Ultrastructural maturation of the endbulb of Held active zones comparing wild-type and otoferlin-deficient mice

Summary

Endbulbs of Held are located in the anteroventral cochlear nucleus and present the first central synapses of the auditory pathway. During development, endbulbs mature functionally to enable rapid and powerful synaptic transmission with high temporal precision. This process is accompanied by morphological changes of endbulb terminals. Loss of the hair cell-specific protein otoferlin (Otof) abolishes neurotransmission in the cochlea and results in the smaller endbulb of Held terminals. Thus, peripheral hearing impairment likely also leads to alterations in the morphological synaptic vesicle (SV) pool size at individual endbulb of Held active zones (AZs). Here, we investigated endbulb AZs in pre-hearing, young, and adult wild-type and Otof^−/− mice. During maturation, SV numbers at endbulb AZs increased in wild-type mice but were found to be reduced in Otof^−/− mice. The SV population at a distance of 0–15 nm was most strongly affected. Finally, overall SV diameters decreased in Otof^−/− animals during maturation.

Graphical abstract

Highlights

• •Maturation of wt endbulb of Held active zones leads to more synaptic vesicles
• •At endbulbs of otoferlin knockout mice, synaptic vesicles decline with age
• •Mainly two distinct synaptic vesicle populations are affected
• •Synaptic vesicles sizes are reduced in six-month-old otoferlin knockout animals

Audiology; Molecular Neuroscience; Cellular Neuroscience

Introduction

The mammalian auditory system encodes acoustic stimuli with exquisitely high temporal precision (Moser et al., 2006; Safieddine and Wenthold, 1999; Wang et al., 2011). Various processes such as directional sound localization as well as interindividual communication essentially rely on this fidelity (Ashida and Carr, 2011; Eddins et al., 2018; Grothe et al., 2010).

In mammals, the conversion of a physical sound wave into neural code is mediated via ribbon synapses between cochlear inner hair cells (IHCs) and primary afferent spiral ganglion neurons (Moser et al., 2006). Here, type I afferent auditory nerve fibers (ANFs) (Liberman and Oliver, 1984) project into the cochlear nucleus (CN) as a first central relay station (Nicol and Walmsley, 2002; Ryugo and Fekete, 1982). Consequently, auditory synaptopathy at IHC ribbon synapses leads to functional impairments downstream on CN level (Moser and Starr, 2016). For example, the proper function of exocytosis at IHC ribbon synapses relies on the hair-cell-specific C₂ domain protein otoferlin (Otof) (Pangrsic et al., 2010; Roux et al., 2006). In humans, mutations in OTOF cause recessive, non-syndromic hearing impairment DFNB9 (Rodríguez-Ballesteros et al., 2008; Varga et al., 2006; Yasunaga et al., 1999). Likewise, genetic deletion of Otof (Otof^−/−) in mice abolishes exocytosis at IHC ribbon synapses. Therefore, the peripheral auditory pathway fails to transduce acoustic input and auditory brain stem responses are absent in Otof^−/− mice (Roux et al., 2006). This functional impairment has previously been shown to also affect CN structure (Wright et al., 2014).

Early reports noted that multipolar cells within the CN comprised morphologically and functionally distinct classes (Harrison & Irving, 1965, 1966; Lorente de Nó, 1981; Osen, 1969; Ostapoff et al., 1994). Specifically, in the anteroventral cochlear nucleus (AVCN), large presynaptic terminals are found that wrap around the cell body of two types of bushy cells (BCs). The primary task of this cell type is to encode precise timing information of the sound stimulus, which is ensured due to BCs' outstanding ability of phase locking (Joris et al., 1994; Paolini et al., 2001; Smith et al., 1993; Wang and Manis, 2006). Structural classifications of BCs into globular bushy cells (GBCs) and spherical bushy cells (SBCs) were mainly drawn based on data from cats (Brawer and Morest, 1975; Osen, 1969; Wu and Oertel, 1984). GBCs are innervated by smaller endings, which were termed as modified endbulbs (Rouiller et al., 1986) and project into the contralateral medial nucleus of the trapezoid body forming the calyx of Held (Ryugo and Spirou, 2009; Borst and Soria van Hoeve, 2012; Joris and Trussell, 2018). SBCs are enveloped by the large endbulbs of Held, as it was found for different vertebrates including rodents (Ryugo and Spirou, 2009; Willard and Ryugo, 1983). They transmit the signal to the contralateral and ipsilateral medial superior olive and to the ipsilateral lateral superior olive (Cant and Casseday, 1986; Warr, 1966). In mice, it remains difficult to distinguish between SBCs and GBCs via their morphology. Rather, regional differences in the synaptic organization of ANF inputs to BCs have been reported (Lauer et al., 2013).

Endbulbs of Held develop from small entities to large finger-like terminals (Limb and Ryugo, 2000; Ryugo and Fekete, 1982; Ryugo and Spirou, 2009). At mature endbulbs, 500–2000 active zones (AZs) (Ryugo et al., 1996) contribute to the high reliability of signal transmission (Babalian et al., 2003; Brenowitz and Trussell, 2001; Nicol and Walmsley, 2002; Pfeiffer, 1966), which is characterized by highly synchronous release (Yang and Xu-Friedman, 2010). Before hearing onset, the auditory pathway exhibits stereotyped bursts of action potentials (Babola et al., 2018; Sonntag et al., 2009; Tritsch et al., 2010a, 2010b), which change to sensory-driven higher spiking rates combined with irregular spiking after hearing onset. Morphologically, endbulbs of Held AZs show presynaptic clusters of clear synaptic vesicles (SVs) and typically a curvature of the postsynaptic density (PSD) (Ryugo et al., 1997). For most synapse types, SVs are sorted in three functional pools: (i) the readily releasable pool (RRP) of SVs, (ii) the recycling pool, and (iii) the reserve pool (reviewed in [Rizzoli and Betz, 2005]). The RRP is widely believed to correspond to the fraction of morphologically docked SVs (Kaeser and Regehr, 2017; Schikorski and Stevens, 2001). Consequently, their number at the AZ determines synapse strength (Murthy et al., 2001; Schikorski and Stevens, 1999; Walmsley et al., 1998). Hence, the determination of ultrastructural SV pools can yield insight into mechanisms of synapse function. The advent of cryo-electron microscopic techniques allowed a novel view on the SV organization at presynaptic AZs. Using these powerful methods, SVs could be assigned to distinct populations based on not only their distance to the AZ membrane but also due to the complement of their filamentous connections i.e., molecular tethers — which play a major role determining SV distance to the AZ membrane (Cole et al., 2016; Fernández-Busnadiego et al., 2010, 2013; Landis et al., 1988; Siksou et al., 2007; Wesseling et al., 2019).

We hypothesize that long-term activity changes due to maturation or sensory deprivation induce alterations of the spatial arrangement and size of morphological SV pools at endbulb AZs. Precisely, (i) we expected that the number of SVs per AZ increases during development to support high transmission fidelity at mature AZs, also taking into account the transition from pre-hearing to hearing, which is accompanied by changes in activity patterns. After hearing onset, a larger amount of SVs might be required to reply to more diverse activity patterns. (ii) We further predict that in deaf Otof^−/− mice the SV numbers at endbulb AZs might be reduced due to the functional impairment caused by the lack of exocytosis at IHC ribbon synapses. (iii) These potential changes in SV numbers might then be restricted to specific SV populations. Finally, (iv) SV diameters are expected to negatively correlate with activity levels as shown for the endbulb of Held upon monaural conductive hearing loss (Clarkson et al., 2016). Therefore, we employed high-pressure freezing (HPF) with subsequent freeze substitution (FS) of acutely prepared AVCN vibratome slices in order to preserve the tissue in a near-native state. We used electron tomography (ET) to investigate the AZs in 3D with the highest axial resolution as it has previously be done for other central synapse preparations (Cole et al., 2016; Imig et al., 2014; Maus et al., 2020; Siksou et al., 2007). This way, distances of SVs as well as the number of docked SVs could be determined unambiguously across the AZ. Hearing onset in mice occurs at the age of P11/12 (Mikaelian and Ruben, 1965). Thus, we chose three different age-groups, pre-hearing (P10), young post-hearing (P21/22), and adult mice (six months) to investigate structural hallmarks of endbulb synaptic maturation. We compared wild-type (wt) C57BL6/J endbulb of Held AZs with AZs of deaf Otof^−/− mice, in which peripheral sound encoding is disrupted while central processing remains unaffected.

We found a steady increase of SV number per AZ in wt, but the average number of docked SVs did not change during maturation. Moreover, specific SV pools at a distance of 5–15 nm from the plasma membrane might contribute to reliable release at wt AZs. In contrast, the number of SVs per AZ strikingly declined in aging Otof^−/− mice. In conclusion, upon a decrease of sound evoked activity, fewer SVs are available at the endbulb of Held presynaptic AZs.

Results

Structural preservation of acutely sliced and cryo-preserved AVCNs

We performed HPF/FS followed by ET to determine the SV distances to the AZ membrane, applying the same categorization criteria as previous studies investigating other synapse types (Cole et al., 2016; Fernández-Busnadiego et al., 2013; Imig et al., 2014; Maus et al., 2020). We focused on the excitatory endbulbs of Held, which harbor so-called asymmetric AZs (Gray, 1959), with a clear PSD (Figure 1B; García-Hernández et al., 2017; Nicol and Walmsley, 2002; O'Neil et al., 2011; Redd et al., 2000). Before subjected to quantification, all samples and tomograms underwent rigorous quality checks in order to determine the freezing quality. Next to the excitatory endbulbs, BCs also receive inhibitory input (Gómez-Nieto and Rubio, 2011; Keine and Rübsamen, 2015; Kuenzel, 2019; Lauer et al., 2013; Spirou et al., 2005). However, these presynaptic terminals are ultrastructurally distinguished from endbulbs as they are considerably smaller (Gulley and Reese, 1981; Spirou et al., 2005), and inhibitory synapses do not exhibit an elaborate PSD (Figure 1C; Colonnier, 1968; Harris and Weinberg, 2012; Tao et al., 2018) and are thus classified as symmetrically synapses (Gray, 1959). Classically, inhibitory synapses have been described to contain flattened SVs (Lauer et al., 2013; Spirou et al., 2005; Tolbert and Morest, 1982). However, after rapid freezing and cryo-electron microscopy (Tao et al., 2018) or FS (Tatsuoka and Reese, 1989), SVs of inhibitory synapses remain spherical. Consequently, we had to thoroughly check for the presence or absence of a PSD to distinguish between excitatory and inhibitory AZs and excluded the latter tomograms from our analysis. Furthermore, only AZs from large terminals were considered, and the presence of parallel pre- and post-synaptic membranes with a regular synaptic cleft served as a required criterion to enable a reliable quantification of SV distances to the AZ membrane (Figure 1).

Figure 1.

Morphological criteria used for AZ quantification

(A) Electron micrograph of a BC with several presynaptic contacts (∗). The large terminal represents an endbulb of Held (∗∗). Scale bar: 1 μm.

(B) Left: example virtual section of an excitatory AZ with a continuous - in HPF/FS samples fuzzy appearing - PSD that were included into the analysis, right: enhanced contrast of the same section, the rim of the PSD is highlighted (yellow line), see also inset.

(C) Left: virtual section of an asymmetrical contact, lacking the PSD, right: contrast enhanced same section as left, as also seen in the inset, a typical PSD is not present. These inhibitory AZs were excluded from the study after inspection of the whole tomogram.

(D and E) The criteria for including a tomogram into the analysis (presence of a PSD, presence of parallel pre- and postsynaptic membranes with a regular synaptic cleft) as well as for the SV quantification are shown on top of a virtual section (D) and in a scheme (E). All SVs (green) within a distance of 200 nm from the AZ membrane (blue) over the extent of the PSD (yellow) that was used to determine the AZ size were included in the analysis. SVs outside of this area (transparent green in [E]) were excluded. The shortest SV (outer leaflet)-AZ membrane distances were divided into 40-nm bins. The SVs from 2 to 40 nm were considered as membrane-proximal SVs. Additionally, from 0 to 40 nm, 5 nm steps were investigated. Morphologically docked SVs were considered separately and are depicted in gray. Scale bars in (B–D): 100 nm.

At wt AZs, the SV numbers increase during maturation

In this study, we set out to investigate (i) the changes of morphological SV pools at endbulb AZs during maturation from pre-hearing to adult stages and (ii) in response to auditory deprivation. Therefore, the morphological SV pools prior to hearing onset (P10), just after hearing onset (P21/22), and of adult (6 months [6M]) wt and Otof^−/− mice were determined (Figure 2A).

Figure 2.

SV numbers of wt and Otof^−/− endbulb AZs during maturation

(A) Single virtual sections and corresponding models of representative tomograms of AZs from P10, P21/22, and 6M wt and Otof^−/− mice showing the AZ membrane (blue), SVs (green), and morphologically docked SVs (gray). Scale bars: 100 nm.

(B) Number of SVs within 200 nm of the AZ membrane. SV numbers are reduced in Otof^−/− mice (wt 6M, 63.52 ± 5.63; Otof^−/− 6M, 39.61 ± 4.68; ∗∗p = 0.0020 and Otof^−/− P22, 66.64 ± 8.73; Otof^−/− 6M, 39.61 ± 4.68; ∗p = 0.0163).

(C) AZ size determined by following the PSD extent in each tomogram.

(D) Number of SVs within 200 nm of the AZ membrane normalized to the AZ area, which is decreased at Otof^−/− AZs (Otof^−/− P22, 726.0 ± 76.14; Otof^−/− 6M, 431.3 ± 46.38; ∗∗p = 0.0099; wt 6M, 801.5 ± 65.67; Otof^−/− 6M, 431.3 ± 46.38; ∗∗∗∗p < 0.0001).

(E) Number of SVs within bins of 5 and 40 nm from the AZ normalized to the AZ area, see also Table 1. Data are presented as box and whisker plots that indicate median, lower/upper quartiles, 10–90th percentiles (B–D), or as bar graph that represent mean ± SEM (E).

∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001. wt P10 (N = 4; n = 24), Otof^−/− P10 (N = 3; n = 22), wt P21 (N = 3; n = 22), Otof^−/− P22 (N = 3; n = 23), wt 6M (N = 3; n = 29), and Otof^−/− 6M (N = 3; n = 23), (N, number of animals; n, number of AZs). All values see Tables S1–S5.

In wt, the total SV number at a distance up to 200 nm from the plasma membrane increased slightly during maturation (Figure 2B; P10 with ∼46 vs. ∼64 at 6M; exact values and statistical details can be found in Table S1). Since tomograms were generated from 250-nm semithin sections, which typically do not include the whole AZ, we normalized the SV counts to the respective AZ area within each tomogram (Figure 2C). A significant increase of the SV numbers per μm² AZ area could be observed comparing P10 (∼477 SVs/μm²) and adult 6M wt endbulb AZs (∼802 SVs/μm²) (Figure 2D; Table S1).

Next, we determined the spatial distribution of SVs per area in membrane proximity, using 5-nm bin sizes from 0 to 40 nm (Figure 1E). We found an accumulation of SVs close to the membrane within 5–15 nm and more distant to the membrane above 80 nm. Interestingly, both fractions increased with advancing age at wt endbulb AZs (Figure 2E; values of the significant findings can be found in Table 1, all values see Table S1).

Table 1.

Significant differences in the distribution of SVs

	Distance to AZ membrane	wt P10	wt P21	wt 6M	p value
wt P10 vs. P21 vs. 6M	5-10 nm	13.00 ± 3.04 (9.17)	29.83 ± 6.18 (21.09)	41.31 ± 5.71 (39.50)	P10 vs. P21: ∗p = 0.0236 P10 vs. 6M: ∗∗∗p = 0.0002
	10-15 nm	18.01 ± 4.17 (13.03)	17.70 ± 3.23 (16.37)	31.66 ± 3.85 (28.60)	P10 vs. 6M: ∗p = 0.0171
	80-120 nm	89.10 ± 10.28 (82.06)	114.6 ± 20.4 (82.79)	154.0 ± 12.56 (165.8)	P10 vs. 6M: ∗∗p = 0.0051
	120-160 nm	68.29 ± 11.34 (68.50)	88.30 ± 16.71 (69.24)	153.2 ± 18.75 (130.4)	P10 vs. 6M: ∗∗p = 0.0010 P21 vs. 6M: ∗p = 0.0181
	160-200 nm	39.61 ± 8.66 (27.61)	78.09 ± 25.73 (35.87)	120.9 ± 16.6 (98.5)	P10 vs. 6M: ∗∗p = 0.0010
		Otof−/−P10	Otof−/− P22	Otof−/−6M
Otof−/− P10 vs. P22 vs. 6M	120-160 nm	106.1 ± 19.4 (66.7)	124.9 ± 18.58 (109.0)	63.03 ± 8.99 (62.64)	P22 vs. 6M: ∗p = 0.0230
	160-200 nm	83.14 ± 17.60 (52.28)	92.32 ± 16.93 (62.86)	33.19 ± 8.12 (16.15)	P22 vs. 6M: ∗p = 0.0151
		wt 6M	Otof−/−6M
6M wt vs. Otof−/−	0-5 nm	71.81 ± 11.68 (49.23)	32.30 ± 7.15 (21.43)		∗∗p = 0.0036
	5-10 nm	41.31 ± 5.71 (39.50)	20.83 ± 4.82 (13.33)		∗p = 0.0110
	10-15 nm	31.66 ± 3.85 (28.60)	18.02 ± 3.52 (14.96)		∗p = 0.0117
	40-80 nm	152.3 ± 13.4 (145.3)	108.4 ± 12.08 (85.71)		∗p = 0.0231
	80-120 nm	154.0 ± 12.56 (165.8)	92.95 ± 13.17 (77.47)		∗∗p = 0.0016
	120-160 nm	153.2 ± 18.75 (130.4)	63.03 ± 8.99 (62.64)		∗∗∗∗p < 0.0001
	160-200 nm	120.9 ± 16.6 (98.5)	33.19 ± 8.12 (16.15)		∗∗∗∗p < 0.0001

The number of SVs within the respective bins is normalized to the AZ area. Data are presented as mean ± SEM (median in brackets). More details about statistics can be found in Tables S1, S2, and S5.

A decrease of SV numbers in aging Otof^−/− mice

Endbulbs of Held AZs in congenitally deaf Otof^−/− mice do not receive auditory input, which results in gross morphological changes of the endbulb terminals as described previously (Wright et al., 2014). In the current study, we report an increase of SV numbers at wt AZs during maturation. At Otof^−/− endbulb AZs, the opposite trend becomes obvious. A significant decrease of the total SV counts and the SV numbers per μm² AZ area was detected when comparing young P22 to adult 6M mutant animals (Figures 2B and 2D; Table S2). However, closer inspection of different SV populations revealed an initial increase of SV numbers within 5–15 nm during the transition from P10 to P22, which declines significantly at 6M of age (Figure 2E; Table S2). Additionally, a significant reduction of SVs at a distance of 80–200 nm was observed in 6M Otof^−/− mice compared to P22 Otof^−/− AZs.

In a direct comparison to the corresponding wt controls, SV counts at 6M Otof^−/− AZs were found to be significantly reduced — though being initially comparable at pre-hearing and early post-hearing stages (Tables S3–S5). This was also observed when normalizing the SV count to the AZ area (Figures 2D and 2E; Tables S3–S5; Table 1).

In order to visualize the overall SV organization, we employed quantification of 10-nm bins throughout 0–200 nm, which revealed for both genotypes that the SV density was high within 0–10 nm to the AZ membrane and decreased with distance. Comparable to the results presented above, upon aging, the number of SVs increased at wt AZs close to the membrane, while at Otof^−/−, an initial increase occurred from P10 to P22 AZs but a strong reduction was observed in adult mutant mice (Figure S1; Tables S6–S10).

In summary, developmental maturation of wt endbulbs of Held is accompanied by an increase in SV numbers at individual AZs. In deaf 6M old Otof^−/− mice, a reduction of SVs was found. Both changes affected mainly two SV populations: (i) SVs at a distance of up to 15 nm to the AZ membrane and (ii) SVs that are located further away (≥80 nm) (SV pool changes are schematized in Figure 3D; Tables S1–S5).

Figure 3.

Docked and membrane-proximal SV at wt and Otof^−/− AZs

(A) Number of morphologically docked SVs (0–2 nm from the AZ membrane) normalized to the AZ area.

(B) Number of SVs within 2–40 nm from the AZ membrane normalized to the AZ area.

(C) Proportion of docked SVs within all SVs from 0 to 40 nm distance. The ratio is smaller in 6M Otof^−/− (Otof^−/− P10, 0.2326 ± 0.0334; Otof^−/− P22, 0.1102 ± 0.0276; ∗p = 0.0168; Otof^−/− P10, 0.2326 ± 0.0334; Otof^−/− 6M, 0.1203 ± 0.0304; ∗p = 0.0308). Data are presented as box and whisker plots that indicate median, lower/upper quartiles, 10–90th percentiles.

(D) Schematic summary of the SV pool changes. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001.

All values see Tables S1–S5.

Decreasing proportion of docked SVs in membrane proximity at adult Otof^−/− endbulb AZs

We distinguished in our analysis between two SV sub-populations, physically docked SVs (0–2 nm from the presynaptic AZ membrane) and membrane-proximal SVs (2–40 nm from the presynaptic AZ membrane, docked SVs are excluded) — analogous to previous studies performing HPF/FS and subsequent ET (Imig et al., 2014; Maus et al., 2020; Watanabe et al., 2013). These definitions are based on different properties of the SVs, (i) physically docked SVs reduce upon stimulation, as demonstrated, for example, in experiments that combined optogenetic stimulations with rapid freezing and ET (Watanabe et al., 2013). Moreover, previous ET or cryo-ET studies on different synapse types identified a specific population of SVs, which are connected via one single long tether of a maximum length of 45 nm to the AZ membrane (Cole et al., 2016; Fernández-Busnadiego et al., 2010, 2013; Siksou et al., 2007, 2009).

At wt endbulbs of Held, the number of docked SVs per μm² AZ area remained unaltered during maturation from pre-hearing to adult animals, while the number of membrane-proximal SVs steadily increased (Figures 3A and 3B). At Otof^−/− AZs, the docked SV count was similar in all age-groups, but membrane-proximal SVs declined significantly at 6M of age (Figure 3B). Comparing both genotypes, significantly fewer docked SVs were observed at 6M Otof^−/− endbulb AZs, which harbored on average ∼22 docked SVs/μm² compared to ∼37 in age-matched wt (Figure 3A). Finally, we analyzed the proportion of docked SVs relative to all SVs within 40 nm of the AZ membrane. This proportion decreased continuously upon maturation in both genotypes, which was statistically significant for Otof^−/− P10 vs. 6M AZs, reducing from 0.23 to 0.12 (Figure 3C; Table S2).

The frequency distribution of docked SVs remained unaltered

Consistent with our observations, a previous study had already reported a highly variable number of docked SVs at endbulb AZs (Nicol and Walmsley, 2002). The total number of docked SVs per endbulb AZ was only reduced in Otof^−/− 6M compared to wt 6M (Figure 4B), but their number ranged from 0 to 14 docked SVs at wt P10 AZs and from 0 to 16 at P22 Otof^−/− AZs. However, most AZs of wt and Otof^−/− endbulbs of all age-groups harbored 0–3 docked SVs (Figure 4D) as also shown in representative AZ top views of 3D models (Figure 4E). We analyzed the frequency distribution comparing maturation in wt and in Otof^−/− mice as well as between genotypes. The corresponding density and relative frequency distributions are depicted in Figures 4C and 4D, which were not significantly different (Table S11).

Figure 4.

Frequency distribution of docked SVs

(A) Scheme of morphologically docked SVs (0–2 nm from the AZ membrane).

(B) Number of docked SVs per AZ. See also Tables S1–S5. The number of docked SVs is lower at 6M Otof^−/− compared to 6M wt (wt 6M: 2.97 ± 0.51; Otof^−/− 6M: 1.83 ± 0.56; ∗p = 0.0379). Data are presented as box and whisker plots that indicate median, lower/upper quartiles, 10–90th percentiles.

(C) Density distribution of the total number of docked SVs per AZ for all conditions.

(D) Histograms showing the relative frequency of AZs with 0–16 docked SVs at wt P10, P21, and 6M as well as at Otof^−/− P10, P22, and 6M AZs.

(E) Top views of representative 3D models of AZ membranes with their most frequent number of docked SVs. Scale bars: 100 nm.

The number and distribution of membrane-proximal SVs remained largely unaffected

We next investigated the relative frequency distribution of the membrane-proximal SV pool (2–40 nm). At wt AZs, the membrane-proximal SV numbers remained unaltered at all ages and compared to Otof^−/− AZs (Figure 5B). However, the difference between P10 Otof^−/− AZs with ∼9 membrane-proximal SVs in comparison to P22 AZs with ∼15 SVs reached significance. The density and relative frequency distributions are shown in Figures 5C and 5D and the application of an F-test revealed a significantly broader distribution of membrane-proximal SVs in Otof^−/− P22 compared to P10 and 6M (Table S11).

Figure 5.

Frequency distribution of membrane-proximal SVs

(A) Scheme of membrane-proximal SVs (2–40 nm from the AZ membrane).

(B) Number of membrane-proximal SVs per AZ. Values see Tables S1–S5. Data are presented as box and whisker plots that indicate median, lower/upper quartiles, 10–90th percentiles. Membrane-proximal SVs increase at Otof^−/− between P10 and P22 AZs (Otof^−/− P10, 8.86 ± 1.13; Otof^−/− P22, 15.39 ± 2.11; ∗p = 0.0104).

(C) Density distribution of the total number of membrane˗proximal SVs per AZ for all conditions.

(D) Histograms showing the relative frequency of AZs with 0–44 SVs within 2–40 nm at wt and Otof^−/− of all ages.

SV diameters are decreasing upon maturation at Otof^−/− endbulb AZs

Acoustic deprivation was previously shown to decrease SV diameters at endbulbs of Held (Clarkson et al., 2016). Therefore, we compared the SV diameters in both genotypes and all age-groups (Figure 6, Tables S1–S5).

Figure 6.

SV diameters change in Otof^−/− endbulb AZs

(A, C, and E) Mean diameters averaged per tomogram for each condition. The mean diameters of all SVs decrease in Otof^−/− mice (A, Otof^−/− P10, 54.87 ± 0.90; Otof^−/− 6M, 50.63 ± 1.03; ∗p = 0.0128), as well as of morphologically docked (C), and membrane-proximal (E) SVs per AZ. All values see Tables S1–S5. Data are presented as box and whisker plots that indicate median, lower/upper quartiles, 10–90th percentiles. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001. wt P10 (N = 4; n = 24), Otof^−/− P10 (N = 3; n = 22), wt P21 (N = 3; n = 22), Otof^−/− P22 (N = 3; n = 23), wt 6M (N = 3; n = 29), and Otof^−/− 6M (N = 3; n = 23), (N, number of animals; n, number of AZs).

(B, D, and F) SV diameters were divided in 5 nm bins and the frequency distribution of SV diameters from all (B), morphologically docked (D), and membrane-proximal (F) SVs is shown as histograms.

Indeed, in Otof^−/− mice, the SV diameters decreased in the whole SV population from ∼55 nm in P10 to ∼50 nm at 6M of age (Figure 6A). A separate analysis of SV diameters of docked and membrane-proximal SVs revealed that this decrease appears in both populations (Figures 6C and 6E; Table S2). When comparing both genotypes, significantly larger SVs were found at Otof^−/− P10 compared to age-matched wt. Again, this applies to all analyzed SV populations (Figures 6A, 6C, and 6E; Table S3). Finally, at 6M Otof^−/− endbulb AZs, docked SVs were found to be significantly smaller (∼46 nm) compared to age-matched wt (∼50 nm) (Figure 6C; Table S5). The frequency distributions of SV diameters are shown in Figures 6B, 6C, and 6F.

In summary, we found significant changes in SV numbers at wt and mutant endbulb of Held AZs that depend on the developmental stage. In wt, maturation resulted in an increase of SVs at endbulb AZs. This increase was not evenly distributed over the full quantified spatial range; instead, a prominent SV pool accumulated at 5–15 nm. Moreover, SVs further away (≥80 nm) to the membrane were also increased with age, indicating the presence of different morphological SV pools at endbulb AZs. The opposite trend was found at Otof^−/− endbulbs, in that there was a reduction of SVs within 5–15 nm. In comparison to wt, the docked SV numbers decreased in adult mutants. Finally, the overall SV diameters decreased at Otof^−/− AZs, while at wt, maturation did not affect the diameters. In conclusion, during maturation, ultrastructural parameters show opposite trends when comparing wt to congenitally deaf Otof^−/− endbulb AZs. Therefore, the lack of Otof upstream of the CN influences morphological SV pools at individual AZs.

Discussion

In the current study, we tested the hypothesis that functional changes are directly reflected in the presynaptic morphology of the endbulb of Held AZs. Using state-of-the-art ultrastructural methods, we could show that corrupted sound encoding in the cochlea also affects downstream SV numbers and their spatial arrangement at the first central synapse of the auditory pathway. Other key findings of our analysis included increased SV numbers upon maturation in wt and a clear reduction in SV numbers upon a longer period of deafness in Otof^−/− mice, which was accompanied by smaller SVs in adult mutants.

SV populations at the endbulb of Held AZs

It has been proposed that the SV pool sizes increase during development in the calyx of Held (Iwasaki and Takahashi, 2001; Taschenberger et al., 2002; Taschenberger and von Gersdorff, 2000). Recent serial 3D reconstructions of developing calyces of Held comparing P7 to P21 revealed smaller AZs with a higher SV density in P21 mice (Thomas et al., 2019). In several aspects, the calyx of Held is comparable to the “smaller cousin”, the endbulb of Held. Both synapses are part of the auditory pathway and perform ultrafast and temporally precise neurotransmission (Lin et al., 2011). Both comprise large presynaptic terminals that contain hundreds of individual AZs (Nicol and Walmsley, 2002; Ryugo et al., 1996; Sätzler et al., 2002; Wichmann, 2015). In the current study, we found significantly more SVs at adult wt endbulb of Held AZs compared to pre-hearing AZs, which consisted of two distinct SV populations: (i) SVs at a distance of 5–15 nm and (ii) ≥80 nm from the AZ membrane. The 5–15 nm SV population was previously described as a pool of SVs, which are tethered to the AZ membrane via the neuronal soluble N-ethylmaleimide-sensitive factor attachment receptor complex (Cole et al., 2016; Imig et al., 2014; Maus et al., 2020). Studies of hippocampal synapses revealed that this pool was pronounced in docking and priming-deficient mutants, which specifically lack morphologically docked SVs (Imig et al., 2014; Siksou et al., 2009). In Munc13-deficient synapses, SVs accumulated at distances of 8–10 nm from the AZ and for SNAP-25- or Syb-2-deficient synapses from ∼4–8 nm (Imig et al., 2014). Moreover, this specific SV pool was interpreted as a supply pool of tethered or loosely docked SVs that likely contribute to synaptic short-term plasticity characteristics of mossy fiber synapses (Maus et al., 2020). Finally, using cryo-ET, SVs connected via multiple ≥5 nm filaments were less abundant at Rab3-interacting molecule (RIM)1-deficient AZs (Fernández-Busnadiego et al., 2013). RIM is known as a critical organizer of SVs in membrane proximity that acts independently of Munc13-1 (Fernández-Busnadiego et al., 2013; Zarebidaki et al., 2020). Therefore, it is conceivable that an accumulation of this specific SV population at 5–15 nm at adult wt endbulb AZs might contribute to an efficient SV recruitment that is required to support reliable neurotransmission with exceedingly high membrane turnover rates.

Docked SVs at wt endbulb AZs during maturation

At wt endbulb of Held AZs, docked SVs numbers and densities remained remarkedly stable throughout maturation. We observed ∼3 docked SVs at both, P21 and 6M AZs and ∼35–37 docked SVs per μm² for all ages. In contrast to previous studies that employed aldehyde-based fixation protocols and a serial reconstruction approach in P25 rats and reported ∼15 docked SVs per AZ and 54–83 SVs/μm² (Nicol and Walmsley, 2002), we found fewer docked SVs at murine endbulb of Held AZs. This divergence might be explained by either a species difference and/or the different methodologies used: Serial reconstructions have the undisputed advantage that whole AZs can be reconstructed that clearly exceed the volume of our 250 nm tomograms; however, ET enables a higher resolution along the z axis (Harlow et al., 1998).

Consistent with a previous report (Nicol and Walmsley, 2002), we observed highly variable docked SV numbers in all experimental groups. In vitro electrophysiological experiments on endbulbs of Held revealed a high variability of functional properties such as the synaptic release probability (P_r) (Oleskevich et al., 2000), which was also observed at the calyx of Held (Fekete et al., 2019). Early on, it was proposed that synaptic P_r may be influenced by the number of docked SVs per AZ (Schikorski and Stevens, 1999; Walmsley et al., 1998), presumably scaling linearly at different synapse types (Branco et al., 2010; Holderith et al., 2012; Murthy et al., 2001). Recent work on hippocampal mossy fiber synapses and Schaffer collateral CA1 synapses revealed a correlation between initial synaptic P_r of the respective synapse type and the number of docked SVs (Maus et al., 2020). At endbulbs of Held AZs, synaptic P_r mainly depends on activity, as shown in a study that investigated short-term depression at endbulbs in different acoustic environments (Zhuang et al., 2020). Thus, it is tempting to speculate that a large range of docked SVs numbers, as found in our study, might contribute to the variability in synapse P_r as found in a previous functional study on endbulbs (Oleskevich et al., 2000).

We want to mention further factors that could influence the variability in SV numbers. The method used in this study might have its limitations. Tomograms from 250-nm sections typically do not include full AZs. To exclude potential errors due to different AZ proportions, we therefore normalized the SV numbers to the AZ area. We approached all our samples in a comparable way and sampled over larger AVCN areas, which ensured that tomograms were taken from different tissue regions. However, properties of the ANFs that project onto BCs might have a greater contribution to variable SV numbers. ANFs can be subdivided functionally and biochemically into three distinct groups that have different sensitivities and spontaneous firing rates (SRs) (Liberman, 1978, 1991; Pfeiffer and Kiang, 1965; Shrestha et al., 2018; Sun et al., 2018). Within the AVCN, SBCs are innervated from all groups, namely low, medium, and high SR ANFs (Liberman 1991), which we cannot distinguish morphologically in our study. Finally, C57BL/6 mice are known to show progressive hearing loss starting at the age of three months (Kujawa and Liberman, 2019; Stamataki et al., 2006). The hearing loss first affects the basal and mid-region of the cochlea and a degeneration of ANFs and ribbon synapses can be detected lastly in the apical region (Stamataki et al., 2006). Hence, age-related changes in the innervation pattern of the CN might be present in our 6M age-group and might contribute to AZ diversity.

Alterations of SV pools at Otof^−/− endbulb of Held AZs

Spontaneous activity within the auditory system prior to hearing onset shapes the pathway during its development (Sonntag et al., 2009; Tritsch et al., 2010a, 2010b). This non-sensory spontaneous activity seems to be one requirement to build functional AZs (Kerschensteiner, 2014) and is preserved in immature VGlut3 KO mice, which serve as a model of deafness that lacks IHC-driven activity (Babola et al., 2018). However, type I fibers can be classified molecularly in three subtypes (type Ia, b, c) (Shrestha et al., 2018), and a recent study demonstrated that the subtypes converged at different ratios onto individual target cells (Wang et al., 2021). The molecular differentiation was shown to be disrupted in VGlut3 KO mice (Shrestha et al., 2018). In our present study, we took advantage of deaf Otof^−/− mice to investigate the consequences of sensory deprivation on the endbulb of Held individual AZs without directly altering the molecular organization of the endbulb AZs. Otof is essential for IHC ribbon synapse exocytosis and its expression starts from the embryonic stage e16 onwards, with its peak in immunofluorescence at the age of P6 (Roux et al., 2006). At P4, Ca²⁺⁻evoked exocytosis in mouse IHC switches from an Otof-independent to an Otof-dependent mechanism (Beurg et al., 2010). Therefore, the complete lack of Otof — and the concomitant abolishment of IHC synaptic sound encoding — throughout maturation may exert structural and functional consequences on the subsequent centers of the auditory pathway. The deletion of Otof dramatically reduces, albeit does not fully abolish the spontaneous miniature excitatory postsynaptic current (mEPSC) frequency, as measured in P9 mice using single bouton recordings at IHC afferent fibers (Takago et al., 2019). Despite this residual activity in these mutants, we could demonstrate a reduction of specific SV populations at individual endbulb Otof^−/− AZs in adult mice. This matches ultrastructural studies on other deafness models. In 7-month-old shaker-2 mice, next to larger PSDs, fewer SVs were found compared to age-matched hearing controls (Lee et al., 2003). We observed in Otof^−/− mice significantly fewer docked SVs and a strong reduction of the 5–15 nm SV pool but also SVs at a larger distance to the AZ appeared partially depleted. Additionally, the proportion of docked SVs in membrane proximity was reduced at adult Otof^−/− AZs. Taken together, at endbulb AZs of Otof-deficient mice, synaptic transmission might be less reliable due to a reduction of several SV populations. Indeed, recordings from BCs in Otof^−/− mice showed a stronger depression as well as significantly higher synaptic failure rates (Wright et al., 2014).

SV diameters alter at Otof^−/− endbulb AZs

As a last ultrastructural parameter, we measured the SV diameters, which were previously shown to change upon the lack of activity. Diversity in SV diameters might speak for immaturity or a defective SV formation as shown for a plethora of synapses such as for mammalian ribbon synapses (Strenzke et al., 2016), hippocampal synapses (Imig et al., 2014), and Drosophila neuromuscular junctions (Stevens et al., 2012). Moreover, a defective regulation of autophagy can result in larger, more variable SV sizes at synapses (Lüningschrör et al., 2017). A study on monaural conductive hearing loss in rats revealed smaller SVs in earplugged animals, which again increased in size after earplug removal (Clarkson et al., 2016). At endbulbs of Held of normal hearing P20, P44 to P100 mice, SV sizes increased upon development, which was arrested in knockouts for the essential glutamate receptor subunit GluA3 (Antunes et al., 2020). We found that SV diameters in wt remained stable throughout maturation, while in Otof^−/− endbulbs of Held a decrease in SV sizes was observed. This decrease also applied to docked SVs; therefore, a smaller quantal size could be assumed at Otof^−/− endbulbs. However, in Otof^−/− mice, spontaneous mEPSCs in BCs were found to be significantly larger compared to controls (Wright et al., 2014), which partially contradicts our findings on SV sizes. In contrast, occlusion to abolish auditory input had no significant effect on mEPSC amplitude as recorded at P23 endbulbs (Zhuang et al., 2017) but altered other endbulb properties such as an elevation of P_r (Zhuang et al., 2017, 2020). It should be mentioned that quantal size differences do not always correlate to the SV sizes (Taschenberger et al., 2002) because other mechanisms such as vesicular glutamate concentration can contribute to the measured quantal size (Edwards, 2007; Wu et al., 2007). Future studies are required to tackle these findings using correlative structure-function approaches.

Limitations of the study

In our study, we cannot distinguish which functional properties the analyzed endbulbs of Held initially had (low, medium, high spontaneous rate fibers), which might have contributed to a certain variability of the data. We also did not provide cell physiological experiments along with our morphological findings. Future studies are required to decipher these questions.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Carolin Wichmann (carolin.wichmann@med.uni-goettingen.de).

Materials availability

This study did not generate any new unique reagents.

Data and code availability

Does not apply for this study.

Methods

All methods can be found in the accompanying Transparent methods supplemental file.

Published: April 23, 2021