloured 3D reconstruction of RWM using microCT. (c–g) Cross-sections through the inferior-superior axis of the RWM at different locations indicated in panel a (0 = anterior side of the RWM, 1.0 = posterior side of the RWM). (h–j, k and l) Cross-section through the anterior-posterior axis of the RWM at different inferior to superior locations indicated in panel a (0 = inferior end of the RWM, 1.0 = posterior end of the RWM). ST, scala tympani; ME, middle ear. Arrowhead; the RWM curvature towards the middle ear. Scale bars; 500 μm

Fig. 8.

Analysis of the sheep RWM curvature, attachment and thickness. (a–d) MicroCT reconstruction of the RWM visualized from the middle and inner ear sides. Red lines indicate curvature on the RWM. (e and f) MicroCT reconstruction of the RWM based on zones of different thickness (light blue thinner, dark blue thicker as annotated, μm) viewed from the middle ear (e) and inner ear (f) sides

Table 3.

Quantification of RWM thickness across different points. The RWM was segmented and thickness was measured using ImageJ at 10 different locations across anterior-posterior and inferior-superior axes. Values represent the mean ± SD (n = 3)

Meshwork Structure Adjacent to the RWM

Interestingly, we also observed a meshwork of soft tissue at the inner ear side of the RWM adjacent to the cochlear aqueduct (Fig. 10). This soft tissue (Fig. 10a′, asterisk) may represent a fibrous meshwork previously reported in humans (Gussen 1978; Toriya et al. 1991). This is not a fixation artefact, as the meshwork was also observed in non-fixed cochlea using optical coherence tomography (OCT) (Fig. 10b, asterisk). This meshwork structure was stained with H&E and Van Gieson and is composed of very elongated cells with long lateral extensions that connect to each other, resembling a “net” (Fig. 10c, d). The structure is in direct contact with the inner epithelial layer of the RWM and extends within the ST and covers the opening of the cochlear aqueduct (Fig. 10c, d).

Fig. 10.

The meshwork structure. (a) Fixed and decalcified adult sheep cochlea dissected to visualize the round window niche from the inner ear side. The niche was covered with soft mesh-like structure labelled by asterisk (a′). (b) Optical coherence tomography (OCT) of the fresh sheep cochlea also revealed meshwork-like structure on the inner (scala tympani) side of the RWM (asterisk). (c, d) Histological assessment of meshwork cells with H&E stain and their attachment to the RWM in horizontal (c) and vertical cross-section taken close to the centre of the RWM (d). The meshwork structure is labelled with an asterisk (d′). ME, middle ear; ST, scala tympani; CA; cochlear aqueduct (below the plane). Scale bars, 500 μm (a, b), 100 μm (c), 1 mm (c, d), 50 μm (d′)

Occasionally, we have observed an extra layer of mucosal membrane attached to the sheep RWM, which was assumed to represent a false RWM (data not shown). A false RWM has been described as a mucosal membrane veil that is stretched across the opening of the round window niche obstructing the true RWM. It was documented in both human and animal cochleae with reported frequency of 33 % of temporal bones in humans (Alzamil and Linthicum Jr 2000). In our study, these false RWM were not tightly adhered to the bony niche unlike the true RWM which forms tight adherence to the surrounding bony round window niche, and the false RWMs were lost during tissue processing.

DISCUSSION

The current study sets out to characterize the structural and cellular composition of the sheep RWM using complementary imaging techniques, microCT and histology. Using 3D reconstructions, the curvature and thickness of the RWM were mapped relative to the points of attachment to the bone and surrounding inner ear structures as landmarks. These morphological characteristics and relationship to the surrounding structures are summarized in a contour map of the membrane (Fig. 11). The diagram shows the superior half of the RWM in close proximity to the spiral ligament, while the inferior half of the RWM opens directly to the basal turn of the scala tympani and is in close proximity to the cochlear aqueduct (Fig. 11). The human RWM is slightly convex towards the middle ear (Watanabe et al. 2014), with an average thickness of 70 μm (Luers et al. 2018; Rask-Andersen et al. 2012; Zhang and Gan 2013). The RWM thickness is not uniform in humans being thicker around the periphery and thinner towards the centre of the membrane (Bellucci et al. 1972; Luers et al. 2018; Zhang and Gan 2013). Our 3D analyses showed that the sheep RWM has a similar average thickness and variation in thickness throughout the membrane. The centre of the sheep RWM was the thinnest (55–70 μm), while the periphery of the RWM was much thicker (86–172 μm). Additional detail we uncovered was that the thickness of the sheep RWM is asymmetric along the inferior-superior axis and is thinner in the superior portion, which is closest to the spiral ligament and organ of Corti.

Fig. 11.

Contour map of differential thickness of the RWM relative to surrounding landmarks. Heatmap showing differential thickness across the RWM was generated based on our microCT analysis (Table 2) and super-imposed on the round window niche (a). (b, c) Cross-sectional schematic diagrams along the anterior-posterior axis (b) and inferior-superior axis (c) showing the RWM relative to surrounding landmarks. (d) Detail of the thickness heatmap. CA, cochlear aqueduct; SL, spiral ligament; SM, scala media; ST, scala tympani; SV, scala vestibuli. The superior half of the RWM in close proximity to the SM and SL was relatively thinner. The inferior half of the membrane is thicker, opens to ST and is close to CA

At the cellular level, three layers of the RWM identified in humans (Goycoolea and Lundman 1997) and cats (Goycoolea et al. 1987; Nordang et al. 2003) were also identified in the sheep RWM. In human, the outer epithelial layer of the RWM consists of a single layer of cells with sparsely distributed microvilli at the peripheral and central portions of the RWM (Carpenter et al. 1989; Schachern et al. 1984). The middle fibrous connective tissue layer is the thickest layer of the three and is thought to play an important role in providing structural support to the membrane and contains fibroblasts, collagen and elastic fibres as well as vasculature and lymph vessels (Carpenter et al. 1989; Richardson et al. 1971). A study in human temporal bones (Gussen 1978) reported a varying number of melanocytes in the peripheral region of the RWM; however, their role and physiological properties remain unknown. The inner epithelium comprises flattened squamous epithelial cells with long lateral extensions forming a thin single epithelial layer facing the perilymph of the ST. These layers and the structural features including a small number of melanocytes and vasculature, particularly adjacent to the outer epithelium, were similarly identified in the sheep RWM in this study. While some studies claim three-layered structure in rodent and guinea pig RWM, our studies show that the layered structure in guinea pig is not as evident compared to sheep. Better understanding of junctional complexes and the possible presence of active transporters in the RWM cells will also be important as they may influence drug entry and need to be further studied but were beyond the scope of this structural study. The presence of a fibrous network covering the inner ear side of the RWM and the cochlear aqueduct is another common feature of the human and sheep RWM. The functional role of this meshwork is not clear. It is possible that this meshwork serves to anchor RWM to the walls of ST, providing physical support to the RWM as well as regulating the shape and degree of RWM displacement. The proximity to the cochlear aqueduct may suggest that the meshwork regulates aqueduct flow resistance and thus the volume of inner ear perilymph (Feijen et al. 2004). This cellular meshwork and its function warrant further investigation.

Development of novel pharmacological therapies for hearing loss is challenged by gaining consistent access to the inner ear target tissues and the paucity of effective methods for local drug delivery to the inner ear. The intratympanic approach of drug delivery presents an opportunity to preclude systemic side effects of drugs using a minimally invasive outpatient procedure (El Kechai et al. 2015). The main challenge for intratympanic drug delivery is differential diffusion of therapeutically active molecules across the RWM and throughout the cochlear fluids and tissues. It is generally accepted that there is individual variability in the size and shape of RWM in humans which may affect drug diffusion (Luers et al. 2018; Marchioni et al. 2015; Marchioni et al. 2016; Toth et al. 2006). Some have suggested that perforation of RWM as means to bypass the RWM (Aksit et al. 2018), while others have suggested use of adjuvant reagents (Creber et al. 2019) or ultrasound-aided microbubble (Shih et al. 2013) to increase diffusions of molecules across the RWM. Previous studies in humans and animal models suggest that both acute and chronic changes occur to the RWM in disease; these include central thickening of the RWM within the middle connective tissue layer; infiltration of inflammatory cells (leukocytes and macrophages); and formation of gaps between epithelial cells, which may imply the loss of tight junction integrity (Ikeda et al. 1990; Jiang et al. 2016; Sahni et al. 1987; Yoon and Hellstrom 2002). However, further studies are required to understand how RWM permeability may be altered under these pathological conditions or by ageing. Our results provide comprehensive information on the structure and cellular composition of the sheep RWM and demonstrate considerable similarity of the sheep to the human RWM suggesting it may be an important animal model of the human RWM for drug delivery investigations. However, there are other factors that are also important when choosing an animal model for translational research in this field, such as drug entry through the oval window, drug clearance from the middle and inner ear, drug binding and metabolism in the inner ear (Glueckert et al. 2018; Salt and Plontke 2018). In addition, the difference in size of the RWM and round window niche in animal models and humans may also influence drug formulation properties such as viscosity and hence drug delivery. The sheep has recently attracted considerable interest as an appropriate large animal model for the development of cochlear implant devices (Péus et al. 2017; Schnabl et al. 2012). A study by Kaufmann et al. (2020) reported the first in vivo cochlear implant on adult sheep and examined cochlear function using electrocochleography (ECoG) and auditory brainstem responses (ABR) following implant surgery (Kaufmann et al. 2020). Given the lifespan of the sheep and structural similarities, the sheep may be a useful model for investigating RWM permeability in both healthy and diseased states. Better understanding of the physical and structural properties of the RWM and the establishment of appropriate animal models may lead to development of a novel technological platform for safe and effective delivery of therapeutically active compounds through the RWM.

Abbreviations

μCT

micro-computed tomography

SV

scala vestibuli

ST

scala tympani

SL

spiral ligament

SM

scala media

RWM

round window membrane

RWN

round window niche

BLB

blood-labyrinth barrier

Compliance With Ethical Standards

Conflict of Interest

The authors declare that they have no conflict of interest.