Twenty-seven C57BL/6 mice (weight 20 g each, 2 months, male) were purchased from the Model Animal Research Center of Nanjing University, China. The animal research was completed with the approval of the Ethics Board of the first affiliated hospital of USTC (2021-N(A)-019). These mice were kept in a specific pathogen-free environment, where humidity was maintained in the range of 50%–60% and temperature was at 25°C. These mice were randomly decided into four groups: group I received daily normal saline injection of 0.2 ml as control (n = 8); group II received daily caffeine (Sigma-Aldrich, St. Louis, MO, USA) injection of 120 mg/ kg for 7 days (n = 8); group III received daily caffeine injection of 120 mg/kg for 14 days (n = 8); group IV received daily caffeine injection of 20 mg/ kg for 14 days (n = 3). These were sacrificed by overdose of ketamine after Auditory Brainstem Response (ABR) tests.

Twenty-four Sprague-Dawley (SD) rats [postnatal day 3 (P3)] were purchased from the Model Animal Research Center of Nanjing University, China. These SD rats were sacrificed by overdose of ketamine. Cochlear basilar membranes were cultured in Neurobasal medium (Gibco, Grand Island, NY, USA) after being gently isolated from SD rats. The 24 SD rats were equally allocated to two experiments. In one, 12 neonatal SD rats were divided into four groups: no caffeine treated as control group (n = 3); 1 mM caffeine group (n = 3); 5 mM caffeine group (n = 3); and 10 mM caffeine group (n = 3). In the other, 12 neonatal SD rats were divided into four groups, as shown in Figure 6C: no caffeine treated as control group (n = 3); 10 μM GSK650394 group (n = 3); 10 mM caffeine group (n = 3); and 10 μM GSK650394 + 10 mM caffeine group (n = 3).

The House Ear Institute-Organ of Corti 1 (HEI-OC1) cell line has been extensively used in many previous reports (Guan et al., 2016; He et al., 2016; Yu et al., 2017) and was a gift from professor Hao Xiong in the Sun Yat-sen University (Xiong et al., 2019). HEI-OC1 cells were cultured under permissive conditions (33°C, 10% CO₂) in DMEM (Servicebio, Wuhan, China) supplemented with 10% fetal bovine serum (Gibco, Sydney, Australia). Caffeine was added to the culture medium at a concentration of 1, 5, 10, and 20 mM. In order to confirm the role of SGK1 (serum and glucocorticoid-induced protein kinase 1) in HEI-OC1 cells after caffeine treatment, GSK650394, a SGK1 inhibitor, was used (Peng et al., 2012): no caffeine treated as control group; 10 μM GSK650394 group; 10 mM caffeine group; and 10 μM GSK650394 + 10 mM caffeine group. In order to confirm the role of HIF-1α (hypoxia inducible factor-1) in HEI-OC1 cells, we divided HEI-OC1 cells into four groups (Figure 5F): no caffeine treated as control group; 100 μM CoCl₂ group; 10 mM caffeine group; and 100 μM CoCl₂ + 10 mM caffeine group (Zhao et al., 2020). In order to verify whether autophagy leads to apoptosis, we divided HEI-OC1 cells into four groups (Supplementary Figure S3): no caffeine treated as control group; 5 mM 3-MA group; 10 mM caffeine group; and 5 mM 3-MA + 10 mM caffeine group.

Tests were performed under general anesthesia induced through injection of 100 mg/kg ketamine. ABR tests were conducted using a Tucker-Davis Technology System hardware and software (Alachua, NY, USA). C57BL/6 mice were subcutaneously inserted with subdermal needle electrodes at the vertex, below the left ear (reference), and on the right ear (ground) after being anesthetized. ABR tests were measured at frequencies of 8, 16, and 32 kHz. In order to obtain the average response to 1,024 stimuli, the stimulus sound was decreased from 90 to 10 dB SPL, whose intensity was reduced by 10 dB at intervals near the threshold. As soon as the electrophysiological response to the stimulus sound disappeared, we measured the minimum stimulus sound that evoked a response as the hearing threshold of the mouse tested at this frequency.

Basilar membranes and cells were isolated and fixed in 4% paraformaldehyde. Basilar membranes were divided into three segments (apex, middle, and base) and mounted on glass slides. Then, they were incubated in 0.5% Triton X-100 for 20 min at room temperature. After incubation in 10% goat serum for blocking nonspecific antibody binding for 30 min at room temperature, the samples were incubated with the primary antibody (information in Table1) overnight at 4°C. After washing three times (5 min each), the tissues were incubated with the secondary antibody (Jackson, West Grove, PA, USA, 1:100) at room temperature for 1 h. Basilar membranes needed to be incubated with phalloidin (Yeasen, Shanghai, China, 1:100) containing fluorescein isothiocyanate at room temperature for another 1 h. The nuclei were stained by DAPI (Biosharp, Shanghai, China) for 5 min. Images were captured by using Zeiss LSM800.

The sections were dewaxed and deparaffinized in xylene and rehydrated in graded alcohol solutions. Then, the sections were heated for 30 min in Tris–EDTA buffer by microwave oven. Subsequently, the slides were stained with hematoxylin and eosin (HE) or primary antibodies for SGK1 (information in Table 1) and their respective secondary antibodies. Before dehydration and mounting, the sections were counter-stained with hematoxylin. Images were performed with an Olympus microscope camera (Tokyo, Japan). The stria vascularis thickness of cochlea was measured by Olympus cellSens Standard software (Olympus life science, Tokyo, Japan).

Total RNA was extracted using TRIzol reagent kit (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s protocol. RNA quality was assessed on an Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA, USA) and checked using RNase free agarose gel electrophoresis. After total RNA was extracted, mRNA was enriched by Oligo (dT) beads, while prokaryotic mRNA was enriched by removing rRNA by Ribo-Zero™ Magnetic Kit (Epicentre, Madison, WI, USA). Then the enriched mRNA was fragmented into short fragments using fragmentation buffer and reverse transcripted into cDNA with random primers. Second-strand cDNA were synthesized by DNA polymerase I, RNase H, dNTP, and buffer. Then the cDNA fragments were purified with the QIAquick PCR extraction kit (Qiagen, Venlo, Netherlands), end repaired, poly(A) added, and ligated to Illumina sequencing adapters. The ligation products were size selected by agarose gel electrophoresis, PCR amplified, and sequenced using Illumina HiSeq 2500 by Gene Denovo Biotechnology Co. (Guangzhou, China).

Cell viability was detected using CCK-8 kits (Topscience, Shanghai, China) according to the manufacturer’s protocols. Briefly, 5,000 cells were plated into a 96-well flat-bottom plate and incubated overnight under permissive conditions. After drug treatment in 100 µl culture medium, 10 µl CCK-8 was added for 1 h at 37°C. The optical density (OD) values were measured at 450 nm by an ELISA reader (Thermo Multiskan Mk3, Waltham, MA, USA). The blank underwent the same procedure, but without cell seeding, whereas the negative control was just treated without drugs. The relative viability was calculated as (OD experiment–OD blank)/(OD control–OD blank) × 100%.

HEI-OC1 cells (5 × 10³ cells/well) were seeded into six-well plates overnight. Cells were collected after being digested with trypsin. An Annexin FITC/PI Apoptosis Detection Kit (Yeasen, Shanghai, China) was used to examine the rate of apoptosis. Data analysis was performed by using NovoCyte (Agilent, Santa Clara, CA, USA).

The experiment was conducted with the One Step TUNEL Apoptosis Assay Kit (Beyotime, Shanghai, China) according to the manufacturer’s instructions. After being fixed in 4% paraformaldehyde for 20 min, HEI-OC1 cells were then incubated in 0.5% Triton X-100 for 5 min. Next, samples were labeled with 50 μl TUNEL reaction mixture and incubated at room temperature for 1 h in the dark. After washing, slides were immediately examined under a Leica microscope (DMi8). The percentage of apoptotic cells was calculated as (TUNEL-positive cells/total cells) × 100%. All assays were performed in triplicate.

HEI-OC1 cells were fixed in TEM fixative (Servicebio, Wuhan, China) for 24 h and then fixed at 4°C for preservation and transportation. The 1% agarose solution was prepared by heating and dissolving in advance. Before agarose solidification, the precipitation was suspended with forceps and wrapped in the agarose. Agarose blocks with samples avoid light post fixed with 1% OsO₄ (Ted Pella Inc., CA, USA) for 2 h at room temperature. An ethanol dehydration process (series of 30%, 50%, 70%, 80%, 95% and two changes of 100% ethanol) followed by a 20-min immersion in acetone was performed before the final EPON resin. The resin blocks were cut to 60–80 nm thin on an ultramicrotome (Leica, Germany), and the tissues were fished out onto 150 mesh cuprum grids with formvar film. Two percent of uranium acetate was saturated with alcohol solution to avoid light staining for 8 min, then 2.6% lead citrate to avoid CO₂ staining for 8 min. The cuprum grids are observed under TEM (Hitachi, HT7800), and images were taken.

Total RNA was isolated from cells by using TRIzol reagent (Takara Bio, Tokyo, Japan). On the basis of the instructions of the reverse transcriptase kit (Takara, Tokyo, Japan), cDNA was synthesized using 2 μg of the total RNA in TProfessional Thermocycler (Biometra, Berlin, Germany). Then, cDNA samples were subjected to qRT-PCR for 40 cycles by using TB Green™ Premix Ex Taq™ II (Takara, Tokyo, Japan) in Roche LightCycler 96 (Roche, Basel, Switzerland). Primers (Table 1) were designed with the approval of the Sango Biotech Co. Ltd. (Shanghai, China). GAPDH was used as an internal control. The results were calculated using the comparative cycle threshold (ΔΔCt) method.

First, total protein was extracted from HEI-OC1 cells and basilar membrane by using RIPA buffer (Beyotime, Shanghai, China). Almost 10 μg of crude protein was denatured and electrophoresed on 12.5% SDS-PAGE gels. Proteins were transferred onto PVDF membranes by electro-blotting after electrophoretic separation, followed by blocking for 15 min at room temperature in Protein Free Rapid Blocking Buffer (Epizyme, Shanghai, China). The blots were incubated with SGK1, HIF-1α, P62, Bcl-2, Bax, Cleaved-Caspase3, and LC3B (information in Table 2) primary antibodies at 4°C overnight. After washing with PBS-T, membranes were hybridized with an appropriate secondary antibody (Abmart, Shanghai, China) at room temperature for 1 h. Lastly, images of the Western blot bands were performed with chemi Capture (Clinx, Shanghai, China) and the intensity in each group was measured with ImageJ. GAPDH was used as an internal control.

Data are shown as the mean ± SD, and all experiments were repeated at least three times. Statistical analysis was conducted using GraphPad Prism 6. Two-tailed, unpaired Student’s t-tests were used to determine statistical significance when comparing two groups, and one-way ANOVA followed by a Dunnett multiple-comparison test was used when comparing more than two groups. A value of p < 0.05 was considered statistically significant.

Two-month-old C57BL/6 mice were intraperitoneally injected with caffeine 120 mg/kg/day caffeine for 7 and 14 days, while mice in the control group were treated with normal saline. After 7 and 14 days of injection, auditory threshold shifts in the caffeine group significantly increased at 8, 16, and 32 kHz (Figure 1A). However, compared to the control group, auditory threshold shifts were not significantly difference after 20 mg/kg caffeine injection (Supplementary Figure S1). These results suggested that the effect of caffeine on hearing might be dose-dependent. To explore the position of caffeine damage, basilar membranes were stained with TRITC-phalloidin (Figure 1B), and paraffin sections of cochlea were stained with HE (Figure 1D). According to these results, caffeine caused disorder of the Corti organ and the loss of spiral ganglion neurons and hair cells (Figures 1C,E). Moreover, we also found that caffeine did not lead to significant stria vascularis damage.

To identify the mechanism of caffeine on hearing loss, cochlear tissues from the control and caffeine-treated C57BL/6 mice were collected for RNA sequencing, and a volcano plot (Figure 2A) was constructed to show all the molecules detected. Transcriptome analysis showed 1,300 upregulated and 508 downregulated mRNAs in caffeine-treated mice compared to the control group (Figure 2B). Heat maps (Figure 2C) were generated using the differentially expressed mRNAs. Bubble showed enrichments of various functional categories (Figure 2D). The expression change of SGK1, the most differentially expressed gene, was further confirmed by qRT-PCR (Figure 2E), which was consistent with the RNA-seq analysis. Additionally, the protein expression of SGK1, Cleaved-Caspase3, and LC3B was evaluated by Western blot analysis. The results showed that the expression of SGK1, Cleaved-Caspase3, and LC3B II/I markedly increased in the cochlea after caffeine injection (Figures 2F,G). To evaluate the major expression location of SGK1, we investigated the expression levels of SGK1 in C57BL/6 mouse cochlea through immunohistochemical staining. SGK1 was mainly expressed in the Corti organ, stria vascularis, and spiral ganglia (Figure 2H), which indicated that the expression of SGK1 increased in the cochlea after caffeine injection, suggesting that caffeine may induce autophagy and apoptosis in the cochlea.

HEI-OC1 cells were used to investigate the effect of caffeine. Cells were treated with different caffeine concentrations (0, 0.1, 1, 5, 10, and 20 mM) for different times (12, 24, 48, and 72 h), and the results showed that caffeine inhibited HEI-OC1 cell viability in a time- and dose-dependent manner. When the concentration was greater than 1 mM, the inhibitory effect of caffeine on HEI-OC1 cells was markedly increased (Figure 3A). Moreover, the cell viability inhibition of HEI-OC1 at 10 mM was similar to that of 20 mM when treated for 72 h. Based on these results, treatments with 0, 1, 5, and 10 mM caffeine for 24 h were selected as the conditions in subsequent experiments. Dead cells were labeled by PI, and the cells undergoing apoptosis were labeled by Annexin V (Figures 3B,C). The rate of apoptosis in HEI-OC1 cells treated with 1 mM (14.25 ± 0.55%), 5 mM (36.38 ± 2.30%), and 10 mM (49.40 ± 3.34%) caffeine was significantly higher than that in the control group (1.20 ± 0.57%). HEI-OC1 cells in each group were examined by TUNEL assays in order to confirm the apoptosis effect. The percentage of TUNEL-positive cells was positively correlated with the concentration of caffeine (Figures 3D,E). Transmission electron microscope imaging showed that the caffeine-treated cells had more autophagic vacuoles (double membrane-bound autophagosomes) than the control cells, indicating that autophagy levels increase after caffeine treatment (Figures 3F,G). Taken together, these results suggested that caffeine induces apoptosis and autophagy in HEI-OC1 cells.

We next investigated the role of SGK1 in HEI-OC1 cells by detecting the expression of SGK1 and p62 in HEI-OC1 cells by immunofluorescence. As the concentration of caffeine increased, the fluorescence intensity of SGK1 increased and the fluorescence intensity of p62 decreased, which suggested that the expression of SGK1 and p62 had a consistent trend (Figure 4A). Furthermore, we analyzed the mRNA and protein levels of apoptotic and autophagic markers to explore the involved signaling pathway. After caffeine treatment, qRT-PCR (Figures 4B,C) and Western blot analyses (Figures 5A,B) showed that the expressions of SGK1, LC3B, and caspase3 significantly increased, while the expressions of p62 and the Bcl-2/Bax ratio decreased. Thus, these findings suggested that caffeine may induce apoptosis, induce autophagy, and increase the expression of SGK1. Next, we verified the interaction between SGK1 and HIF-1α by co-IP (Figure 5C). GSK650394 is a known SGK1 inhibitor. To further confirm the role of SGK1, we generated a control group, a GSK650394-treated group, a caffeine-treated group, and a GSK650394 + caffeine-treated group. Compared to caffeine treatment alone, Western blot analyses demonstrated that GSK650394 + caffeine treatment increased the expression of HIF-1α and p62 as well as the Bcl-2/Bax ratio but inhibited the expression of SGK1, Cleaved-Caspase3, and LC3B II/I (Figures 5D,E). Next, we took advantage of CoCl₂, an inducer of HIF-1α, to explore the role of HIF-1α in HEI-OC1 cells after caffeine treatment. The Western blot data showed that CoCl₂ increased the expression of p62 and the Bcl-2/Bax ratio after caffeine treatment, while it inhibited the expression of SGK1, Cleaved-Caspase3, and LC3B II/I (Figures 5F,G). Additionally, we also found that CoCl₂ did not mediate the expression of SGK1 (Supplementary Figure S2), indicating that HIF-1α might be a downstream signaling molecule of SGK1. To explore the relationship between autophagy and apoptosis, 3-methyladenine (3-MA) was used as an autophagy inhibitor. After treatment with 3-MA and caffeine, the expression of Cleaved-Caspase3 decreased and the ratio of Bcl-2/Bax increased, suggesting that autophagy may lead to apoptosis when HEI-OC1 cells are treated with caffeine (Supplementary Figure S3).

Cultured neonatal rat basilar membrane in vitro is an important hearing research model. Cochlear basilar membranes were gently isolated from P3 SD rats and used for experiments after 24 h in culture. After treatment with different concentrations of caffeine, hair cells were labeled with TRITC–phalloidin and labeled auditory nerve fibers were labeled with the 488-TUBB3 antibody. Microscopic analysis showed that caffeine caused disorder and loss of hair cells, especially inner hair cells (Figures 6A,B). However, we also found that caffeine caused disorder of auditory nerve fibers. Additionally, we confirmed the role of SGK1 in the basilar membrane of neonatal SD rats (Figure 6C). GSK650394 effectively protected hair cells from caffeine damage. These results indicated that inhibition of SGK1 might be a potential target for protecting hair cell loss, especially inner hair cells (Figure 6D).