Supplementary Information for

Mapping developmental maturation of inner hair cell ribbon synapses in the apical mouse cochlea

S. Michanski1,2, K. Smaluch2,3,10*, A. M. Steyer4,5*, R. Chakrabarti1,2, C. Setz2,3,6,10, D. Oestreicher3, C.

Fischer7, W. Möbius4,5, T. Moser2,3,5,8,9, C. Vogl3,9,10#, C. Wichmann1,2,9#

# Correspondence should be addressed to:

Carolin Wichmann

Molecular Architecture of Synapses Group, Institute for Auditory Neuroscience and InnerEarLab,

University Medical Center Göttingen, 37075 Göttingen, Germany

Tel.: +49 (0) 551 39-61128, Email: carolin.wichmann@med.uni-goettingen.de

Or:

www.pnas.org/cgi/doi/10.1073/pnas.1812029116

Christian Vogl

Presynaptogenesis and Intracellular Transport in Hair Cells Junior Research Group, Institute for

Auditory Neuroscience and InnerEarLab, Auditory Neuroscience Group, Max Planck Institute for

Experimental Medicine, 37075 Göttingen, Germany

Tel.: +49 (0) 551 39-61945, Email: christian.vogl@med.uni-goettingen.de

This PDF file includes:

Supplementary text: Materials and Methods

Figs. S1 to S7

Tables S1 to S3

Captions for movies S1 and S2

References for SI reference citations

Other supplementary materials for this manuscript include the following:

Movie S1

Movie S2

Supplementary Information Text

Materials and Methods.

Animals C57/BL6 wild-type mice of either sex between the embryonic age (E)14 and the postnatal day (P)48

were sacrificed by decapitation for immediate dissection of the organs of Corti.

Immunohistochemistry, confocal and super-resolution STED microscopy

Acutely dissected cochlear explants from the apical coil were briefly fixed with 4% formaldehyde

containing phosphate-buffered saline on ice as previously described [1], [2]. After extensive washing

and a blocking step with a goat serum containing buffer (16% normal goat serum, 450 mM NaCl, 0.3%

Triton X-100 and 20 mM phosphate buffer at pH 7.4), the following primary antibodies were applied

overnight at 4°C: mouse monoclonal anti-PSD95 (clone 7E3-1B8; Cat.-Nr. P246; Sigma Aldrich) and

anti-CtBP2 (Cat.-Nr. 612044; BD Biosciences), chicken anti-Calretinin (Cat.-Nr. 214 106; Synaptic

Systems) and anti-Homer1 (Cat.-Nr. 160 006; Synaptic Systems), as well as rabbit polyclonal anti-α-

tubulin (Cat.-Nr. 302 203; Synaptic Systems), anti-Ribeye A (Cat.-Nr. 192 103; Synaptic Systems), anti-

KIF1a (Cat.-Nr. ab180153; Abcam), anti-KIF2a (Cat.-Nr. NBP2-55116; Novus Biologicals), anti-

KIF5a-c (Cat.-Nr. ab62104; Abcam) and anti-Aczp18p19 ([3]; a kind gift from Prof. M. Kilimann,

Department of Molecular Neurobiology, Max-Planck-Institute for Experimental Medicine Göttingen).

For visualization, secondary goat anti-chicken Alexa-488 (Cat.-Nr. A11039; Thermo Fisher Scientific),

goat anti-mouse Abberior STAR-580 and goat anti-rabbit Abberior STAR-635p-conjugated antibodies

(Cat.-Nr. 2-0002-005-1 and 2-0012-007-2; Abberior GmbH) were applied for 1 h at room temperature.

After mounting the specimen in Mowiol, image acquisition was performed on an Abberior Instruments

Expert Line STED microscope (Abberior Instruments GmbH; based on an Olympus IX83 inverted

microscope) in confocal and/or STED mode using a 1.4 NA UPlanSApo 100x oil immersion objective.

We employed 561 nm and 640 nm laser lines for excitation and a pulsed 775 nm laser for stimulated

emission depletion. Image stacks were acquired with Imspector Software, xy pixel sizes of 80 x 80 nm

and step sizes of 200 nm (confocal). Pixel sizes in 2D-STED equated to 15 x 15 nm in xy.

Conventional embedding and transmission electron microscopy

Conventional embedding of organs of Corti was largely performed as previously described [4]. In brief,

the apical turn of organs of Corti were processed from mice ranging in age from E14 to P48 in phosphate-

buffer saline (PBS). Specimens were fixed immediately after dissection with 4% paraformaldehyde and

0.5% glutaraldehyde in PBS (pH 7.4) for 1 h on ice followed by a further fixation step overnight with

2% glutaraldehyde in 0.1 M sodium cacodylate buffer (pH 7.2) on ice. Subsequently, samples were

washed in 0.1 M sodium cacodylate buffer and treated with 1% osmium tetroxide (v/v in 0.1 M sodium

cacodylate buffer) for 1 h on ice. After osmium tetroxide treatment, samples were washed twice in 0.1

M sodium cacodylate buffer on ice for 10 min, respectively and further in distilled water (three times

for 5 min each, on ice). Next, en bloc staining with 1% uranyl acetate (v/v in distilled water) was

performed for 1 h on ice and samples were briefly washed three times in distilled water, dehydrated in

an ascending concentration series of ethanol, infiltrated and embedded in epoxy resin (AGAR-100,

Plano) to get finally polymerized for 48 h at 70°C. From the cured resin blocks, ultrathin sections (70-

75 nm) were cut with an Ultracut E microtome (Leica Microsystems) or an UC7 microtome (Leica

Microsystems) equipped with a 35° diamond knife (Diatome AG), mounted on 1% formvar-coated (w/v

in water-free chloroform) copper slot grids (ATHENE, 3.05 mm Ø, 1 mm x 2 mm; Plano) and

counterstained with uranyl acetate replacement solution (EMS, Science Services GmbH) and Reynold’s

lead citrate. Thereafter, sections were examined at 80 kV using a JEM1011 transmission electron

microscope (JEOL) and micrographs were acquired at 10,000-x magnification with a Gatan Orius

1200A camera (GATAN GmbH, using the Digital Micrograph software package).

Electron tomography

Electron tomography was performed according to Strenzke et al. [5]. 250 nm conventionally embedded

semithin sections were cut on an Ultracut E or an UC7 ultramicrotome (Leica Microsystems) with a 35°

diamond knife (Diatome AG). Sections were collected on 1% formvar-coated (w/v in water-free

chloroform) copper 100 mesh grids (ATHENE, 3.05 mm Ø; Plano) and counterstained as described

above. Afterwards, 10 nm gold beads (British Bio Cell) were applied to both sides of the grids to act as

fiducial markers. Tilt series image acquisition was conducted mainly from -60 to +60° with 1°

increments at 10,000-x magnification using a JEM2100 (JEOL) transmission electron microscope at 200

kV with the Serial-EM software [6]. For tomogram alignments, the IMOD software package etomo was

used with the fiducial-guided alignment mode and tomographic reconstructions were generated using

3dmod [7].

Immunogold pre-embedding

Immunogold pre-embedding labeling (Triton X protocol) was adapted from Nieratschker et al. [8].

Organs of Corti were dissected and depending on the protein of interest the following two different

protocols were utilized.

Triton X protocol:

Fixation of the organs of Corti was done with 2% paraformaldehyde and 0.06% glutaraldehyde in PEM

(0.1 M PIPES, 2 mM EGTA, 1 mM MgSO4 x 7 H2O, v/v) for 90 min on ice. After fixation, the samples

were washed twice in PEM for 15 min each and blocked for 1 h in 2% bovine serum albumin (BSA)/

3% normal horse serum (NHS) in 0.2% PBST (0.2% Triton X-100 diluted in PBS, v/v). Next, samples

were incubated with the rabbit anti-piccolo primary antibody (Aczp18p19, polyclonal [3]; 1:500 diluted

in 0.1% PBST), detecting the long and the short (piccolino) isoform of piccolo for 1 h at room

temperature and overnight at 4°C. Subsequently, specimens were washed four times with 0.1% PBST

for 1 h each and incubated for 2 h with the 1.4 nm gold-coupled anti-rabbit secondary antibody

(Nanogold-anti-rabbit, Nanoprobes; 1:30 diluted in 0.1% PBST) followed by another washing step in

0.1% PBST for 30 min and overnight at 4 °C. Further washing steps were performed in 0.1% PBST

(two times for 30 min each) until samples were post-fixated with 2% glutaraldehyde in PBS (v/v) for 30

min and briefly washed four times in distilled water. For silver enhancement, the HQ Silver-

enhancement kit (Nanoprobes) was used for 3 min in the dark and specimens were briefly washed for

additional four times. Further fixation was obtained by the treatment with 2% osmium tetroxide (v/v in

0.1 M cacodylate buffer) for 30 min followed by one washing step in distilled water for 1 h and two

washing steps in distilled water for 30 min, respectively. Finally, samples were dehydrated, embedded

in epoxy resin and further processed as described before for the conventional embedding.

Saponin protocol:

Here, organs of Corti were fixed with 4% paraformaldehyde in PBS for 35 min on ice followed by a

brief washing step in PBS and a permeabilization step with 0.05% saponin for 45 min. After washing in

PBS, samples were first blocked with 2% BSA/ 3% NHS in PBS for 1 h and then incubated with the

mouse anti-CtBP2 primary antibody (CtBP2, monoclonal, BD Biosciences; 1:200 diluted in 5% NHS

in PBS) for 1 h at room temperature and overnight at 4°C. Thereafter, specimens were washed three

times in PBS for 1 h each and incubated for 2 h with the 1.4 nm gold-coupled anti-mouse secondary

antibody (Nanogold-anti-mouse, Nanoprobes; 1:30 diluted in 5% NHS in PBS). Subsequently, several

washing steps were performed in PBS (for 30 min, overnight at 4°C and two times for 30 min each)

until samples were post-fixated with 2% glutaraldehyde in PBS (v/v) for 30 min and briefly washed four

times in distilled water. Further processing of the samples like the silver enhancement, osmium tetroxide

treatment, dehydration and embedding in epoxy resin was performed as described in the Triton X

protocol.

Serial block face (SBF)- and focused ion beam (FIB)-scanning electron microscopy (SEM)

Enhanced en bloc staining was performed according to Deerinck et al. [9]. After fixation (see

conventional embedding), samples were treated with a solution of 1.5% potassium ferrocyanide and 4%

osmium tetroxide (v/v in 0.1 M sodium cacodylate buffer) for 1 h on ice. Next, specimens were briefly

washed (five times) in distilled water and placed in a freshly prepared thiocarbohydrazide solution for

20 min followed by five additional washing steps in distilled water. A second exposure to 2% osmium

tetroxide (v/v in 0.1 M sodium cacodylate buffer) was conducted followed by five brief washing steps

in distilled water before the sample was placed in 2.5% uranyl acetate (v/v in distilled water) overnight

at dark. Subsequently, samples were washed in distilled water and contrasted with Reynold’s lead citrate

for 30 min at 60°C to be finally washed once again in distilled water, dehydrated in increasing ethanol

concentrations, infiltrated and embedded in Durcupan (25%, 50%, 75% Durcupan in Acetone for 1 h

each and 100% Durcupan overnight; Sigma Aldrich) to get polymerized for 48 h at 60°C. 3D image

acquisition of polymerized samples was performed either with a SBF-SEM or FIB-SEM.

The setup for SBF-SEM comprises a FEI Quanta 250 FEG scanning electron microscope (FEI) equipped

with a Gatan 3View (GATAN Inc.) automatic microtome. Samples were coated with a gold layer for

300 sec using the Balzers SCD 050 sputter coater in order to limit charging by the electron beam during

SEM imaging. The system was set to cut sections with 80 nm thickness at activated oscillator. Image

acquisition was done at 30 Pa chamber pressure at 2.50 kV with spot size 3.5, a dwell time of 7 µs/pixel

and an image size of 4,096 x 4,096 pixels. The magnification was set to 7,034 x resulting in images with

a pixel size of 4.9 nm.

For FIB-SEM, samples were trimmed with a 90° diamond trimming knife (Diatome AG) in a way that

the region of interest was very close to the surface. The blocks were attached to the SEM stub (Science

Services GmbH, Pin 12.7 mm x 3.1 mm) by a silver filled epoxy (Epoxy Conductive Adhesive, EPO-

TEK EE 129-4; EMS) and polymerized at 60° overnight. Samples were coated with an 8 nm platinum

layer using the sputter coating machine EM ACE600 (Leica Microsystems) at 35 mA current.

Afterwards, samples were placed into the Crossbeam 540/Crossbeam 340 FIB-SEM (Carl Zeiss

Microscopy GmbH) and positioned at an angle of 54°. To ensure even milling and to protect the surface,

a 300 nm carbon layer was deposited on top of the region of interest. Atlas 3D (Atlas 5.1, Fibics, Canada)

software was used to collect the 3D data. Samples were exposed with a 30 nA current and a 7 nA current

was used to polish the surface. The images were acquired at 1.5 kV with the ESB detector (1100 V ESB

grid, pixel size x/y 3 nm) in a continuous mill and acquire mode using 1.5 nA for the milling aperture

(z-step 5 nm). P9 apical dataset: SEM: 60 µm aperture at 1.5 kV, InlensDuo detector at 400 V, pixel

size x/y 5 nm, FIB: 3 nA current, slice thickness 10 nm.

Data analysis and statistics

Quantitative analysis of electron microscopic random sections was performed with ImageJ as follows:

For synaptic vesicles at the ribbon, the first layer around the ribbon with a maximum distance of 80 nm

from the vesicle membrane to the ribbon were counted per section including vesicles at the AZ-

membrane (Fig. S2A, A'), random section analysis according to [4]. The horizontal and vertical axis

were measured and averaged to calculate the mean synaptic vesicle diameter. For ribbon size, the height

and width were measured taking the longest axis of the ribbon excluding the presynaptic density as well

as manual tracing of the synaptic ribbon was performed in order to determine the ribbon area. To

quantify the occurrence of floating ribbon precursors, we defined “floating ribbons” by a lack of any

physical contact to the presynaptic density, while an “anchored ribbon” exhibited a clear membrane

attachment via a presynaptic density. In addition, the number of the postsynaptic density was counted

and the length was measured along the AZ-membrane.

Semi-quantitative analysis of pre-embedding immunogold data

For quantifications only cross sectioned ribbon synapses with a clear visible IHC AZ membrane were

selected. Micrographs were adjusted to the size of the ribbon scheme without changing the ratio.

Afterwards, different sized gold particles were marked by big, medium or small gray circles. Finally,

overlay of multiple ribbons created a map of the protein disposition.

Quantitative analysis of SBF- and FIB-SEM data

The FIB-SEM data were inverted, cropped and then aligned using the Plugin “Linear Stack Alignment

with SIFT” in Fiji. A smoothing function in Fiji was applied (3x3), followed by local contrast

enhancement using a CLAHE plugin in Fiji.

SBF- and FIB-SEM datasets were reconstructed semi-automatically using 3dmod of the IMOD software

package [6]. Using the imodinfo function of 3dmod, information about the ribbon volume was given.

Distance measurements were performed with the measurement drawing tool along the x, y and z-axis.

To separate modiolar and pillar sides for ribbon localization analysis, the cellular axes were determined

as follows: (i) the cochlear apical-basal axis was defined based on the tissue context, e.g. the position

and orientation of adjacent IHCs; subsequently, (ii) to establish the pillar-modiolar axis, a second line

was drawn perpendicular to the apical-basal axis, starting from the center of the nucleus and ending at

the most basal point of the IHC, thereby allowing the 3D-alignment of the respective IHC. Finally, the

localizations of all synaptic ribbons were transposed into 2D projections for pillar-modiolar assignment

and further analysis.

Data were analysed using Excel, Igor Pro 6 (Wavemetrics) and Java. Normality was assessed with the

Jarque-Bera test and equality of variances in normally distributed data was assessed with the F-test.

One-way ANOVA test followed by Tukey’s test was used to calculate statistical significance in multiple

comparisons for normally distributed data or in case of non-normally distributed data the Kruskal-Wallis

(KW) test followed by non-parametric multiple comparisons test (NPMC) was used.

Employing the analysis tool based on Java Statistical Classes library (JSC) [10] applied in our previous

study [11], we performed the KW test for the identification of significant differences between synaptic

vesicle diameters. Non-significant differences between samples are indicated as n.s., significant

differences are indicated as * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001.

Supplemental Figures

Fig. S1. Random section analysis showed first afferent fiber contact with IHCs at E16. (A-A’)

Representative electron micrograph of an E14 IHC. At this developmental stage, no afferent contacts

could be detected. (B-B’) Afferent fibers (aff) start contacting IHCs at ~E16; however, no ribbons, SVs

or PSDs could be found at this age. (C-C’) In E18 IHCs, afferent fibers have established contact sites

with morphologically distinguishable PSDs (blue arrowheads), presynaptic ribbons and SVs. Inset

depicts two attached ribbons (red dashed lines), which tether SVs. Scale bar in C’ inset: 200 nm. Black

dashed lines outline IHC plasma membranes, blue dashed lines indicate afferent fiber contacts in B’ and

C’.

Fig. S2. SV number is positively correlated with ribbon size. (A, A’) Electron micrographs depicting

the performed analysis routine to assess ribbon size and SV counts per ribbon from EM random ultrathin

sections. All SVs (yellow) were counted in the first layer (≤ 80 nm) around the ribbon also including

those at the AZ-membrane. SV diameter was calculated by the average of the horizontal and vertical

axis. Ribbon size was measured in height, width and area, as indicated by the red arrows and dashed

line. (B, C) Quantifications of ribbon height and width across all age groups depict an increase in ribbon

size upon maturation. N=animal number, n=ribbon number. Data are presented as box plots showing

10, 25, 50, 75 and 90th percentiles with individual data points overlaid (****p < 0.0001). For more

detailed statistical evaluation see Table S1. (D, E) Representative images of synaptic ribbons with a

hollow, i.e. electron-translucent, core at the indicated ages. (F) Quantification of the occurrence of

ribbons with a translucent core across all analyzed age groups. (G) A positive correlation is shown

between the number of SVs and the ribbon area. Black dashed line represents a global linear fit with the

correlation coefficient r = 0.917. For better visualization, color-coded lines highlight the data

distribution for each age group, respectively. (H) Cumulative probability distributions of SV diameters

illustrating the gradual reduction in SV size between immature and mature age groups.

Fig. S3. Postsynaptic developmental refinement leads to an increase in PSD size and formation of a

continuous, single PSD in SGN boutons. (A, B) PSD length and number was examined as depicted,

highlighting the starting (small blue arrowheads from above) and end points (large blue arrowheads

from below) of each individual PSD in an immature (A) and a mature (B) IHC. (C) The number of PSD

appositions per random section of a synaptic contact decreased during development from up to six

separated PSD patches to predominantly single ones. N=animal number, n=PSD number. (D) An

increase in PSD length was observed until a maximum is reached around the end of the third postnatal

week (smallest PSDs at P2-4: 330.00 ± 12.52 nm; largest PSD at P19-20: 626.27 ± 18.97 nm; ****p <

0.0001 NPMC test). Box plots depict 10, 25, 50, 75 and 90th percentiles with individual data points

overlaid. For more detailed statistical analysis see Table S2. (E) Schematic representation of

developmental changes from pre-hearing to hearing in murine apical coil cochlear IHCs illustrating the

synaptic refinement of the PSD.

Fig. S4. Fusion of ribbon material likely contributes to the formation of mature ribbon synapses, DCVs

may assist in initial AZ establishment. (A, B) Depicted are floating spheroid ribbon precursors shortly

after birth in P2 and P4. (C, D) Around the onset of hearing at P12, putative fusion events of synaptic

ribbons occur. 2D random ultrathin sections reveal that small spherical ribbon material is attached to an

already anchored ribbon. (E-E’’ and F-F’’) Single tomogram virtual sections and the corresponding 3D

model (E’’’ and F’’’, respectively). (E’- E’’’) A floating ribbon (highlighted with an arrow) is in close

proximity to the cavity of the attached ribbon. In single virtual sections, it appears like a ribbon precursor

is already fused to the attached ribbon (E, E’). In (F’’’) 'bulges' were found (see arrows). Ribbons: red,

AZ membrane: blue, presynaptic density: magenta. (G-I) DCVs in immature cochlear IHCs. In the age

groups between E18 and P4, several DCVs (orange arrowheads) were found in close proximity to the

AZ membrane (smaller blue arrowheads), ribbons (red dashed lines) and microtubules (larger purple

arrowhead). (J) Box plots reveal a highly variable DCV diameter showing 10, 25, 50, 75 and

90th percentiles with individual data points overlaid (*p < 0.05). For more detailed statistical analysis

see Table S2.

Fig. S5. Microtubule occurrence at the presynaptic IHC AZ and kinesin expression in the developing

organ of Corti. (A, A’) Single tomographic virtual sections of a synaptic ribbon (R) attached to the IHC

membrane (dashed blue line) and surrounded by multiple microtubules (purple arrowheads). (B, B’)

Corresponding 3D model (top and side views) visualizes the microtubule network around the ribbon.

(C-E’) Immunolocalization of kinesins (C-C’) KIF1a, (D-D’) KIF2a and (E-E’) KIF5a-c in IHCs.

Presented are confocal maximum intensity projections (C, D, E) and higher magnification single

sections (C’, D’, E’) counterstained with the ribbon marker CtBP2 and cytosolic Ca2+ buffer calretinin

to label IHCs. While KIF1a spots can be observed to associate with cytoplasmic ribbons, neither KIF2a

nor KIF5a-c immunofluorescence appears to colocalize with CtBP2.

Fig. S6. Consistent labeling pattern of CtBP2 during maturation in comparison to piccolino. (A-D)

Piccolino localizes to IHC ribbons and floating ribbon precursors (B). At P2 (A) and P14 (C), big gold

clusters appear to concentrate at the apical part of attached ribbons. At P21 (D), the staining pattern

distributes also towards the lower ribbon half. (E-J) The ribbon marker CtBP2 labels the whole ribbon

surface in attached and floating (F) ribbons. (K) Bar plot depicting an equal distribution of gold clusters

between the apical and basal ribbon (R) half in immature and mature IHCs (N=animal number, n=ribbon

number). (L) Representation of the CtBP2 labeling (gray circles) in P12 IHCs from various overlaid

synaptic ribbons. Red oval-shaped circle outlines the ribbon, red box below indicates the IHC

membrane.

Fig. S7. FIB- and SBF-SEM evaluation of ribbon and SV number. (A-C) FIB-SEM 3D projection of

the basolateral compartment of a second cochlear IHC from each age group showing synaptic ribbons

with attached SVs. Grey dashed lines demarcate the division into pillar and modiolar sides. (D-H)

Quantifications of FIB-SEM data sets. (D) SV counts per ribbon indicated an increase during maturation

(Table S3). (E) Graph represents absolute ribbon numbers of two IHCs per age group with more multiple

ribbons per afferent contact at the modiolar side of all investigated age groups (P9 modiolar side: 42.86%

single and 57.14% multiple ribbons vs. P9 pillar side: 62.50% single and 37.50% multiple ribbons, P15

modiolar side: 92.30% single and 7.70% multiple ribbons vs. P15 pillar side: 100% single and 0%

multiple ribbons; P34 modiolar side: 72.22% single and 27.78% multiple ribbons vs. P34 pillar side:

83.33% single and 16.67% multiple ribbons). (F) Analysis of distance measurements between ribbons

that are localized at the same afferent (aff.) contact (P9: mean modiolar (green X) = 1.67 ± 0.23 µm,

n=33 ribbons, N=2 IHCs, mean pillar (black X) = 2.45 ± 0.49 µm, n=8 ribbons, N=2 IHCs; P34: mean

modiolar = 0.25 ± 0.02 µm, n=11 ribbons, N=2 IHCs, mean pillar = 0.16 µm, n=2 ribbons, N=2 IHCs).

Individual data points are represented by green circles for modiolar and black squares for pillar sides.

(G) Measurements of the shortest distance between the nearest ribbons of different afferent contacts

(P9: mean modiolar (green X) = 1.62 ± 0.29 µm, n=9 ribbons, N=2 IHCs, mean pillar (black X) = 3.23

± 0.5 µm, n=5 ribbons, N=2 IHCs; P15: mean modiolar = 4.25 ± 1.71 µm, n=10 ribbons, N=2 IHCs,

mean pillar = 4.08 ± 1.20 µm, n=4 ribbons, N=2 IHCs; P34: mean modiolar = 3.09 ± 0.61 µm, n=13

ribbons, N=2 IHCs, mean pillar = 3.17 ± 0.215 µm, n=5 ribbons, N=2 IHCs). (H) Distance

measurements of floating precursor ribbons to the cell membrane at P9 (mean = 117.22 ± 32.51 nm, n=6

ribbons, N=2 IHCs). (I) The P14 dataset was obtained by SBF-SEM, showing slightly more ribbons at

the modiolar side (J). (K) FIB-SEM of the supranuclear region of two P9 IHCs, depicting two different

cutting planes, one right above the nucleus the other one close to the cuticular plate (CP) of both cells.

Table S1. Detailed list of ribbon synapse and SV parameters.

SV

dia

me

ter

(nm

)

(Fig

. 1

H)

47

.00

± 0

.28

46

.76

± 0

.27

46

.85

± 0

.24

44

.25

± 0

.14

41

.78

± 0

.11

38

.26

± 0

.14

37

.32

± 0

.12

34

.86

± 0

.13

*p =

0.0

32

1

(E

18

vs.P

15

)

*p =

0.0

32

1

(P

0-1

vs.P

12

)

*p =

0.0

32

1

(P

0-1

vs.P

15

)

*p =

0.0

29

8

(P

9 v

s.P

48

)

**p

= 0

.005

2

(E

18

vs.P

48

)

*p =

0.0

22

0

(P

0-1

vs.P

48

)

**p

= 0

.004

3

(P

2-4

vs.P

12

)

**p

= 0

.001

4

(E

18

vs.P

19

-20

)

**p

= 0

.002

0

(P

0-1

vs.P

19

-20)

***p

= 0

.000

7

(P2

-4 v

s.P

15

)

***p

= 0

.000

5

(P2

-4 v

s.P

19

-20

)

***p

= 0

.000

5

(P2

-4 v

s.P

48

)

**p

= 0

.002

0

(P

9 v

s.P

19

-20

)

KW

te

st

n

94

17

3

21

6

39

5

25

5

14

7

19

3

17

6

- -

N

2

2

2

2

2

2

2

2

- -

E1

8

P0

-1

P2

-4

P9

P1

2

P1

5

P1

9-2

0

P4

8

p-v

alu

es

Sta

tis

tics

Data are presented as Mean ± SEM, N = number of animals, n = number of ribbons, p-values and

statistical tests of all examined age groups (****p < 0.0001).

SV

co

un

t p

er

rib

bo

n

(Fig

. 1

G)

4.4

8 ±

0.1

6

4.2

1 ±

0.1

1

4.8

0 ±

0.1

3

8.3

8 ±

0.1

6

10

.87

± 0

.24

12

.57

± 0

.42

12

.42

± 0

.29

13

.98

± 0

.36

****

P4

8 v

s.

E1

8,

P0

-1,

P2

-4, P

9,

P12

****

P1

9-2

0 v

s.

E1

8,

P0

-1,

P2

-4, P

9

***

* P

15

vs.

E1

8,

P0

-1,

P2

-4, P

9

****

P1

2 v

s.

E1

8,

P0

-1,

P2

-4, P

9

****

P9

vs.

E1

8,

P0

-1,

P2

-4

NP

MC

te

st

Rib

bo

n w

idth

(n

m)

(Fig

. S

2C

)

63

.14

± 2

.09

66

.91

± 1

.42

10

1.7

9 ±

1.9

7

13

1.4

3 ±

2.3

5

16

0.2

4 ±

4.2

3

14

8.0

2

± 6

.20

14

9.2

9 ±

4.6

8

13

0.8

9 ±

4.9

8

****

P1

2 v

s.

E1

8,

P0

-1,

P2

-4, P

9,

P48

****

P1

5 v

s.

E1

8,

P0

-1,

P2

-4

****

P1

9-2

0 v

s.

E1

8,

P0

-1,

P2

-4

****

P9

vs.

E1

8,

P0

-1,

P2

-4

****

P4

8 v

s.

E1

8,

P0

-1,

P2

-4

****

P2

-4 v

s.

E1

8,

P0

-1

NP

MC

te

st

Rib

bo

n h

eig

ht

(nm

)

(Fig

. S

2B

)

60

.40

± 1

.62

67

.66

± 1

.73

10

8.3

6 ±

2.3

2

13

4.5

7 ±

2.1

3

16

9.1

7 ±

3.5

8

18

4.6

9 ±

4.7

5

17

2.9

9 ±

3.3

6

18

1.4

0 ±

4.2

3

****

P4

8 v

s.

E1

8,

P0

-1,

P2

-4, P

9

****

P1

5 v

s.

E1

8,

P0

-1,

P2

-4, P

9

****

P1

9-2

0 v

s.

E1

8,

P0

-1,

P2

-4, P

9

****

P1

2 v

s.

E1

8,

P0

-1,

P2

-4, P

9

****

P9

vs.

E1

8,

P0

-1,

P2

-4

****

P2

-4 v

s.

E1

8,

P0

-1

NP

MC

te

st

Rib

bo

n a

rea

(n

m2)

(Fig

. 1

F)

31

66

.47

± 1

76

.92

38

67

.46

± 1

70

.98

95

61

.49

± 3

53

.31

15

72

1.6

0 ±

47

4.3

9

22

14

8.8

6 ±

75

6.4

9

22

03

5.3

3 ±

1.1

8

20

98

1.1

2 ±

79

9.7

4

18

56

6.4

5 ±

71

0.2

2

****

P1

5 v

s.

E1

8,

P0

-1,

P2

-4

****

P1

5 v

s.

P9

****

P1

2 v

s.

E1

8,

P0

-1,

P2

-4, P

9

****

P1

9-2

0 v

s.

E1

8,

P0

-1,

P2

-4, P

9

****

P4

8 v

s.

E1

8,

P0

-1,

P2

-4

****

P9

vs.

E1

8,

P0

-1,

P2

-4

****

P2

-4 v

s.

E1

8,

P0

-1

NP

MC

te

st

Table S2. Detailed list of further ultrastructural AZ parameters.

DC

V d

iam

ete

r (n

m)

(Fig

. S

4J

)

94

.51

± 6

.93

83

.67

± 5

.95

12

3.6

6 ±

10

.29

–

–

–

–

–

*p =

0.0

15

(P

2-4

vs.

P0

-1)

KW

te

st

Data are presented as Mean ± SEM, N = number of animals, n = number of PSDs and DCVs, p-values

and statistical tests of all examined age groups (****p < 0.0001).

n

(DC

Vs

)

16

36

7

0

0

0

0

0 - -

PS

D l

en

gth

(n

m)

(Fig

. S

3D

)

35

4.8

0 ±

23

.23

36

6.4

3 ±

17

.50

33

0.0

0 ±

12

.52

37

6.4

3 ±

8.8

6

41

8.3

8 ±

13

.21

53

1.1

1 ±

20

.09

62

6.2

7 ±

18

.97

55

5.3

6 ±

15

.38

****

P1

9-2

0 v

s.

E18

, P

0-1

, P

2-4

, P

9,

P12

****

P4

8 v

s.

E1

8,

P0

-1,

P2

-4, P

9,

P1

2

****

P1

5 v

s.

E1

8,

P0

-1,

P2

-4, P

9,

P1

2

****

P1

2 v

s.

P2

-4

NP

MC

te

st

n

(PS

Ds

)

85

19

2

25

0

54

5

33

3

15

0

19

7

17

2

- -

N

2

2

2

2

2

2

2

2 - -

E1

8

P0

-1

P2

-4

P9

P1

2

P1

5

P1

9-2

0

P4

8

p-v

alu

es

Sta

tis

tics

Table S3. Detailed list of FIB-SEM data analysis parameters.

To

tal

rib

bo

n v

olu

me

(n

m3

x 1

06)

pe

r a

ff.

co

nta

ct

(Fig

. 7

M)

8.3

1 ±

1.4

0

6.5

0 ±

1.7

7

11

.26

± 1

.36

5.5

7 ±

1.3

5

8.5

2 ±

0.8

6

6.6

2 ±

0.8

0

n.s

.

AN

OV

A t

est

Nu

mb

er

of

SV

s p

er

rib

bo

n

(Fig

. S

7D

)

18

.12

± 1

.11

17

.54

± 2

.77

79

.29

± 9

.48

58

.29

± 6

.92

55

.83

± 4

.39

49

.14

± 4

.10

****

p =

1.6

7E

-11

(P

15 m

od

iola

r vs.

P9

pill

ar)

****

p =

2.0

4E

-13

(P

15 m

od

iola

r vs.

P9

mo

dio

lar)

****

p =

7.1

9E

-05

(P

15 p

illa

r vs.

P9

pill

ar)

****

p =

4.8

0E

-06

(P

15 p

illa

r vs.

P9

mo

dio

lar)

****

p =

2.1

2E

-07

(P

34 m

od

iola

r vs.

P9

pill

ar)

****

p =

8.2

0E

-11

(P

34 m

od

iola

r vs.

P9

mo

dio

lar)

***p

= 7

.29

E-0

4

(P

34

pill

ar

vs.

P9

mo

dio

lar)

**p

= 4

.02

E-0

3

(

P3

4 p

illa

r vs.

P9

pill

ar)

**p

=

6.1

3E

-03

(P1

5 m

od

iola

r vs.

P3

4 p

illa

r)

**p

=

2.5

9E

-03

(P1

5 m

od

iola

r vs.

P3

4 m

od

iola

r)

Tu

key´s

te

st

Ind

ivid

ua

l ri

bb

on

vo

lum

e

(nm

3 x

10

6)

(Fig

. 7L

)

4.7

0 ±

0.6

6

3.4

4 ±

0.9

9

10

.46

± 1

.23

5.5

7 ±

1.3

5

6.3

9 ±

0.6

9

5.6

8 ±

0.4

5

****

p =

8.5

3E

-05

(P1

5 m

od

iola

r vs. P

9 m

od

iola

r)

***p

= 1

.41

E-0

4

(P1

5 m

od

iola

r vs. P

9 p

illa

r)

*p =

0.0

3

(P1

5 m

od

iola

r vs. P

34

mo

dio

lar)

Tu

key´s

test

n

43

13

14

7

24

7 - -

N

2

2

2

2

2

2 - -

P9

mo

dio

lar

P9

pil

lar

P1

5 m

od

iola

r

P1

5 p

illa

r

P3

4 m

od

iola

r

P3

4 p

illa

r

p-v

alu

es

Sta

tis

tics

Table continued

Dis

tan

ce b

etw

ee

n r

ibb

on

s (

µm

) a

t

dif

fere

nt

aff

. c

on

tac

ts (

Fig

. S

7G

)

1.6

2 ±

0.2

9

3.2

3 ±

0.4

9

4.2

5 ±

1.7

1

4.0

8 ±

1.2

0

3.0

9 ±

0.6

1

3.1

7 ±

0.2

1

n.s

.

AN

OV

A t

est

n

43

13

14

7

24

7 - -

N 2

2

2

2

2

2 - -

Data are presented as Mean ± SEM, N = number of animals, n = number of ribbons, p-values and

statistical tests of all examined age groups (n.s. > 0.05).

To

tal

SV

nu

mb

er

pe

r a

ff.

co

nta

ct

(Fig

. 7

N)

35

.95

± 5

.0

31

.5 ±

7.2

5

85

.38

± 9

.91

58

.29

± 6

.92

74

.44

± 5

.91

57

.33

± 5

.82

****

p =

4.8

5E

-06

(P1

5 m

od

iola

r vs. P

9 m

od

iola

r)

***p

= 4

.30

E-0

4

(P1

5 m

od

iola

r vs. P

9 p

illa

r)

***p

= 1

.00

E-0

4

(P3

4 m

od

iola

r vs. P

9 m

od

iola

r)

**p

= 5

.11

E-0

3

(P3

4 m

od

iola

r vs. P

9 p

illa

r)

Tu

key´s

te

st

P9

mo

dio

lar

P9

pil

lar

P1

5 m

od

iola

r

P1

5 p

illa

r

P3

4 m

od

iola

r

P3

4 p

illa

r

p-v

alu

es

Sta

tis

tics

Movie S1. Movie scanning through a representative FIB-SEM z-stack of a P15 IHC, showing 3D

reconstruction of the basolateral IHC contour, part of the nucleus, mitochondria, innervating nerve

fibers, synaptic ribbons and their corresponding SVs. The movie starts at the nuclear region and extends

towards the base of the cochlear IHC. Higher resolution of this video can be found using this link:

http://www.innerearlab.uni-goettingen.de/materials.html

Movie S2. Movie scanning through a FIB-SEM z-stack of the supranuclear part of two neighboring P9

IHCs, shown in Fig. S7K. No ribbon precursors or ribbons could be observed in this dataset. Scale bar:

5 µm. Higher resolution of this video can be found using this link:

http://www.innerearlab.uni-goettingen.de/materials.html

References

[1] A. C. Meyer et al., “Tuning of synapse number, structure and function in the cochlea,” Nat.

Neurosci., vol. 12, no. 4, pp. 444–453, Mar. 2009.

[2] C. Vogl et al., “The BEACH protein LRBA is required for hair bundle maintenance in cochlear

hair cells and for hearing,” EMBO Rep., vol. 18, no. 11, pp. 2015–2029, Sep. 2017.

[3] C. Limbach et al., “Molecular in situ topology of Aczonin/Piccolo and associated proteins at the

mammalian neurotransmitter release site,” Proc. Natl. Acad. Sci., vol. 108, no. 31, pp. E392–

E401, 2011.

[4] A. B. Wong et al., “Developmental refinement of hair cell synapses tightens the coupling of Ca2+

influx to exocytosis,” EMBO J., vol. 33, no. 3, pp. 247–264, Feb. 2014.

[5] N. Strenzke et al., “Hair cell synaptic dysfunction, auditory fatigue and thermal sensitivity in

otoferlin Ile515Thr mutants,” EMBO J., vol. 35, no. 23, pp. 2519–2535, Dec. 2016.

[6] D. N. Mastronarde, “Automated electron microscope tomography using robust prediction of

specimen movements,” J. Struct. Biol., vol. 152, no. 1, pp. 36–51, Oct. 2005.

[7] J. R. Kremer, D. N. Mastronarde, and J. R. McIntosh, “Computer visualization of three-

dimensional image data using IMOD,” J. Struct. Biol., vol. 116, no. 1, pp. 71–76, Feb. 1996.

[8] V. Nieratschker et al., “Bruchpilot in ribbon-like axonal agglomerates, behavioral defects, and

early death in SRPK79D kinase mutants of Drosophila,” PLoS Genet., vol. 5, no. 10, p.

e1000700, Oct. 2009.

[9] T. J. Deerinck, T. M. Shone, E. A. Bushong, R. Ramachandra, S. T. Peltier, and M. H. Ellisman,

“High‐performance serial block‐face SEM of nonconductive biological samples enabled by focal

gas injection‐based charge compensation,” J. Microsc., Dec. 2017.

[10] A. Bertie, “Java applications for teaching statistics,” Mathematics, Statistics and Operational

Research Connect., vol. 2, no. 3, pp. 78–81, 2002.

[11] P. Jean et al., “The synaptic ribbon is critical for sound encoding at high rates and with temporal

precision,” eLife, vol. 7, 2018.