Using cortical neuron markers to target cells in the dorsal cochlear nucleus

The dorsal cochlear nucleus (DCN) is the first auditory region that integrates somatosensory and auditory inputs. The region is of particular interest for auditory research due to the large incidence of somatic tinnitus and increased aberrant activity in other forms of tinnitus. Yet, the lack of useful genetic markers for in vivo manipulations hinders the elucidation of the DCN contribution to tinnitus pathophysiology. In this work, we assessed whether adeno-associated viral vectors (AAV) containing the calcium/calmodulin-dependent protein kinase 2 alpha (CaMKIIα) promoter and our mouse line of nicotinic acetylcholine receptor alpha 2 subunit (Chrna2)-Cre can be used to target specific DCN populations. The CaMKIIα promoter is usually applied in studies of principal neurons of neo and paleocortex while Chrna2-cre mice express Cre recombinase in cortical dendrite inhibiting interneurons. We found that CaMKIIα cannot be used to specifically target excitatory fusiform DCN neurons. EYFP expression driven by the CaMKIIα promoter was stronger in the fusiform layer but labelled cells showed a diverse morphology indicating that they belong to different classes of DCN neurons. Light stimulation after driving Channelrhodopsin2 (ChR2) by the CaMKIIα promoter generated spikes in some units but firing rate decreased when light stimulation coincide with sound presentation. Expression and activation of eArch3.0 (CaMKIIα driven) in the DCN produced spike inhibition in some units but, most importantly, sound-driven spikes were delayed by concomitant light stimulation. We explored the existence of Cre+ cells in the DCN of Chrna2-Cre mice by hydrogel embedding technique (CLARITY). There were almost no Cre+ cell bodies in the DCN; however, we observed profuse projections arising from the ventral cochlear nucleus (VCN). Anterograde labeling Cre dependent AAV injected in the VCN revealed two main projections: one arising in the ipsilateral superior olive and the contralateral medial nucleus of the trapezoid body (bushy cells) and a second bundle terminating in the DCN, suggesting the latter to be excitatory Chrna2+ T-stellate cells). Stimulating ChR2 expressing terminals (light applied on the DCN) of VCN Chrna2+ cells increased firing of sound responding and nonresponding DCN units. This work shows that molecular tools intensively used in cortical studies may be useful for manipulating the DCN especially in tinnitus studies.


30
The dorsal cochlear nucleus (DCN) of the auditory brainstem is the first integrator of auditory and 31 multisensory signals and has been pointed as a key structure in tinnitus physiopathology (Kaltenbach 32 et al., 2005;Tzounopoulos, 2008;Baizer et al., 2012). Cells in the DCN receives direct or indirect 33 (e.g. relayed by the ventral cochlear nucleus -VCN) sound input onto different cell populations in a 34 layer arrangement. The most cell-populated DCN field is the fusiform cell layer formed by excitatory 35 fusiform cells intercalated with interneurons (Oertel and Young, 2004). An interesting aspect of the 36 DCN is its architectural similarity to the cerebellum (Devor, 2000) that is thought to be responsible 37 for integrative processing (e.g. sound/somatosensory, Oertel and Young, 2004). 38 Abnormal sensory integration in the DCN is clinically relevant due to the prevalence of temporo-39 mandibular tinnitus (Levine, 1999;Grossan and Peterson, 2017). Other forms of mechanical tinnitus 40 are also attributed to aberrant activity in the DCN (Han et al., 2009). Also, a large number of studies 41 have shown altered synaptic and intrinsic cellular properties within the DCN circuit relating to noise-42 induced tinnitus (reviewed by Shore et al., 2016) yet tinnitus treatments to date do not specifically 43 target this region. The ventral cochlear nucleus (VCN) can also contribute to noise-induced tinnitus 44 (Kraus et al., 2011;Coomber et al., 2015). Aberrant activity in any cochlear nucleus subregions 45 can trigger upstream changes as cochlear nucleus neurons relay auditory signals to higher areas of 46 the auditory pathway (Kraus et al., 2011;Coomber et al., 2015). Abnormal activity from auditory 47 cortex and inferior colliculus can also produce downstream alterations in the DCN as its cells receive 48 feedback through descending auditory fibers (Winer and Prieto, 2001;Milinkeviciute et al., 2016). 49 Despite its physiological importance and its well accepted role in tinnitus, the contribution of specific 50 DCN populations to hearing and tinnitus pathophysiology are largely unknown.

51
Due to its variety of cell types and its cerebellum like structure, DCN circuit studies could benefit 52 from identifying key neuronal markers . The Calcium/calmodulin-dependent 53 protein kinase 2 alpha (CaMKIIα) promoter is widely used for targeting cortical pyramidal cells. 54 Immunohistochemical data from rats has shown CaMKIIα expression in the DCN molecular and 55 fusiform cell layers (Ochiishi et al., 1998). In mice CaMKIIα RNA is widely distributed in the 56 fusiform layer (Lein et al., 2007). Hence, viral vectors to express of reporter or optogenetic proteins 57 by CaMKIIα promoter may be applied to DCN manipulation. Cortical interneuron markers could 58 also be used to tag DCN cells. The calcium buffer protein parvalbumin (PV) is used for targeting fast 59 spiking interneurons in studies of the hippocampus/neocortex (Kawaguchi and Kondo, 2002;Courtin 60 et al., 2013) and PV expression specific to inhibitory neurons have also been described in some 61 subcortical nuclei (Unal et al., 2015). In the DCN, PV is distributed across layers without population 62 specificity. Moreover, somatostatin expression (found in dendrite targeting cortical/hippocampal 63 interneurons) does not appear to follow a layer/cell specific expression (Lein et al., 2007), similar to 64 the cortex/hippocampus where PV and somatostatin expression can be quite promiscuous (Kawaguchi 65 and Kondo, 2002;Mikulovic et al., 2015). Recently, the nicotinic acetylcholine receptor alpha 2 66 subunit (chrna2) has been described as a marker for highly specific interneuron populations (CA1 67 2/26 oriens-lacunosum moleculare cells LM in the ventral hippocampus or L5 Martinotti cells in the 68 neocortex; Leão et al., 2012;Hilscher et al., 2017). Our labs have developed (Kullander) and evaluated 69 (Leão, Leão and Kullander) a Chrna2-cre mouse line that significantly leaped the study of dendritic 70 targeting interneuron populations (Leão et al., 2012;Enjin et al., 2017;Hilscher et al., 2017). Cre+ 71 cells in Chrna2-cre mice seem to belong to single populations in several subcortical nuclei (Siwani 72 et al., 2018). Expression of Chrna2 in the DCN is not described but a glance in whole brain imaging 73 (clarity) evince almost no cell body expression in the DCN with Cre+ cell clusters in its vicinity and 74 Chrna2+ positive axonal terminals that profusely target the DCN Siwani 75 et al., 2018). 76 Here, we test if adeno-associated viral vectors (AAV) with the CaMKIIα promoter can be used 77 for manipulating DCN circuits in vivo. AAV encoded optogenetic protein expression and light 78 stimulation paired with brief sound presentation was used to functionally identify cells and assess the 79 effect of optical depolarization/hyperpolatization in input/output functions in CaMKIIα+ neurons 80 in combination with brief sound stimulation. Lastly, we examined how activation of Chrna2+ dendrites spreading towards the molecular layer (possible fusiform cells, Figure 1B, arrows). Smaller 95 neuronal somas were also labeled with eYFP in the fusiform and deeper layers, as well as several 96 large neuronal somas of the deep layer of the DCN with dendrites stretching along the internal edge 97 of the DCN (possible giant cells, Figure 1C, arrow 80dB, 5∼15kHz noise pulses presented at 10Hz) and response to sound was quantified and visualized 109 by peristimulus time histograms (PSTHs; Figure 2B and E). Blue light stimulation (473nm, 10ms 110 duration, at 10Hz with intensity of 5mW/mm² at fiber tip) delivered by a glass optic fiber to the 111 DCN (Ø200µm, inserted in a 45°angle from the contralateral side; Figure 2A), elicited increased 112 firing of units immediately following blue light stimulation ( Figure 2C and E). We found 25% (  Example of a unit with its waveform shown at higher magnification (center), that responded to sound stimulation by briefly (∼20ms) increasing its firing. C) An example of a unit that does not respond to sound stimulation but increases firing in response to blue light stimulation. D) Example of a unit responding to sound (center) and blue light (right) stimulation. E) Group mean number of spikes for all units (n=76) showing a significant increase after sound (S) comparing to baseline (B; left) or blue light (L; center) stimulation (p = 2.6e-5 and 0.016) and a significant decrease after concomitant sound and light stimulation (S+L; right; p=4.2e-5). F) Group mean number of spikes for all units (n=76) after sound stimulation is significantly higher than after concomitant sound and light stimulation (p=3.8e-4).

5/26
Out of these, 58% (11/19) responded exclusively to light stimulation, 26% (5/19) responded to 115 either sound or light stimulation ( Figure 2D), and 16% ( 1.5min including stimulus epochs, and comparing to specific stimuli showed a significant increase in 121 response to sound (p=2.6e-5; Figure 2E left) or light (p=0.016; Figure 2E center) but, interestingly, a 122 decrease in response to concomitant sound and light stimulation (p=4.2e-5; Figure 2E right). Also, 123 the mean number of spikes in response to sound were significantly higher than to concomitant sound 124 and light (p=3.8e-4; Figure 2F). This shows that optogenetic excitation using ChR2 can increase 125 firing of DCN units, even in units not directly responding to sound or light, but controversially, 126 presence of simple sounds during optogenetic stimulation can decrease the over all unit firing.  Units are in the same DCN region (strongest signal at ∼-3.5mm DV), that have different waveforms and different baseline firing rates. Even though none of the units responded to sound stimulation, one of the units decreased firing rate (A) while and the other (B) increased its firing rate upon inhibition of CaMKIIα-eArch3.0 DCN cells. C) Group firing rate of all non-responding units (n=69) under sound (red) or concomitant sound and green light (olive). Units were divided into high and low firing rate, and a significant increase in firing was found for low firing units (p=0.03). Units were also divided into units that decrease or increase firing rate comparing both stimulations, and a significant decrease and increase was found (p = 4.7e-4 and 0.01, respectively). that decreased firing to 70% of the initial frequency (23 ± 6.68Hz) under green light stimulation, 159 while 15 units were low frequency firing (2.19 ± 0.57Hz), decreased firing frequency by 58% (1.26 160 ± 0.48Hz) upon green light stimulation. On the contrary, 42 units increased firing frequency upon 161 green light stimulation ( Figure 4B and C) where 11 high frequency firing units increase in firing 162 frequency to ∼ double the initial frequency (16.4 ± 4.38Hz to 34.11 ± 6.76 Hz) while 31 low firing 163 units on average increased firing from 0.33 ± 0.11 Hz to 1.54 ± 0.29 Hz ( Figure 4B and C). Overall, 164 non-responding units that had low firing rate in response to sound showed a significant increase in 165 firing rate (n=46/69 units, p=0.03). Also, CaMKIIα-eArch3.0 inhibition caused a bidirectional effect, 166 8/26 with 27/69 decreasing (p=4e-4) and 42/69 increasing (p=0.01) firing rate ( Figure 4C). This highlights 167 the complexity of the DCN and how precaution must be taken when attempting to decrease neuronal 168 activity in vivo of the auditory brainstem using tools such as CaMKIIα-eArch3.0. and ABR peaks, where I corresponds to the auditory nerve; II, cochlear nuclei; III, superior olivary 177 complex; IV and V, inferior colliculus (Henry, 1979) is useful for verifying an intact auditory brainstem 178 system. We found all animals to display normal mean ABR waveforms (n=13) in response to 80 179 decibel sound pressure level (dBSPL) stimulation ( Figure 1D). Also, ABR mean amplitude and 180 latency was not affected by concomitant sound and light stimulation ( Figure 1E). Together this show 181 that the animal's hearing at the stimulus intensity was not impaired by the viral vector injection 182 procedure or by concomitant blue light stimulation.  T-stellate cells (Oertel et al., 2011). T-stellate cells also projects to the ipsilateral LSO (Oertel et al.,204 2011), which supports the strong labeling of ChR2 in the ipsilateral LSO ( Figure 5D). The ipsilateral 205 LSO is also labeled by VCN bushy cells as projections to the contralateral medial nucleus of the 206 trapezoid body (MNTB, Figure 5F) as well as the ipsilateral LSO were apparent, thereby indicating 207 9/26 that Chrna2-cre labels both globular and spherical bushy cells. DCN unit with response to VCN light stimulation that was prolonged and appeared increased in 220 the presence of sound ( Figure 6C). Also, one DCN unit with highly temporally precise responses to 221 VCN light showed a loss of response in the presence of sound ( Figure 6D). In summary, 4/15 units 222 responded only to light stimulation, 2/15 units responded only to sound stimulation ( Figure 6B and 223 E), and 13/15 units responded to light or combined light and sound stimulation ( Figure 6C and E). 224

Excitation of VCN Chrna2+ cells significantly increased firing rate in DCN units, both responding 225
and not responding to sound ( Figure 6F). Still, the identity of these different units has to be further 226 investigated but highlights that the Chrna2-cre line has potentials for auditory research.

13/26
Attempting to silence DCN units responding to sound, using light stimulation of CaMKIIα-287 eArch3.0 expressing neurons showed that units could be inhibited using eArch3.0, but also that green 288 light exposure generated a distinct delay in response-onset to sound. The delay was consistently 289 around 20ms suggesting polysynaptic activity, possibly from the recruitment of complex spiking 290 cartwheel cells that can respond with 20ms delay to pure tone stimulation (Parham et al., 2000). 291 Also PSTH of unidentified neurons with a 40ms delay between initial and basal firing have been 292 reported for guinea pigs (Robertson and Mulders, 2018). Our experiment could not identify the type 293 of unit responsible for this delay, but it shows that silencing CaMKIIα-eArch3.0 expressing neurons 294 is not enough to disrupt sound generating activity in the DCN circuit. Some studies have pointed 295 to technical problems when using the proton pump eArch3.0, as it may affect intracellular pH of 296 presynaptic membranes and promote neurotransmitter release if light is applied continuously for 297 several minutes (Mahn et al., 2016). Mahn et al. (2016) showed that 5 min of continuous eArch3.0 298 activity significantly increased the EPSC rate (Mahn et al., 2016 the placement of the recording electrode could influence our findings as we are only sampling local 307 neurons according to the probe location. As we describe in methods, we adjusted our coordinates to 308 the animal's skull size and recorded at three different depth to cover an as large as possible region of 309 the DCN for each animal, without inserting the probe at multiple ML/AP locations or aspirating the 310 cerebellum, as done in other rodent studies (Kaltenbach and Zhang, 2007;Shore et al., 2007;Finlayson 311 and Kaltenbach, 2009;Koehler et al., 2010;Dehmel et al., 2012;Manzoor et al., 2012

322
A limitation of our study is that we did not assess the best frequency of units. Thereby the 323 sound stimulus will not display full firing potential nor specific firing patterns (such as pauser, onset, 324 build-up units). However, as light stimulation also was brief, it allows for more direct comparison 325 between modulation of unit activity in responses to sound or after exciting DCN units with light. We 326 speculate that the decrease in firing when stimulating CaMKIIα-ChR2 positive neurons with blue 327 light pulses during sound stimulation could be part of motifs of feed-forward inhibition (Roberts and 328 14/26 Trussell, 2010); or light masking (Hernandez et al., 2014), where the cell do not respond to sound 329 because it is in refractory period after responding to light stimulation.

330
Our work also show for the first time that the Chrna2-cre line targets two different cell population 331 of the VCN; both putative T-stellate cells and bushy cells of the VCN. Using cre-dependent viral 332 constructs we could excite DCN units that did not respond directly to sound, but responded temporally 333 precise to light stimulation of the VCN, suggesting that specific circuits may be targeted using these 334 animals. We found no disruption of hearing upon optogenetic stimulation, but specifically altered 335 network activity compared to brief sound stimulation, including delayed responses and disinhibition 336 of activity. Interestingly, we also found altered spiking activity for units not responding directly to 337 sound. Furthermore, stimulation of bushy cells may especially be useful for studies of the calyx of 338 Held presynaptic release and/or sound localization studies using the Chrna2-cre line. Together, these 339 results opens up for more detailed control of DCN circuit output in vivo and novel tools for studying 340 tinnitus mechanisms. can be used to manipulate unit firing of the DCN and to modulate response to sound. In general, 344 we found the CaMKIIα and Chrna2 promoter to be interesting tools as smaller and more specific 345 network activity modulation was achieved compared to sound stimulation. eYFP-WPRE-hGH (Vector core, at 1x10¹ 3 vm/ml). The eArch3.0 used was rAAV5/CamK2a-362 eArch3.0-eYFP (Vector core, at 2.5x10¹² vm/ml). In detail, mice were anesthetized with ketamine-363 xylazine at 90/6 mg/kg intraperitoneal (i.p.). If required, additional ketamine was re-administered (as 364 15/26 half the dose of the previous injection, i.e. 45 mg/kg and 22.5 mg/kg) during surgery. The mouse was 365 mounted into a stereotaxic device while resting on a heating block at 37°C. Eye gel (dexpanthenol) 366 was applied to avoid drying of eyes during surgery. The head was wiped with polyvidone-iodine (10%) 367 to avoid infections. The skin was anesthetized with lidocaine hydrochloride 3% before a straight 368 incision was made. After the incision, hydrogen peroxide 3% was applied onto the exposed skull to 369 remove the connective tissue and to visualize sutures.

370
The DCN coordinates were taken from Franklin and Paxinos (Franklin et al., 2008). Specifically, we 371 used -6.1mm anteroposterior (AP), -2.3mm mediolateral (ML), and -4.3mm and -3.8mm dorsoventral 372 (DV, two steps). For each animal, those coordinates were corrected by multiplying by the normalized 373 bregma-lambda distance (mouse's bregma-lambda in mm divided by 4.2 -the average bregma-lambda 374 distance from the mice used in Paxinos and Franklin's atlas), to account for head size differences. 375 Additionally, the vertical distance between the bregma and the point in the skull at the AP and 376 ML coordinate was subtracted from the DV coordinate, so that the DCN can be reached using the 377 brain surface as reference. A small mark was made at the AP and ML coordinates and a small 378 hole was carefully drilled with a dental microdrill (Beavers Dental, Morrisburg, Canada). Next 379 pre-aliquoted virus (20% for Cre-dependent, 30% for CaMKIIα-dependent vectors) was rapidly 380 thawed and withdrawn into a 10µl Nanofil syringe with a 34-gauge removable needle, at the speed of 381 of transparent viral solution that sometimes fails to be withdrawn due to technical issues), and 2) 394 animals can be sacrificed after only a few days (compared to waiting 2-4 weeks for viral expression) 395 to confirm the appropriate location of fluorescent signal.

Sound calibration and sound stimulation 397
As different sound devices can have inherent shifts in unit level, and thereby in the signal generation, 398 the sound card was initially calibrated using an oscilloscope. A 10kHz sine wave of 1V amplitude was 399 written to the card, and the sound card output amplification factor was recorded as 1 divided by the 400 amplitude of the output signal. All sound signals were multiplied by the output amplification factor 401 before being written to the card. We connected the sound card output to the sound card input, and 402 a 1V 10kHz sine wave was played and recorded. The input amplification factor was measured as 1 403 divided by the amplitude of the recorded signal, and signals read from the board were multiplied by it 404 before any further processing. A loudspeaker (Super tweeter ST400 trio, Selenium Pro) was calibrated 405 16/26 using a microphone (4939-A-011, Brüel and Kjaer, Denmark) 4.5-10 cm in front of the speaker. Sound 406 pulses (2s duration) were generated at the desired frequency bands with logarithmically decreasing 407 amplification factors (voltage output to the speaker) and simultaneously recorded using a personal 408 computer, and the power spectral density (PSD) of the recorded signal was calculated using a Hann 409 window with no overlap. Root mean square (RMS) was calculated as where P SD is a 1×n array and BinSize is the spectral resolution. The intensity in decibels sound 411 pressure level (dBSPL) was calculated as where MicSens vpa is the microphone sensitivity in V/Pa, 0.004236V/Pa for our microphone. All data 413 was saved to disk and loaded to provide the correct amplification factors for each sound intensity 414 used for sound stimulation. The frequency band generated corresponds to the frequency band of 415 greatest power in the signal spectrum, with border frequencies strongly attenuated (Supplementary 416 Figure S1). Sound calibrations were routinely repeated before every beginning of an experimental 417 group. The full sound calibration tests 300 amplification factors for each frequency band, providing 418 0.5 dBSPL precision. All hardware described here are outlined in Supplementary Table S1.

419
Sound stimulation consisted of sound pulses of gaussian white noise filtered from 5 to 15kHz, 420 intensity of 80dBSPL and duration of 3ms, presented at 10Hz (3ms of sound pulse followed by 97ms 421 of silence), repeated for 5 blocks of 200 pulses. were presented 4ms before the sound pulses, so that these are embedded in the light pulse.

453
For experiments using sound synchronized with light stimulation three outputs would be required 454 (one carrying the sound signal to a speaker, one carrying the sound square waves /timestamps and 455 the third carrying square waves for light trigger/timestamps). Here a square wave of twice the length 456 for light stimulation was used, so when simultaneous sound and light stimulation is required, the 457 sound pulse is written to channel 1 while the sum of the sound and light square waves are written to 458 channel 2 (Supplementary Figure S2D). Thereby channel 2 triggers the laser (amplitude >3.3V), as 459 well as provides edges for timestamp detection. Animals were anesthetized with ketamine-xylazine (90/6 mg/kg i.p.) and an additional injection 462 of (ketamine 45 mg/kg) if surgery required. The anesthetized mouse was placed on an electric 463 thermal pad (37°C) and fixed into a stereotaxic frame with ear bars holding in front of and slightly 464 above ears, on the temporal bone, to not block the ear canals. The skin over the vertex was 465 removed and hydrogen peroxide (3%) was applied on the skull to visualize sutures. All coordinates 466 were corrected as for the virus injection procedure. Next, three small holes were drilled: at AP=-467 6.1mm ML=-2.3mm (left DCN, for probe placement); at AP=-6.1mm ML=2mm, (for optic fiber 468 placement); and at AP=-2mm ML=1mm (for reference). Next, a micro screw was fixed in the 469 reference coordinate using Polymethyl methacrylate. The optic fiber was inserted into the brain using 470 a micromanipulator positioned in a 45°angle to a dept of 5. Finally, the p-value is calculated as 497 p = (g + 1)/(N + 1) where g is the number of simulated histograms with corrected spike count bigger than the real unit 498 spike count, and N is the total number of simulated histograms, which here is the number of trials 499 presented at that unit recording. A cell was classified as responsive for a stimulation if the resulting 500 p-value was <0.05. Cells were classified as responsive to sound only, light only, sound+light only or 501 sound and light. Additionally, cells responding to sound stimulation were classified as light-masked 502 when they respond to sound stimulation but do not respond to sound+light stimulation. Spike rate 503 was calculated as spike events per second along all the recording (including the stimulation period). 504 The threshold of 9 Hz was considered to separate between slow-and fast-spiking neurons, since ∼88% 505 of neurons had firing rate < 6.42Hz and the remaining ∼12% had firing rate > 9.24Hz. Student's 506 t-test, two-tailed, unequal variance was applied to compare firing rate between neurons, and all firing 507 rate values are represented as frequency ± standard error of the mean (s.e.m).

509
Auditory brainstem responses (ABRs) were extracted from the same extracellular unit recordings. 510 Data was filtered (4th order butterworth digital bandpass filter from 500 to 1500Hz), sliced (3ms 511 before and 12ms after each sound pulse) and averaged. ABR peaks were detected as a positive value 512 19/26 one standard deviation (SD) above the mean, larger than the previous value, and larger or equal the 513 next value. ABR peak values and latencies were then grouped by sound or light stimulation (see 514 Figure 8).

516
Intracardial perfusion was carried out by deeply anesthetizing mice with ketamine/xylazine (180/12 517 mg/kg). Animals were fixed in a polystyrene plate, and a horizontal incision was made in the skin at 518 the level of the diaphragm. Thoracic cavity was open by cutting the ribs laterally and the sternal 519 medially. A 30G needle was inserted into the left ventricle for perfusion with cold phosphate buffered 520 saline (PBS) and an incision was made in the right atrium to allow for out-flow. In total, 20-30ml 521 of cold PBS followed by 20-30ml of fixative (4% paraformaldehyde in 0.1M phosphate buffer; pH 522 7.4) was used. Next, the brain was dissected and stored in 4% paraformaldehyde overnight. For 523 free-floating vibratome (OTS-4000, EMS, Hatfield) sections the brain was stored in PBS before 524 slicing; and for cryostat sections, the brain was kept in PBS with 30% sucrose until dehydrated 525 (visualized by the brain sinking to the bottom of the solution), and frozen using isopentane at -60°C. 526 Horizontal sections (120µm thick) of the brainstem, containing the DCN, were collected on glass 527 slides and kept dark until examination of fluorescent expression by neurons. Cell nuclei were stained 528 with 4',6-diamidino-2-Phenylindole (DAPI) (Sigma) to visualize cell layers and borders of the DCN 529 and VCN. Expression of optogenetic proteins was visualized by detection of genetically expressed 530 eYFP. Images were collected using Zeiss Observer Z1 fluorescence microscope or a Zeiss Examiner Z1 531 confocal microscope. The objectives N-Achroplan 5x/0.15; N-Achroplan 10x/0.25; Plan-Apochromat 532 20x/0.8; and Plan-Neochromat 40x/0.75 were used. Images were collected using AxioVision and Zen 533 software, respectively, and edited for brightness and contrast in ImageJ (NIH, Schneider et al., 2012). 534

535
The CLARITY procedure followed standard protocol and was previously describe for another brain 536 region . Data from the auditory brainstem was collected during the same 537 experiment as previously published  while the video attached here (Supplementary 538 Video S1) was compiled specifically for the cochlear nucleus containing region. were done using Open-ephys GUI (Siegle et al., 2017). Calculations were done using Scipy (Jones 542 et al., 2001) and Numpy (Van Der Walt et al., 2011), and all plots were produced using Matplotlib 543 v2.2.4 (Caswell et al., 2019;Hunter, 2007). Spikes were detected and clustered using Klusta, and 544 visual inspection was performed using Phy (Rossant et al., 2016). All scripts used for stimulation 545 control and data analysis are available at https://gitlab.com/malfatti/SciScripts.

20/26
The authors declare that the research was conducted in the absence of any commercial or financial 548 relationships that could be construed as a potential conflict of interest.