Imaging the Aging Cochlea with Light-Sheet Fluorescence Microscopy

Summary

A light-sheet microscope was developed to image and digitize whole cochlea.

Abstract

Deafness is the most common sensory impairment, affecting approximately 5% or 430 million people worldwide as per the World Health Organization¹. Aging or presbycusis is a primary cause of sensorineural hearing loss and is characterized by damage to hair cells, spiral ganglion neurons (SGNs), and the stria vascularis. These structures reside within the cochlea, which has a complex, spiral-shaped anatomy of membranous tissues suspended in fluid and surrounded by bone. These properties make it technically difficult to investigate and quantify histopathological changes. To address this need, we developed a light-sheet microscope (TSLIM) that can image and digitize the whole cochlea to facilitate the study of structure-function relationships in the inner ear. Well-aligned serial sections of the whole cochlea result in a stack of images for three-dimensional (3D) volume rendering and segmentation of individual structures for 3D visualization and quantitative analysis (i.e., length, width, surface, volume, and number). Cochleae require minimal processing steps (fixation, decalcification, dehydration, staining, and optical clearing), all of which are compatible with subsequent high-resolution imaging by scanning and transmission electron microscopy. Since all the tissues are present in the stacks, each structure can be assessed individually or relative to other structures. In addition, since imaging uses fluorescent probes, immunohistochemistry and ligand binding can be used to identify specific structures and their 3D volume or distribution within the cochlea. Here we used TSLIM to examine cochleae from aged mice to quantify the loss of hair cells and spiral ganglion neurons. In addition, advanced analyses (e.g., cluster analysis) were used to visualize local reductions of spiral ganglion neurons in Rosenthal's canal along its 3D volume. These approaches demonstrate TSLIM microscopy's ability to quantify structure-function relationships within and between cochleae.

Introduction

The cochlea is the peripheral sensory organ for hearing in mammals. It has a complex spiral anatomy of repeating sensory and supporting cells that are anatomically specialized to detect sound vibrations and transmit them to the brain for the perception of hearing. The main sensory elements are the inner and outer hair cells and their innervating nerve fibers whose cell bodies compose the spiral ganglion, which resides within Rosenthal's canal (Figure 1). These sensory and neural structures are tonotopically arranged such that high-frequency sounds are transduced in the cochlear base and low-frequency sounds are transduced in the cochlea apex². An anatomical map of this sensory cell distribution along the spiral length of the supporting basilar membrane is called a cytocochleogram³ and can be compared with hearing loss as a function of frequency as depicted in an audiogram.

The membranous labyrinth of the cochlea, which is surrounded by dense bone, makes it technically difficult to examine more than one cochlear structure at a time. Therefore, the rationale for developing a light-sheet microscope is to produce well-aligned serial sections of the complete cochlea so that all cochlear structures can be examined relative to one another in 3D reconstructions. Voie et al.⁴ and Voie and Spelman⁵ designed the first light-sheet microscope, called orthogonal plane fluorescence optical sectioning (OPFOS) microscope, to optically section the whole cochlea. However, this microscope was never commercially developed; so, our aim was to construct a light sheet microscope called a thin sheet laser imaging microscope (TSLIM; Figure 2). The design and construction details for TSLIM have previously been published⁸. TSLIM made several improvements over the OPFOS, including using a low-light digital camera versus a CCD camera for image collection, optically encoded micropositioners for accurate and reproducible movement of the specimen through the light-sheet, use of a commercially available, optically clear specimen chamber, and Rhodamine staining in ethanol rather than in the clearing solution to prevent stain precipitation within the tissue. Commercial development of light-sheet microscopes such as SPIM⁶ have focused on high-resolution imaging of live, small transparent specimens but are unsuitable for whole cochlear imaging as they lack adequate working distance. A review of the development of other light-sheet microscopes was published by Santi⁷. TSLIM's primary advantage over other histological methods to examine the cochlea is to optically section tissues for 3D reconstruction while preserving the integrity of the specimen so that it can be used by other histological methods. Another advantage of TSLIM imaging is that only a thin light-sheet produced by a laser is exposed to the tissue, compared with whole tissue thickness exposure to the laser as in confocal microscopy. Tissue clearing to minimize light scatter and the fact that only a small portion of the tissue is exposed to the laser results in minimal fluorochrome fading (photobleaching) with light-sheet laser imaging. However, the process of fixation, dehydration, and clearing does alter the morphology of cochlear structures and results in tissue shrinkage compared with living tissue. The actual amount of tissue shrinkage that occurs was not determined.

TSLIM was developed by Shane Johnson and eight German optical engineering students (see Acknowledgements). TSLIM construction details were provided by Santi et al.⁸ and a scanning version (sTSLIM) by Schröter et al.⁹. TSLIM functions as a nondestructive microtome to optically section specimens and as a microscope to collect 2D serial sections through the full width and thickness of the cochlea. TSLIM can image both small (mm) and large (cm), thick specimens. Lenses are air mounted to allow for long working distances with collection objectives of 1x and 2x on a dissection microscope. The dissection microscope also has zoom optics that allow TSLIM to resolve subcellular and synaptic structures on cells. TSLIM is equipped with both a blue (473 nm) and green (532 nm) laser for illumination that allows for a variety of fluorescent probes to be used for imaging. The goal of TSLIM is to produce well-aligned 2D optical sections through a whole cochlea for a complete digital reconstruction of cochlear tissues. Since it is a fluorescent method, ligands and immunohistochemistry can also be used to identify specific cochlear structures.

Initially, a cylindrical lens was used to produce two opposed Gaussian light sheets, but it produced absorption imaging artifacts. Due to the work of Keller et al.¹⁰, the fixed cylindrical lens was replaced by a scanning galvanometer mirror to produce the light sheet⁹. In addition, since the center of the light sheet is the thinnest at the beam waist, sTSLIM 2D images are produced by collecting a composite of X-axis columns of data across the specimen's width (Figure 3). This method was first described by Buytaert and Dircks¹¹. TSLIM custom software to drive and collect images was developed using a graphical program for instrument control. The light sheet travels through the specimen and illuminates a fluorescent plane within the tissue. This fluorescent plane is projected orthogonally through the transparent specimen and is collected by a dissection microscope. Optically encoded micropositioners allow scanning through the beam waist in the X-axis to collect a single composite 2D image and, subsequently, the Z-axis micropositioner moves the specimen to a deeper plane within the tissue to obtain a stack of serial, sectioned 2D images (Video 1, Figure 4). A stack of translational images is collected through the entire width, thickness, and length of the cochlea, and stitching of images is not required (Video 2). The image stack is transferred to another computer and loaded into a 3D rendering program for 3D reconstruction and quantification. The image stacks contain all the digital information about the morphology of a cochlea at the resolution of the microscope. However, if a higher resolution is required, the intact cochlea can be further processed by destructive histological methods such as microtome sectioning, scanning, and transmission electron microscopy.

The 3D rendering program is used to segment different cochlear structures for 3D rendering and quantitative analysis. For segmentation, each structure in every 2D image of the stack is traced using a different color by a graphics tablet and pen (Figure 5). To date, 20 different cochlear structures have been segmented (Figure 6). After segmentation, a variety of 3D analyses can be performed. For example, 3D rendering software can virtually resection the cochlea in any plane along the structure's centroid. Video 3 shows sectioning tangential to the organ of Corti, which reveals the hair cells along the length of the basilar membrane. This process first requires manual segmentation of the structure of interest. Next, the structure's centroid is calculated based on the least squares fit of spline points placed along the center of the structure from its base to its apex, thus allowing an approximation of the structure's length (Video 4). A similar process called skeletonization can be used to visualize the radial width of the structure along its length using a color map (Video 4). The total volume of each structure is calculated by the program after segmentation, but relative distances can also be quantified and visualized with color maps in a 3D rendering software (Figure 7). Segmented structures can also be exported to produce enlarged, solid-plastic model renderings (Figure 8). In addition, semi-automated cell counting can also be performed using 3D rendering software (Figure 9). Immunohistochemistry and ligand binding can be used to stain specific cochlear structures and these structures can be isolated from other cochlear structures for morphometrical assessment such as producing a cytocochleogram (Figure 10). Length, width, surface, volume, and number of all cochlear structures can be determined from the 3D models, making this approach ideal for mapping cochlear damage to functional impairments. Specifically, cochlear damage due to aging, noise-induced trauma, or other insults can be shown and quantified in 3D cochlear reconstructions from 2D optical sections. Once a cochlea has been digitized there are numerous imaging algorithms that can be used to assess cochlear damage of any tissue within the cochlea in the anatomical registry to other cochlear tissues.

Protocol

All the procedures and the use of live animals have been reviewed and approved (Protocol ID #2010-38573A) by the University of Minnesota Institutional Care and Use Committee (IACUC) and investigators who use these animals have been thoroughly trained and tested by the Research Animal Resources (RAR) Veterinarians before they have access to the animal facilities. Both male and female mice were used in this study.

1. Cochlea removal for fixation and tissue processing for imaging

1. Euthanize a mouse using CO₂ inhalation. Decapitate the mouse with scissors and make a dorsal-ventral incision through the brain to hemisect the skull. Remove the brain, identify the round bulla in the baso-ventral part of the skull, open the bulla with rongeurs, and visualize and remove the cochlea.
2. Fixation: Perform this procedure under a fume hood and using a dissection microscope at 5x magnification. Wear gloves and protective clothing. Puncture the oval window and remove the stapes with a sharp pick. Insert a pick into the round window to puncture the membrane.
3. Cover the opened round window with the cut tip of an infusion set attached to a 1 mL syringe filled with 2 mL of formalin. Slowly infuse formalin through the perilymphatic spaces of the cochlea over a 2 min period, noting that formalin is exiting the cochlea via the opened oval window. Trim excess tissue off the cochlea and immerse in a bottle containing 10% formalin, and place on a rotator overnight.
4. Decalcification: Rinse the cochlea in PBS 3x for 5 min each and immerse in a bottle containing 10% solution of disodium ethylenediaminetetraacetic acid (EDTA) with rotation for 4 days, changing the solution daily.
5. Dehydration: Perfuse the cochlea with PBS 3x and immerse for 15 min between changes. Dehydrate the cochlea with ascending concentrations of ethanol 10%, 50%, 70%, 95%, 95%, 100%, 100%; for 30 min in each concentration.
   ​NOTE: It is important to remove all the EDTA before dehydration as EDTA precipitates in ethanol. Also, cochleae can be left in any concentration of ethanol greater than 70% overnight.
6. Staining: Stain the whole cochlea by immersion in a solution of Rhodamine B isothiocynate (5 µg/mL in 100% ethanol) overnight with rotation. Remove excess dye from the cochlea with two changes of 100% ethanol, 5 min each change.
7. Clearing: Transfer the stained cochlea into two changes of Spalteholz¹² solution (5:3 methyl salicylate:benzyl benzoate), 30 min each change and leave overnight in the clearing solution with rotation. Cochleae can be left in Spalteholz solution indefinitely.

2. Imaging of cochleae

1. Attach the cochleae to a specimen rod at the oval and round window membrane end so that the clearing solution remains within the cochlea and bubbles are not formed (Figure 2). Care must be taken not to let bubbles form within the cochlea as they are difficult to remove and if left in the tissue, they will cause imaging artifacts.
2. Use a UV activating glue to attach the wet cochlea to the dry specimen rod (Figure 2). Attach the cochlea at the oval and round window ends. Cure the UV glue for 10 s by moving around the cochlea with the UV light.
   NOTE: A loose attachment of the cochlea to the specimen rod will result in imaging defects. This rod is specifically manufactured for this protocol (see Santi et al.⁸ for details) and is specific to our light-sheet microscope.
3. Suspend the cochlea into the imaging chamber filled with Spalteholz solution for imaging. The specimen chamber is an optically clear quartz fluorometer cell (Video 1).
4. Attach the specimen rod to a rotating holder that is also attached to XZ translation stage. Most stacks are obtained by translating the specimen in the XZ planes, but rotational stacks can also be obtained.
5. TSLIM optical sectioning: Use a blue or green laser for excitation depending upon the type of fluorochrome staining. Position the light sheet in the middle of the tissue for focusing and determine the magnification that will be used to illuminate the full width of the cochlea. Then, use a custom-designed program to move the specimen through the light-sheet across the specimen in the X-axis (stitching the image) and in Z-steps to make a stack of 2D images throughout the cochlea.
6. For the first image, the beam waist of the light-sheet is positioned at the edge of the specimen and the program scans the full width of the specimen collecting columns of data (see image stitching; Santi et al.⁸) that are the width of the confocal parameter (Figure 3) to produce a composite 2D image of maximum resolution across the width of the specimen. The program automates image stitching for each Z-step until the specimen is completely imaged.
7. Image processing: Transfer the image stack to another computer and load it into a 3D rendering program for 3D reconstruction and quantification.

Representative Results

Since the theme of this special issue is imaging the effects of aging in the cochlea, a young (3-month-old, HS2479, CBA strain mouse) and aged (23-month-old, HS2521, C57 strain mouse) cochleae will be used as examples. It should be noted that TSLIM is capable of imaging a variety of specimens, including cochleae from humans, mammals, other rodents, and fish, as well as other organs such as the brain.

Johnson et al. ¹³ published an article on SGNs in young (3-week-old) CBA mice using TSLIM. All SGNs were counted in five mice and there was a range of 8,408-8,912 SGNs with an average Rosenthal's length of 2.0 mm. In the present study, we counted the SGNs in a 3-month-old CBA mouse that contained 7,913 SGNs in a Rosenthal's canal length of 2.0 mm. The 23-month-old C57BL6 mouse contained 6,521 SGN with a Rosenthal's length of 2.11 mm. A volume rendering shows all SGNs in both cochleae, but SGN loss was not revealed in the older mouse. However, SGN cell counts as a function of Rosenthal's canal distance, depicted in a linear plot, showed greater SGNs in the younger animal in the middle and apical ends of the cochlea as compared to the older animal (Figure 11). This apparent loss of SGNs was further visualized by using cluster analysis in 3D rendering software. Here, SGN density differences were depicted on a color scale where high-density clusters are yellow, and blue represents lower density regions (Figure 12). Although it is not possible to identify specific cells that have been lost, the cluster analysis clearly visualizes regions of lower cell density along the length of Rosenthal's canal (Figure 12).

Figure 1: Direct volume rendering of a cochlea. A composite figure from a volume rendering of a mouse cochlea showing the oval (O) and round (R) windows, organ of Corti with hair cells (blue line, OC) and Rosenthal's canal that has been segmented and volume rendered that contains spiral ganglion neurons (SGNs). Bar = 200 µm. Please click here to view a larger version of this figure.

Figure 2: Specimen holder. A wood prototype of a custom specimen holder for attachment of the cochlea (arrowhead) to a specimen rod (arrow) using UV-activated glue while keeping the clearing solution within the cochlea. Please click here to view a larger version of this figure.

Figure 3: Light sheet cross-section. A cross-section through an X-axis scanned light sheet showing the beam waist in the middle of the light sheet and the confocal parameter (CP), which is a region where the beam thickness is relatively constant. The upper solid rendering shows the width of the beam without scanning and the lower solid rendering shows the width of the scanned beam that remains thin across the full width of the cochlea. Please click here to view a larger version of this figure.

Figure 4: A mid-modiolar 2D cross-section through a mouse cochlea. Four cochlear turns are sectioned. The basal turn of Rosenthal's canal can be seen containing spherical-shaped spiral ganglion neurons. Bar = 100 µm. Please click here to view a larger version of this figure.

Figure 5: Two cross sections of the cochlea. On the left is a 2D cross-section and on the right is the same section with several structures that have been segmented using different colored tracings. Bar = 200 µm. Please click here to view a larger version of this figure.

Figure 6: Cochlear structures segmented. (A) This figure shows a segmented cochlea with the surrounding otic capsule. (B-K) These show several different cochlear structures that have been segmented and the centroids (white line) are determined for most structures. Cochlear structures have been segmented with different colors for their identification: (A) composite figure with bone (white), (B) turquoise (spiral ligament), (C) maroon (stria vascularis), (D) yellow (basilar membrane), (E) red (organ of Corti), (F) yellow-green (Rosenthal's canal), (G) blue (scala tympani), gold (stapes footplate), (H) white (scala media), green (ductus reuniens), (I) red (scala vestibuli), purple (round window membrane), gold (cochlear aqueduct), (J) green (tectorial membrane), (K) purple (spiral limbus). Bar = 400 µm. Please click here to view a larger version of this figure.

Figure 7: Comparison of the same cochlear structure in two different cochleae. Using the Procrustus method in 3D rendering software, the Scala media from two different mice have been compared. The left panel shows the fitting of a least-squares fit of two structures and the right panel shows their quantitative differences using a color map. Please click here to view a larger version of this figure.

Figure 8: Solid 3D plastic model of the cochlea. The left panel shows the Scala tympani of a mouse cochlea with the basilar membrane, organ of Corti, and helicotrema segmented. The right panel shows a solid plastic model of the structures on the left. Please click here to view a larger version of this figure.

Figure 9: Spiral ganglion neuron counting. SGNs have been identified and marked by the 3D rendering software program on a cross-section of the Scala media of the mouse cochlea. Please click here to view a larger version of this figure.

Figure 10: Immunohistochemical labeling of outer hair cells. Anti-prestin antibodies and fluorescent secondary antibodies that were perfused through the peri lymphatic Scala of a mouse cochlea have labeled all the outer hair cells that have been volume rendered using 3D rendering software. Please click here to view a larger version of this figure.

Figure 11: Density of SGNs along the length of Rosenthal's canal. The greatest loss of SGNs appears to be in the middle and apical ends of Rosenthal's canal in the 23-month-old mouse. Please click here to view a larger version of this figure.

Figure 12: Cluster analysis of SGN cell density. The left panel shows a cluster analysis of SGNs in a 3-month-old, normal hearing mouse. A color map shows that the greatest density of cells within a cluster of a given size is in the middle Rosenthal's canal, where mouse hearing thresholds are lowest. The right panel shows the loss of SGNs throughout Rosenthal's canal with the greatest loss of SGNs in the middle of Rosenthal's canal. Please click here to view a larger version of this figure.

Video 1: X-axis scanning video of the cochlea. X-axis scanning of the specimen through the light sheet showing the cochlea suspended by the specimen rod in the clearing solution that is contained within a glass chamber. An orthogonal 2D fluorescent cross-section of the cochlea is shown due to illumination by the light sheet. Please click here to download this Video.

Video 2: Stack of serial sections of the cochlea. A short stack of 2D cross sections is obtained by X-axis scanning and 10 µm Z-steps through the cochlea. Notice the presence of horizontal lines throughout the stack, which are absorption artifacts due to the use of a cylindrical lens to produce the light sheet. Please click here to download this Video.

Video 3: Resectioning of a cochlea in a different plane. A stack loaded into the 3D rendering software program and virtual resectioning in a plane tangential to the organ of Corti showing the hair cells along the length of the basilar membrane. This video has been modified from⁸. Please click here to download this Video.

Video 4: Skeletonization of the Scala media. The left panel shows the calculation of the centroid within the Scala media of a mouse cochlea. The right panel shows a color map of the radial dimensions of the Scala media along its length. Please click here to download this Video.

Discussion

Optical sectioning by light sheet microscopy for examination of cochlear structures is not mechanically destructive like other more traditional histological methods, and it provides a complete digital view of cochlear structures relative to one another. Previous methods such as surface preparations of the organ of Corti¹⁴ provided a map of hair cell loss along the length of the basilar membrane, but SGN loss could not be assessed since the tissue had been dissected away to reveal the organ of Corti. Alternatively, mid-modoilar microtome sections of the cochlea provide only a very small sample of the condition of the SGNs throughout the length of Rosenthal's canal. With TSLIM, all cochlear structures are digitized at the level of the resolution of the instrument (i.e., subcellular). Complete image stacks of the cochlea allow for quantification and visualization of multiple cochlear structures, that are shown here, that could not be obtained by traditional histological methods. For example, using a clustering algorithm, the present investigation showed a completely new way to view and quantitate SGN cell density within Rosenthal's canal and characterize cell loss due to aging (Figure 12).

TSLIM is an early⁸ development of a light-sheet microscope that was inspired by the work of Voie and colleagues^4,5. Specifications for the construction of TSLIM were included in the paper by Santi et al.⁸. In fact, Brown et al.¹⁵ stated that they constructed a similar light sheet microscope for imaging the cochlea based on TSLIM's design. As mentioned in the Introduction and reviewed by Santi⁷ development of other light sheet microscopes (e.g., SPIM) was directed at investigating live cell development and gained better resolution by immersion of high N.A. objectives within the specimen chamber, which required small working distances and is unsuitable for imaging whole cochleae. Commercial development of light sheet microscopes continues, and this type of imaging is extremely useful for preparing well-aligned, serial sections of tissues and animals made transparent by chemical clearing methods.

For any light-sheet microscope application, it is critical that the specimen be made transparent to prevent light scatter and absorption. Decalcification is the first step in the process to remove calcium from the cochlea for transparency. If EDTA is not rinsed completely out of the tissue before dehydration, it will precipitate within the tissue and good imaging will not be possible. Dehydration by ethanol is necessary for infiltration of the Spalteholz clearing solution. If a specimen is heavily pigmented, then bleaching of the pigment may help make the specimen more transparent. There are many other chemicals and methods that can be used to make a specimen transparent and users may wish to try different methods to determine which method is most suitable for their specimen. Although light-sheet microscopy produces 2D images of tissue, its resolution is not as great as can be obtained from thin mechanical sections of tissue (especially plastic sections) and brightfield microscopy. Its primary advantage is to produce well-aligned, serial sections of tissue for 3D reconstruction of structures.

The two mice cochleae that we show as examples of SGN loss due to aging reveal loss of neurons due to aging. An unexpected finding was the loss of SGNs in the middle and apical ends of the cochlea rather than a basal loss that would correspond to a loss of hair cells in the base. However, we did not assess hair cell loss in these cochleae and the two samples are simply examples rather than an investigation of the aging effects on SGNs. In a paper by White et al.¹⁶, they described SGN loss in 18-month-old C57 mice but did not determine the distribution of the loss along the length of Rosenthal's canal. In another paper by Grierson et al.¹⁷, using 24-28-month-old-mice they described massive degeneration of OHC efferents, especially in the apical half of the cochlea, where there was also a significant loss of OHCs. Indeed, the future holds many possibilities for extracting new data and obtaining a better understanding of pathological processes within the cochlea due to aging and other insults. Furthermore, cochlear image stacks have been provided to many other investigators and it will be interesting to see how these investigators mine data to answer their research questions.

Materials

Name	Company	Catalog Number	Comments
Amira 3D Rendering Software	ThermoFisher Scientific		Address: 501 90th Ave NW, Coon Rapids, MN 55433
benzyl benzoate (W213810)	Sigma-Aldrich, Inc.		Address: PO Box, 14508, St. Louis, MO 68178
Bondic	Bondic		Address: 235 Industrial Parkway S., Unit 18 Aurora, ON L4G 3V5 Canada
Ethanol 95% and 100%	University of Minnesota		Address: General Storehouse, Minneapolis, MN 55455
Ethylenediaminetetraacetic acid disodium salt dihydrate (EDTA)  (E5134)	Sigma-Aldrich, Inc.		Address: PO Box, 14508, St. Louis, MO 68178
LabVIEW graphical program and Vision	National Instruments		Address: 11500 N Mopac Expwy Austin, TX 78759-3504
methyl salicylate (M6742)	Sigma-Aldrich, Inc.		Address: PO Box, 14508, St. Louis, MO 68178
Olympus MVX10 dissection microscope	Olympus Corp		Address: 3500 Corporate Parkway, Center Valley, PA 18034
Rhodamine B isothiocynate, (283924)	Sigma-Aldrich, Inc.		Address: PO Box, 14508, St. Louis, MO 68178
Starna Flurometer Cell (3-G-20)	Starna Cells		Address: PO Box 1919, Atascadero, CA 82423