NMNAT1 triggers cell recycling to slow aging caused by D-galactose in inner ear hair cells

Friend Virus B‐type (FVB) mice were purchased from SiPeiFu Biotechnology (Beijing, China) and maintained at the Core for Laboratory Animal Medicine, Institute of Health and Medicine, Hefei Comprehensive National Science Center, China. Postnatal day 3 (P3) FVB mice were used for cochlear explant culture. FVB mice were selected based on their well‐defined genetic background, high fecundity, and large litter size, ensuring a consistent and sufficient supply of experimental animals. The cochleae were dissected and immediately placed in precooled Hank's balanced salt solution (HBSS, 14175095; Gibco). The cochlear volute was carefully removed under a dissecting microscope using forceps, and the basement membrane along with the spiral ligament was gently isolated. Subsequently, the basement membrane was separated from the spiral ligament, adhered to an 8 mm glass slide pre‐coated with Cell‐Tak (354240; Corning), and transferred into a 4‐well culture dish. Each well contained 2 mL Dulbecco's modified Eagle's medium (DMEM) (C11995500BT; Gibco) supplemented with 1% N2 (A1370701; Gibco), 2% B27 (17504044; Gibco), and 50 μg/mL ampicillin (A0166; Sigma‐Aldrich). The cultures were incubated at 37°C with 5% CO₂. All animal experiments were conducted in strict accordance with the ethical guidelines for laboratory animals established by the Anhui Medical University, China.

HEI‐OC1 cells (M8‐0401) were obtained from Cyagen (Guangzhou, China). Cells were maintained in DMEM supplemented with 10% fetal bovine serum (FBS) and 50 μg/mL ampicillin at 33°C in a 10% CO₂ incubator. When the cells were approximately 90% confluence, they were digested into single‐cell suspensions using 0.25% trypsin‐ethylenediaminetetraacetic acid (EDTA) (25200054; Thermo Fisher Scientific) for passaging and subsequent experiments.

HEI‐OC1 cells were plated in 96‐well plates (100 μL per well) and cultured for 12–24 h. Once the cell density reached 60%–70%, they were treated with DMEM containing 2, 5, 10, 20, and 40 mg/mL D‐gal (G7050; Sigma‐Aldrich) for 72 h. Subsequently, 100 μL DMEM supplemented with 10% Cell Counting Kit‐8 solution (HY‐K0301; MCE) was added to each well. After incubation at 37°C for 2 h, the absorbance at 450 nm was measured using a microplate reader (Bio‐Rad). Cell viability was calculated using the formula: Cell viability (%) = ([As−Ab]/[Ac − Ab]) × 100%—where As represents the absorbance of the experimental wells (cells, culture medium and CCK‐8 solution, with drugs); Ac represents the absorbance of control wells (cells, culture medium and CCK‐8 solution, without drugs); and Ab represents the absorbance of blank wells (medium, CCK‐8 solution, without cells and drugs).

The MitoBright ROS Deep Red‐Mitochondrial Superoxide Detection Kit (MT16; DOJINDO) and ROS Assay Kit‐Highly Sensitive DCFH‐DA (R252; DOJINDO) were purchased from DOJINDO. HEI‐OC1 cells were treated with D‐gal (5 and 20 mg/mL) for 72 h. After treatment, the culture medium was removed, and the cells were washed twice with phosphate‐buffered saline (PBS). Following incubation with Mito‐SOX or DCFH‐DA reagent at 37°C for 30 min. Subsequently, the cells were mounted on slides for confocal imaging (LSM900; ZEISS).

HEI‐OC1 cells were treated with D‐gal (5 and 20 mg/mL) for 72 h. The cells were harvested using trypsin (15090046; Gibco), incubated with Mito‐SOX or DCFH‐DA reagent at 37°C for 30 min. Then, the cells were resuspended with PBS for flow cytometry analysis.

Proteins were extracted from HEI‐OC1 cells and cochlear explants using RIPA buffer (P0013B; Beyotime) supplemented with a protease inhibitor cocktail (04693132001; Roche, Basel, Switzerland). Protein concentration was estimated using the bicinchoninic acid Enhanced Protein Assay Kit (P0010; Beyotime). The primary antibodies used in this study were as follows: rabbit anti‐β‐tubulin (1:40000, 5568; CST), rabbit anti‐β‐Actin (1:40000, 20536‐1‐AP; Proteintech), mouse anti‐GAPDH (1:4000, ab8245; Abcam), rabbit anti‐senescence marker protein‐30 (SMP30) (1:1000, 17947‐1‐AP; Proteintech), rabbit anti‐lamin B1 (1:1000, 13435; CST), rabbit anti‐γ‐H2A.X (1:400, 2577; CST), rabbit anti‐P21 (1:1000, 37543; CST), rabbit anti‐P16 (1:1000, 29271; CST), rabbit anti‐NMNAT1 (1:1000, 98354; CST), rabbit anti‐NMNAT1 (1:400, 11399‐1‐AP; Proteintech), rabbit anti‐LC3B (1:1000, 3868; CST), rabbit anti‐P62 (1:1000, 23214; CST), mouse anti‐autophagy related 7 (ATG7) (1:1000, 67341; Proteintech), and rabbit anti‐BECLIN1 (1:1000, 3495; CST). The secondary antibodies used were anti‐rabbit IgG (1:4000, CST; 7074) and anti‐mouse IgG (1:4000, CST; 7076). Protein bands were visualized using the Tanon Gel Imaging System (Tanon, China), and quantified using ImageJ software, with tubulin or actin serving as an internal reference.

The cells or cochlear explants were fixed using 4% paraformaldehyde (P6148; Sigma‐Aldrich). The samples were permeabilized by incubation in PBST (PBS containing 1% Triton X‐100 (1231O021; Solarbio)) for 30 min, and blocking with 10% FBS (SL050; Solarbio) for 1 h, and incubated overnight at 4°C with appropriately diluted primary antibodies. After three 10‐min washes with PBST, the samples were incubated for 2 h at RT with appropriately diluted secondary antibodies. Nuclei were stained using DAPI (1:1000, C0065; Solarbio) before mounting. The primary antibodies used were as follows: rabbit anti‐NMNAT1 (1:400, 11399‐1‐AP; Proteintech), rabbit anti‐myosin7a (1:1000, 25‐6790; Proteus Bioscience), mouse anti‐myosin7a (1:400, 138‐1; DSHB), and rabbit anti‐γ‐H2A.X (1:400, 2577; CST). The secondary antibodies used were Alexa Fluor 555 donkey anti‐rabbit (1:400, Invitrogen, A‐31572), Alexa Fluor 488 donkey anti‐rabbit (1:400, Invitrogen, A‐21206), and Alexa Fluor 488 donkey anti‐mouse (1:400, Invitrogen, R37114). Fluorescence images were photographed using a confocal microscope (LSM900; ZEISS).

Semi‐quantification of immunolabeled signals is a well‐accepted method (He, Pan, et al. 2021; Yuan et al. 2015). NMNAT1 immunolabeling in OHCs was semi‐quantified from confocal images captured with a × 63 lens under equal parameter settings. The fluorescence linear ranges of NMNAT1 were semi‐quantified using the ImageJ software (National Institutes of Health, Bethesda, MD, USA). Individual OHCs were outlined using the ImageJ circle tool based on myosin7a staining. NMNAT1 fluorescence in grayscale OHCs was measured, and background was subtracted. The average grayscale intensity of the NMNAT1 fluorescence was calculated for each OHC. The relative grayscale intensity was normalized to that of the control group.

Cellular Senescence β‐Galactosidase Staining Kit (C0602) was purchased from Beyotime (China). Cells and cochlear explants were exposed to D‐gal for 72 h, followed by removal of the culture medium and washing with PBS. Subsequently, added 1 mL β‐galactosidase staining fixative for 15 min at RT. The cells were aspirated and washed thrice with PBS or HBSS for 3 min each. Subsequently, 1 mL of staining solution was added, and the samples were incubated overnight at 37°C.

NAD⁺/NADH levels were measured using an Enhanced NAD⁺/NADH Assay Kit (S0176S; Beyotime). The cell culture medium was aspirated and discarded, and added 200 μL of NAD⁺/NADH extract buffer per million cells. After cell lysis, the cell supernatant was collected. The samples were diluted 5‐fold with NAD⁺/NADH extract and transferred to a 96‐well plate to determine the total NAD⁺ and NADH levels. A 100 μL aliquot of the diluted sample was incubated at 60°C for 30 min to degrade NAD⁺. Subsequently, 20 μL of the extract buffer was added to measure the NADH levels. The NAD⁺ concentration was calculated as follows: [NAD⁺] = ([NAD⁺] + [NADH]) − [NADH].

Proteomic analysis was conducted in collaboration with LUMINGBIO (Shanghai, China). For sample preparation, HEI‐OC1 cells were exposed to D‐gal (20 mg/mL) for 72 h. Then, the cells were washed thrice with 10 mL PBS on ice, and after trypsin‐EDTA (25200072; Gibco) digestion, PBS was added to the cell suspension. Following centrifugation (4°C, 1000 × g, 5–10 min), the PBS was removed. The cell pellets were flash‐frozen in liquid nitrogen, and stored at −80°C until shipment on dry ice for proteomic analysis. For proteomic profiling, total proteins were extracted from the samples. An aliquot was reserved for protein quantification and quality control by SDS‐PAGE. The remaining portion was digested with trypsin and labeled using Tandem Mass Tag (TMT) reagents (Thermo Fisher Scientific). Equal amounts of each TMT‐labeled sample were combined and then fractionated by chromatography. The resulting peptides were analyzed by liquid chromatography–tandem mass spectrometry (LC–MS/MS) using an Easy‐nLC 1100 system (Thermo Scientific) coupled to an Orbitrap Fusion Lumos Tribrid mass spectrometer (Thermo Scientific). Raw data were processed with Proteome Discoverer 2.4 (Thermo Fisher Scientific), applying a false discovery rate (FDR) threshold of 0.01 at both the protein and peptide levels. Following quality control and preprocessing, we performed analyses at the expression and functional levels. Specifically, differentially expressed proteins were identified and subsequently subjected to Gene Ontology (GO) enrichment analysis, pathway analysis, and protein–protein interaction (PPI) network analysis. All identified proteins are listed in Table . S1

Nuclear and cytoplasmic protein extraction kits (Beyotime P0028) were purchased from Beyotime (Shanghai, China). HEI‐OC1 cells were washed with precooled PBS and collected using cell scrapers. Nucleoplasmic separation experiments were performed using the kit, following the manufacturer's protocol.

HBLV‐mCherry‐GFP‐LC3B lentivirus was obtained from HANBIO, China. HEI‐OC1 cells were infected with HBLV‐mCherry‐GFP‐LC3B lentivirus (multiplicity of infection = 10) at a density of 30%–40%. After 48 h, HEI‐OC1 cells were cultured in a medium containing puromycin (5 μg/mL) and screened for at least 48 h. When all the puromycin‐treated untransfected cells in the control group died, all remaining cells in the infected virus group were positive. Subsequently, a stably expressing mCherry‐GFP‐LC3B cell line was established. Autophagic flux was assessed by monitoring mCherry‐GFP‐LC3B expression. Green fluorescence protein (GFP) fluorescence is quenched in acidic environments. When the lysosomes fuse with the autophagosomes to form autolysosomes, GFP is quenched, whereas mCherry remains stable. Yellow puncta (GFP⁺ mCherry⁺) represent autophagosomes. Red puncta (GFP⁻ mCherry⁺) indicate acidic autolysosomes formed after autophagosomes fuse with lysosomes (GFP signal is quenched). Alterations in the number of red spots or the red/yellow spot ratio are reliable indicators of autophagic flux.

HEI‐OC1 cells were exposed to 100 nM rapamycin (HY‐10219; MedChemExpress), an autophagy activator, for 12 or 24 h. Autophagy levels were assessed using western blotting. Additionally, autophagy levels were determined after treating the cells with 100 nM bafilomycin A1 (HY‐100558; MedChemExpress), an autophagy inhibitor that targets the late stage of autophagy, for 12 h.

The recombinant adeno‐associated virus (AAV) vector pAAV‐CMV‐Nmnat1‐EGFP plasmid was generated by Tsingke Biotech (Beijing, China). It was transfected into cells by co‐incubating with Lipofectamine 2000 transfection reagent (11668027; Invitrogen). The AAV2.7m8‐Ctrl and AAV2.7m8‐Nmnat1 (containing pAAV‐CMV‐Nmnat1‐3xFLAG‐EF1‐GdGreen‐WPRE) were constructed by Hanbio Biotechnology. After culturing the cochlear explants for 12 h, they were infected with AAV2.7m8‐Ctrl and AAV2.7m8‐Nmnat1 at a titer of 5 × 10¹⁰ for 36 h. Subsequently, the explants were exposed to D‐gal for 72 h, followed by immunofluorescence analysis.

The HEI‐OC1‐Nmnat1‐KO cell line was generated using CRISPR/Cas9 technology in collaboration with EDITGENE (Guangzhou, China). Based on Nmnat1 gene sequence data, two single‐guide RNAs (sgRNAs) were designed at the 5′ end of the open reading frame. sgRNA 1 (5′→3′): gaacagcctgaggtgcatgt and sgRNA 2 (5′→3′): tgaggtgcatgttggtgatg. The efficiency of Nmnat1 knockout (KO) was verified using RT‐qPCR and western blot. The off‐target sites of sgRNA1 and sgRNA2 were predicted using the CCTop‐CRISRP/Cas9 target online prediction website. The design of primers targeting potential off‐target sites and off‐target efficiency was verified using PCR and q‐PCR. The PCR primer sequences for predicting the off‐target sites are as follows: Prkar1b‐forward (5′ → 3′): GAGTTGTCCAAAGTATGCAGTAGT, Prkar1b‐reverse (5′ → 3′): AGAGTACCTG CCTAGCCTATGT; WD repeat domain 92 (Wdr92)‐forward (5′ → 3′): AGGGTGTGGTAGGAGAGACTG, Wdr92‐reverse (5′ → 3′): GGCCCAGGCTAAAAACCTTC; basonuclin 2 (Bnc2)‐forward (5′ → 3′): AGCATTTCCCTTCCCAATAAGAGA, Bnc2‐reverse (5′ → 3′): TGTTCTGATTTGA CCAGAGATTTCG; predicted gene 4117 (Gm4117)‐forward (5′ → 3′): GGCCAAACCCTCCCATATCAA, Gm4117‐reverse (5′ → 3′): AAGAGAGGCGTGTTTTTCCCA; and Uhrf2‐forward (5′ → 3′): GCTAGGTCCACTCAGACACAC; Uhrf2‐reverse (5′ → 3′): GGAGAGAG GCTCAGGTCAGA. The q‐PCR primer sequences for predicting the off‐target sites are as follows: Prkar1b‐forward (5′ → 3′): GAGTTGTCCAAAGTATGCAGTAGT, Prkar1b‐reverse (5′ → 3′): AGAGTACCTG CCTAGCCTATGT; and Uhrf2‐forward (5′ → 3′): GCTAGGTCCACTCAGACACAC; Uhrf2‐reverse (5′ → 3′): GGAGAGAG GCTCAGGTCAGA.

Metabolomic analysis was conducted in collaboration with LUMINGBIO (Shanghai, China). For sample preparation, added 1 mL mixture (methanol: water, V: V = 4:1) to the samples in two portions and transferred to a glass vial. Subsequently, 200 μL of chloroform was added, followed by ultrasound treatment in an ice bath (500 W; 3 min, 6 s On, 4 s Off). The liquid was aspirated into centrifuge tubes, and subjected to ultrasound in an ice bath for 20 min before being stored overnight at −40°C. The samples were centrifuged at 12,000 × g for 10 min at 4°C, and 400 μL of the supernatant was collected, concentrated, and dried using a centrifugal concentrator. Then, 80 μL of methoxamine salt pyridine solution (15 mg/mL) was added, and the mixture was incubated for 60 min at 37°C. Then, equilibrated at RT for 30 min. Chromatographic separation was performed using a DB‐5MS capillary column (30 m × 0.25 mm × 0.25 μm, Agilent J&W Scientific, Folsom, CA, USA). High‐purity helium (≥ 99.999%) was used as the carrier gas at a flow rate of 1.0 mL/min. The injection port temperature is 260°C; injection volume, 1 μL, split less injection; and solvent delay, 5 min. Temperature program: initial column temperature, 60°C (maintained for 0.5 min); increase temperature at 8°C/min to 125°C; then to 210°C, followed by 15°C/min to 270°C; and 20°C/min to 305°C (maintained for 5 min). Mass spectrometric detection was performed using electron impact ion source at 70 eV, with the ion source temperatures at 230°C and quadrupole temperatures at 150°C. Full‐scan acquisition was conducted over the m/z range of 50–500. Biologically significant differential metabolites were identified, and the significance of the differences between groups was assessed using variable importance in projection (VIP) values and Student's t‐test. Metabolites with p value < 0.05 and VIP > 1 were considered indicative of significant differences. The raw data (list) of all individual metabolites in the supplemental information Table S2.

Statistical analyses were performed using GraphPad Prism version 9. Western blot results are presented as the mean ± standard deviation (SD), and group comparisons were analyzed using a one‐sample t‐test. Immunofluorescence staining results are presented as the mean ± standard error of the mean. Student's t‐test was used to compare the number of γ‐H2A.X aggregation spots and myosin7a⁺ cells. Significance was determined based on the following thresholds: *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001, p value greater than 0.05 was considered not statistically significant.

Natural aging is a gradual physiological process. To assess ARHL, an aging model was generated by treating HEI‐OC1 cells with various concentrations of D‐gal (2, 5, 10, 20, and 40 mg/mL) for 72 h. CCK‐8 assay showed that with an increase in D‐gal concentration, HEI‐OC1 cell viability gradually reduced (Figure 1A,B). Based on these results, HEI‐OC1 cells were treated with low (5 mg/mL) and high (20 mg/mL) concentrations of D‐gal to determine the levels of reactive oxygen species (ROS). Immunofluorescence staining and flow cytometry with MitoSOX and DCFH‐DA indicated increased levels of ROS in D‐gal‐treated HEI‐OC1 cells after 72 h (Figure 1C–J). Subsequently, aging‐related marker expression levels of SMP30, Lamin B1, γ‐H2A.X, P21, and P16 were detected in HEI‐OC1 cells exposed to different D‐gal concentrations (Matias et al. 2022; Shao et al. 2021; Lopez‐Otin et al. 2023). SMP30 and Lamin B1 were downregulated in HEI‐OC1 cells exposed to high concentrations of D‐gal. However, the expression levels of SMP30 and Lamin B1 increased abnormally after treatment with low D‐gal concentrations (Figure 1K,L,N,O, Figure S1A,B,D,E), indicating that low D‐gal levels may activate cellular self‐protection mechanisms. With the increasing concentration of D‐gal, the expression of γ‐H2A.X, P21, and P16 was increased gradually (Figure 1M,P, Figure S1C,F,G–I,K). Immunofluorescence staining revealed that the number of γ‐H2A.X aggregation spots in D‐gal‐treated HEI‐OC1 cells increased significantly (Figure S1J,L), consistent with the Western blotting results. Additionally, the SA‐β‐gal staining results revealed that cellular aging was obviously elevated in HEI‐OC1 cells treated with D‐gal (Figure S1M,N). These results indicate that treatment with 20 mg/mL D‐gal for 72 h effectively induces cellular aging in HEI‐OC1 cells, providing a robust model for studying aging.

To establish a cochlear explant aging model, the cochlear basilar membrane was dissected from P3 mice and cultured in vitro. Based on a previous report (Guo et al. 2020), cochlear explants were exposed to various concentrations of D‐gal for 72 h. Cochlear explants were subjected to immunofluorescence staining for myosin7a—a marker of hair cells, revealing that no significant change in the number of myosin7a⁺ cells was observed in the low D‐gal concentration (20 mg/mL) group, indicating that low concentration (20 mg/mL) did not cause the loss of cochlear hair cells. However, the number of myosin7a⁺ cells in the high‐concentration D‐gal treatment group did not alter significantly in the apical region but reduced significantly in the middle and basal regions (Figure 2A,B), indicating that high concentration of D‐gal (40 mg/mL) induced an aging phenotype and caused cochlear hair cell damage. SMP30 was downregulated in cochlear explants exposed to high concentration of D‐gal (40 mg/mL), whereas it was abnormally upregulated after treatment with low D‐gal concentration (Figure 2C,F, Figure S2A,D). With an increase in D‐gal concentration, the expression level of Lamin B1 was reduced (Figure 2D,G, Figure S2B,E), whereas that of γ‐H2A.X, P21, and P16 were increased significantly (Figure 2E,H, Figure S2C,F,G–J). Additionally, SA‐β‐gal staining intensity was significantly upregulated in cochlear explants exposed to D‐gal (Figure 2I,J). These findings indicated that exposure to 40 mg/mL D‐gal for 72 h effectively induces aging in cochlear hair cells in vitro. D‐gal‐treated HEI‐OC1 cell and cochlear explant aging models offer a valuable framework for assessing the mechanisms of aging cochlear hair cell damage and developing potential therapeutic approaches for ARHL.

NAD⁺ levels are obviously decreased in the cochlear cells of aged mice (Okur et al. 2023). Additionally, we identified that NAD⁺ levels in HEI‐OC1 cells exposed to D‐gal (20 mg/mL) were significantly reduced (Figure 3A). Moreover, NAD⁺ levels were significantly reduced with an increase in D‐gal concentrations in the D‐gal‐treated cochlear explant aging model (Figure 3B). To assess alterations in gene expression in D‐gal‐induced aging cells, we used proteomics to detect protein expression in HEI‐OC1 cells exposed to 20 mg/mL D‐gal. The proteomic results revealed that NMNAT1 expression was significantly reduced in the D‐gal‐induced aging model (Figure 3C).

NMNAT1 is a core enzyme involved in NAD⁺ biosynthesis and is essential for maintaining cellular NAD⁺ levels. Alterations in NAD⁺ levels are associated with a reduced NMNAT1 expression. Human Nmnat1 is predominantly expressed in the nucleus (Lau et al. 2010); however, its localization in mice remains unclear. To determine the subcellular localization of NMNAT1 in mouse‐derived HEI‐OC1 cells, we performed a nucleoplasmic separation assay and observed that NMNAT1 was expressed in both the cytoplasm and nucleus, with a predominant localization in the nucleus (Figure S3A,B). Additionally, immunofluorescence staining revealed that NMNAT1 was expressed in the nucleus and cytoplasm of mouse cochlear hair cells (Figure S3C), consistent with the results of the nucleoplasmic separation assay.

To verify the proteomics results, we performed western blot and observed that NMNAT1 was slightly upregulated in HEI‐OC1 cells exposed to low concentration of D‐gal (5 mg/mL), but significantly downregulated at high concentration of D‐gal (20 mg/mL) (Figure 3D,E). Alterations in NMNAT1 expression were further confirmed in D‐gal‐exposed aging cochlear hair cells, with significantly reduced NMNAT1 levels in the high‐concentration D‐gal (40 mg/mL) treatment group (Figure 3F,G). These results indicate that NMNAT1 levels were significantly reduced in D‐gal‐treated aging HEI‐OC1 and cochlear hair cells.

To further confirm these findings, immunofluorescence staining revealed that NMNAT1 expression levels in cochlear OHCs in the apical, middle, and basal regions were significantly reduced in the high‐concentration D‐gal (40 mg/mL) group (Figure 3H,J,K,M,N,P). Specifically, the expression level of NMNAT1 in cochlear hair cells in the basal region gradually reduced with increasing D‐gal concentrations (Figure 3N,P), indicating that cochlear basal hair cells were more sensitive to D‐gal induction and more susceptible to damage. These results indicated that D‐gal induces cochlear hair cell damage and alters NMNAT1 expression in aging hair cells, supporting the potential role of NMNAT1 in aging hair cells.

Autophagy levels gradually decline with age, and there is an association between autophagy dysfunction and aging (Aman et al. 2021; Kaushik et al. 2021; Wong et al. 2020). To assess whether autophagy levels were altered in the D‐gal‐induced aging model, we determined that LC3B‐II, ATG7, and BECLIN1 expression levels were increased after treatment with low D‐gal concentration (5 mg/mL) but were obviously reduced after exposure to high D‐gal concentration (20 mg/mL) in HEI‐OC1 cells (Figure 4A,B, Figure S4A,C,D,F,G,J). P62 expression levels gradually increased after exposure to D‐gal for 72 h (Figure S4B,E). Additionally, the expression levels of LC3B‐II, P62, ATG7, and BECLIN1 in D‐gal‐induced aging cochlear explants aligned with the results obtained from D‐gal‐treated aging of HEI‐OC1 cells (Figure 4C,D, Figure S4H,I,K–P). These results indicated that low D‐gal concentrations activate autophagy, whereas high concentrations suppress it.

To assess alterations in autophagy levels, HEI‐OC1 cells were infected with HBLV‐mCherry‐GFP‐LC3B lentivirus to establish a stable cell line. Western blot results confirmed that the expression of endogenous LC3B was not affected by exogenous mCherry‐GFP‐LC3B in the stable cell line (Figure S5F). Following D‐gal‐induced aging, autophagic flux was assessed by monitoring mCherry‐GFP‐LC3B expression. The number of yellow (autophagosomes) and red puncta (autolysosomes) significantly increased in cells treated with a low D‐gal concentration (5 mg/mL). In contrast, treatment with a high D‐gal concentration (20 mg/mL) significantly reduced both types of puncta compared with the low concentration group (Figure 4E–G). These results align with our previous findings (Wei et al. 2025), indicating that low D‐gal concentration activates autophagy, whereas high D‐gal concentration significantly impairs autophagic activity in aging HEI‐OC1 cells. These results indicate that autophagy was impaired in D‐gal‐induced aging of HEI‐OC1 cells and cochlear explants.

hNmnat1 Overexpression in AD facilitates autophagy and reduces amyloid plaques (Zhu et al. 2022). NMNAT1 activates autophagy by regulating the SIRT1/mTOR pathway in aged rats (Wang et al. 2019). To assess the relationship between NMNAT1 and autophagy in the D‐gal‐treated aging model, HEI‐OC1 cells were treated with 100 nM rapamycin for 12 h. Western blotting revealed a significant increase in LC3B‐II expression (Figure 3H,I). However, NMNAT1 expression was unaffected (Figure 4H–J). Similar results were obtained after rapamycin treatment for 24 h (Figure S5A–C). These results indicated that NMNAT1 expression was not affected by autophagic activation.

To confirm the role of NMNAT1 in autophagy regulation, NMNAT1 was overexpressed in HEI‐OC1 cells by transfection with the pAAV‐CMV‐Nmnat1‐EGFP plasmid for 12 h, followed by treatment with D‐gal (20 mg/mL) for 72 h. Subsequently, HEI‐OC1 cells were treated with bafilomycin A1 (100 nM) for 12 h before cell collection. Notably, western blotting revealed that Nmnat1 overexpression upregulated LC3B‐II (Figure 5A,B) and downregulated P62 (Figure 5A,C). However, BECLIN1 levels remained unaltered (Figure S5D,E). Compared with bafilomycin A1 treatment alone, combined treatment with Nmnat1 overexpression and bafilomycin A1 further increased LC3B‐II levels (Figure 5A,B) and slightly reduced P62 levels (Figure 5A,C). To further affirm these findings, autophagic flux was evaluated using mCherry‐GFP‐LC3B expression. Nmnat1 overexpression increased the number of yellow (autophagosomes) and red puncta (autolysosomes) in D‐gal‐induced aging HEI‐OC1 cells, and co‐treatment with bafilomycin A1 further enhanced their accumulation (Figure 5D–F). These findings indicated that Nmnat1 overexpression enhances autophagic flux and may modulate the aging process by regulating autophagy in the D‐gal‐induced aging model.

Autophagy activation in aging cochlear hair cells facilitates their survival (He, Li, et al. 2021). To confirm the effect of NMNAT1 on aging, HEI‐OC1 cells were transfected with the Nmnat1 plasmid for 12 h and then treated with D‐gal (20 mg/mL) for 72 h. Western blotting revealed that Lamin B1 was significantly upregulated in the NMNAT1 overexpression group compared to the D‐gal‐treated group alone (Figure 5G–I), whereas the expression level of γ‐H2A.X was significantly reduced (Figure 5G,J). These results demonstrate that Nmnat1 overexpression can delay D‐gal‐induced aging by activating autophagy.

To verify the role of NMNAT1 in cochlear hair cells, we constructed AAV2.7m8‐Nmnat1 virus. AAV2.7m8 is highly efficient in infecting cochlear hair cells (Isgrig et al. 2019). Cultured cochlear explants were infected with AAV2.7m8‐Nmnat1 in vitro for 36 h, followed by treatment with 40 mg/mL D‐gal for 72 h. Immunofluorescence staining revealed that the number of myosin7a⁺ hair cells was obviously decreased in the middle and basal turns of the D‐gal‐alone treatment group (Figure 6A–F). However, the number of myosin7a⁺ hair cells was higher in the middle and basal turns of the AAV2.7m8‐Nmnat1 and D‐gal treatment groups than that in the D‐gal‐alone treatment group (Figure 6A–F). These results indicate that Nmnat1 overexpression attenuates D‐gal‐treated cochlear hair cell damage and increases cochlear hair cell survival.

NAD⁺ is a primary coenzyme in oxidation–reduction reactions, whereas NADH is the reduced form that is essential for maintaining energy metabolism and regulating various dehydrogenases involved in metabolic pathways (Verdin 2015; Ross 2021). To assess the biological function of Nmnat1 in HEI‐OC1 cells, we constructed an Nmnat1‐KO cell line using CRISPR/Cas9 technology in HEI‐OC1 cells. RT‐qPCR and western blot analyses indicated that Nmnat1 expression was successfully reduced at both the mRNA and protein levels in the Nmnat1‐KO cells (Figure S6A–C). To predict the potential off‐target sites of CRISPR/Cas9 technology, we predicted the off‐target sites of sgRNAs using the CCTop‐CRISRP/Cas9 target online predictor website (Figure S6D,E). Analysis of potential off‐target sites confirmed the absence of unintended editing by sgRNA1 and sgRNA2 in Nmnat1‐KO cell lines (Figure S6F–H). These results indicate that Nmnat1 was effectively knocked out without off‐target effects.

To assess the effects of NMNAT1 on cellular metabolic pathways, we determined the differential metabolite profiles of HEI‐OC1 and Nmnat1‐KO cells using untargeted metabolomics. Volcano plots revealed the differential expression of various metabolites in Nmnat1‐KO cells (Figure 7A). Pathway enrichment analysis revealed that multiple metabolic pathways were affected in Nmnat1‐KO cells, including amino acid and glucose metabolism (Figure 7B). Notably, the TCA cycle was disrupted in Nmnat1‐KO cells, with an abnormal accumulation of multiple TCA cycle intermediates, including citric acid, D‐threo‐isocitric acid, oxoglutaric acid, and malic acid (Figure 7B–E). The TCA cycle is the hub of intracellular material and energy metabolism (Martinez‐Reyes and Chandel 2020). The conversion of citric acid and D‐threo‐isocitric acid to oxoglutaric acid, oxoglutaric acid to succinyl coenzyme A, and malic acid to oxaloacetic acid requires NAD⁺ (Figure 7C). NMNAT1 is the primary rate‐limiting enzyme in NAD⁺ biosynthesis, and its deletion blocks NAD⁺ synthesis, disrupting the TCA cycle and causing the accumulation of citric acid, D‐threo‐isocitric acid, oxoglutaric acid, and malic acid owing to impaired subsequent metabolism (Figure 7F–I). These results revealed a significant regulatory role of NMNAT1 in cellular metabolism.