Postsynaptic GluA3 subunits are required for the appropriate assembly of AMPA receptor GluA2 and GluA4 subunits on mammalian cochlear afferent synapses and for presynaptic ribbon modiolar-pillar morphological distinctions

The encoding of acoustic signals in the cochlea depends on α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid receptors (AMPARs), but relatively little is known about their reliance on specific pore-forming subunits. With 5-week-old male GluA3KO mice, we determined cochlear function, synapse ultrastructure, and AMPAR subunit molecular anatomy at ribbon synapses between inner hair cells (IHCs) and spiral ganglion neurons (SGNs). GluA3KO and wild-type (GluA3WT) mice reared in ambient sound pressure level (SPL) of 55-75 dB had similar ABR thresholds, wave-1 amplitudes, and latencies. Ultrastructurally, the IHC modiolar-pillar differences in presynaptic ribbon size and shape, and synaptic vesicle size seen in GluA3WT were diminished or reversed in GluA3KO. The quantity of paired synapses (presynaptic ribbons juxtaposed with postsynaptic GluA2 and GluA4) was similar, however, GluA2-lacking synapses (ribbons paired with GluA4 but not GluA2) were observed only in GluA3KO. SGNs of GluA3KO mice had AMPAR arrays of smaller overall volume, containing less GluA2 and greater GluA4 immunofluorescence intensity relative to GluA3WT (3-fold difference in mean GluA4:GluA2 ratio). The expected modiolar-pillar gradient in ribbon volume was observed in IHCs of GluA3WT but not GluA3KO. Unexpected modiolar-pillar gradients in GluA2 and GluA4 volume were present in GluA3KO. GluA3 is essential to the morphology and molecular composition of IHC-ribbon synapses. We propose the hearing loss seen in older male GluA3KO mice results from progressive synaptopathy evident in 5-week-old mice as increased abundance of GluA2-lacking, GluA4 monomeric, Ca2+-permeable AMPARs.


63
In the cochlear ganglion and the ascending central auditory system, hearing relies on fast 64 excitatory synaptic transmission via unique α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic processing due to effects on synaptic transmission associated with altered ultrastructure of 109 Antunes et al., 2020). Therefore, we also examined presynaptic ribbon morphology in relation to

132
To test the auditory output of the GluA3 WT and GluA3 KO mice, we performed ABR as previously

134
Recordings were conducted under isoflurane anesthesia in a soundproof chamber and using a 135 Tucker-Davis Technologies (Alachua, FL) recording system. Click or tone stimuli were 136 presented through a calibrated multi-field magnetic speaker connected to a 2-mm diameter 137 plastic tube inserted into the ear canal. ABR were recorded by placing subdermal needle 138 electrodes at the scalp's vertex, at the right pinna's ventral border, and the ventral edge of the 139 left pinna. ABR were recorded in response to broadband noise clicks (0.1 ms) or tone pips of 4, 7 8, 12, 16, 24, and 32 kHz (5 ms). Stimuli were presented with alternating polarity at a rate of 21 141 Hz, with an inter-stimulus interval of 47.6 ms. The intensity levels used were from 80 dB to 10 142 dB, in decreasing steps of 5 dB. The waveforms of 512 presentations were averaged, amplified 143 20x, and digitalized through a low impedance preamplifier. The digitalized signals were 144 transferred via optical port to an RZ6 processor, where the signals were band-pass filtered (0.3 145 -3 kHz) and converted to analog form. The analog signals were digitized at a sample rate of 146 ~200 kHz and stored for offline analyses. Hearing threshold levels were determined from the 147 averaged waveforms by identifying the lowest intensity level at which clear, reproducible peaks 148 were visible. Wave 1 amplitudes were compared between GluA3 WT and GluA3 KO mice. For 149 measurements of amplitudes, the peaks and troughs from the click-evoked ABR waveforms 150 were selected manually in BioSigRZ software and exported as CSV files. The peak amplitude 151 was calculated as the height from the maximum positive peak to the next negative trough.

153
Immunohistochemistry and immunofluorescence 154 A total of 14 mice (GluA3 WT n= 7; GluA3 KO n=7) were anesthetized with a mixture of ketamine 155 (60 mg/kg) and xylazine (6.5 mg/kg) and were transcardially perfused with 4% 156 paraformaldehyde (PFA) in 0.1M phosphate buffer (PB) pH= 7.2. After 10 minutes of 8 one month. Brains were cryoprotected in the same sucrose dilutions gradient and frozen on dry 167 ice. Cochleae were cut at a 20 um thickness section with a cryostat and were mounted on glass 168 slides. Brains were cut with a slicing vibratome at 50-60 um thickness and collected on culture 169 wheel plates containing 0.1M PBS. Cochlea and brain sections followed standard 170 immunofluorescence and immunohistochemistry protocols described in Douyard et al. (2007)    The other ten cochleae (5 of each genotype) were shipped overnight to Washington University 183 in Saint Louis in 0.1M PBS containing 5% glycerol for wholemount immunolabeling and confocal 184 analysis of presynaptic ribbons (CtBP2/Ribeye) and postsynaptic AMPAR subunits GluA2,

331
We first determined whether the cohort of 5-week-old C57BL/6J GluA3 WT and GluA3 KO differed 332 in their auditory sensitivity. Our ABR analysis showed no differences between genotypes in 333 clicks or pure tone thresholds or wave-1 amplitude or latency (Fig. 1A). We note that male

334
GluA3 KO  observed GluA2 in the SGNs of both genotypes (Fig. 1B, upper left). We found that SGNs 342 lacked GluA1 in GluA3 WT mice, as expected, and we did not observe compensatory GluA1 343 expression in SGNs of GluA3 KO (Fig. 1B, lower left). We also checked the immunolabeling of GluA3 KO lacked immunolabeling for GluA3 (Fig. 2). This confirmed the deletion of GluA3

363
we found no significant differences (Fig. 1C). In addition, using the PCR gels, we calculated the

367
Overall, this shows that lack of GluA3 did not affect hearing sensitivity at 5-weeks of age, in   (Figs. 3 and 4).

386
A total of 27 synapses of GluA3 WT mice were analyzed in 3D using serial sections (on average, 387 7 ultrathin sections per PSD). Of this total, 18 were on the modiolar side and 9 on the pillar side 388 of the IHCs (Fig. 3A-B). We found the PSD surface area to be larger (p= 0.017, Mann-Whitney 389 U test, U: 23) for synapses on modiolar side (mean: 0.50 + 0.1 µm 3 ) when compared to the pillar 390 side (mean: 0.33 + 0.09 µm 3 ) (Fig. 3C, left). Presynaptic ribbon volume was similar between

405
From GluA3 KO , a total of 30 synapses were analyzed in 3D with serial sections (on average, 7 406 ultrathin sections per PSD). Of this total, 20 were on the modiolar side and 10 on the pillar side 407 of the IHCs (Fig. 4A-B). Analysis showed that PSD surface area and ribbon volume were GluA3 KO (Fig. 4D, left and center). Also, opposite to the pattern in GluA3 WT , SVs of modiolar-

425
We then compared PSDs and ribbons among GluA3 WT and GluA3 KO mice on the modiolar and 426 pillar sides (Fig. 5A). Overall PSD surface area was significantly different between genotypes (p 427 = 0.027, Kruskal-Walli's test). Considering modiolar-side and pillar-side synapses separately 428 with multiple (paired) comparisons revealed the PSD surface areas of modiolar-side synapses 429 to be similar among GluA3 KO and GluA3 WT (p = 0.91). In contrast, the mean PSD surface area of 430 the pillar-side synapses was larger in GluA3 KO than GluA3 WT (p = 0.007) (Fig. 5A, left). Synaptic 431 ribbon volume differed between GluA3 WT and GluA3 KO (p< 0.0001, one-way ANOVA).

432
Comparison analyses showed that the ribbon volumes of modiolar-side synapses were similar 433 between GluA3 WT and GluA3 KO (p= 0.99). In contrast, the pillar-side synapses were larger in 434 GluA3 KO than GluA3 WT (p = 0.0006) (Fig. 5A, right). Differences between the ribbon major axis 435 were found between WT and KO (p< 0.0001; one-way ANOVA). On the modiolar side, analysis 436 of the ribbon major axis length showed that those of the GluA3 KO were significantly smaller than 437 GluA3 WT (p < 0.0001), whereas pillar-side synapses were similar in major axis length (Fig. 5B,   438 left) (p > 0.5). Differences in ribbon circularity were also found between genotypes (p < 0.0001, 439 one-way ANOVA). Paired comparisons showed that modiolar-side ribbons were significantly 19 less circular in GluA3 WT (p < 0.0001), whereas pillar-side ribbons were of similar circularity (p = 441 0.31) (Fig. 5B, center). SV size differed between genotypes (p < 0.0001, one-way ANOVA).

445
Altogether, our data of 5-week-old male mice show the AMPAR subunit GluA3 is essential to GluA2 labeling and increase in GluA4 labeling relative to GluA3 WT (Fig. 2; 6A). Despite this, the

475
Grayscale and color images ( Fig. 2; Fig. 6A the reduction in GluA2 (Fig. 7E-F). Relative to the mean GluA puncta volume per GluA3 WT 490 synapse, the mean volumes of GluA2, GluA4, and GluA Sum were all reduced in GluA3 KO (Fig.   491   7G). When normalized to the mean puncta volume per image in either group, the distributions of synapse volumes were broadened for GluA2 and GluA4 subunits in GluA3 KO relative to GluA3 WT 493 (Fig. 7H). For each image, we calculated the coefficient of variation (CV = SD / mean) in puncta

504
To test the statistical significance and to confirm the differences observed in Figure 7 in a larger 505 data set from a replication cohort, we next assessed mean synaptic CtBP2, GluA2, and GluA4 506 volume and intensity per image in 14 image stacks from each genotype. In image stacks of 507 sufficient quality, we also measured synapse position on the IHC modiolar-pillar axis to sort 508 them into modiolar and pillar groups. Image means and group means are displayed in Figure 8.

533
In GluA3 WT , we commonly observed apparent oscillations in synapse volume as a function of 534 position in the Z-axis of the confocal microscope when the modiolar-pillar dimension was 535 approximately parallel to the Z-axis (Fig. 9A, lower). These spatial oscillations were clearer 536 when measured as sphericity (Fig. 9A, upper), which was inversely related to volume (Fig. 9A,   537 right). We observed a similar phenomenon in GluA3 KO (Fig. 9B), although the synapses 538 resided in a smaller range along the Z-axis. GluA2 and GluA4 intensities per synapse were 539 positively related in both genotypes (Fig. 9C, left). GluA2 and GluA4 intensities were positively 540 related with CtBP2 intensities, but the relationships were less apparent in GluA3 KO (Fig. 9C,   541 center and right), consistent with the increase in CV measured for GluA2 and GluA4 intensities 542 23 per synapse in GluA3 KO relative to GluA3 WT (Fig. 7). Plotting the GluA4:GluA2 intensity ratio as 543 a function of Z-position revealed that increases of the GluA4:GluA2 intensity ratios in GluA3 KO 544 tended to be greater for synapses on the pillar side than the modiolar side of the IHC relative to 545 GluA3 WT (Fig. 9D).

638
The ABR wave-1 amplitude is unaltered in male GluA3 KO at 5-weeks of age suggesting a similar 639 hearing sensitivity to WT mice. However, ABR peak amplitudes are reduced in the male KO at 8