Morphological variability in the inner ear of mice with osteogenesis imperfecta

Osteogenesis imperfecta (OI) is known to cause hearing loss in ~60% of the affected human population. While OI‐related pathologies have been studied in the middle ear, the development of cochlear pathologies is less well understood. In this study, we examine OI‐related pathologies of the cochlea in a mouse model of OI to (1) document variation between OI and unaffected mice, and (2) assess the intrusion of the otic capsule onto the cochlea by analyzing differences in duct volumes. Juvenile and adult OIM C57BL/6mice were compared to unaffected wildtype (WT) mice using three‐dimensional models of the cochlea generated from high resolution micro‐CT scans. Two‐tailed Mann–Whitney U tests were then used to investigate duct volume differences both within and between the OI and WT samples. Areas of higher ossification were observed at the cochlear base in the OI sample. OI mice also had significant intraindividual differences in duct volume between right and left ears (4%–15%), an effect not observed in WT mice. WT and OI duct volumes showed a large degree of overlap, although the OIM volumes were more variable. Our findings indicate that OIM mice are likely to exhibit more asymmetry and variation in cochlear volume despite minor differences in sample cochlear volumes, possibly due to bony capsule intrusion. This suggests a potential mechanism of hearing loss, and a high potential for cochlear and otic capsule alteration in OIM mice.


| INTRODUCTION
Osteogenesis imperfecta (OI) is a congenital disease affecting the formation, integrity, and amount of type I collagen in the body (Swinnen et al., 2012).Lack of structural integrity and flexibility normally afforded by collagen leads to increased long bone fractures and muscle weakness.This collagen deficiency may also lead to craniofacial defects, such as dental malocclusions, midfacial hypoplasia, and cranial base abnormalities (Gentry et al., 2010).These craniofacial phenotypes are also observable in the osteogenesis imperfecta murine (OIM) model of OI (Berman et al., 2020;Menegaz et al., 2020).Hearing loss is a commonly reported symptom among patients with this disease (Carre et al., 2019) with 62% of OI patients diagnosed with mild to profound hearing loss.In patients with type I OI, the most common form, a 95% rate of hearing loss has been reported after the age of 30 (Pillion et al., 2011).Identified forms include conductive hearing loss (17.4% of OI patients), involving loss of function within the ossicular chain, and sensorineural hearing loss (25.8%), resulting from damage to the cochlea and other parts of the neural pathway, with the most predominant form being mixed hearing loss (56.8%), involving damage to both the cochlea (the sensory hearing organ) and ossicles (the conductive hearing bones) (Hartikka et al., 2004).Within individuals, there has been some suggestion that the type of hearing loss (conductive, sensorineural, or mixed) can be asymmetrical, varying between right and left ears within the same patient (Hartikka et al., 2004).
Previous studies that endeavor to analyze hearing loss with OI are not discriminatory by phenotype of OI.Generally, as type I OI is the most common, and mildest form of the disease, most adults with OI that have hearing loss are type I.In studies including type III OI individuals, the population often consists of mixed OI phenotypes.Therefore, there is limited information on the rate of hearing loss in type III OI individuals alone.
Most research on OI-related hearing pathologies has focused on the middle ear and has identified alterations in bone makeup and structure of the ossicular chain (Santos et al., 2012).The stapes is the key ossicle affected by OI in humans and shows increased porosity of the crura (bend) and increased otosclerotic fixation of the stapedial footplate in those affected with OI (Waissbluth et al., 2020).Less is known about the pathological state of the cochlea, although increased bone density and porosity have been observed in the surrounding otic capsule of human patients (Swinnen et al., 2012) and mice with OI (de Paolis et al., 2021).Several hypotheses have been proposed regarding the cause of loss of cochlear function in OI, which include intrusion of petrous bone on the cochlea and subsequent soft tissue damage (i.e., hemorrhage of blood into the labyrinth, atrophy of the stria vascularis, the blood supply of the cochlea, or atrophy of the cochlear hair cells) (De Paolis et al., 2021;Shapiro et al., 1982).However, research on this topic is limited.While bone density likely plays a role in OI's effects on the otic capsule, associated changes to the cochlear morphology are still unknown.
This study quantifies OI-related changes in the cochlea in osteogenesis imperfecta murine (OIM) mice.
Our aims are to (1) characterize morphological differences in the inner ear for juvenile and adult OIM mice compared to unaffected mice to determine the anatomy of the disease state, and (2) measure intraindividual variation between cochlea duct volumes in OIM mice and unaffected mice to determine potential asymmetry in the etiology of hearing loss.We hypothesize that cochleae in OIM mice will have more morphological variation than their unaffected counterparts, likely due to abnormal growth of the bony capsule.

| Mouse colony and developmental time points
The osteogenesis imperfecta murine (OIM) is a mouse strain with a nonlethal recessive inherited mutation of the COL1A2 gene.Homozygous OIM À/À mice (B6C3FE a/a-Col1a2 OIM/OIM ) are a model for the human presentation of OI type III.The wild type (WT) controls in this experiment are OIM +/+ mice (B6C3FE a/a-Col1a2 J/J).WT and OIM littermates were weaned and genotyped during week 4 (post-weaning juveniles) and raised until early adulthood (week 16) (Menegaz et al., 2020) (Table 1).In the week 4 age group, there was a total of 14 OIM and 17 WT.In the week 16 age group, there was a total of 11 OIM and 12 WT, due to attrition of the specimens.Roughly equal males and females were used across both age groups, and in WT and OI groups.

| Animal husbandry
All procedures and animal care were approved by the Indiana University School of Medicine Institutional Animal Care and Use Committee and Use Committee (IACUC) approved protocol.WT and homozygous OIM mice were bred in-house at the Indiana University School of Medicine and maintained on a C57BL/6J background (Berman et al., 2020).Mice were group housed in a facility with 12-h light/dark cycles and access to food and water ad libitum.Environmental enrichment objects were of a size and shape (e.g., large spheroids like toy balls) chosen to limit opportunities for non-feeding-related biting and chewing behaviors.

| Digital data collection
Study animals were imaged in vivo using micro-computed tomography (microCT) during both weeks 4 and 16 in the Indiana University Center for Musculoskeletal Health.Scans were made using a Skyscan 1176 microCT machine (Bruker Corp, MA, USA) operated at 41-65 kV and 385-500 μA.Scans were reconstructed using Skyscan NRecon software (Bruker Corp) into 16-bit TIF stacks using 0.008 or 0.017 mm 3 voxels, depending on cranial length.(see SI Table 1 for individual scanning parameters).
During imaging, animals were anesthetized via inhalation anesthesia using an isoflurane nonrebreathing system with a 3%-5% flow rate and maintained at 1.5% for the duration of the scan.Animals were secured to the scanning bed using standard non-porous medical tape to minimize movement during imaging.During scanning, animals were monitored using a camera inside of the microCT machine.Vitals (heart rate, respiration rate) were also monitored from the control computer and isoflurane flow rate adjusted as needed.Following the scan, animals were allowed to fully recover on room air in an empty cage.Once sternal recumbency was achieved, the animal was returned to its home cage.

| Segmentation and cropping protocol
To assess changes to the bony anatomy, the inner ear was digitally segmented from CT scans to create 3D endocasts of the bony cochlea.All segmentation was done using the imaging software 3D Slicer and ImageJ (Fedorov et al., 2012;Schindelin et al., 2012) (Figure 1).Within the Segment Editor module, the "Paint" tool was used to create a 3D model of the hollow space within the bony cochlea by selecting areas of low Hounsfield units (HU) values.The "Scissor" and "Erase" tools were used to remove the modiolus (the bony core of the cochlea) so that a volumetric measure of only the cochlea was captured.All cochlear segments were cropped horizontally at the inferior most point of the oval window to separate the cochlea from the semicircular canals.

| Data collection and analyses
Morphological differences were assessed both qualitatively and quantitatively.Observations were made of gross morphological changes and any potential bony intrusions within the cochlea, seen as high HU-value areas within the cochlea.Cochlear volumes were taken from the endocasts of the bony cochlea, and do not necessarily represent the soft tissue components of the cochlear duct (i.e., scala media).Endocasts were quantified using the "Segment Statistics" module in 3D Slicer to assess differences in cochlear morphology between the genotypes.
To examine the effects of body size on morphology and cochlear volume, each specimen was scaled to its centroid size derived from 72 cranial landmarks (Steele, 2020).Centroid size is a proxy for skull size, defined as the square root of the sum of squared distances of all the landmarks of an object, in this case the skull from their center point (Klingenberg, 2016).Because this data are corrected by centroid size, the scaled values are unitless, and not representative of raw cochlear size in mm 3 .Mann-Whitney U tests were used to compare cochlear volumes between OIM and WT groups at each timepoint.To compare intraindividual differences between left and right cochleae, an intraindividual symmetry variable was calculated by creating a ratio of bilateral duct volumes.

¼ intraindividual variation
Mann-Whitney U tests were used to compare intraindividual symmetry variables (ISV) between genotypes at each time point.An ANOVA was also performed on the week 16 specimens to address a wider range observed in the cochlear volume of the OI population.

| Gross cochlear morphological changes
Cochlear morphology: In OIM mice, areas of higher ossification were observed throughout the cochlea, which we describe henceforth as cochlear intrusions (Figure 2).
Laterality of intrusions: Cochlear intrusions in OIM mice occurred mostly unilaterally, with 15/25 (60%) OIM mice exhibiting unilateral incidence and 7/25 (28%) OIM mice exhibiting bilateral incidence.Of the 15 unilateral intrusions in OIM mice, six occurred on the right side and nine occurred on the left side.Comparatively, 7/29 (24%) WT mice had unilateral intrusions and 1/29 (3%) WT mice had bilateral intrusions.Of the 7 unilateral intrusions in WT mice, three occurred on the right side and four occurred on the left side.

| WT and OIM comparisons
No statistical difference was observed in scaled cochlear volumes between OIM and WT mice at either week 4 ( p = 0.703; Figure 3a) or week 16 ( p = 0.704; Figure 3b).The volumetric ranges of the WT and OIM samples overlap significantly.A wider range of volumes was observed within the OIM volumes in week 16 (Table 2).However, an ANOVA indicated that the variance in cochlear volume was not significantly different between the genotypes at week 16 ( p = 0.542).
F I G U R E 2 Examples of cochlear endocasts derived from microCT in OIM (top) and WT (bottom) mice.Areas of higher ossification (yellow arrow) were observed near the base of the cochlea more often within the OIM cochlea compared to the WT cochlea.OIM, osteogenesis imperfecta murine model; WT, wildtype model.

| Intra-individual comparisons
Mann-Whitney U tests of the intraindividual symmetry variable (ISV) between WT and OIM showed significantly more asymmetry in OIM in both age groups, indicating that within an OIM individual, one cochlea was often larger than the other (p < 0.05, Figure 4).The range in ISVs among the OIM specimens evidences an erratic and potentially pathological pattern of bone growth.This is seen in Figure 4, as the ISV range is larger in the OIM group than in their WT counterparts.Across OIM samples, the pattern of laterality did not exhibit a clear sidedness (Figure 5).

| DISCUSSION
Osteogenesis imperfecta (OI) is a disease that significantly impacts hearing.Most research on the effects of OI in the ear has focused on the ossicular chain in the middle ear.However, the effects of OI on the cochlea have only been vaguely described and not quantified despite a clinical indication that cochlear damage affects 82.6% individuals with OI-related hearing loss (sensorineural or mixed) (Hartikka et al., 2004).As such, this study's aim was to document OI-related pathology on cochlear shape to better understand the factors that might contribute to hearing loss.
The project was divided into two aims.The first aim was to establish the morphological differences in the inner ear for adult OIM mice compared to WT mice in order to determine the morphology associated with the disorder.The second aim was to examine intraindividual variation between cochlea of OIM and WT mice to determine potential asymmetry in the inner ear.While there are minimal differences between cochlear volumes between OIM and WT mice, we observed a significant degree of intraindvidual asymmetry in the cochlea of OIM mice, emphasizing OI's asymmetrical spread.

| Gross cochlear morphological changes
A qualitative analysis revealed that macroscopic cochlear intrusions were present within the majority of the OIM specimens (58%) and tended to only appear in one ear.The appearance of these intrusions suggests that the otic capsule of the surrounding petrous bone grows into the space comprising the hollow bony cochlea, and may be responsible for decreasing the total volume of this region.The errant bony growth leading to intrusions could potentially affect the function of the membranous cochlea within the bony otic capsule and result in sensorineural hearing loss.It also likely affects how sound waves move through the scala vestibuli and scala tympani, impacting how sound waves are received by the vestibulocochlear nerve.

| WT and OIM comparison
Overall cochlear volumes were not significantly different between OIM and WT mice.However, while there was observable overlap in cochlear volumes, OIM cochlear volumes occupied a wider range than WT cochlear volumes.This suggests that the pathological OI state introduces more variation to cochlear morphology.This finding supports evidence from new research suggesting that the OIM phenotype is associated with increased morphological variance in skull morphology (Steele, 2020).

| Intraindividual differences
Intraindividual variation in cochlear volumes provided the strongest evidence for OI-related changes to the inner ear.At both the juvenile and adult stages, OIM mice had a wider range of intraindividual symmetry variables (ISVs) than their unaffected counterparts.This significant difference in volume between right and left cochleae can potentially be attributed to OI's etiology, considering that errant and asymmetrical bone growth is typical of the disease (Gentry et al., 2010).These results indicate that OIM mice are much more likely to have unilateral cochlear volume changes, despite minor differences in overall volume, possibly due to bony capsule intrusion.In postmortem studies, humans with asymmetrical OI-related hearing loss due to damage to the cochlea exhibited increased porosity and fragility of the bony capsule (Santos et al., 2012).This poor bony quality could be a contributor to asymmetrical improper growth, leading to damage to the soft tissue components of the cochlea.Additionally, as demonstrated by Hartikka et al. (2004), individuals with OI may experience different kinds of hearing loss between ears.The asymmetry we observed in these cochlear volumes of OIM mice may be a contributing factor to this asymmetrical hearing loss.

| Comments on biomechanics and sensorineural hearing
The results of this project suggest that the deleterious effects of OI on the cochlea are due to an intrusion of the otic capsule into the hollow bony coil of the cochlea.These intrusions have the potential to disrupt the movement of soundwaves through the perilymph, which transmit soundwaves throughout the cochlea which elicit neural responses (Swinnen et al., 2012).Displacement of this fluid is responsible for bending cilia on hair cells that line the membranous cochlea, which send nervous impulses to the brain to be interpreted as sound.
Uninhibited movement of fluid within the membranous cochlea is essential for the accurate transmission of soundwaves.Therefore, interruptions to this movement, such as errant bony intrusions as seen in OI, will negatively affect the transmission of sound within the cochlea (Gentry et al., 2010).
Another possible problem is that the intrusions could cause damage to soft tissue components of the cochlea.This damage could distress and rupture the membranous labyrinth, cause a hemorrhage of cochlear blood supply, a loss of either perilymph or endolymph, or cause damage to hair cells lining the membranous cochlea.Loss of blood supply could lead to atrophy of the membranous cochlea, and damage to hair cells could lead to diminished neural impulses related to hearing.Any damage to either the bony or membranous cochlea could alter the interpretation and/or transfer of sound waves to the brain, resulting in loss of sensorineural hearing, and overall hearing loss.
Finally, the asymmetric nature of these bony intrusions in the OIM mouse model could help to explain to the observed diminished hearing in human populations with OI.Approximately 60% of OIM mice in this study exhibited unilateral intrusions, and 28% exhibited bilateral intrusions of the hollow cochlea.In clinical studies, approximately 25% of patients with OI related hearing loss exhibited unilateral sensorineural or mixed hearing loss and 42% exhibited bilateral sensorineural or mixed hearing loss (Hartikka et al., 2004).Understanding the mechanism behind this bony intrusion on the cochlea may provide new avenues for the prevention and recovery of hearing loss in patients with OI.

| Comments on clinical treatments in humans
The primary goal of this study was to understand the etiology of hearing loss in OI.The significant and asymmetric presence of cochlear intrusions in OIM mice provides support for a similar mechanism of sensorineural hearing loss in human patients with OI.The primary treatment for sensorineural hearing loss, cochlear implants, requires electrodes to be placed directly on the cochlea.However, if the cochlea is damaged due to bony intrusions, this could potentially alter the possible treatment options available to those with mixed or sensorineural hearing loss caused by OI.Further analysis of the effects of OI on the bony capsule of the cochlea and surrounding bone is needed in order to make a definitive statement regarding impacts on treatment outcomes.

| CONCLUSIONS
While there are minimal differences between relative cochlear volumes between OIM and WT mice, we observed a significant rate of intraindvidual asymmetry in the cochlea of OIM mice.The majority of OIM mice had either unilateral or bilateral intrusions, and the observed fluctuating asymmetry in intraindividual cochlear volumes can also be attributed to unilateral errant bony growth.These bony intrusions may potentially affect the soft tissue and fluid components of the cochlea, leading to damage to the sensory components of hearing.
The results of this study indicate that there is a high potential for sensorineural (damage to the cochlea) hearing loss in OIM mice and elucidates at least one mechanism behind how this type of hearing loss might be occurring.Future research focusing on the middle ear could explore the potential for conductive (damage to the middle ear) or mixed hearing loss and correlate morphological findings with behavioral testing to confirm hearing loss in OIM mice.
3D rendering of a WT mouse skull.(b) Segmentation of the hollow bony cochlea in coronal/transverse view.(c) 3D rendering of the cochlear endocast.WT, wildtype model.

F
I G U R E 3 Box-&-whisker plots showing the range of cochlear volumes scaled to centroid size in OIM and WT mice at (a) week 4 and (b) week 16.No significant differences in cochlear volume were found between the genotypes.OIM, osteogenesis imperfecta murine model; WT, wildtype model.T A B L E 2 Cochlear volumes and comparisons, separated by weeks and disease state.

F
I G U R E 4 Box-&-whisker plots of cochlear intraindividual asymmetry in OIM and WT mice at (a) week 4 and (b) week 16.OIM mice have significantly greater intraindividual asymmetry ( p < 0.001) in cochlear volume than do WT mice.OIM, osteogenesis imperfecta murine model; WT, wildtype model.F I G U R E 5 Frequency of laterality for the larger cochlea (determined by volume) in OIM mice.No statistical differences in laterality were observed.OIM, osteogenesis imperfecta murine model.
Sample sizes for each scanned time point.