DNA damage signaling regulates age-dependent proliferative capacity of quiescent inner ear supporting cells

Adenoviral tropism to postnatal inner ear supporting cells

In the present study, the sensory epithelia from postnatal day 6 (P6) utricles and cochleas, and from P50 utricles were prepared for organotypic cultures, as previously described [4]. The P6 and P50 SCs were used to represent the juvenile and adult SCs, respectively. Explants were maintained for 2 to 14 days in vitro (DIV). The auditory sensory epithelium, the organ of Corti, was not studied at adulthood due to difficulties in preserving in vitro the normal cytoarchitecture of the mature organ and the survival of its hair cells. SCs were marked by antibodies against Sox9 and Sox2 [4, 17]. In postnatal utricles, Sox2 is expressed in both SCs and hair cells. However, the nuclei of two cell types are located at different heights in the sensory epithelium and have different morphology, allowing cell type-specific analysis in whole mount surface preparations (Fig. 1A,B). In some experiments, hair cell-specific markers, parvalbumin and myosin 6 (myo6), were used.

Our previous work has established optimal conditions for transduction by adenoviruses encoding cD1 (AdcD1) and β-galactosidase (AdβGal) in adult utricular explants [4]. In the present study, also AdGFP reporter viruses were used to investigate viral tropism, an important issue, because our model organ comprises different cell types and because we studied different ages. AdGFP viruses transduced P6 and P50 utricular SCs, as detected by the presence of GFP+/Sox2+ (Fig. 1C,D) and GFP+/Sox9+ cells (data not shown) at 3 DIV. Transduction efficiency varied between individual explants, ranging from 20 to 50%. Only occasional AdGFP-infected hair cells were found in adult utricles (data not shown). P6 utricles showed higher amount of infected hair cells, based on quantification of parvalbumin+/GFP+ cells. The average infection rate of hair cells was 10% (10.1 ± 0.7, n = 3, total number of hair cells counted = 843). Together, even though infected hair cells were present in juvenile utricles, their amount was clearly outnumbered by infected SCs (Fig. 1E,E') [18].

In AdGFP- or AdβGal-infected P6 cochleas analyzed at 3 DIV, transgenes expressions were concentrated to Deiters' cells, a specific subtype of auditory SCs (Fig. 1F,F'). This expression was concentrated to the upper half of the cochlear duct, transduced Deiters' cells being often arranged in small patches (Fig. 1F',G). Hair cells were not transduced, based on the absence of GFP+/parvalbumin+ cells (data not shown). In the AdβGal-infected P6 cochlea shown in Fig. 1G, the boxed area represents the cochlear region analyzed in the present study. Taken together, under the experimental conditions used, the adenoviral serotype 5 vector (Ad5) with the CMV promoter preferentially transduces SCs in the juvenile and adult inner ear sensory epithelia, with an interesting Deiters´ cell-specific pattern in the cochlea.

Response of juvenile and adult utricular supporting cells to AdcD1 infection

We used ectopic cD1 expression as a tool to force SCs into the cell cycle, based on the fact that many proliferation-promoting signaling pathways target this core cell cycle component. Particularly, cD1 is a central mediator of the proliferative response following activation of the Wnt/β-catenin pathway. It has been shown in mutant mouse models that Wnt/β-catenin activation increases proliferative activity of neonatal inner ear SCs [19]. To study cell cycle activity of SCs transduced by AdcD1, we pulsed explants with the thymidine analogue EdU that incorporates into replicating DNA, for 24 h between days 2 and 3. EdU labeling revealed high proliferative activity in AdcD1-infected P6 utricles at 7 DIV. Numerous clumps of small-sized, EdU+/Sox2+ nuclei of SCs were seen (Fig. 2A,A'), suggesting complete cell cycles. In AdcD1-infected P50 utricles, the amount of EdU+ SCs was clearly smaller, and pairs or clumps of EdU+ SCs were infrequent (Fig. 2B,B'). Combined with the prior findings of G2/M arrest of a large part of cell cycle reactivated adults SCs [4], these results point to inefficient expansion of the adult SC population. Non-infected and reporter virus-infected P6 and P50 explants were devoid of EdU+ SCs (Fig. 2C,C'; data not shown). Quantification revealed that on the average 20% (19.7 ± 2.4, n = 3, total number of SCs counted = 1194) of SCs were EdU-labeled in AdcD1-infected adult utricles at 3 DIV. In P6 utricles, this value was 50% (50.1 ± 2.2, n = 5, total number of SCs counted = 3631), reflecting the robust cell cycle activity of this juvenile cell population. Both P6 and adult utricles showed EdU-labeled SCs with cD1 expression, linking ectopic proliferation with adenovirus-mediated cD1 overexpression (Fig. 2D,D'; data not shown). Of note, because cD1 expression is rapidly downregulated in cells that initiate replication and because we retrospectively analyzed EdU-pulsed cells, SCs did not always show colabeling of cD1 and EdU.

AdcD1-infected utricular supporting cells show a DNA damage response

Oncogene-induced aberrant proliferation stimulates replication stress that can lead to DSB formation and DDR activation [20, 21]. We hypothesized that this is the case also with SCs following ectopic expression of the cD1 proto-oncogene. Furthermore, unresolved DSBs or other type of DNA damage might explain the low-level cell cycle activity of AdcD1-infected adult SCs compared to juvenile SCs (Fig. 2) [4]. To find out whether DSBs are formed in SCs, we compared using confocal microscopy the expression pattern of γH2AX, a DNA damage marker, in SCs challenged with AdcD1 and ionizing radiation (IR). γH2AX marks various forms of DNA damage, but the characteristic foci-like expression pattern seen in non-replicating cells following IR exposure represents DSBs [22]. Therefore, we used the γH2AX expression pattern seen 1 h after 2 gray (Gy) dose of IR as an established template of DSB profiles. Comparative analysis after 3 DIV showed that, similar to IR-treated SCs, AdcD1-infected and EdU-labeled (pulse between culture days 2 and 3) SCs of both ages expressed γH2AX as abundant small nuclear foci (Fig. 3A-D). These findings suggest that DSBs are induced in juvenile and adult SCs upon cell cycle re-entry.

To determine whether culture conditions and adenoviral transduction itself trigger DNA damage, non-infected as well as AdβGal- and AdGFP-infected utricles were analyzed after 3 DIV. One or two large γH2AX foci were found in SCs of both non-infected and reporter virus-infected utricles of both ages (Fig. 3E-G; data not shown). Thus, in vitro conditions induce low-level DNA damage, and viral transduction does not markedly alter this basal pattern. However, the DNA damage profile in these control explants is clearly different from the DSB-like profile with abundant small γH2AX foci seen following AdcD1-infection and IR. To confirm that the baseline γH2AX expression arising in vitro does not negatively affect proliferation, we prepared utricular cultures at P2, at the stage when a part of utricular SCs still naturally proliferates in vivo [16]. Both non-infected and AdGFP-infected P2 cultures showed EdU-labeled SCs after a 3-day-long pulsing period (data not shown), suggesting that the baseline γH2AX expression in explant cultures does not prevent natural SC proliferation.

As an additional marker for activated DDR, we used p53 binding protein 1 (53BP1), a component of the DDR pathway that localizes to the area of DSBs [23]. Similar to IR-challenged SCs, EdU-labeled SCs in AdcD1-infected utricles of both ages displayed 53BP1 foci (3H-I; data not shown). These foci were not detected in Sox2+/GFP+ SCs in explants infected with reporter viruses (Fig. 3J). We conclude that DDR is activated in juvenile and adult SCs upon forced cell cycle re-entry, with DSBs as a major type of DNA lesion.

γH2AX foci resolution in AdcD1-infected utricular supporting cells

We next examined the temporal pattern of γH2AX expression in AdcD1-infected and EdU-labeled SCs (Fig. 4). In P6 utricles, the first replicating SCs were found at 2 DIV (EdU pulse between days 1 and 2). These cells showed γH2AX foci (data not shown), but the intensity of this expression was not as prominent as one day later when widespread γH2AX expression was found in the SC population, concomitantly with the high proliferative activity of these cells. Based on quantification at 3 DIV, 51% of EdU+ SCs (EdU pulse between culture days 2 and 3) showed the typical DSB pattern of γH2AX (SC numbers counted per explant: EdU+ 326.8 ± 27.8; γH2AX+/EdU+ 164 ± 12.2). This value was only 18% at 7 DIV (SC numbers counted per explant: EdU+ 274.8 ± 38.3; γH2AX+/EdU+ 48.6 ± 8.6) (Fig. 4A). The disappearance of SCs with γH2AX foci was not caused by cell death, based on the findings of only rare cleaved caspase-3-positive profiles in the juvenile explants analyzed at 7 DIV (data not shown). Of note, it is likely that the rather low percentage of EdU+ SCs with γH2AX foci at 3 DIV is an underestimate, because a part of the foci had already been resolved by this time point. These data on γH2AX resolution in AdcD1-infected juvenile SCs suggested successful DNA repair.

Next, using the same EdU pulsing regimen, the temporal pattern of γH2AX expression was analyzed in AdcD1-infected adult utricles (Fig. 4B). At 3 DIV, 83% of EdU+ SCs displayed γH2AX foci (SC numbers counted per explant: EdU+ 68.6 ± 4.1; γH2AX+/EdU+ 56.6 ± 2.4). This value was 54% at 7 DIV (SC numbers counted per explant: EdU+ 40.6 ± 10.8; γH2AX+/EdU+ 22.6 ± 6.5). By 14 DIV, the percentage of double-positive cells had decreased to 16% (SC numbers counted per explant: EdU+ 39.5 ± 3.7; γH2AX+/EdU+ 6 ± 0.4), a value comparable to that seen in juvenile utricles at 7 DIV (Fig. 4B). Compared to the juvenile specimens, these results showed that the γH2AX foci resolution is prolonged in adult utricles and, thus, suggested slower DNA repair.

At 7 DIV, in addition to EdU+ SCs with γH2AX foci, AdcD1-infected adult utricles displayed EdU+ SCs with intensive, pan-nuclear γH2AX staining (Figs. 3B, 4C). This γH2AX pattern has been previously associated with severe DNA damage and cell death [22]. Consistently, the majority of SCs with pan-nuclear γH2AX co-expressed cleaved caspase-3, a marker for apoptotic death (Fig. 4C,C'). Quantification at 7 DIV showed a highly significant difference (p < 0.005) between the relative amount of γH2AX+ SCs with intensive, pan-nuclear staining in P6 and P50 utricles (Fig. 4D). This γH2AX expression profile was absent in explants transduced by reporter viruses (data not shown). Together, these data suggest that prolonged DNA damage resolution in cell cycle reactivated adult SCs predisposes these cells to apoptosis.

The repair protein Rad51 is dynamically expressed in AdcD1-infected utricular supporting cells

Rad51 is a DNA repair protein engaged in homologous recombination repair (HRR) and is detected in the respective repair foci. HRR is relevant in the context of cell cycle reactivated SCs investigated in the present work, because it predominates in S and G2 phases of the cell cycle [24]. We used Rad51 as a marker for HRR, and examined whether its expression correlate with the observed γH2AX foci resolution.

At 3 DIV, Rad51 was upregulated in AdcD1-infected P6 utricles, specifically in SCs with γH2AX foci (Fig. 5A,A'). This was readily seen under light microcopy and low magnification. The nuclear Rad51 expression appeared fuzzy at low magnification, because the expression was formed by numerous small foci, as revealed by confocal imaging at high magnification (Fig. 5B,B'). Interestingly, in juvenile SCs, Rad51 expression was downregulated by 7 DIV, paralleling the resolution of γH2AX foci (Fig. 5C,C'). AdβGal-infected explants did not show Rad51 expression (Fig. 5D,D'). As an additional control, IR-challenged, non-proliferating SCs lacked Rad51 expression (Fig. 5E,E'), consistent with the predominance of non-homologous end-joining rather than HRR in DSB repair in non-proliferating cells. Together, in juvenile SCs, HRR was induced upon ectopic proliferation and DDR activation, and was down-regulated concomitantly with successful DNA repair.

Also AdcD1-infected adult utricles showed Rad51 induction at 3 DIV, but as opposed to juvenile explants, downregulation was not observed before 14 DIV. This is consistent with the resolution of γH2AX foci by this late post-infection time point (Fig. 5F-H'). Similar to younger SCs, confocal microscopy revealed Rad51 expression as small nuclear foci in adult SCs (Fig. 5I,I'). Quantification showed that Rad51 expression closely paralled that of γH2AX both in juvenile and adult SCs (Fig. 5K). As the presence of γH2AX foci matched with EdU labeling (Fig. 4A,B), we conclude that Rad51+ cells represent cell cycle reactivated and DNA damaged SCs. Interestingly, at 7 DIV, SCs with the strong, pan-nuclear γH2AX expression lacked or showed only very weak Rad51 expression (Fig. 5G,G',J,J'). Because many of these cells were also positive for cleaved caspase-3 (Fig. 4C,C'), these data link delayed DNA repair to cell death.

AdcD1-triggered cell cycle re-entry of auditory supporting cells

We next studied cell cycle activity in cochlear SCs transduced by AdcD1 at P6. Consistent with the transduction pattern revealed by reporter viruses (Fig. 1F',G), the upper half of AdcD1-infected cochlear explants showed EdU labeling exclusively in Sox2+ Deiters' cells (EdU pulse between days 2 and 3). These double-positive cells were found both at 3 and 7 DIV (Fig. 6A-B'). Non-infected and reporter virus-infected cochleas were devoid of EdU+ cells (Fig.6C; data not shown). Following AdcD1 infection, similar to utricles, cochleas showed EdU-labeled Deiters' cells with cD1 upregulation (Fig. 6D,D').

Several EdU+ doublets of Deiters' cells were found in AdcD1-infected P6 cochleas. In many cases, the nuclei of these doublets appeared to be firmly connected with each other, suggesting for incomplete cytokinesis (Fig. 6A-B'). Consistently, the Deiters' cell population did not show clear signs of expansion, based on morphological examination and comparison to the control (non-infected and reporter virus-infected) cochleas at 7 DIV (Fig. 6A-C). An underlying reason might be the induction of p53, an anti-proliferative transcription factor often activated upon unscheduled proliferation [25]. Indeed, we found p53 upregulation in Deiters' cells of AdcD1-infected cochleas, particularly in cell doublets, but also in cells with large nuclei, typical to replicating cells (Fig. 7A,A). Interestingly, SCs of age-matched, AdcD1-infected utricles did not show p53 upregulation (Fig. 7B,B), consistent with their high proliferative activity (Fig. 2A,A'). p53 expression was undetectable in control explants of both the cochlea and utricle (Fig. 7C,D).

Thus, these data suggested an inverse correlation between the proliferative activity and p53 response in the cochlear Deiters' cells.

γH2AX foci accumulate in cell cycle reactivated auditory supporting cells in vitro and in vivo

We next studied the dynamics of γH2AX expression in AdcD1-infected Deiters' cells. At 3 DIV, P6 cochlear explants showed EdU+/Sox2+ Deiters' cells with γH2AX foci, typical to DSBs (Fig. 8A,A'). At 7 DIV, most of these cells, including most of the cell doublets, lacked γH2AX foci (Fig. 8B,B'). Combining these results to the utricular data, we conclude that the dynamics of DSB resolution in the inner ear SCs is age-dependent.

To reveal whether the accumulation of γH2AX foci is a specific response to cD1 overexpression or whether it is a general response to cell cycle re-entry, the retinoblastoma (Rb) gene was inactivated in vivo in auditory SCs. We generated Rb^loxP/loxP;Fgfr3-iCre-ER^T2 mutant mice and injected the pups with tamoxifen in order to inactivate Rb in auditory SCs from P2 onward. The Fgfr3-iCre-ER^T2 transgenic mouse line mediates efficient and cell type-specific recombination in the cochlea, as previously shown [26]. Cross-sections were prepared from the cochleas of mutant animals and littermate controls at P7, and stained for γH2AX and Ki-67. Pillar cells, a subtype of auditory SCs that similar to Deiters' cells do not proliferate postnatally, showed unscheduled proliferation in the mutant mice during the first two postnatal weeks (Fig. 8C). We focused our analysis on pillar cells, because only a few Deiters' cells of the Rb^loxP/loxP;Fgfr3-iCre-ER^T2 mice showed cell cycle reactivation, consistent with prior data [27]. In contrast to control specimens, pillar cells of mutant animals showed induction of γH2AX foci (Fig. 8D,E). Thus, these data show that both ectopic cD1 in vitro and genetic inactivation of a negative cell cycle regulator in vivo can trigger cell cycle re-entry of auditory SCs and that this event is linked with DSB accumulation.

Supporting cell-to-hair-cell transdifferentiation does not trigger DNA damage

Having found that DSBs are formed in cell cycle reactivated SCs and that DNA damage resolution directs proliferative activity, we investigated whether SC-to-hair-cell transdifferentiation, the other main step in the regeneration process, triggers DNA damage as well. Transdifferentiation can be triggered by the γ-secretase inhibitor DAPT that inactivates Notch signaling [28, 29]. In addition to P6 explants, we treated P2 cochlear explants with DAPT, because SC's response to DAPT is more pronounced at this earlier age. Cochleas were analyzed after a 3-day-long DAPT incubation. It is known that DAPT stimulates new hair cell formation during this whole culturing period [30]. Unlike control specimens (treated with DMSO), DAPT-treated cochleas showed supernumerary, myo6-positive hair cells, based on surface views where the normal mosaic-like organization of the sensory epithelium (Fig. 1B) was in places lost. In these areas, hair cells were in contact with each other, unlike in control explants where hair cells were contacted by SCs. Notably, these ectopic hair cells did not show γH2AX induction above the level seen in normal hair cells of the same explants or of control cultures (Fig. 9A-C'). Similar results were obtained in P6 and adult utricles after a 3-day-long DAPT treatment. Regions were found in these specimens where myo6-positive hair cells lacked mature stereociliary bundles, based on phalloidin staining that marks filamentous actin-rich stereocilia, suggesting that these cells had been generated through trans-differentiation (Fig. 9D-E''; data not shown). Thus, as opposed to forced cell cycle re-entry, stimulation of transdifferentiation of SCs does not trigger DNA damage.

Uncoupling between forced proliferation and transdifferentiation of utricular supporting cells

Next we wanted to investigate whether forced cell cycle re-entry is coupled to transdifferentiation, because natural regeneration in non-mammalian inner ears often involves SC divisions, followed by transdifferentiation of a part of the progeny into hair cells [1]. We used AdcD1-infected P6 utricles for these experiments, applying a 24-h-long EdU pulse between days 2 and 3. We took into account the fact that, in addition to SCs, a small part of hair cells of juvenile utricles are infected by adenoviruses (Fig. 1E,E'). At 6 DIV, the amount of EdU+/myo6+ hair cells was less than 10% in AdcD1-infected utricles, and this value corresponded to the amount of hair cells in the explants infected with reporter viruses (Figs. 1E,E' and 10A,A'). These results argue against coupling between forced proliferation and transdifferentiation of postmitotic SCs.

We then asked whether forced proliferation can be combined with DAPT-induced transdifferentiation. AdcD1-infected P6 utricles were cultured for 3 days (EdU pulse between days 2 and 3) and then treated with DAPT for another 3 days. After 6 DIV, a robust, threefold increase in EdU+/myo6+ hair cells was found in these specimens compared to only AdcD1-infected explants (AdcD1: number of myo6+ cells counted 768, number of EdU+/myo6+ cells 22, n = 3 explants; AdcD1 plus DAPT: corresponding values 1096 and 109, n = 3) (Fig. 10A-B'). Thus, preceding cell cycle reactivation does not prevent pharmaco-logical stimulation of SCs to transdifferentiate into hair cells.

Mice

Explant cultures of the utricular and cochlear sensory epithelia were prepared from NMRI mice, and histological sections from the inner ears of Rb^loxP/loxP;Fgfr3-iCre-ER^T2 mice. Conditional, inducible inactivation of Rb in auditory SCs of these mutant mice was induced by tamoxifen at P2 and P3, as described previously [26]. Rb^loxP/loxP and Fgfr3-iCre-ER^T2 mice were genotyped by PCR as described in the original publications [46, 47]. The day of birth was considered as P0. All animal work has been conducted according to relevant national and international guidelines.

Explant cultures and viral infections

Explant cultures were established at P2 and P6 from the cochlea, and at P2, P6 and P50 from the utricle. Explants were maintained on pieces of Nuclepore filter membrane (Whatman) placed on a metal grid in Dulbecco's modified Eagle's DMEM/F-12 medium supplied with 2 mM L-glutamine and penicillin (100 U/mL) (Gibco/Invitrogen) and 10% fetal bovine serum (FBS) (HyClone/Thermo Scientific). Incubations were done in a humidified 5% CO₂ atmosphere at 37°C. Ad5 vectors harbouring the βGal, GFP or cD1 transgenes and CMV promoter were used. Cloning and propagation of the viruses have been described previously [48]. Cochlear and utricular explants were infected for 16 and 7 h, respectively, with AdβGal (4.2 × 10⁷ pfu/ml), AdGFP (3.6 × 10⁷ pfu/ml) or AdcD1 (2.9 × 10⁷ pfu/ml). Infections were done in 30 µl drops of medium containing 5% FBS. Thereafter, explants were maintained for 2 to 14 DIV in medium containing 10% FBS. Medium was changed every other day.

Whole mount specimens

Explants were fixed with 4% paraformaldehyde (PFA) in phosphate buffered saline (PBS) for 10 min and permeabilized in PBS containing 0.25% Triton-X-100 (PBS-TX). Double- and triple-immunofluorescent stainings were performed in PBS-TX and 10% normal serum, using the following antibodies: goat parvalbumin (Swant, #PVG-214, 1:2000); rabbit cleaved caspase-3 (Asp-175) (Cell Signaling Technology, #9661, 1:250); rabbit Sox9 (Millipore, #AB5535, 1:2000); mouse γH2AX (Millipore, #05-636, clone JBW301, 1:500); goat Sox2 (Santa Cruz Biotechnology, #SC-17320, 1:1000); rabbit Rad51 (H-92) (Santa Cruz Biotechnology, #SC-8349, 1:800), rabbit GFP (Molecular Probes/Invitrogen, #A6455, 1:2000); rabbit p53 (Novocastra, #NCL-p53-CM5p, 1:500); rabbit myosin 6 [49]; rabbit 53BP1 (Cell Signaling Technology, #4937, 1:350). Following antibody incubations, filamentous actin-containing stereociliary bundles were visualized using Oregon Green-labeled phalloidin (Molecular Probes/Invitrogen, 1:400). Secondary antibodies conjugated to Alexa 350, 488, 568, 594 and 647 were used for visualization, and ProLong Gold anti-fade reagent (Molecular Probes/Invitrogen) for mounting. β-Galactosidase staining was performed by incubating fixed explants in X-gal staining solution [1 mg/ ml X-gal (5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside; Promega), 5 mM K₃Fe(CN)₆, 5 mM K₄Fe(CN)₆, 2 mM MgCl₂, 0.025% sodium deoxycholate and 0.025% Nonidet P-40] for 15 min at room temperature and mounted in ProLong Gold medium.

EdU pulsing and labeling

Adenovirus-infected explants were given a single pulse of 10 µM EdU (5-ethynyl-2'-deoxyuridine; Molecular Probes/Invitrogen) for 24 h between days 2 and 3, and fixed at 3, 7 and 14 DIV. A part of infected explants were given a single pulse of EdU for 24 h between days 1 and 2, and fixed immediately thereafter, at 2 DIV. Non-infected P2 utricles were given a single pulse of EdU every 24 h for three consecutive days, followed by fixation at 3 DIV. Specimens were stained according to manufacturer's instructions using the Click-iT EdU Imaging Kit (Invitrogen) and mounted in ProLong Gold medium.

Histological sections

Inner ears from Rb^loxP/loxP;Fgfr3-iCre-ER^T2 mice were perilymphatically fixed with 4% PFA at P7, and immersed in the fixative overnight. Specimens were decalcified for 6 h in 0.5 M EDTA, pH 8.0, followed by embedding in paraffin and cutting to 5-μm-thick sections. Epitopes were unmasked by microwave heating (800 W) in 10 mM citrate buffer, pH 6.0, for 10 min. Sections were stained overnight with the γH2AX and rabbit monoclonal Ki-67 (LabVision/Thermo Scientific, #RM-9106, clone SP6, 1:250) antibodies in PBS-TX. Detection was done using the Mouse-on-Mouse (γH2AX) and Vectastain Elite ABC (Ki-67) kits and the DAB Detection kit (Vector Laboratories). Sections were counterstained with 3% methyl green and mounted in Permount (Fisher Scientific).

DAPT treatment

Cochlear (P2 and P6) and utricular (P6 and P50) explants were treated with the γ-secretase inhibitor DAPT (N-[N-(3,5-Difluorophenacetyl)-L-alanyl]-S-phenylglycine t-butyl ester; Sigma Aldrich, #D5942) or the DMSO vehicle for 72 h. A part of P6 utricles were first infected with AdcD1, pulsed with EdU for 24 h between days 2 and 3, and thereafter treated with DAPT for 72 h, followed by fixation at 6 DIV. DAPT was dissolved in DMSO (concentration in the medium 0.1%) and used at the concentration of 10 µM. The compounds were added to the culture medium every 24 h.

Irradiation of whole mount specimens

Utricular (P6 and P50) explants were irradiated with 2 Gy using calibrated ¹³⁷Cs γ-ray source (BioBeam 8000, STS GmbH). Utricles were cultured thereafter for 1 or 4 h.

Imaging

Histological cross-sections as well as wholemount surface preparations were analyzed under Axio Imager.M2 microscope (Zeiss) using brightfield and epifluorescence optics. Images were acquired through CCD colour camera (AxioCam HRc, Zeiss) and ZEN software (Zeiss), and processed using Adobe Photoshop CS6 (Adobe Systems). Confocal images were acquired using a Leica TCS SP5 microscope with Plan Apochromat 63×/1.3 NA objective. The acquisition software was Leica LAS AF. Z-projections were processed with Imaris 7 (Bitplane Scientific Software). Blind 3D deconvolution was made for Z-projections with AutoQuant X2 (Media Cybernetics).

Quantification

To quantify the amount of adenovirus-infected utricular hair cells, parvalbumin+/GFP+ hair cells were counted from one random 40x microscopic field per AdGFP-infected P6 utricular explant at 3 DIV, each field covering striolar and non-striolar areas. Data are shown as the percentage of parvalbumin+/GFP+ hair cells (mean ± SEM).

To quantify the amount of proliferating cells following AdcD1 infection, Sox2+ and EdU+/Sox2+ SCs were counted from one random 40x microscopic field per infected P6 and P50 utricular explant at 3 DIV, each field covering striolar and non-striolar areas. Data are shown as the percentage of EdU+/Sox2+ SCs out of Sox2+ SCs (mean ± SEM).

To quantify the resolution of γH2AX+ foci, EdU+ and EdU+/γH2AX foci+ SCs were counted from one random 40x microscopic field per AdcD1-infected P6 and P50 utricular explant at 3, 7 and 14 DIV, each field covering striolar and non-striolar areas. Data are shown as the average percentage of EdU+/γH2AX+ cells out of EdU+ SCs (mean ± SEM).

To determine the relative proportion of SCs with pan-nuclear γH2AX profiles, Sox2+/ total γH2AX+ (foci plus pan-γH2AX) and Sox2+/pan-γH2AX+ SCs were counted from one random 40x microscopic field per P6 and P50 utricle at 7 DIV, each field covering striolar and non-striolar areas. Data are shown as the relative numbers of Sox2+/pan-γH2AX+ SCs out of Sox2+/total γH2AX+ SCs (mean ± SEM).

To determine the relative proportion of γH2AX+/Rad51- SCs, γH2AX+/Rad51+ and γH2AX+/Rad51- cells were counted from one random 40x microscopic field per P6 and P50 AdcD1-infected utricle at 3, 7 and 14 DIV, each field covering striolar and non-striolar areas. Data are shown as the percentage of Rad51+/γH2AX- SCs out of Rad51+/γH2AX+ SCs (mean ± SEM).

To quantify the amount of EdU+ hair cells following AdcD1-infection alone and in combination with DAPT treatment, EdU+/myo6+ cells were counted from one random 40x microscopic field per P6 AdcD1-infected and, AdcD1-infected and DAPT-treated utricle at 6 DIV, each field covering striolar and non-striolar areas. Data are shown as the total number of myo6+ and EdU+/myo6+ cells.

Three or five individual explants (n) were used for each quantification. Two-tailed Student's t test and one-way ANOVA were used for statistical analysis. A P-value ≤0.005 was considered highly significant.