A. The synaptopathy model
The model of synaptopathy used here is one we have studied extensively at the light-microscopic level [15, 16, 37, 38]. It is designed to produce only a transient elevation of cochlear thresholds, but massive and permanent loss of the synapses between auditory nerve fibers (ANFs) and cochlear sensory cells, despite no loss of the hair cells themselves. After this exposure to an 8-16 kHz noise band at 98 dB SPL for 2 hrs, the loss of synapses, as seen via immunostaining of pre-synaptic ribbons and post-synaptic glutamate receptors (Fig. 1C), peaks at the 32 kHz region, where only about 50% remain, when viewed either at 1 day or 1 week post exposure (Fig. 1A).
In prior studies, the sampling power of confocal microscopy allowed us to extract synaptic counts from literally hundreds of hair cells at each of multiple cochlear regions, from many animals in each experimental group. In the present study, the labor-intensive nature of ultrastructural investigation limits us to a much smaller sample, i.e. one 20-mm cochlear span (~2 inner hair cells wide) from each of four ears. Thus, the consistency of the synaptopathic outcome in randomly selected 20 mm samples is key to evaluating the reliability of the present results. To this end, we re-analyzed data from confocal image stacks taken at the 32 kHz region, the heart of the synaptopathic region, extracting synaptic counts from contiguous 20 mm samples from the relevant z-stacks in prior studies, As shown in Figure 1B, the synaptic counts per 20 mm from the confocal data are non-overlapping between control and exposed ears at either survival time. Furthermore, the synaptic counts from the present study (see below) all fit well within the prior light-microscopic data.
B. Synaptic analysis
In the present study, we analyzed four FIB-SEM blocks, all from the 32 kHz region, two from control ears, one from an exposed ear at 1-day post exposure and one from an exposed ear at 1-week post exposure. In each case, we segmented the inner hair cells (IHCs), each ANF all the way to its terminus at or near the IHC, and all of the mitochondria they contained. (Fig. 2A,B).
In the two control ears, a total of 75 ANFs were reconstructed, of which 71 formed synapses with an IHC and 4 did not. Of the non-synaptic fibers, two touched an IHC while the other two ended blindly without doing so. Of all control fibers from two separate blocks, 73/75 were unbranched, forming a single synapse via a single terminal swelling. One fiber branched to make synaptic contact with two adjacent IHCs, the other made two synapses with the same IHC.
The morphology of the presynaptic ribbons in control ears can appear heterogeneous (Fig. 3A1-A4 1st column); however, when virtually re-sectioned in planes customized to the synapse’s orientation around the roughly hemispherical basal pole of the IHC, most of the heterogeneity disappears (Fig. 3A1-A4 columns 2 - 4). In the cross-sectional views (2nd column), all the classic features of the ribbon synapse are seen: an electron-dense presynaptic ribbon in the IHC, surrounded by a halo of vesicles, and closely apposed to a region of thickened post-synaptic membrane [20]. In three dimensions, each ribbon is a flattened cylinder, significantly longer than its diameter, and oriented with its long axis parallel to the IHC membrane, and some appear to have an electron-lucent core (Fig. 3C,D,E). In control ears, almost all synapses had a single ribbon at the active zone (69/71), while only two (3%) had a double ribbon.
In exposed ears, we reconstructed 59 ANFs at each of the two survival times evaluated. Multiple-ribbon synapses were much more common in these exposed ears (Fig. 3B,D,E). In the 1-day case, 7 of 22 synapses (32%) had multiple ribbons (5 doubles and two triples), and in the 1-week case, 9/28 synapses (32%) had multiple ribbons (8 doubles and 1 triple). Some examples are shown in Figure 3B. In addition to the multiplication of ribbons per synapse, there was a pronounced elongation of the ribbons’ long axes, typically associated with a decrease in the short-axis diameter (Fig. 3,C,D,E). Although the multiplication of ribbons was apparently long lasting, the elongation of the ribbons had largely reversed by 1 week post exposure. Some of the ribbons at the 1-day survival were also abnormally small (Fig. 3D).
In normal ears, spatial gradient of ribbon sizes has been seen via confocal microscopy, larger on the modiolar side than the pillar side. Our 3D reconstructions of ribbon volume showed the same trend in both control and exposed ears, however the modiolar vs. pillar group difference only reached statistical significance for the 1-day stack (Fig. 4).
The most striking abnormality in the exposed ears was the large number of non-synaptic ANFs, i.e. fibers that failed to form any of the classic pre- or post-synaptic specializations, despite the fact that many of these non-synaptic terminals came in intimate contact with the IHC. One of these non-synaptic IHC contacts is illustrated in Figure 5, where we show every 5th slice through the entire contact zone. Although there is a complex web of juxta-membranous cisternae within the IHC that seems localized to the contact zone, there is no pre-synaptic ribbon, no cluster of synaptic vesicles and no obvious zone of post-synaptic membrane specialization. Its possible that the cloud of large irregular vesicles and nearby diffuse electron density represents the former site of the pre-synaptic ribbon (red dashed circles in panels 26 and 31). In the 1-day stack, 37/59 ANFs (62%) were non-synaptic (21 of which (57%) were still in intimate contact with an IHC), and in the 1-week stack, 31/59 (53%) were non-synaptic (15 of which (48%) were still in intimate contact with an IHC). Among the non-synaptic fibers no longer in contact with the IHCs, most had not retracted very far: the average distance to the nearest IHC was 2.1 mm for both the 1-day and the 1-week survivals.
As seen in the 3D reconstructions, most of these non-synaptic fibers were on the modiolar side of the IHCs (Fig. 6). The total number of ANFs in each stack, i.e. synaptic plus non-synaptic, was within the range expected in this region in a normal ear (Fig. 1B). Thus, we infer that virtually no unmyelinated terminals have disappeared, at either the 1 day or the 1 week post-exposure time.
In prior confocal studies, ~10% of the ribbons are “orphaned” at the 24 hr post-exposure time point, i.e. they appear near the IHC membrane without any visible post-synaptic marker. Orphans are extremely rare in control ears or in exposed ears at longer post-exposure times [16]. Here, we saw a single orphan ribbon at the IHC membrane in the 1 day stack, without any ANF terminal closeby. There were no orphan ribbons in either of the control stacks or in the 1-week stack.
C. Mitochondrial content
Mitochondrial content of ANFs is particularly relevant, because it is related to threshold sensitivity, spontaneous discharge rate and vulnerability to synaptopathy [6, 20]. Exemplar 3D renderings in Figure 7A show that, although mitochondria are present throughout much of the extent of all ANFs, there is often a clustering of mitochondria in the terminal swelling just below the extreme terminus. This accumulation of mitochondrial area in the terminal vs. the pre-terminal regions of the fibers is illustrated in Figure 7B, where we plot the cross-sectional area of all the mitochondria in every slice from every fiber in all four FIB-SEM stacks. The clearly non-uniform distribution of mitochondria along each fiber’s length suggest it could be useful to consider the terminal vs. pre-terminal regions separately when analyzing ANFs and the effects of noise.
In control ears, as expected based on prior ultrastructural studies in cat [20], the fibers synapsing on the modiolar side of the IHC tend to be mitochondrion-poor compared to those on the pillar side, whether looking at the terminal or pre-terminal regions (Figs. 8C and D, respectively). The data in Figure 7B also suggest that there is a transient enhancement of mitochondrial area evident at 1 day post exposure that largely recovers 1 week later. The further analyses of Figure 8 show that this is indeed the case. Mean mitochondrial area per slice increases one day post exposure, in both the terminal and pre-terminal regions (Figs. 8E and F, respectively). 1 week later, mitochondrial content is smaller than pre-exposure in the terminal region, but statistically indistinguishable from control in the pre-terminal region (Figs. 8G and H, respectively). Interestingly, among the ANFs in exposed ears, the non-synaptic fibers have a clear tendency towards lower mitochondrial content than the synaptic fibers.
The increases in total mitochondrial cross-sectional area could arise by swelling of individual mitochondria, or by increase in the number of mitochondrial profiles per section via elongation, division or migration [44]. The micrographs in Figure 8A,B, suggest that the increases in mitochondrial area per slice are not due to gross swelling of individual mitochondria, indeed, the mitochondrial morphology looks quite normal in noise-exposed ANFS. To be more quantitative, we manually counted mitochondrial profiles in every 10th slice of the ANFs from a control and the 1-day post-exposure cases. Results showed an 50% increase in the number of mitochondrial profiles per section (3.75 ± 0.29 vs 2.48 ±0.13, p < 0.001) with a smaller (15%), but significant, increase in the mean size of each profile (6.22 ± 0.2 vs 5.41 ± 0.115, p = 0.002), suggesting that mitochondria are both hypertrophying and multiplying, elongating or migrating from the cell body to the unmyelinated terminals within the organ of Corti by 1 day post exposure.
D. Efferent Innervation
In addition to the synaptic connection to sensory cells, ANFs in mammalian ears are innervated by terminals from the lateral olivocochlear (LOC) pathway. These efferent neurons contain several neurotransmitters including acetylcholine and dopamine [7, 32] and provide both excitatory and inhibitory feedback, respectively [8]. The dopaminergic fibers have been implicated in the protection of ANFs from noise-induced synaptopathy [4].
In the neuropil beneath the IHCs, LOC terminals are easily identified (Fig. 9A,B,C,D) based on their content of large clusters of synaptic vesicles [19]. LOC neurons spiral for long distances within the inner spiral bundle (Fig. 2A) and send off numerous small and large side branches to innervate many different ANFs [1]. Thus, these vesicle-filled terminals are almost always easily followed back to larger branches of highly complex neuronal structures. There are many places in the neuropil beneath the IHCs where LOC terminals come in close contact with ANFs, however, in the present study, we considered these contacts to be synaptic in nature only if the following criteria were met: 1) the membranes of the afferent and efferent fibers were in direct apposition (i.e. as close to each other as at an afferent synapse), 2) a cluster of at least 5 vesicles was present within 2 vesicle-diameters from the efferent membrane, 3) there was membrane thickening and/or intermembranous darkening at the region of apposition, and 4) criteria 1-3 were met for a minimum of 5 consecutive slices. Some of the synaptic and non-synaptic interactions between afferent and efferent fibers are illustrated in Figure 9A,B,C,D.
The overall distribution of LOC synapses is well seen in the 3D reconstruction of Figure 6 (bottom row). As expected from confocal microscopy of cochleas immunostained for cholinergic and dopaminergic markers, these synapses are concentrated in a cloud near the bases of the IHCs. In the control ears, there is a tendency for synapses on pillar-side fibers to extend farther from the IHC than on modiolar-side fibers. These controlled transparency superpositions make it difficult to appreciate the quantitative differences in efferent innervation. These are better illustrated by summing the surface area of synaptic contact per fiber. In control ears, the modiolar-side fibers have lower areas of synaptic contact than the pillar-side fibers (Fig. 9E). In exposed ears, there is a transient increase in the area of efferent synaptic contact, that reverts back to control values at 1 week post-exposure (Fig. 9F). Interestingly, at both 1 day and 1 week post-exposure, there is significantly less efferent innervation on the non-synaptic ANFs than on the synaptic ANFs (Fig. 9G). For each of these analyses, we evaluated whether the differences in total efferent synaptic contact were due to changes in the number of contacts or the size of the existing contacts: the data suggest it is the latter. For example, the average plaque size in control vs. 1-day ANFs was 0.496 mm2 ± 0.034 vs. 0.666 mm2 ± 0.052 (a significant increase of 34%; p = 0.003). whereas the average number of plaques per ANF was 3.78 ± 0.17 vs. 3.98 ± 0.23 (an insignificant increase of 5%; p = 0.57), respectively.