Fast Ca2+ Transients of Inner Hair Cells Arise Coupled and Uncoupled to Ca2+ Waves of Inner Supporting Cells in the Developing Mouse Cochlea

Before the onset of hearing, which occurs around postnatal day 12 (P12) in mice, inner hair cells (IHCs) of the immature cochlea generate sound-independent Ca2+ action potentials (APs), which stimulate the auditory pathway and guide maturation of neuronal circuits. During these early postnatal days, intercellular propagating Ca2+ waves elicited by ATP-induced ATP release are found in inner supporting cells (ISCs). It is debated whether IHCs are able to fire Ca2+ APs independently or require a trigger by an ISC Ca2+ wave. To identify the Ca2+ transients of IHCs underlying Ca2+ APs and to analyze their dependence on ISC Ca2+ waves, we performed fast Ca2+ imaging of Fluo-8 AM-loaded organs of Corti at P4/P5. Fast Ca2+ transients (fCaTs) generated by IHCs were simultaneously imaged with Ca2+ waves in ISCs. ISC Ca2+ waves frequently evoked bursts consisting of >5 fCaTs in multiple adjacent IHCs. Although Ca2+ elevations of small amplitude appeared to be triggered by ISC Ca2+ waves in IHCs of Cav1.3 knockout mice we never observed fCaTs, indicating their requirement for Ca2+ influx through Cav1.3 channels. The Ca2+ wave-triggered Ca2+ upstroke in wildtype IHCs occurred 0.52 ± 0.27 s later than the rise of the Ca2+ signal in the adjacent ISCs. In comparison, superfusion of 1 μM ATP elicited bursts of fCaTs in IHCs starting 0.99 ± 0.34 s prior to Ca2+ elevations in adjacent ISCs. PPADS irreversibly abolished Ca2+ waves in ISCs and reversibly reduced fCaTs in IHCs indicating differential involvement of P2 receptors. IHC and ISC Ca2+ signals were however unaltered in P2X2R/P2X3R double knockout or in P2X7R knockout mice. Together, our data revealed a fairly similar occurrence of fCaTs within a burst (56.5%) compared with 43.5% as isolated single fCaTs or in groups of 2–5 fCaTs (minibursts). We provide evidence that IHCs autonomously generate single fCaTs and minibursts whereas bursts synchronized between neighboring IHCs were mostly triggered by ISC Ca2+ waves. Neonatal IHCs thus spontaneously generate electrical and Ca2+ activity, which is enhanced and largely synchronized by activity of ISCs of Kölliker’s organ indicating two sources of spontaneous activity in the developing auditory system.

Although IHC Ca 2+ APs, which are based on regenerative opening of Ca v 1.3 Ca 2+ channels, K v and SK2 K + channels, have been analyzed in depth by patch-clamp recordings (Brandt et al., 2003;Marcotti et al., 2003b;Johnson et al., 2007Johnson et al., , 2011Johnson et al., , 2013Johnson et al., , 2017Sendin et al., 2014) the accompanying Ca 2+ transients in IHCs have not been studied so far. Moreover, we still lack simultaneous imaging data resolving the temporal and spatial correlation between fast Ca 2+ transients in IHCs and Ca 2+ waves in ISCs. We therefore conducted Ca 2+ imaging in the immature mouse organ of Corti to determine whether Ca 2+ transients in IHCs occur independent of ISC Ca 2+ waves.

Animals
All experimental procedures were conducted in agreement with the European Communities Council Directive (2010/63/EU) in accordance with the German law on the use of laboratory animals and were approved by the Saxonian District Government, Leipzig (TVV59/16 and T34/16), and the regional board for scientific animal experiments of the Saarland. Prehearing mice (P4-P6) of either sex were studied. NMRI mice (Charles River, Sulzfeld, Germany and bred in-house; 64 animals) were used unless stated otherwise and are designated as controls. Ca v 1.3 −/− mice (Platzer et al., 2000), which were bred to either C57BL6/N (seven animals) or NMRI background (13 animals) for >10 generations, mice deficient for the purinergic receptor P2X7 (P2X7R −/− mice; The Jackson Laboratories, Bar Harbor, ME, USA; 21 animals) and P2X2R/P2X3R double knockout mice (P2X2R −/− /P2X3R −/− ; 20 animals; Cockayne et al., 2005) both on C57BL6/N background were used. Every experimental observation was confirmed in at least three experiments using different animals.

Tissue Preparation
Neonatal mice were sacrificed by decapitation. Cochleae were immediately dissected in ice-cold physiological solution mimicking perilymph (in mM): 143 NaCl, 5.8 KCl, 1.3 CaCl 2 , 0.9 MgCl 2 , 5.6 glucose, 10 HEPES and 0.9 NaH 2 PO 4 , 305 mOsm/l, pH 7.35. The organ of Corti was dissected out of the cochlea with great care to preserve the integrity of the tissue. The apical turn was cut and placed in a recording chamber underneath a soft thread of a nylon mesh, which was attached to a steel ring (Figure 1). Nylon threads had an average distance to the recording window of 253 ± 47 µm (determined for 26 recordings). After cautiously removing the tectorial membrane, the specimen was incubated in the perilymph-like solution containing 10 µM of the cell-permeable Ca 2+ indicator Fluo-8 AM (AAT Bioquest Inc., Sunnyvale, CA, USA) that had been dissolved in a Pluronic-DMSO mixture (100 mg/ml; Thermo Fisher Scientific Inc., Waltham, MA, USA). To allow for uptake of Fluo-8 AM and hydrolysis, the recording chamber with the tissue was kept in a dark, humid plastic box for 40 min at room temperature. The chamber was mounted on the stage of the upright laser scanning microscope LSM 710 (Zeiss Microscopy GmbH, Göttingen, Germany) and perfused with perilymph-like solution for at least 7 min to remove any Fluo-8 AM prior to the experiment. Only healthy looking epithelia with smooth surfaces, absence of vacuoles and low background fluorescence in both IHCs and ISCs were used for imaging (Figure 2A). IHCs were recorded within a fractional distance of 10-25% from the apex, which corresponds to a frequency range of about 7-10 kHz in the adult animal (Müller et al., 2005). Agonists and antagonists (ATP; 1 or 10 µM; PPADS; 100 µM; both Sigma-Aldrich, St. Louis, MO, USA) were diluted in perilymph-like solution FIGURE 1 | Configuration for imaging inner supporting cell (ISC) Ca 2+ waves and inner hair cell (IHC) Ca 2+ transients in the immature cochlea of the mouse. Transmitted light image of an acutely dissected whole-mount preparation of the apical turn from a P4 control mouse organ of Corti. The focus is set on the hair cell region (HC) with three rows of outer hair cells and one row of IHCs and the great epithelial ridge with the ISCs. Because of the upward slope of the epithelium towards the modiolus the outer sulcus region (OS) and interdental cells (IDCs) are out of focus. The outlet of a gravity-fed application system made from tapered PVC tubing with an inner diameter of 400 µm was placed at the outer sulcus (pillar) side for gentle application of solutions with blockers or agonists. The tissue was held by two soft nylon threads (one is visible as the rightmost black shadow) fixed to a stainless steel ring that fit into a recording chamber. Care was taken that only the outermost edges of the tissue were held down to the glass bottom of the chamber. One to two recordings were made per apical cochlear turn. Recording frames are depicted by the rectangles labeled 1 and 2. The distance from the thread to the recording site was >200 µm. Scale bar, 50 µm.
(see above) and applied by a gravity-driven custom-made application system with a multi-barreled pipette (Figure 1), which led to a change of the solution above ISCs and IHCs within 3-6 s. Unless stated otherwise, the specimen was constantly superfused with either perilymph-like solution, the same solution containing the respective agonist/antagonist, or with Ca 2+ -free solution, respectively, via the application system at 0.29 ml/min. To avoid movement artifacts caused by changes in fluid pressure, the switch between two solutions was performed by simultaneously closing respectively opening the valves of the respective reservoirs. The Ca 2+ -free solution containing the following (in mM): 143 NaCl, 5.8 KCl, 2.2 MgCl 2 , 5.6 glucose, 10 HEPES, 0.9 NaH 2 PO 4 and 0.5 EGTA acid, 305 mOsm/l, pH 7.35. The distance between the scan field and the lateral edges of the tissue combined with the application of perilymph-like solution without/with agonist/antagonists from the pillar side prevented spillover of ATP from potentially damaged cells and ISCs. The delay in solution flow from reaching the IHCs to reaching the ISCs was estimated to be <20 ms.

Ca 2+ Imaging, Analysis and Data Handling
Data were recorded using a LSM 710 based on an upright AxioExaminer microscope with 40×/1.0 NA or 20×/1.0 NA Plan Apochromat water immersion objectives and Zen-2012 SP1 (Black edition) software (all Zeiss). Fluo-8 fluorescence was excited with the 488 nm line of an argon laser. Laser intensity was set to very low levels to allow for long continuous recordings (>10 min, up to 30 min) while avoiding photobleaching and strong phototoxicity. We opened up the pinhole to maximal seven airy units accepting a reduced spatial resolution in favor of increased fluorescence intensity at the chosen low laser power and a higher scan speed. Recordings were performed at room temperature (21 • C ± 1 • C). Because Fluo-8 based on Fluo-4 (Gee et al., 2000) is a non-ratiometric Ca 2+ indicator with a high quantum yield and a K d of 389 nM we determined the relative fluorescence changes ∆ F F 0 corresponding to [Ca 2+ ] i changes in ISCs and IHCs off-line: where F(t) is the fluorescence intensity of a pixel at a given time t and F 0 the fluorescence of this pixel averaged over the first 10 images without obvious Ca 2+ changes.
For imaging the spread of ISC Ca 2+ waves in Kölliker's organ, scan fields of about 200 µm × 100 µm were chosen resulting in frame rates of ≤5 Hz (scan times >200 ms). To resolve the much faster fCaTs in IHCs, we increased the frame rate by: (i) imaging smaller scan fields; and (ii) increasing the pixel size to 0.4 µm × 0.4 µm (up to 0.6 µm × 0.6 µm). Scan fields were 140 µm ± 21 µm in width (axis parallel to the IHCs) and 27 µm ± 5 µm in height (Figures 1, 2A) and comprised the basolateral poles of 7-9 (up to 23) IHCs with a stripe of ISCs extending up to 36 µm in the modiolar direction. This led to frame rates of >31 Hz and to corresponding shorter scan times of ≤32 ms (except Figure 6, 36.2 ms), which allowed simultaneous imaging of Ca 2+ waves in ISCs and fCaTs in IHCs. For achieving an even better temporal resolution of fCaTs, a subset of control IHCs was imaged at 175 Hz (scan time of a single frame: 5.73 ms). With these settings, photobleaching amounted to 0.5 F/F 0 at maximum during periods of ≥6 min. Offline analysis and graphical representation was performed using FIJI (Schindelin et al., 2012), IGOR 6.12 (Wavemetrics, Inc., Lake Oswego, OR, USA) and Inkscape TM 0.91. Regions of interest (ROIs) in a time series (t-series) of fluorescence images were defined by activity patterns within IHCs and ISCs and were validated using a high-resolution transmitted light image acquired prior to the t-series with a 100-fold smaller pixel area. Transmitted light images were also taken after a t-series to check for a potential tissue drift. Data are reported as mean ± standard deviation (SD) except data in Figure 10B, where mean ± SEM is given.

RESULTS
Using Fluo-8 AM-incubated explants of the apical organ of Corti, which were dissected with greatest care, we observed spontaneously occurring Ca 2+ waves and corresponding waves of light refraction (crenation waves) in the region of ISCs. This is consistent with previous studies of ISCs of Kölliker's organ in the immature organ of Corti (Tritsch et al., 2007;Anselmi et al., 2008;Dayaratne et al., 2015;Wang et al., 2015). ISCs generated diverse Ca 2+ phenomena ranging from Ca 2+ elevations confined to a single ISC (Ca 2+ transients) to Ca 2+ waves extending over various distances. Small Ca 2+ waves propagated radially to maximum two neighboring ISC circles comprising about 200-600 µm 2 , whereas large waves covered areas from 600 µm 2 upto >2200 µm 2 within Kölliker's organ involving about 75 ISCs. Sources and directions of the waves were random, indicating that the tissue had not been impaled in or near the field of view and that ATP potentially leaking out from injured or dying cells at the outermost edges of the tissue piece did not affect ISCs and IHCs in the field of view.
Simultaneous Imaging of Ca 2+ Waves in ISCs and Fast Ca 2+ Transients in IHCs  (Figure 2A). IHCs are enwrapped by specialized ISCs, the inner phalangeal cells, which stay throughout maturity. We do not differentiate here between transient ISCs of Kölliker's organ and inner phalangeal cells because they cannot be unambiguously discriminated in DIC images. IHCs numbered 6-8 within the Fluo-8 AM-loaded specimen showed a slightly elevated initial raw fluorescence at t = 0 s (green; Figure 2B). Usually care was taken to eliminate frames in which IHCs showed transient fluorescence elevations for determining the baseline (the first 10 frames) to avoid F/F 0 values <0 at later time points. Here, we show such an example on purpose to demonstrate that the calculation of relative fluorescence changes (  (Figures 2G,H). fCaTs occurred in IHCs 1, 3, 4, 5, 7. Averaging F/F 0 in ROIs indicated in Figure 2A as a function of time ( Figure 2I) revealed fCaTs in IHCs 3, 4 and 7, and a large and long-lasting Ca 2+ transient in ISC III triggering the ISC Ca 2+ wave seen at maximum spread in Figure 2H. IHC 3 was active before its adjacent ISC III generated a Ca 2+ wave ( Figure 2I, inset i), whereas the fCaTs in IHCs 4 and 7 started after Ca 2+ had risen in the nearby ISC III. While IHC 7 produced 4 fCaTs shown at the onset of the recording, no elevation in Ca 2+ could be observed in ISCs (Figure 2I, left).

Properties and Incidence of Fast Ca 2+ Transients in IHCs
To study the temporal properties of fCaTs in detail, IHCs were imaged at 175 Hz (scan time of a single frame: 5.73 ms), which required a smaller scan field not allowing for simultaneous imaging of ISCs. During a 573 s-long recording of eight IHCs, seven of them generated fCaTs. Single fCaTs, which were not generated within a burst of activity, had an average half-width of 32.5 ± 9.8 ms (n = 18) and decayed to baseline after 895.9 ± 381.2 ms (n = 17; Figure 3A). Time to peak, measured from 10% of the peak amplitude to the maximum fluorescence, amounted to 16.3 ± 5.5 ms (n = 19). Decay time defined as time from peak back to 10% above baseline was 879.7 ± 379.1 ms (n = 17), thus revealing a much slower decay in fluorescence compared to the rise ( Figure 3A, Table 1). Besides single isolated fCaTs, IHCs generated fCaTs within events we designate as minibursts (2-5 fCaTs in close sequence, Figure 3B) and within bursts with more than five consecutive fCaTs ( Figure 3C, top). The mean frequency of fCaTs was 0.32 ± 0.11 Hz, but they were not evenly distributed. Within bursts, frequencies varied from 4 Hz to 10 Hz ( Figure 3C, bottom) similar to previously reported frequencies in bursts of IHC Ca 2+ APs (Johnson et al., 2011(Johnson et al., , 2017Sendin et al., 2014). A total of 49 bursts and 54 minibursts were generated in the seven active IHCs during 573 s recording time. Of those, 42 bursts and 18 minibursts occurred simultaneously in adjacent IHCs with at least two IHCs involved (mean number of simultaneously active IHCs: 4.62 ± 1.85). Notably, F/F 0 continuously exceeded values of 1.5 only within bursts of fCaTs ( Figure 3C, top).
To study the incidence of the three types of IHC fCaTs (single events, minibursts and bursts) perfusion was switched off in a set of experiments to exclude the possibility that the bath flow may have caused an interaction between ISCs and IHCs. Fifty IHCs from four control mice were monitored in seven recordings lasting between 544 s and 1120 s. The incidence of fCaTs was analyzed during a total observation time of 598 min (total number of IHCs multiplied by the length of respective recordings), irrespective of Ca 2+ elevations in ISCs. IHCs generated 9668 fCaTs with 18.3% being single fCaTs, 25.2% occurring within minibursts of 2-5 fCaTs and 56.5% occurring within bursts of 6-67 fCaTs ( Figure 3D). A detailed analysis showed that bursts consisted on average of 15.3 ± 9.5 fCaTs (n = 353) whereas minibursts contained an average number of 2.8 ± 1.0 fCaTs (n = 867; Table 1). During bursts of fCaTs, F/F 0 generally increased to >2 with maximal values <4 and declined slowly within 10 s. In contrast, F/F 0 reached the baseline within 1-2 s in events with ≤5 fCaTs (see Figures 3B,C). These data indicate that although the majority of fCaTs (56.5%) were embedded in a burst, a comparable number (43.5%) appeared outside of bursts as single fCaTs or as part of a miniburst ( Figure 3D).   of ISCs, however, did not vanish in 0 mM Ca 2+ (Figure 4A, black trace). The decrease in the baseline of F/F 0 in both IHCs and ISCs was likely caused by a reduction in the intracellular Ca 2+ concentration in the Ca 2+ -free solution.
Because Ca 2+ APs of IHCs crucially depend on Ca v 1.3 channels (Brandt et al., 2003), which account for more than 90% of voltage-gated Ca 2+ currents in neonatal IHCs (Platzer et al., 2000), Ca 2+ measurements were conducted in IHCs and ISCs of Ca v 1.3 −/− mice in standard (1.3 mM) extracellular Ca 2+ concentration (Figures 4D,E). In 103 IHCs from three Ca v 1.3 −/− mice we did not observe a single fCaT ( Figure 4E) during a total recording time of 787 min (recording length multiplied by number of IHCs). However, those IHCs contacted by an ISC Ca 2+ wave typically showed slowly rising Ca 2+ elevations of small amplitude ( F/F 0 <1) temporally matching the Ca 2+ signals of the ISC (Figure 4E, left). This slow IHC signal may have resulted from: (i) potential bleedthrough of the fluorescence signal of the ISCs; (ii) Ca 2+ influx through P2X receptors; or (iii) Ca 2+ influx through the small percentage of Ca 2+ channels other than Ca v 1.3 (mostly Ca v 1.2 and some Ca v 2.3-mediated currents; Brandt et al., 2003). We never saw this kind of small slow Ca 2+ elevation in IHCs in the absence of an invading ISC Ca 2+ wave. On rare occasions, IHCs of Ca v 1.3 −/− mice did not respond to an invading Ca 2+ wave with small and long-lasting Ca 2+ elevations (Figure 4E, left). Summarizing the results from control mice in 0 mM Ca 2+ solution and those from Ca v 1.3 −/− mice in 1.3 mM extracellular Ca 2+ we conclude that fCaTs in IHCs reflect the Ca 2+ increases mediated by Ca v 1.3 channels.

Synchronized Bursts of fCaTs in Adjacent
IHCs Were Triggered by ISC Ca 2+ Waves or by Extracellular ATP IHCs were able to generate fCaTs autonomously without being triggered by ISC Ca 2+ waves (see Figure 3 and below). However, when an ISC Ca 2+ wave approached the IHC region it typically evoked bursts of fCaTs in a number of adjacent IHCs (Figures 5A-C). The spontaneous Ca 2+ wave involving ISC I (at t ∼70 s) triggered bursts in IHCs 1 and 2, and although ISC IV was silent, IHCs 3-5 were active at the same time. The Ca 2+ elevation of ISC I preceded that of the IHCs (Figure 5C). Analyzing 12 waves causing burst events in IHCs from a larger data set, the Ca 2+ wave-evoked fluorescence elevation in ISCs lining the IHC row preceded the onset of fCaTs within bursts in the closest IHC by 0.52 ± 0.27 s (mean ± SD). Though less frequent, synchronized IHC bursts were also observed without any ISC activity within the recording frame as e.g., in IHCs 1, 2, 4, 5, 6, 8 at t ∼90 s (Figure 5B and see below).
To validate previous results showing ATP-induced inward currents in neonatal IHCs  in our preparation, we tested the effect of ATP on IHC fCaTs. Application of 1 µM ATP for 5 s or 10 s from the pillar side ( Figure 5B) strongly increased F/F 0 in all cells of the sensory epithelium (cells of the inner and outer sulcus, interdental cells (IDCs), spiral ganglion tissue and IHCs), indicating an overall expression of purinergic receptors at the age of P4/P5. The ATP-evoked increase in F/F 0 in ISCs resembled that induced by a spontaneous Ca 2+ wave, lasting for 10-15 s (Figure 5B; black traces). In IHCs, ATP-evoked bursts of fCaTs strongly increased F/F 0 up to 4. Repetitive ATP stimulation (five times at intervals >30 s) did not deplete the burst responses in IHCs (see Figure 5B). However, fCaTs within ATP-induced bursts were smaller in amplitude compared with fCaTs triggered by a Ca 2+ wave and were challenging to detect ( Figure 5D). Together, these experiments corroborate earlier studies showing that ATP generates Ca 2+ waves in ISCs (Tritsch et al., 2007. To test the hypothesis if ATP can activate IHCs independently of ISC Ca 2+ elevations, the ATP had been applied from the pillar side (Figures 5B,D,E). Such a direct stimulation of purinergic receptors on IHCs evoked bursts of fCaTs that preceded Ca 2+ elevations in ISCs by 0.99 ± 0.34 s (n = 23 recordings). When taking into account that the solution flow reached the IHCs ∼20 ms earlier than the ISCs, these results still indicate that ATP can elicit bursts in IHCs independent of ISC Ca 2+ elevations.

Block of P2 Receptors by PPADS Irreversibly Suppressed Ca 2+ Waves in ISCs and Reversibly Reduced Fast Ca 2+ Transients in IHCs
To further analyze the importance of purinergic signaling on fCaTs, we tested the effect of 100 µM PPADS, a broad spectrum blocker of P2Y and P2X receptors (Abbracchio et al., 2006;Coddou et al., 2011). Before PPADS application, control measurements were conducted for 180 s in order to simultaneously record Ca 2+ waves in ISCs and single fCaTs, minibursts and bursts in IHCs (Figures 6B-D). PPADS superfusion (100 µM) completely abolished fCaTs in 33 of 44 IHCs. The remaining 11 IHCs still generated single fCaTs, but neither minibursts nor bursts. Single fCaTs, minibursts or bursts were recorded after washout in 27 IHCs, albeit with strongly reduced amplitude (Figures 6C,E). Notably, blocking P2 receptors had an even stronger effect on ISCs (Figures 6B,E): the incidence of ISC Ca 2+ waves was reduced from 5.4 waves/100 s (35 waves in 645 s total recording time, three control mice) to 1.0 waves/100 s (five waves during 509 s in total). After removal of PPADS, the ability of ISCs to generate Ca 2+ waves was very weak, as only three Ca 2+ waves occurred during 1910 s washout (0.2 Ca 2+ waves/100 s; Figures 6B-E).

Fast Ca 2+ Transients Can Occur Independently of ISC Ca 2+ Waves
It has been suggested previously that IHCs can generate Ca 2+ APs independently of Ca 2+ wave activity in adjacent ISCs (Johnson et al., 2011(Johnson et al., , 2017Sendin et al., 2014). These studies are based on electrical recordings where patch pipettes could have potentially liberated ATP from injured cells at the recording site, making the interpretation of ''spontaneous activity'' challenging. Performing imaging only, we were able to simultaneously observe IHC fCaTs and ISC Ca 2+ waves and their potential coupling (Figure 8). Though autonomously generated fCaTs were mostly solitary fCaTs or minibursts, spontaneous bursts were observed, too. An example is shown in Figures 8A,B where bursts of fCaTs appeared in IHC 2 without Ca 2+ elevation in any ISC as indicated by asterisks referring to the IHC 2-trace in red. Yet, bursts with >5 fCaTs were more frequently triggered by a Ca 2+ wave reaching the IHC region ( Figure 8B, indicated by triangles, and Figure 8C) with an overall delay of 0.52 ± 0.27 s (n = 12 waves from four animals, see above). However, not every ISC Ca 2+ wave elicited burst-like activity in the adjacent IHC 2 (Figure 8D). This lack of activity of IHC 2 was transient because it autonomously generated bursts (Figure 8B * * ) or single fCaTs and minibursts ( Figure 8E) as well as ISC-evoked activity ( Figure 8B, single and double triangle) thereafter. Together, our data support the hypothesis that isolated fCaTs and minibursts, but also bursts can be generated in IHCs irrespective of ISC Ca 2+ waves.

Simultaneous Bursts in Neighboring IHCs Were Mostly Triggered by ISC Ca 2+ Waves
Due to their spatial spread, most Ca 2+ waves that invaded the IHC area evoked bursts of fCaTs in more than one IHC, causing synchronous Ca 2+ activity in two to eight IHCs. The upper number may be larger, but was limited by the size of the scan field. An example in Figure 9 shows the spatiotemporal coupling of a wave starting from ISC II, which elicited bursts in IHCs 1, 2, 3, 4 (Figures 9A-C,F). Notably, IHCs also generated simultaneous bursts without being triggered by an ISC Ca 2+ wave. IHCs 4-8 responded with bursts to a wave starting in ISC VIII (Figure 9D), but IHCs 1-3 at 10-30 µm distance from the ISC Ca 2+ signal also responded with bursts (Figures 9E,F) suggesting direct coupling of activity between IHCs. This means that simultaneous Ca 2+ activity in adjacent IHCs can occur without a Ca 2+ elevation in the ISCs contacting these IHCs or their direct IHC neighbors.
We analyzed burst activity synchronized between IHCs in periods of 15 s, the approximate duration of a large Ca 2+ wave at a given ISC ROI, for data from four control mice (seven FIGURE 8 | Fast Ca 2+ transients can occur independently from ISC Ca 2+ waves. (A) Transmitted light image of two rows of ISCs and the basolateral poles of nine IHCs, which are outlined by white dotted shapes; two ROIs are labeled in red (IHC 2) and black (ISC II). Scale bar = 10 µm. (B) Changes in [Ca 2+ ] i as F/F 0 in ROI "IHC 2" and in ROI "ISC II" comprising three ISCs adjacent to IHC 2 during a period of 17 min 38 s. ROI "ISC II" was invaded by several Ca 2+ waves (black trace) whereas IHC 2 (red trace) showed bursts independent of Ca 2+ waves from ISC II (labeled by * and * * ), bursts triggered by Ca 2+ waves from ISC II (labeled by (C) and triangles), minibursts and single fCaTs. Three events indicated by black bars are shown at larger temporal resolution in (C-E). (C) A Ca 2+ wave caused a large [Ca 2+ ] i elevation in ISC II, which preceded the burst activity of IHC 2. (D) Another Ca 2+ wave caused a [Ca 2+ ] i elevation in ROI "ISC II" whereas IHC 2 remained inactive. (E) IHC 2 showed minibursts and single fCaTs whereas the adjacent ISC II did not show any activity. Data file 20141127_002, frame rate = 34 Hz (single frame = 29.4 ms), 40×/1.0 NA objective.
recordings comprising a total of 50 IHCs). The relative incidence of synchronous miniburst activity in at least two adjacent IHCs without being triggered by an ISC Ca 2+ was 40.3% (149 cases) compared with 59.7% (221 cases) with a Ca 2+ wave as trigger ( Figure 9G). Spontaneous synchronous bursts in at least two adjacent IHCs had a relative incidence of 32.1% (92 cases), which FIGURE 9 | Continued every IHC showed moderate (IHCs 1-3) to strong Ca 2+ elevations (IHCs 4-8).
(F) F/F 0 for ROIs of IHCs 1-3 (colored) and the ISC origin of wave 1 (ISC II, black, top panel); of IHCs 4-6 (colored) and ISC V adjacent to IHC 5 (dark gray, middle panel), and of IHC 7 and IHC 8 (colored) and the ISC origin of wave 2 (ISC VIII, gray, bottom). Wave 1 originating from ISC II appeared to synchronize IHCs 1-5 whereas wave 2 originating from ISC VIII synchronized all IHCs in the scan field. The Ca 2+ increase of ISC V (middle panel, dark gray) was delayed compared with the bursts in adjacent IHCs 4-6. Vertical dotted lines and labels (B-E) in the top row refer to the time points at which images (B-E) were captured. Data file 20150114_006, frame rate = 32 Hz (single frame = 31.1 ms), 20×/1.0 NA objective. (G) Summary of the incidence of synchronized IHC activity occurring simultaneously in at least two adjacent IHCs as minibursts and bursts, respectively, without (top, "spontaneous") or with an invading ISC Ca 2+ wave as trigger (bottom). Absolute number of synchronized IHC events is given in the bars.
increased to 67.9% (195 cases) when triggered by a Ca 2+ wave ( Figure 9G). A possible explanation for the coupling of Ca 2+ activities between IHCs in the absence of Ca 2+ waves might be an activity-driven K + efflux from IHCs through their voltage-gated K + channels. If an increase in the extracellular K + concentration around IHCs was not effectively prevented by inner phalangeal cells/ISCs, neighboring IHCs would become depolarized, which in turn would trigger Ca 2+ APs and hence fCaTs. A similar K + accumulation-based activation of IHCs was proposed as coupling mechanism between Ca 2+ waves and IHC activity, although the latter originates from K + efflux of ISCs including inner phalangeal cells (Wang et al., 2015).

Efficacy of ISC Ca 2+ Waves in Triggering fCaTs in IHCs
To quantify the efficacy of a Ca 2+ wave invading the IHC area in generating fCaTs, ISC Ca 2+ signals were classified into four groups ( Figure 10A). Wave class 0 reflects the lack of ISC activity during a 15 s period, which is the approximate duration of a large ISC Ca 2+ wave. Class 1 defines a non-propagating Ca 2+ transient restricted to a single ISC characterized by a mean duration of 8.5 ± 3.7 s (n = 12 Ca 2+ transients). Class 2 denotes a small ISC Ca 2+ wave that radially propagated ≤2 cells from its origin (mean duration of 10.1 ± 1.9 s, n = 9 waves), whereas class 3 defines a large ISC Ca 2+ wave propagating >2 cells from its origin (mean duration of 12.5 s ± 2.0 s, n = 19 waves; Figure 10A). Small and large Ca 2+ waves were only included into the analysis if they originated within the scan field. A rank order of IHC activities was defined as follows: burst > miniburst > single fCaT > inactive IHC. For a given wave class, the IHC event type with the highest rank was determined for each IHC in the scan field during the respective time window. These event numbers were added, normalized to the total number of IHCs in the scan field and averaged between recordings, yielding the relative incidence of IHC events for the respective wave class ( Figure 10B).
Relating the class of ISC activity to the highest rank of the simultaneous IHC activity revealed striking differences ( Figure 10B): bursts were extremely rare when ISCs were silent (4% ± 0.8%; mean ± SEM, n = 196 time windows without ISC activity during 15 s) or showed class 1 events (2% ± 1.5%, n = 13 FIGURE 10 | Efficacy of different classes of ISC Ca 2+ waves in triggering fCaT activity in IHCs. (A) Sketch to illustrate classification of ISC activity, with a stripe of four rows of ISCs and one row of IHCs. Green color indicates elevated [Ca 2+ ] i ; wave origin is depicted by a red dot. Class 0: no ISC activity, but autonomous IHC activity; class 1: a non-propagating Ca 2+ transient restricted to a single ISC; class 2: a small ISC Ca 2+ wave that traveled to maximally two ISCs radially; class 3: a large ISC Ca 2+ wave traveling >2 cells from its origin. (B) Average incidence ± SEM of IHC Ca 2+ events during a defined period of time as a function of the ISC wave class. For each wave class, only the highest rank of the activity of an individual IHC was considered, in the order burst > miniburst > fCaT > inactive IHC (see "Results" section).

DISCUSSION
Spontaneous activity is a hallmark of developing sensory systems including the auditory system (Blankenship and Feller, 2010;Leighton and Lohmann, 2016). Spontaneously occurring Ca 2+ signals in IHC and their dependance on ISC Ca 2+ waves was presently studied with Fluo-8 AM Ca 2+ imaging on freshly dissected explants of organs of Corti. Since ATP can trigger Ca 2+ waves in Kölliker's organ of the immature organ of Corti (Tritsch et al., 2007;Anselmi et al., 2008;Majumder et al., 2010) and also Ca 2+ APs in IHCs (Tritsch et al., 2007;Johnson et al., 2011) it was necessary to avoid unintended release of ATP and putative ATP-dependent signaling. Great care was taken to keep the epithelium as intact as possible. Moreover, the recordings were performed from free-running tissue in the middle of the explant far from the lateral edges. By using small scan fields and a pixel size of ∼0.17 µm 2 it was possible to temporally resolve fCaTs in IHCs and classify them into single fCaTs, mini-bursts (2-5 fCaTs) and bursts of fCaTs. Our F/F 0 values of fCaTs in Fluo-8 AM-loaded IHCs varied from 0.5 up to 5 (in bursts), providing a much higher amplitude resolution than previously reported using Fluo-4 AM (Ceriani et al., 2016a).

IHC fCaTs Are Consistent With Ca 2+ APs
In the first postnatal week, mouse and rat IHCs generate Ca 2+ APs that vanish shortly before the onset of hearing (Kros et al., 1998;Brandt et al., 2003Brandt et al., , 2007Marcotti et al., 2003b;Johnson et al., 2005Johnson et al., , 2007Johnson et al., , 2011Johnson et al., , 2012Johnson et al., , 2013Sendin et al., 2014;Iosub et al., 2015). Ca 2+ APs appear as single events, but more often in bursts of 10-20 single APs. They are reversibly abolished in nominally Ca 2+ -free solution (Marcotti et al., 2003b) and missing in Ca v 1.3 −/− mice (Brandt et al., 2003). Similarly, the IHC fCaTs recorded in this study were reversibly suppressed in nominally Ca 2+ -free solution and absent from IHCs of Ca v 1.3 −/− mice. Time-to-peak of fCaTs measured at the highest scan rate was 16 ms, which is in the range of the half-width of 15 ms for a Ca 2+ AP at room temperature (Marcotti et al., 2003b). It is not surprising that a single fCaT lasts longer than the electrical AP because (i) Ca 2+ influx into an IHC extends to the repolarizing phase of the AP due to the large driving force for Ca 2+ and (ii) AP repolarization by delayed rectifier K + and SK2 channels (Marcotti et al., 2003a;Mammano and Bortolozzi, 2018) is, particularly in minibursts and bursts, faster than clearance of Ca 2+ from the cytosol by Ca 2+ ATPases (Grati et al., 2006) or by uptake into intracellular stores and mitochondria (Kennedy, 2002), see Figure 3C. Ca 2+ -induced Ca 2+ release from ryanodine-sensitive stores (Iosub et al., 2015) may additionally increase and prolong the Ca 2+ signal in IHCs. Further, Fluo-8 acts as a Ca 2+ buffer itself and could slow the Ca 2+ decay kinetics. The dissociation time constant of the chemically related Fluo-3 is ∼5 ms (Eberhard and Erne, 1989), much faster than the average decay time of fCaTs of 880 ms. Further, our fCaTs resemble the events measured with simultaneous imaging of Fluo-4FF-filled IHCs and Ca 2+ APs recordings in current clamp (Iosub et al., 2015). Taken together, IHC fCaTs imaged under our conditions reflect the Ca 2+ signals resulting from IHC APs.
It might be argued that due to the limited z-resolution of a confocal LSM compared with its resolution in x and y, we recorded signals from inner phalangeal cells rather than from IHCs, especially when IHCs were oriented steeply within the organ of Corti. However, Ca 2+ signals of ROIs taken well inside those IHCs that had a shallow position clearly showed a different temporal pattern including fCaTs compared with adjacent ROIs likely involving inner phalangeal cells, which always lacked fCaTs (e.g., Figures 2A,I, ISC III). However, some bleed-through of fluorescence from inner phalangeal cell signals to IHCs in the z-direction cannot be excluded.

ATP Independently Triggers Ca 2+ Elevations in Both ISCs and IHCs
Applying 1 µM ATP from the pillar side towards the modiolus elicited bursts of fCaTs that were highly synchronized between neighboring IHCs and appeared before Ca 2+ had risen in adjacent ISCs (Figure 5). This indicates that ATP can trigger IHC Ca 2+ APs independently of ISC Ca 2+ waves. Moreover, the large reduction of Ca 2+ wave-independent IHC fCaTs under the broad spectrum P2R blocker PPADS (Figure 6) favors a role of P2 receptors in increasing the IHC's susceptibility to generate Ca 2+ APs. These results are consistent with ATP-induced inward currents through P2X receptors in IHCs , which according to mRNA/protein expression data or PPADS sensitivity suggests contribution of P2X2, P2X3, P2X4, P2X7 before the onset of hearing (Nikolic et al., 2003;Huang et al., 2005;Coddou et al., 2011), yet contribution of P2Y receptors cannot be ruled out (Abbracchio et al., 2006;Huang et al., 2010). Regarding PPADS-sensitive ATP-mediated Ca 2+ signals in ISCs, P2Y receptors such as P2Y1, P2Y2, P2Y4 and P2Y6 may be involved (Abbracchio et al., 2006;Huang et al., 2010) but P2X receptors may additionally contribute (Tritsch et al., 2007;Coddou et al., 2011).
Electrical recordings of IHC Ca 2+ APs, which require access of the patch pipette to the body of the IHC, inevitably cause tissue damage and release of ATP, either during ''cleaning'' of an IHC from ISCs at one side before establishing the pipette seal, or when approaching the IHC under pressure similar as in a tissue slice. Due to local ATP release in patch-clamp experiments, the interpretation of IHC AP activity as spontaneously occurring should be taken with caution. Release of ATP also occurs when recording from ISCs (Tritsch et al., 2007;Wang et al., 2015), which makes the interpretation of the players in the interaction between Ca 2+ waves and IHC Ca 2+ APs and a differentiation of putative roles of ATP rather challenging. Our free-running whole-mount preparation has the advantage that the epithelium was left intact within a region much larger than the field of view, which allowed imaging the interaction of ISC Ca 2+ waves and IHC fCaTs without local tissue damage and contamination with ATP. The fact that ISC Ca 2+ waves did not repeatedly originate from the same foci suggests that the tissue in the field of view was not injured.

Electrical Properties of IHCs Sustain Autonomous Ca 2+ AP Activity
Frequently, single fCaTs and minibursts but sometimes also bursts were observed in IHCs in the absence of Ca 2+ waves in adjacent ISCs (Figures 5, 8, 9; summarized in Figure 10), which were termed here spontaneous or autonomous IHC events. These events reflecting Ca 2+ APs were therefore neither the consequence of extracellular K + accumulation through K + efflux nor of ATP release from the ISCs (see below). It is rather likely that sizes and properties of IHC currents, such as Ca v 1.3 currents, various K + currents including SK2, transducer currents at rest, purinergic currents, cholinergic currents (Brandt et al., 2003;Marcotti et al., 2004;Housley et al., 2006;Johnson et al., 2007Johnson et al., , 2012 favor to the ability of the IHC to produce regenerative Ca 2+ APs.

Coupling of ISC Ca 2+ Waves to IHC Ca 2+ Signals
When an ISC Ca 2+ wave invaded the IHC region it typically elicited bursts of fCaTs in IHCs. The sizes of ISC Ca 2+ waves correlated with the number of activated IHCs; large waves activated more IHCs that responded by synchronously generating bursts with an average delay of 0.52 s (Figure 10). The mechanism of coupling between ISCs and IHC APs is not entirely clear. One proposed mechanism is the regenerative efflux of ATP through connexin hemichannels into the endolymph and activation of P2X2 and P2X3 receptors (perhaps also P2X7) at the IHC (Anselmi et al., 2008;Majumder et al., 2010;Johnson et al., 2011;Mammano and Bortolozzi, 2018). As the ATP superfused from the pillar side evoked bursts in IHCs before eliciting Ca 2+ transients in the ISCs (Figures 5D,E), a direct ATP activation of IHCs could possibly occur in vivo. However, the fact that IHC fCaT activity was unaffected in P2X2R −/− /P2X3R −/− and in P2X7 −/− mice indicates that neither of the receptors was essential for activating IHCs by incoming Ca 2+ waves. Because IHCs from P2X4R −/− mice produced normal Ca 2+ APs (Sendin et al., 2014), an essential role of P2X4 receptors for IHC Ca 2+ AP activity can be excluded, too. Recently, an alternative, ATP-independent mechanism of IHC activation has been proposed, in which the Ca 2+ elevation in ISCs opens Ca 2+dependent Cl − channels resulting in a coupled efflux of Cl − , K + and water (Wang et al., 2015). The accumulation of K + ions around the IHCs in turn depolarizes the IHC over the threshold of AP generation (Wang et al., 2015).
We repeatedly observed that neighboring IHCs synchronously generated Ca 2+ signals (from single fCaTs to bursts) in the absence of ISC Ca 2+ waves (Figure 5A,B, 9E). Because AP activity of an IHC itself causes efflux of K + ions (Marcotti et al., 2003a(Marcotti et al., , 2004, a sizeable amount of K + ions may have been able to depolarize neighboring IHCs over the threshold for generating Ca 2+ APs. The basolateral part of an IHC is surrounded by specialized ISCs, the inner phalangeal cells, leaving a small extracellular volume in which K + efflux from the IHC may lead to a substantial increase in the extracellular K + concentration. Inner phalangeal cells around the IHCs normally remove K + ions by means of Kir4.1 channels (Eckhard et al., 2012), but their uptake rate might be insufficient at times, leading to a series of fCaTs in an IHC and its neighbors.

Spontaneous Activity in the Immature Cochlea
The limited number of fCaTs in a burst (mean: 15.3 ± 9.5) and its limited duration suggest that a rising Ca 2+ level probably activated SK2 channels, which in turn hyperpolarized the membrane potential of the IHCs and prevented further Ca 2+ APs (Johnson et al., 2007), shaping a temporally structured pattern of Ca 2+ APs (Johnson et al., 2011;Sendin et al., 2014). In vivo, AP patterning in the central auditory system is additionally shaped by inhibition of IHC Ca 2+ AP activity through efferent cholinergic fibers acting via α9/α10 acetylcholine receptors coupling to SK2 channels (Glowatzki and Fuchs, 2000;Clause et al., 2014). Our data support the existence of two independent types of spontaneous activity in the developing auditory system, autonomous generation of Ca 2+ APs in IHCs independent of ISC Ca 2+ waves and the stochastic generation of Ca 2+ transients in ISCs that elicit Ca 2+ waves inducing synchronized bursts of IHC Ca 2+ APs. The role of these different types of spontaneous activity for the firing patterns of spiral ganglion neurons with a functioning inhibitory efferent system in vivo needs to be established.

AUTHOR CONTRIBUTIONS
JE and TE designed the study. TE and KB acquired Ca 2+ imaging data. TE, KB and IM analyzed the data. TE, KB, IM and JE prepared the manuscript.