Immunolabeling and Counting Ribbon Synapses in Young Adult and Aged Gerbil Cochleae

Summary

A protocol for processing young adult and aged gerbil cochleae by immunolabeling the afferent synaptic structures and hair cells, quenching autofluorescence in aged tissue, dissecting and estimating the length of the cochleae, and quantifying the synapses in image stacks obtained with confocal imaging is presented.

Abstract

The loss of ribbon synapses connecting inner hair cells and afferent auditory nerve fibers is assumed to be one cause of age-related hearing loss. The most common method for detecting the loss of ribbon synapses is immunolabeling because it allows for quantitative sampling from several tonotopic locations in an individual cochlea. However, the structures of interest are buried deep inside the bony cochlea. Gerbils are used as an animal model for age-related hearing loss. Here, routine protocols for fixation, immunolabeling gerbil cochlear whole mounts, confocal imaging, and quantifying ribbon synapse numbers and volumes are described. Furthermore, the particular challenges associated with obtaining good material from valuable aging individuals are highlighted.

Gerbils are euthanized and either perfused cardiovascularly, or their tympanic bullae are carefully dissected out of the skull. The cochleae are opened at the apex and base and directly transferred to the fixative. Irrespective of the initial method, the cochleae are postfixed and subsequently decalcified. The tissue is then labeled with primary antibodies against pre- and postsynaptic structures and hair cells. Next, the cochleae are incubated with secondary fluorescence-tagged antibodies that are specific against their respective primary ones. The cochleae of aged gerbils are then treated with an autofluorescence quencher to reduce the typically substantial background fluorescence of older animals' tissues.

Finally, cochleae are dissected into 6-11 segments. The entire cochlear length is reconstructed such that specific cochlear locations can be reliably determined between individuals. Confocal image stacks, acquired sequentially, help visualize hair cells and synapses at the chosen locations. The confocal stacks are deconvolved, and the synapses are either counted manually using ImageJ, or more extensive quantification of synaptic structures is carried out with image analysis procedures custom-written in Matlab.

Introduction

Age-related hearing loss is one of the world's most prevalent diseases that affects more than one-third of the world's population aged 65 years and older¹. The underlying causes are still under debate and actively being investigated but may include the loss of the specialized synapses connecting inner hair cells (IHCs) with afferent auditory nerve fibers². These ribbon synapses comprise a presynaptic structure that has vesicles filled with the neurotransmitter glutamate tethered to it, as well as postsynaptic α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) glutamate receptors^3,4,5. In the gerbil, ~20 afferent auditory nerve fibers contact one IHC^6,7,8. Fibers on the IHC facing the modiolus are opposed to large synaptic ribbons, while the fibers connecting on the pillar side of the IHC face small synaptic ribbons (i.e., in cats⁹, gerbils⁷, guinea pigs¹⁰, and mice^{3,11,12,13,14}). Furthermore, in the gerbil, the size of the presynaptic ribbons and the postsynaptic glutamate patches are positively correlated^7,14. Fibers that are opposed to large ribbons on the modiolar side of the IHC are small in caliber and have low spontaneous rates and high thresholds¹⁵. There is evidence that low spontaneous rate fibers are more vulnerable to noise exposure¹⁰ and ototoxic drugs¹⁶ than high-spontaneous low-threshold fibers, which are located on the pillar side of IHCs¹⁵.

The loss of ribbon synapses is the earliest degenerative event in cochlear neural age-related hearing loss, while the loss of spiral ganglion cells and their afferent auditory nerve fibers lags behind^17,18. Electrophysiological correlates include recordings of auditory brainstem responses¹⁷ and compound action potentials⁸; however, these do not reflect the subtleties of synapse loss, since low spontaneous rate fibers do not contribute to these measures¹⁶. More promising electrophysiological metrics are the mass potential-derived neural index¹⁹ and the peristimulus time response²⁰. However, these are only reliable if the animal has no other cochlear pathologies, beyond auditory nerve fiber loss, that affect the activity of the remaining auditory nerve fibers⁸. Furthermore, behaviorally assessed thresholds in the gerbil were not correlated with synapse numbers²¹. Therefore, reliable quantification of surviving ribbon synapses and, thus, the number of functional auditory nerve fibers is only possible by direct examination of the cochlear tissue.

The Mongolian gerbil (Meriones unguiculatus) is a suitable animal model for studying age-related hearing loss. It has a short life span, has low-frequency hearing similar to humans, is easy to maintain, and shows similarities to human pathologies related to age-related hearing loss^2,22,23,24. Gerbils are considered aged when they reach 36 months of age, which is near the end of their average life span²². Importantly, an age-related loss of ribbon synapses has been demonstrated in gerbils raised and aged in quiet environments^8,21.

Here, a protocol to immunolabel, dissect, and analyze cochleae from gerbils of different ages, from young adults to aged, is presented. Antibodies directed against components of the presynapse (CtBP2), postsynaptic glutamate receptor patches (GluA2), and IHCs (myoVIIa) are used. An autofluorescence quencher is applied that reduces the background in aged cochleae and leaves the fluorescence signal intact. Further, a description is given of how to dissect the cochlea to examine both the sensory epithelium and the stria vascularis. The cochlear length is measured to enable the selection of distinct cochlear locations that correspond to specific best frequencies²⁵. Quantification of synapse numbers is carried out with the freely available software ImageJ²⁶. Additional quantification of synapse volumes and locations within the individual HC is performed with software custom written in Matlab. This software is not made publicly available, as the authors lack the resources to provide professional documentation and support.

Protocol

All protocols and procedures were approved by the relevant authorities of Lower Saxony, Germany, with permit numbers AZ 33.19-42502-04-15/1828 and 33.19-42502-04-15/1990. This protocol is for Mongolian gerbils (M. unguiculatus) of both sexes. Young adult refers to the age of 3-12 months, while gerbils are considered aged at 36 months and older. When not stated otherwise, buffers and solutions can be prepared and stored in the fridge for up to several months (4-8 °C). Before use, ensure that the buffers and solutions have not precipitated.

1. Fixation and organ collection

NOTE: If only the cochleae are needed, it is recommended to carry out the somewhat simpler procedure of fixation by immersion. However, if a well-preserved brain is also needed, then transcardial perfusion is the only option. The fixative in both cases is 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS). This should be freshly made but can be stored frozen until use. Use aliquots of ~300 mL for a transcardial perfusion or ~50-100 mL per cochlea for fixation by immersion.

CAUTION: PFA is a hazardous substance; handle it according to general lab safety procedures.

1. Fixation by transcardial perfusion
   1. Flush the perfusion setup until the tubing is clear of any air bubbles. Fill the perfusion tubing with PBS containing heparin (0.2 mL in 100 mL of PBS) to prevent blood clotting. Stop the flow once the tubing is filled with PBS and the fluid storage bottle has just emptied; pour 200 mL of thawed PFA into it.
      NOTE: The setup is now ready for the animal. The remaining PBS in the tubing is sufficient to flush out the blood and will automatically be followed by the fixative once the flow is resumed.
   2. Ensure that a waste collection system is in place for the outflow of PFA. Use a fresh cannula (19 G) and have large scissors, forceps (with a flattened tip), hemostats, scalpel or sharp cannula, and a rubber mat with pins handy.
   3. Euthanize the gerbil with an intraperitoneal overdose of pentobarbital (160 mg/mL, 0.3 mL per animal, body weight range: 50-120 g). Put the animal back into its cage. When respiratory arrest sets in such that the breathing becomes irregular with intervals of 30 s or longer, place the gerbil on its back on the rubber mat and fix both fore-paws and one hind-paw with pins (leave one hind-paw free to move for better judgment of perfusion success; see note after step 1.1.7).
   4. To open the thoracic cavity, lift the skin above the sternum with forceps and cut the skin approximately 0.5 cm below the sternum with scissors until the white-colored processus xiphoideus is visible. Hold the sternum at the processus xiphoideus with forceps and cut along the diaphragm to get a good view into the thoracic cavity. Cut the ribs laterally on both sides until there is good access to the heart. Ensure that the angle of the scissors is parallel and flat in relation to the body of the gerbil to avoid damage to organs and, thus, ensure a closed circulatory system.
   5. Clamp a hemostat onto the sternum, lift the ribcage, and place the hemostat over the shoulder of the gerbil without resting the hemostat on any body parts to avoid blocking blood flow.
   6. Open the fluid flow slightly until one drop of PBS flows out of the cannula (19 G) approximately every 2 s. Hold the heart with forceps and turn it a bit to the left such that the left ventricle can be clearly seen. Insert the cannula into the left ventricle at an angle that avoids penetrating the septum. Hold the needle in place either by hand or with another hemostat.
      NOTE: The left and right ventricles can be differentiated by their different color hues; the left ventricle appears lighter in color than the rest of the heart tissue.
   7. Slowly increase the pressure of the fluid flow and open the right atrium either with fine scissors, a scalpel, or another cannula. Open the fluid flow further until approximately 2-3 drops/s are observed at the drip chamber, or set a flow of 4 mL/min for the pump. Once fixation of the heart sets in, ensure that the perfusion cannula stays in place without being held any further.
      NOTE: Signs of successful perfusion are slow muscle contractions, stiffening of the neck and extremities, the liver going pale, and rosy lungs. The sign of unsuccessful perfusion is whitening of the lungs, which indicates that the pulmonary circulation is perfused due to puncture of the septum.
   8. In case of unsuccessful perfusion, try to move the cannula further up, into the aorta, and clamp it in place with a hemostat.
   9. Decapitate the gerbil. Remove the bullae and cut and break away its bone as described in step 1.2.2.
      NOTE: If the tissue is not sufficiently fixed, the sensory epithelium will dislodge from the organ of Corti during the dissection.
   10. If the perfusion shows no signs of fixation within 5-10 min, abort it and immediately follow the procedure below for fixation by immersion. For that, treat the cochlea as in step 1.2.2 and postfix the tissue in approximately 50-80 mL of 4% PFA in screw cap containers for 1-2 days on a 2D-shaker at 8 °C.
       NOTE: Since it can be difficult to ultimately rate the quality of fixation during perfusion, it is recommended to routinely postfix the cochleae as described.
2. Fixation by tissue immersion
   1. Euthanize the gerbils with an intraperitoneal overdose of pentobarbital (160 mg/mL, 0.3 mL per animal, body weight range: 50-120 g). When the gerbil has stopped breathing, decapitate the animal.
   2. Remove the bullae and cut and break away its bone with scissors and forceps, respectively, to reach the cochleae. Remove the maellus, incus, and semicircular canals. Make small holes at the apex and in the basal turn by carefully scratching over the cochlear bone with forceps. Immediately transfer the bullae into an excess of cold fixative (at least 50 mL) and fix the tissue in the cold (4-8 °C) under gentle agitation (2D-shaker [100 rpm]) for 2 days.
      ​NOTE: After fixing the tissue, the cochleae can either be processed immediately or stored in PBS with 0.05% sodium azide. CAUTION: Sodium azide is toxic. Note that the length of storage may negatively influence the quality of the staining. Limited trials have suggested that the immunostaining appeared weaker after 2 years of storing the tissue in sodium azide.

2. Tissue preparation and immunolabeling

1. Dissolve ethylenediaminetetraacetic acid (EDTA) powder in PBS to produce a 0.5 M solution at pH 8. For this, place a beaker onto a magnetic stirrer, fill it with approximately half of the total amount of the PBS, and add the appropriate amount of EDTA powder, resulting in an acidic suspension. Carefully add concentrated sodium hydroxide solution (NaOH) while monitoring the pH with a pH meter. Fill up with PBS to the desired final volume and filter the solution to prevent microbial contamination.
   NOTE: The EDTA powder will only completely dissolve once the suspension has reached a neutral pH value.
2. For decalcification, transfer the cochleae to 80 mL of 0.5 M EDTA in PBS. Incubate the tissue in the cold (4-8 °C) under gentle agitation on the 2D-shaker (100 rpm) for 2 days.
   NOTE: After the decalcification step, the tissue can be stored in PBS for several days up to 3 weeks before continuing with the remaining processing steps. For antibodies that label the stria vascularis, it is advisable to cut the cochleae in half along the modiolar axis at this point (i.e., prior to immunostaining) to ensure uniform access of antibodies to their respective targets. This is antibody-dependent and does not apply to the antibodies used here to label synaptic structures and IHCs. Use a piece of breakable razorblade and a blade holder (as described in step 4.2) for the cutting.
3. Perform the following steps (up to step 3.2) in 2 mL safe-seal reaction tubes. If the tissue piece is too large to fit inside the tube, trim the excess tissue with scissors. To improve penetration of the antibodies, first permeabilize the tissue in 1 mL of 1% triton (Triton X-100) in PBS on the 2D-shaker (100 rpm) at room temperature for 1 h. Wash the tissue 3x with 1 mL of 0.2% triton (Triton X-100) in PBS on the 2D-shaker (100 rpm) at room temperature for 5 min each.
4. To block nonspecific antigens, incubate the cochleae in 1 mL of blocking solution (3% bovine serum albumin [BSA], 0.2% triton, in PBS) on the 2D-shaker (100 rpm) at room temperature for 1 h.
   NOTE: The blocking solution can be prepared in advance but should not be older than ~10 days.
5. Dilute the following primary antibodies freshly in the same aliquot of blocking solution: anti-myoVIIa (myosin VIIa) to label IHCs (IgG polyclonal rabbit), diluted 1:400; anti-CtBP2 (C-terminal binding protein 2) to label presynaptic ribbons (IgG1 monoclonal mouse), diluted 1:400; and anti-GluA2 to label postsynaptic receptor patches (IgG2a monoclonal mouse), diluted 1:200. Make sure that the cochleae are fully covered with the antibody solution (typically 0.4 mL) and incubate them at 37 °C for 24 h.
6. Next, wash the tissue 5x with 0.2% triton in PBS on the 2D-shaker (100 rpm) at room temperature for 5 min each. Choose secondary antibodies to match the host species of their primary counterparts and again freshly dilute them in 3% BSA, 0.2% triton, in PBS: goat anti-mouse (IgG1)-Alexa fluorophore (AF) 488, diluted 1:1,000; goat anti-mouse (IgG2a)-AF568, diluted 1:500; and donkey anti-rabbit-AF647 (IgG), diluted 1:1,000. Wrap the tube in aluminum foil to prevent bleaching of the fluorescence. Incubate the cochleae in 0.4 mL of secondary-antibody solution at 37 °C for 24 h.
   NOTE: Limited trials have indicated that incubation of cochlear tissue at 37 °C for 24 h with primary and secondary antibodies resulted in brighter immunostaining than following the more common incubation procedure with lower temperatures and shorter durations.
7. Wash the cochleae 2x with 1 mL of 0.2% triton in PBS for 5 min each and 3x with PBS for 5 min each on the 2D-shaker (100 rpm) at room temperature.
   ​NOTE: After these washing steps, the cochleae can remain in PBS in the fridge at ~4 °C for several days.

3. Treatment with autofluorescence quencher (optional)

NOTE: Cochleae from middle-aged and aged gerbils show extensive background autofluorescence. In young adult tissue, treatment with an autofluorescence quencher is not necessary. It is, in principle, possible to apply the autofluorescence quencher before the immunostaining procedure, which then avoids any inadvertent reduction of the desired antibody fluorescence. However, according to the manufacturer's datasheet, the use of detergents (such as Triton X-100 in the current protocol) is no longer possible as they remove the quencher from the tissue.

1. Cut the cochleae in half under a stereomicroscope, as described in step 4.2.
2. Mix the autofluorescence quencher with 70% ethanol to obtain a 5% solution and incubate the cochleae in this solution on the 2D-shaker at room temperature for 1 min. Wash the cochleae 3x with 1 mL of PBS on the 2D-shaker at room temperature for 5 min each.
   ​CAUTION: The autofluorescence quencher is hazardous and harmful. Wear gloves when handling this substance.

4. Final fine dissection

1. Dissect the cochlea under a stereomicroscope. Fill a polystyrole Petri dish and its lid with PBS and keep two fine forceps, Vannas spring scissors, a blade holder, and a breakable razorblade handy. Break pieces from the razorblade to obtain a cutting surface of ~2-4 mm depending on the dissection step. Prepare a microscope slide by placing three drops of mounting medium in a row.
   NOTE: The cutting surface of the razor blade wears off quickly. The blade must be exchanged after dissecting and mounting approximately every second piece of the cochlea.
2. If not already done in step 3.1, first cut the cochlea in half along the modiolus under a stereomicroscope. For this, position a piece of a razorblade longer than the coiled length of the cochlea into a blade holder. Place the cochlea in the Petri dish and cut away excess tissue with the piece of razorblade. Hold the cochlea in place with fine forceps and cut in half along the modiolus.
3. Start with one half, but leave the other half in the Petri dish as well. Carefully fixate the cochlear half with forceps such that the cutting edge is facing upwards. Use fine spring scissors to cut away the bone of the cochlea above the helicotrema, covering the apex.
4. To begin the separation of the cochlear pieces, isolate the middle turn by cutting with scissors through the modiolus and auditory nerve, above (scala vestibuli) and below (scala tympani).
5. Cut through the cochlear bone covering the stria vascularis within the cochlear duct. Make two cuts on both sides of the cochlear bone above the organ of Corti and along the stria vascularis and use the razorblade to eventually separate the cochlear pieces.
   NOTE: The stria vascularis is easily visible as a dark stripe. Cutting along the stria vascularis leaves the organ of Corti intact.
6. Optional: To collect the stria vascularis, carefully cut between the organ of Corti and the stria vascularis to separate the two. Leave the stria vascularis connected to the spiral ligament (attached to the bone covering the outer surface of the cochlea) and remove the decalcified bone using forceps. Place the piece with the stria side up on the microscope slide in a drop of mounting medium.
   NOTE: If the collected piece is curved too strongly, it may be necessary to divide it into smaller pieces for it to be flat enough for mounting on the slide.
7. Transfer the cochlear pieces to the PBS-filled lid of the polystyrole Petri dish, ensuring that the cochlear whole mounts lie as flat as possible on the slide. Remove excess tissue, such as parts of the spiral ligament on the abneural side and parts of the spiral limbus on the neural side. Carefully remove the tectorial membrane with super fine forceps.
8. Place the cochlear pieces onto the slide into a drop of mounting medium. Place the cochlear whole mounts with the organ of Corti facing up on the slide to avoid obscuring the IHCs in the optical path for imaging. Look for invagination of the spiral limbus in close proximity to the IHCs, which is visible by shifting the cochlear piece into the sagittal plane to identify the side to face upward.
   NOTE: To digitally reconstruct the complete cochlea from its individual pieces during further processing, it is strongly recommended to document sketches of the pieces and note landmarks. Furthermore, the pieces should ideally be arranged in the correct order on the slide.
9. Repeat steps 4.3 to 4.8 until the entire cochlea is transferred onto the microscope slide. If necessary, add more mounting medium, coverslip the slide, and seal the coverslip in place with black nail polish painted around the edges. Let it dry in the dark at room temperature and then store the slide in the dark at 4 °C.
   ​NOTE: Even if parts of the sensory epithelium itself are accidentally lost, it is important to nevertheless mount what is left of the cochlear piece for correct length estimation.

5. Cochlear length measurement

1. Measure the length of the cochlea from brightfield images of its pieces using an epi-fluorescence microscope system and associated software. Save low-magnification images (4x lens) from every cochlear piece and use the lasso measurement tool of the microscope software to draw a line along the row of IHCs in each of the images. Calculate the total length by adding the lengths of all the pieces.
   NOTE: When parts of the sensory epithelium are missing within a cochlear piece, interpolate the line.
2. To define the cochlear locations that should be analyzed in an individual cochlea, calculate their corresponding distances from the apex using the equation given by Müller²⁵. Mark these locations, for instance, on a printout of the cochlear pieces.

6. Image acquisition with a confocal microscope

1. Use a confocal microscope with an oil-immersion 40x objective (numerical aperture 1.3) and the appropriate oil for high-resolution imaging.
   NOTE: If the confocal microscope has an inverted light path, the slide must be positioned upside down. In this case, give the specimen approximately 30 min to sink and rest stably on the coverslip before starting the final scan. Alternatively, use a mounting medium that is fluid throughout the duration of dissection but solidifies later and simultaneously conserves the fluorescence.
2. Activate the appropriate lasers: two optically pumped semiconductor lasers with wavelengths of λ = 488 nm and λ = 522 nm, and a diode laser with a wavelength of λ = 638 nm are used in this protocol. Choose the emission range of the fluorescence tags (AF488: 499-542 nm, AF568: 582-621 nm, AF647: 666-776 nm). As a hybrid detector counts the released photons, place the laser line at least 10 nm away from the emission curve.
   NOTE: It is important to carry out preliminary checks with single-labeled tissue to ensure that the chosen detector bandwidths cleanly separate the color channels (i.e., do not lead to channel crosstalk). Radio waves interfere with the hybrid detector, resulting in maximum photon count, and thus cause artefactual stripes in the stack. Therefore, avoid using a mobile phone near the microscope.
3. Zoom in on the tissue until the image spans 10 IHCs. Choose the resolution of the image according to Nyquist sampling, which is typically ~40-60 nm/pixel. Set the step size in the z-direction to 0.3 µm and the imaging speed to 400-700 Hz.
   NOTE: Both these settings (step size and imaging speed) are compromises between using optimal imaging parameters and saving scanning time.
4. Perform bidirectional sampling to shorten imaging time. Accumulate the frame for the 488 nm and 638 nm channel 3x and for the 522 nm channel 6x; additionally, average the lines 3x for each channel. Choose the beginning and the end of the stack in the z-dimension.
5. Set the gain to 100 when using the hybrid detector (counting mode) to avoid decreased signal-to-noise ratio. Set the laser power so that no pixel is saturated in the region of interest but the structures cover mostly the full 8 bit range. Start with a low laser power of 0.5% and increase it until structures are visible.
   NOTE: Good staining typically needs a laser power between 0.1% and 5%.
6. Use deconvolution software for post hoc processing of the image stacks to remove the blurring halo around small fluorescent structures using a theoretical point-spread function. Use the same settings for each stack within an experiment. Save the deconvolved images as .tif or .ics.
   ​NOTE: The myoVIIa-label (IHCs) does not benefit from deconvolution and may be omitted to save time. The software uses meta data of the image files; however, several parameters such as the light path, the embedding medium, or the immersion medium need to be specified.

7. Synapse quantification

1. Open a copy of the deconvolved stacks in the freely accessible software ImageJ with the additional Biovoxxel-plugin, which is also available on their website.
2. Adjust the colors of individual channels by splitting the channels (Image | Color | Split Channels) and merging (Image | Color | Merge Channels) them again, assigning different colors. Convert the image to an RGB-stack (Image | Color | Stack to RGB) and adjust the brightness and contrast (Image | Adjust | Brightness/Contrast | either Auto or slider regarding the Maximum), if needed, so that the pre-and postsynaptic structures and the IHC label are pleasantly distinct from the background.
3. Choose five IHCs and label them with the text tool (IHC1-IHC5) by clicking on the desired location within the stack. Open the ROI manager (Analyze | Tools | ROI Manager); activate the tickbox Labels to label the points with a number. Zoom in on the IHC of interest.
4. Use the point/multipoint tool in multipoint mode (right-click on the tool to choose between point- or multipoint mode). Click onto a functional ribbon synapse (i.e., a presynaptic ribbon in close juxtaposition to a postsynaptic glutamate patch) while scrolling through the z-dimension.
   NOTE: Depending on their amount of overlap and the color chosen for each channel, the shared pixels appear in mixed color. Usually, the distinction between individual functional synapses is simple because they are sufficiently distant from each other.
5. When all the structures of interest are ticked, click Add on the ROI manager's graphical user interface. Click on the arbitrary name and choose Rename.
   NOTE: The point-labels still stay through all planes of the stack when the Show all checkbox is selected in the point tool options menu (double-click on the point tool icon), which helps avoid counting the same ribbon synapse multiple times.
6. When choosing the next IHC to count, avoid inadvertently adding counts by adjusting the image to fully display the next IHC with the hand tool. Change from multipoint tool to point tool to avoid adding more puncta to the previously stored data. Click on a structure of interest within the next IHC and change back to multipoint tool. Repeat steps 7.4 to 7.5 until all the IHCs of interest are evaluated.
7. To save data, click on a dataset in the ROI manager and then on measure. Wait for a new window to appear, listing the measured data points. Save this list as a spreadsheet file (File | Save as). Save the image by activating Show All in the ROI manager. Click on Flatten to permanently add the point-labels to the stack. When closing the image stack, agree to save the changes.

8. Analysis of synapse volume and position on the hair cell

NOTE: The authors used a custom-programmed procedure based on Matlab. Since it is not publicly available, it is outlined here only in broad terms (see also⁷). Please contact the corresponding author if interested in using it. The procedure expects a triple-labeled (IHCs, pre- and postsynaptic) image stack in TIFF format as input, guides the user through the various steps of analysis via a graphical interface, and provides extensive output of the results in spreadsheet format.

1. Normalize the position of the synaptic structures to a 3D-coordinate system defined by the individual IHC's extent on the pillar-modiolar axis and cochlear apical-basal axis and the IHC's top (cuticular plate) to bottom (synaptic pole) axis.
   NOTE: The volumes of synaptic elements (both pre- and postsynaptic, and the combined volume of functional synapses) are provided in µm³ and normalized to the respective median value³.

Representative Results

Cochleae were either harvested after cardiovascular perfusion with fixative of the whole animal or rapidly dissected after euthanizing the animal and immersion-fixed. With the latter method, the IHCs stayed in place during dissection, whereas, in cases of unsuccessful perfusion and thus insufficiently fixed tissue, the sensory epithelium was often destroyed. Note that the authors encountered cases where fixation of the cochleae after transcardial perfusion was insufficient while fixation of the brain was still adequate. Tissue from an inferior perfusion could still be saved by opening holes into the cochlear apex and base and postfixing the cochleae by immersion in 4% PFA for 2 days.

Length determination of the whole cochlea was conducted according to Müller²⁵ to evaluate IHCs and their synapses at specific cochlear positions, equivalent to distinct characteristic frequencies (Table 1). For this, the desired cochlear positions were calculated as percentages of basilar membrane length from the cochlear base. The median cochlear length of 13 cochleae from young adult gerbils was 11.3 mm (interquartile range: 11.02-11.52 mm). The median cochlear length of 24 cochleae from aged gerbils was 11.5 mm (interquartile range: 11.24-11.73 mm). There was no significant difference between the length of young adult and aged cochleae (Mann-Whitney U-test: U = -1.62, p = 0.105); the overall median was 11.48 mm. One may argue that this variation in the length of individual cochleae was small and using fixed length positions (in mm) is adequate. However, the place-frequency function is non-linear. Therefore, a deviation from the median length causes a larger error for basal cochlear locations than for relatively more apical cochlear locations. For example, for the cochlear location equivalent to a frequency of 1 kHz, calculated based on the median cochlear length (i.e., 11.48 mm), the corresponding frequency ranged from 1.16 kHz to 0.91 kHz for the shortest (10.36 mm) and the longest cochlea (12.28 mm) in this sample, respectively. For a basally located frequency (e.g., 32 kHz), the frequency range was 51.62 kHz to 23.97 kHz, for the shortest and longest cochlea, respectively. Therefore, it is advisable to calculate the individual locations on each cochlea, in particular when examining basal cochlear locations.

Figure 1 depicts maximum intensity z-dimension projections of cochleae from a young adult (10 months, Figure 1A) and an aged gerbil (38 months, Figure 1B), which are examples of ideal immunolabels. In the image depicting the young adult gerbil's cochlea, the myoVIIa-labeled IHCs, here shown in blue, are in sharp contrast to the black background. Therefore, individual IHCs are easily detectable. The pre- and postsynaptic structures are clearly visible as green and red elements, respectively. They are assumed to form a functional synapse whenever they are in close juxtaposition (Figure 1A',B'). There were rarely unpaired presynaptic structures (orphaned ribbons) apparent in the cochleae of aged gerbils. The cochlea from the old gerbil (Figure 1B) was treated with the autofluorescence quencher. The IHC label appears weaker, although the laser power used in this specimen was three times higher than that used for the cochlea from the young adult gerbil. Nevertheless, the IHCs are still distinct from the background. The pre-and postsynaptic structures are clearly visible, and the laser power used for both those channels was similar or even below that for the young adult specimen. Tissue from aged animals typically showed significantly more nonspecific-looking fluorescent signal. Therefore, the main difference in the protocol for young adult and aged material is the treatment with an autofluorescence quencher. Note that this was carried out after the immunostaining with fluorescence-tagged antibodies and might, in principle, also affect the intended antibody label. However, the treatment effectively reduced extraneous fluorescence of nonspecific origin, while leaving sufficient signal of the specific label of the structures of interest (compare Figure 1B with Figure 2B). Preliminary results indicated, however, that, in the stria vascularis, extraneous fluorescence did not manifest in samples from aged gerbils.

Examples of stacks that were suboptimally processed are shown in Figure 2. Figure 2A depicts the maximum-intensity z-projection of a stack from an old gerbil (36 months), where the IHCs, which were also either unnaturally bent or ripped apart, were not scanned in their entirety. As a result, only the apical and basal poles of IHCs are visible in this scan. It cannot be excluded that more synapses were located in the missing middle part of the IHCs and, therefore, a reliable analysis is not possible. Thus, it is crucial that all evaluated IHCs are scanned completely in the confocal stack. Figure 2B shows the maximum-intensity z-projection of a stack sampled in an aged gerbil (38 months). The fluorescence of unclear origin is high as the autofluorescence quencher was not used in this case. However, the extraneous fluorescence was largely confined to the (red) channel associated with the GluA2-label, which is quite typical. In such cases, it may still be possible to count functional synapses if the CtBP2-label of the ribbons is relatively clean and specific. Figure 2C depicts the maximum-intensity z-projection of a stack obtained from an aged gerbil (42 months). Here, the synapses cannot be allocated to individual IHCs as the IHCs were largely disintegrated; indeed, it is difficult to be sure how many IHCs are represented. In the example shown in Figure 3A, a mobile phone was used in close vicinity to the confocal microscope during the scan. Stripes are visible in the blue channel (IHC label). Fortunately, in this case, this did not affect the synaptic channels and affected only the upper part of the IHCs, thus, an analysis was still viable.

Figure 3 displays maximum-intensity z-projections of stacks taken from the same piece of cochlea from a young adult gerbil, acquired 3 months (Figure 3A,A') and 29.5 months (Figure 3B,B') after immunostaining. To remain comparable, the images were not taken from the exact same location (which might have suffered bleaching from the previous scan) but from the same cochlear piece. For the scan taken at the later time point, the laser power had to be increased ~2- to 3-fold. Nevertheless, the labeled structures are still clear. The signal-to-noise ratio had decreased, especially in the channels displaying postsynaptic and IHC structures, while the channel with the presynaptic label was less affected. Quantification of functional synapses per IHC is still possible.

Typical results for the analysis of synapse volume and position on the IHC are illustrated in Figure 4, using the confocal stack shown in Figure 1A. Individually defined IHCs are graphically displayed as grid reconstructions in different colors, with or without the functional synapses (defined by colocalization of pre- and postsynaptic labels) allocated to each. This display can be freely rotated in 3D to obtain various viewing angles (Figure 4A,B). IHCs can also be singled out for graphical display (Figure 4B). A large variety of data graphs, illustrating different aspects of the distribution of synaptic volumes within the IHC-centered coordinate system, is produced (examples in Figure 4C-E). The raw quantitative data for each IHC and each synaptic element are also available as spreadsheets which can then, for example, be combined across several confocal stacks for further statistical analysis.

Figure 1: Examples of successfully processed cochleae. Maximum-intensity z-projections of confocal stacks from (A) a young adult gerbil (10 months), obtained at a cochlear location corresponding to 1 kHz, and (B) an aged gerbil (38 months, panel B), obtained at a cochlear location equivalent to 500 Hz. IHCs were stained with an antibody against myoVIIa (blue), presynaptic ribbons were labeled with anti-CtBP2 (green), and postsynaptic glutamate patches were labeled with anti-GluA2 (red). The cochlea of the aged gerbil was treated with an autofluorescence quencher after the immunostaining. For clarity, the IHCs' outlines are indicated by dashed lines. Panels (A') and (B') display an enlargement of the areas outlined by the squares in the corresponding cochleae from panels (A) and (B). Yellow arrowheads point to functional synapses. The channels displaying the pre- and postsynaptic structures (but not the IHC channel) underwent deconvolution. Scale bars = 10 µm (A,B), 1 µm (A',B'). Abbreviations: IHC = inner hair cell; myoVIIa = myosin VIIa; CtBP2 = C-terminal binding protein 2; GluA2 = ionotropic glutamate receptor. Please click here to view a larger version of this figure.

Figure 2: Examples of suboptimally processed cochleae. Maximum-intensity z-projections of confocal stacks for which the processing was suboptimal, from gerbils that were (A) 36 months and (B, C) 38 months old and from cochlear locations corresponding to 16 kHz, 8 kHz, and 32 kHz, respectively. Gerbils from which the cochleae in panels (A) and (C) were derived were transcardially perfused, and the cochlea from panel (C) was additionally postfixed for 3 days in 4% PFA. The cochlea from (B) was immersion-fixed in 4% PFA for 2 days. Cochleae from (A) and (C) were treated with the autofluorescence quencher. IHCs were stained with an antibody against myoVIIa (blue), presynaptic ribbons were labeled with anti-CtBP2 (green), and postsynaptic glutamate patches were labeled with anti-GluA2 (red). All channels were deconvolved. Brightness and contrast were further adjusted after the confocal scan. Scale bars = 10 µm. Abbreviations: PFA = paraformaldehyde; IHC = inner hair cell; myoVIIa = myosin VIIa; CtBP2 = C-terminal binding protein 2; GluA2 = ionotropic glutamate receptor. Please click here to view a larger version of this figure.

Figure 3: Stability of immunolabel after prolonged storage. Maximum-intensity z-projections of confocal stacks from a young adult gerbil (10 months), taken at the 16 kHz cochlear location, (A) 3 months after staining and (B) at a slightly more basal cochlear location on the same cochlear piece 29.5 months after staining. The cochlea was immersion-fixed in 4% PFA for 2 days. For clarity, (A') and (B') zoom in on only a few functional synapses from the areas indicated by squares in (A) and (B), respectively. IHCs were stained with an antibody against myoVIIa (blue), presynaptic ribbons were labeled with anti-CtBP2 (green), and postsynaptic glutamate patches were labeled with anti-GluA2 (red). The laser power for the IHC channels was 1.3% and 3%, for the presynaptic channels 0.4% and 1.1%, and for the postsynaptic channels 1.1% and 2% in (A) and (B), respectively. Note that in (A), blue stripes are visible in the upper part of the IHCs, which result from the use of a mobile phone in close vicinity to the confocal microscope. Scale bars = 10 µm (A, B), 1 µm (A', B'). Abbreviations: PFA = paraformaldehyde; IHC = inner hair cell; myoVIIa = myosin VIIa; CtBP2 = C-terminal binding protein 2; GluA2 = ionotropic glutamate receptor. Please click here to view a larger version of this figure.

Figure 4: Representative results of the quantification of synapse volume. (A) Ten IHCs shown as differently colored grid reconstructions based on the myoVIIa immunolabel. Their associated functional synapses, defined by colocalized CtBP2-(green) and GluA2-labeled elements (red), are displayed together with the IHCs, and also separately below, for clarity. Note that the angle of view was chosen to be perpendicular to the IHCs' long axis and thus differs from the original confocal image (Figure 1A). (B) One of the IHCs singled out and shown rotated 90°. The black plane was manually defined by the user and bisects the IHC along its pillar-modiolar axis. (C) Bubble plot showing the location of all functional synapses in this confocal stack, relative to the normalized three axes of their respective IHC. size of the symbols is proportional to the volume of the presynaptic element. Note that the angle of view was chosen to be similar to panel (B). (D) Boxplot of the normalized volumes of synaptic elements, separately for the presynaptic (left 2 boxes) and postsynaptic (right 2 boxes) partners of functional synapses and further separated according to their position in the modiolar or pillar half of their respective IHC (different colors). Boxes represent interquartile ranges, with the medians indicated by lines. Dashed whiskers indicate 1.5 times the interquartile range, and the crosses indicate outlying values beyond that. (E) Scatterplot of the normalized volumes of pre- vs. postsynaptic partners of functional synapses. Different symbols indicate the position in the modiolar or pillar half of their respective IHC. Abbreviations: IHC = inner hair cell; myoVIIa = myosin VIIa; CtBP2 = C-terminal binding protein 2; GluA2 = ionotropic glutamate receptor. Please click here to view a larger version of this figure.

Table 1: Distances from the apex that are equivalent to specific target frequencies in cochleae of different lengths. The equivalent best frequencies were calculated based on the equation given by Müller²⁵. Please click here to download this Table.

Discussion

With the method outlined in this protocol, it is possible to immunolabel IHCs and synaptic structures in cochleae from young adult and aged gerbils, identify presumed functional synapses by co-localization of pre- and postsynaptic elements, allocate them to individual IHCs, and quantify their number, volume, and location. The antibodies used in this approach also labeled outer hair cells (OHCs; myoVIIa) and their presynaptic ribbons. Furthermore, a viable alternative for immunolabeling of both IHCs and OHCs is an antibody against otoferlin, with OHCs appearing much fainter than IHCs.

Perfusion may be carried out using two different setups. 1) A gravity-fed drip-line system in which a single bottle feeding a commercial drip line is suspended on a pulley wheel approximately 1.5 m above the animal. Fluids are introduced successively after lowering the bottle, which is open at the top. 2) A system using a digital variable-speed peristaltic pump, with a thin, long tube open at one end to take in the fluid and a needle that can be easily attached at the other end. Both work equally well, and the subtle differences will not be elaborated upon here. The authors specifically recommend, however, a down draft-style workbench, with the animal on a perforated platform and the fumes drawn off below. This enables good access to the surgical area without compromising the exhaust function (unlike working in an open fume cabinet).

Proper fixation of the tissue is of critical importance, since otherwise the sensory epithelium will detach and disintegrate during dissection. In the gerbil, more prolonged exposure to the fixative is necessary than commonly used (e.g., for mice^17,27,13 or guinea pigs²⁸). The preferred method for gerbils is rapid extraction of the cochlea after the animal's death and immersion-fixation for at least 1.5 days. If cardiovascular perfusion is preferred, it is crucial that fixation sets in within a few minutes and proceeds well. Since it can be difficult to ultimately rate the quality of fixation during perfusion, it is recommended to routinely postfix the cochleae as described.

It is important to comply with the washing steps, whereby the last washing step (step 2.7) is the most important one. If not adequately washed, the tissue is sticky and adheres to the dissection instruments, which makes the dissection difficult. It is also recommended to use an autofluorescence quencher in cochleae harvested from aged gerbils to reduce nonspecific fluorescence. Autofluorescence might originate from lipofuscin, which is common in tissue from aged animals^29,30,31, and appears to be broadly excited as well as broadly emitting. When excited with a wavelength in the UV -spectrum (λ = 364 nm), lipofuscin has a broad emission range (λ = 400-700 nm, with a maximum at ~λ = 568 nm)³². In human myocardial tissue, lipofuscin is visible with an excitation of λ = 555 nm and emission of λ = 605 nm³³. Similarly, in the IHCs of gerbils, the authors noticed an increase in nonspecific fluorescence most prominently in the channel used for the AF568 antibody, which suggests autofluorescence from lipofuscin granules. A general recommendation when working with tissue from animals is thus to use the excitation bandwidth around 550-600 nm for the least critical immunolabel.

The measurement of total cochlear length and the correct identification of specific cochlear positions is only possible if its entire length is preserved. If parts of the sensory epithelium are lost in dissection, it is recommended to still mount the remaining part or even only the spiral ganglion piece, and keep detailed notes. It is then usually possible to estimate the missing section with reasonable accuracy. If the apical portions of the cochlea are completely preserved, but the basal end is missing, specific cochlear locations on the apical part might be definable by calculating the positions based on the median cochlear length within the used gerbil population because the deviation from the real value is small.

An important limitation of the synapse volume quantification is that the resulting absolute volumes are not comparable across different confocal stacks. The critical step that determines the resulting volumes of synaptic elements is the choice of the intensity threshold for initial detection. However, the choice of this intensity threshold is made subjectively by the user in the current protocol and many others^3,27,34. Furthermore, the brightness of the immunofluorescence in the confocal image depends on many factors, which are very difficult to standardize across specimens, such as tissue thickness, orientation of tissue relative to the optical path, duration of laser exposure, and precise confocal and deconvolution settings. All these caveats are exacerbated if rare material is acquired over a considerable length of time, as is typical for aging gerbils. Together, this means that biases are very difficult to exclude in pooled data, and the best-case scenario is a dataset with high variance. It is thus recommended to routinely normalize synaptic volumes to the respective median volume of the individual confocal stack, as introduced by Liberman et al.³. The obvious downside is that synaptic volumes are then only comparable within a given image stack, e.g., between different locations on the IHC but not between different cochlear locations or specimen ages.

The dual labeling of pre- and postsynaptic structures and the requirement of co-localization allows for a more reliable estimate of the number of functional synapses than using either label alone. If only one label is feasible, it is recommended to use the presynaptic anti-CtBP2, coupled to AF488 or fluorescent tags with similar wavelength specifics, for two reasons. First, orphaned ribbons, that is, CtBP2-labeled elements without a co-localized postsynaptic label, appear to be rare¹⁷, and the authors confirmed this for aged gerbils. Thus, the error introduced by not being able to verify co-localization with a postsynaptic partner is small. It should be noted, however, that this becomes a more significant issue when examining noise-exposed ears (e.g.,²⁸). Second, in cases with high fluorescence signal of unclear origin, the noise typically affected the channel using an excitation wavelength around 568 nm to the greatest extent (example in Figure 2B, representing the GluA2 label). At least part of this noise may, in tissue from aged animals, be lipofuscin autofluorescence. Thus, to maximize the chance for a clean, single anti-CtBP2 label, it is advisable to avoid this wavelength channel. Finally, it was shown that, with proper cool and dark storage conditions, the immunolabeling used here is stable over long periods of time, up to at least 2.5 years (Figure 3B), which makes the re-evaluation of valuable tissue, such as from aged gerbils, possible.

The quantification of IHC afferent synapse number has become an important metric for evaluating the state of the peripheral auditory system in many contexts, among them age-related hearing loss. There are various protocols published now (e.g., gerbils²¹, mice^17,13, guinea pigs^10,28, and humans³⁵). The method outlined here is, in its details, specific to young adult and aged gerbils but some general recommendations have been derived that may be of use in other species too.

Materials

Name	Company	Catalog Number	Comments
Albumin Fraction V biotin-free	Carl Roth	0163.2
anti-CtBP2 (IgG1 monoclonal mouse)	BD Biosciences, Eysins	612044
anti-GluA2 (IgG2a monoclonal mouse)	Millipore	MAB39
anti-mouse (IgG1)-AF 488	Molecular Probes Inc.	A21121
anti-MyosinVIIa (IgG polyclonal rabbit)	Proteus Biosciences	25e6790
Blade Holder & Breaker - Flat Jaws	Fine Science Tools	10052-11
Bonn Artery Scissors - Ball Tip	Fine Science Tools	14086-09
Coverslip thickness 1.5H, 24 x 60 mm	Carl Roth	LH26.1
Disposable Surgical Blade	Henry Schein	0473
donkey anti-rabbit (IgG)-AF647	Life Technologies-Molecular Probes	A-31573
Dumont #5 - Fine Forceps	Fine Science Tools	11254-20
Dumont #5SF Forceps	Fine Science Tools	11252-00
Ethanol, absolute 99.8%	Fisher Scientific	12468750
Ethylenediaminetetraacetic acid	Carl Roth	8040.2
Excel	Microsoft Corporation
Feather Double Edge Blade	PLANO	112-9
G19 Cannula	Henry Schein	9003633
goat anti-mouse (IgG2a)-AF568	Invitrogen	A-21134
Heparin	Ratiopharm	N68542.04
Huygens Essentials	Scientific Volume Imaging
ImageJ	Fiji
Immersol, Immersion oil 518F	Carl Zeiss	10539438
Intrafix Primeline Classic, 150 cm (mit Datamatrix Code auf der Sterilverpackung)	Braun	4062957E
ISM596D	Ismatec		peristaltic pump
KL 1600 LED	Schott	150.600	light source for stereomicroscope
Leica Application suite X	Leica Microsystem CMS GmbH
Leica TCS SP8 system	Leica Microsystem CMS GmbH
Matlab	The Mathworks Inc.
Mayo Scissors Tungston Carbide ToghCut	Fine Science Tools	14512-17
Mini-100 Orbital-Genie	Scientific Industries	SI-M100	for use in cold environment
Narcoren (pentobarbital)	Boehringer Ingelheim Vetmedica GmbH
Nikon Eclipse Ni-Ei	Nikon
NIS Elements	Nikon Europe B.V.
Paraformaldehyde	Carl Roth	0335.3
Petri dish without vents	Avantor VWR	390-1375
Phosphate-buffered saline:
Disodium phosphate	AppliChem	A1046
Monopotassium phosphate	Carl Roth	3904.1
Potassium chloride	Carl Roth	6781.1
Sodium chloride	Sigma Aldrich	31434-M
Screw Cap Containers	Sarstedt	75.562.300
Sodium azide	Carl Roth	K305.1
Student Adson Forceps	Fine Science Tools	91106-12
Student Halsted-Mosquito Hemostat	Fine Science Tools	91308-12
Superfrost Adhesion Microscope Slides	Epredia	J1800AMNZ
Triton  X	Carl Roth	3051.2
TrueBlack Lipofuscin Autofluorescence Quencher	Biotium	23007
Vannas Spring Scissors, 3mm	Fine Science Tools	15000-00
Vectashield Antifade Mounting Medium	Vector Laboratories	H-1000
Vibrax VXR basic	IKA	0002819000
VX 7 Dish attachment for Vibrax VXR basic	IKA	953300
Wild TYP 355110 (Stereomicroscope)	Wild Heerbrugg		not available anymore