Exploring the bioactive ingredients of three traditional Chinese medicine formulas against age-related hearing loss through network pharmacology and experimental validation

Abstract

Traditional Chinese medicine (TCM) formulas, including the Er-Long-Zuo-Ci pill, Tong-Qiao-Er-Long pill, and Er-Long pill, have long been utilized in China for managing age-related hearing loss (ARHL). However, the specific bioactive compounds, pharmacological targets, and underlying mechanisms remain elusive. This study aims to find the shared bioactive ingredients among these three formulas, uncover the molecular pathways they regulate, and identify potential therapeutic targets for ARHL. Furthermore, it seeks to validate the efficacy of these major components through both in vivo and in vitro experiments. Common bioactive ingredients were extracted from the TCMSP database, and their putative target proteins were predicted using the Swiss Target Prediction database. ARHL-related target proteins were collected from GeneCards and OMIM databases. Our approach involved constructing drug-target networks and drug-disease-specific protein–protein interaction networks and conducting clustering, topological property analyses, and functional annotation through GO and KEGG enrichment analysis. Molecular docking analysis was utilized to delineate interaction mechanisms between major bioactive ingredients and key target proteins. Finally, in vivo and in vitro experiments involving ABR recording, immunofluorescent staining, HE staining, and quantitative PCR were conducted to validate the treatment effects of flavonoids on the declining auditory function in DBA/2 J mice. We identified 11 common chemical compounds across the three formulas and their associated 276 putative targets. Additionally, 3350 ARHL-related targets were compiled. As an intersection of the putative targets of the common compounds and ARHL-related proteins, 145 shared targets were determined. Functional enrichment analysis indicated that these compounds may modulate various biological processes, including cell proliferation, apoptosis, inflammatory response, and synaptic connections. Notably, potential targets such as TNFα, MAPK1, SRC, AKT, EGFR, ESR1, and AR were implicated. Flavonoids emerged as major bioactive components against ARHL based on target numbers, with molecular docking demonstrating diverse interaction models between these flavonoids and protein targets. Furthermore, baicalin could mitigate the age-related cochlear damage and hearing loss of DBA/2 J mice through its multi-target and multi-pathway mechanism, involving anti-inflammation, modulation of sex hormone-related pathways, and activation of potassium channels. This study offers an integrated network pharmacology approach, validated by in vivo and in vitro experiments, shedding light on the potential mechanisms, major active components, and therapeutic targets of TCM formulas for treating ARHL.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00210-024-03464-2.

Introduction

Age-related hearing loss (ARHL), also known as presbycusis, manifests as a progressive, irreversible deterioration of the auditory system and binaural hearing ability with advancing age. It stands as the most prevalent sensory impairment among the elderly, impacting approximately one-third of individuals aged 65 years and above (Chadha et al. 2021). This progressive degeneration of auditory function not only leads to social isolation but also correlates with various comorbidities, such as frailty, falls, and depression (Kamil et al. 2016; Rutherford et al. 2018; Sharma et al. 2021). Recent findings underscore the association between ARHL and long-term risk cognitive decline and incident dementia, highlighting its multifaceted nature (Mosnier et al. 2015; Chern and Golub 2019; Livingston et al. 2020; Powell et al. 2021). Present treatment modalities such as hearing aids and cochlear implants offer limited efficacy and fail to address the underlying physiological alterations associated with aging. As of now, there is no pharmacological intervention specifically designed to prevent or treat ARHL effectively (Wang and Puel 2020). Human and animal studies have elucidated several age-related changes in auditory structures, including the diminution of the cochlear inner hair cells (IHCs) and outer hair cells (OHCs), atrophy of the stria vascularis (SV), loss of fibrocytes within the spiral ligament (SL), degeneration of spiral ganglion neurons (SGNs), and secondary alterations in central auditory system (Mazelova et al. 2003; Ohlemiller 2004; Lin et al. 2014; Bowl and Dawson 2019). Mounting evidence implicates oxidative stress, chronic inflammation, cellular apoptosis, and ischemia as pivotal pathophysiological mechanisms underlying the structural and functional impairments observed in ARHL (Wang and Puel 2020; Chester et al. 2021; Yang et al. 2023). As our understanding of ARHL’s pathogenesis evolves, pharmacologic intervention strategies have emerged. Given the intricacy of the underlying mechanisms, treatments targeting multiple potential pathways present a promising avenue. In recent years, there has been a surge in interest in exploring alternative and complementary approaches to managing ARHL. Traditional Chinese medicine (TCM), characterized by the synergistic effects of various botanical, mineral, or animal-derived substances to treat complex diseases, holds promise as a therapeutic resource. Over 20 TCM formulas and traditional herbal medicines have been employed to prevent and treat ARHL in clinical settings, primarily relying on antioxidant, anti-inflammatory, and autophagy-regulating properties of the active ingredients (Hu et al. 2023; Yan et al. 2023; Wu et al. 2024). Despite the rich empirical knowledge within TCM, rigorous scientific investigation is needed to establish the safety, efficacy, and scientific basis of specific herbal formulations for ARHL. Furthermore, the bioactive ingredients of TCM formulas and herbs against diseases need to be identified and potentially developed.

The present study focuses on three TCM formulas (Er-Long-Zuo-Ci (ELZC), Tong-Qiao-Er-Long (TQEL) pill, and Er-Long (EL) pill)—cataloged in Chinese Pharmacopoeia for the treatment of hearing-related disorders—including age-related hearing decline. Previous studies have reported the therapeutic benefits of ELZC in delaying age-related hearing threshold elevation in C57BL/6 J mice, potentially involving increased ERK1/2 phosphorylation (Dong et al. 2016; Liu et al. 2021a). However, the complex composition of TCM formulas remains the major obstacle in elucidating their mechanisms against diseases. Additional challenges include discerning inhibitory effects from activated effects of the targets, and understanding the varying impacts of compound dosages on targets. Therefore, this study aims to elucidate the potential common mechanisms of the three formulas by decoding their common bioactive ingredients and core targets against ARHL and seeking new substances and avenues for ARHL treatment. The study’s conceptual framework is illustrated in Fig. 1.

Fig. 1.

Flowchart illustrating the technical strategy employed in this study. Experimental methodologies included target prediction for the eleven common active ingredients from three TCM formulas against ARHL, construction and topological analysis on compound-target network and PPI network, enrichment analysis on GO and KEGG pathway, molecular docking, and in vivo and in vitro experiments comprising ABR recordings, H&E staining, immunofluorescent staining, and quantitative PCR

Materials and methods

Screening the common compounds of the three TCM formulas

Traditional Chinese medicine (TCM) formulas used to treat ARHL in the clinic include the Er-Long-Zuo-Ci pill, Tong-Qiao-Er-Long pill, and Er-Long pill. The compositions of the three TCM formulas from the Chinese Pharmacopoeia (2020) are listed in Table S1. The known ingredients of herbs from the three TCM formulas were screened in the Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform (TCMSP) with two criteria (Xu et al. 2012; Liu et al. 2013): oral bioavailability (OB%) value ≥ 30% and drug similarity (DL) ≥ 0.18, with isomer excluded. Finally, 11 common compounds from the three TCM formulas were identified and further analyzed.

Prediction of targets of common compounds and ARHL-associated targets

To detect potential targets of the 11 common compounds from the three TCM formulas, the Simplified Molecular Input Line Entry System (SMILES) information of the 11 common compounds was searched in the PubChem database and subsequently imported into the Swiss Target Prediction (http://www.swisstargetprediction.ch) to mine the targets based on a combination of two-dimensional and three-dimensional similarity measurements with known ligands. The screening criterion was probability > 0, and duplicates were excluded (Gfeller et al. 2014; Daina et al. 2019). Currently, known ARHL-related targets (disease-target) were obtained from two databases: the GeneCards database, a searchable comprehensive resource providing information on annotated and predicted human genes, and the Online Mendelian Inheritance in Man (OMIM), which offers descriptions of human genes, phenotypes, and their interrelationships (Amberger et al. 2015). Two keywords, “age-related hearing loss” and “presbycusis,” were searched in the two databases, and duplicates were removed. The Venny 2.1 software was employed to map the overlapping targets between the 11 common compounds from the three TCM formulas and ARHL.

Construction and cluster analysis of the PPI network

Protein–protein interaction (PPI) of drug-disease targets was derived from STRING (http://stringdb.org/ver.11), an online resource for known and predicted protein–protein interactions (Szklarczyk et al. 2023). To obtain the PPI network, the organisms were set to Homo sapiens, and a high confidence level (score > 0.7) was set to ensure data reliability, the independent nodes were hidden. In the current study, 23 nodes that were independent of the network or formed separate clusters under the condition of score > 0.7 were removed. The PPI network was demonstrated in Cytoscape software for visualization and subsequent analysis. The network analyzer module was used to identify key nodes based on their network topology. Three parameters including “Degree”, “Betweenness Centrality,” and “Closeness Centrality” were adopted to assess the centrality properties of nodes in the PPI network. The Disease-Herbal Medicine-Active Ingredient-Common Target illustrated the intricate relationships between the bioactive ingredients of herbal medicine and their target interactions in the context of the disease. To further identify the core PPI network and core genes, the Molecular Complex Detection (MCODE) plugin of the Cytoscape software was employed to identify compact regions and perform cluster analysis within the PPI network. This analysis utilized specific criteria: degree cutoff = 2, node score cutoff = 0.2, and K-core = 2. Subsequently, the clusters were ranked based on their score values. Using the MCODE algorithm to identify the resulting significant clusters from PPI networks facilitates further investigation of their underlying mechanisms and makes visualization of large interacting networks more manageable.

GO and KEGG enrichment analysis

To investigate how active components influence signaling pathways and gene functions relevant to ARHL treatment, we imported the common targets identified by STRING into the Database for Annotation, Visualization, and Integrated Discovery (DAVID) Bioinformatics Resources (https://david.ncifcrf.gov/) for Gene ontology (GO) enrichment analysis and the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis(Ashburner et al. 2000; Kanehisa and Goto 2000). DAVID integrates various biological model databases and research results, allowing for large-scale analysis of the cellular component (CC), molecular function (MF), and biological process (BP). GO enrichment analysis helps us understand the biological functions, pathways, and cellular locations of gene enrichment, while KEGG enrichment analysis provides a macro perspective on gene functions and helps explore drug-disease signaling pathways. We selected the top five GO terms and the top five KEGG pathways with the smallest p-values for further analysis. These were visualized using bar graphs and bubble charts generated through the bioinformatics mapping website (http://www.bioinformatics.com.cn/). A p-value < 0.05 was considered significant in both GO and KEGG analyses.

Molecular docking

Ligand and protein preparation

The chemical structures of nine compounds (quercetin (CID, 5280343), baicalin (CID, 64982), kaempferol (CID, 5280863), leucanthoside (SWERTIAJAPONIN) (CID, 442659), stigmasterol (CID, 5280794), Diop (CID, 395120), campesterol (CID, 173183), mairin (CID, 64971), and sitosterol (CID, 12303645)) were downloaded from the PubChem database. The chemical structures of the other two compounds (Ethyl Oleate (ID, ZINC000008214560) and Mandenol (ID, ZINC000004474575)) were downloaded from the ZINC database. The 3D structures of ten core targets (AKT1(1UNQ), SRC(2BDJ), MAPK1(2OJG), EGFR(1IVO), PIK3R1(4WAF), RXRA(1MVC), TNFα(6OOY), ESR1(1A52), PTPN11(7JVN), AR(4OJ9)), three specific modulators of three targets (IP4, an activator of AKT (1UNQ); BMS649, an activator of RXRA (1MVC); UCB-6876, an inhibitor of TNFα (6OOY)), KCNMA1 (6V35), KCNQ4 (7VNP), 4-chloro-7-(trifluoromethyl)-10H-benzofuro[3,2-b] indole-1-carboxylic acid, (CTBIC, an activator of BKCa potassium channel), and ML213 (an activator of Kv7.4 potassium channel, 7VNP) were downloaded from the PDB database. To ensure the validity and reliability of the molecular docking results, the specific protein ligands (IP4, BMS649, UCB-6876, CTBIC, and ML213) were employed as positive controls (Chen et al. 2021; Ts et al. 2024).

Docking parameter and criterion

Molecular docking was performed using the Libdock docking module in Discovery Studio 4.0 (Accelrys Software Inc., San Diego, CA, USA). A semi-flexible approach was employed for Libdock docking. Ligand preparation: the “Prepare Ligands” module was used to preprocess small molecules, generate valid 3D conformations, add hydrogen atoms, and perform energy minimization through Chemistry at Harvard Macromolecular Mechanics (CHARMm) force field. Receptor preparation: after removing water molecules from the crystal structure, the “Clean Protein” module was employed to fill incomplete amino acid residues, followed by the addition of hydrogen atoms and CHARMm electric field. The details of the binding pockets of ten core targets and two potassium channels are shown in Table S10 and Table S12. Among them, three target proteins (AKT1, RXRA, and TNFα) and two potassium channels (BKCa and Kv7.4) have specific agonists and antagonists, with clear docking pockets in existing literatures. These known agonists and antagonists and their docking pockets were regarded as positive control. Selection rules of docking pockets for target proteins without positive control: by analyzing the surface of receptor molecules, the software is capable of identifying significant cavity structures, which may represent known or potential ligand binding sites. Subsequently, the software computes the volume of these cavities. Users can view the volume and other attributes of binding sites through the software interface, and analyze these sites using visualization tools that represent them as point sets and transparent red spheres. The details of the molecular docking are revealed using the Discovery Studio Visualizer. A LibDock score of ≥ 90 indicates a strong affinity of the small molecular ligand for the receptor and more effective binding (Wang et al. 2022). This study utilized an integrated experimental protocol combining network pharmacology and molecular docking, based on methodologies established in previous research (Pathania et al. 2020; Guzman-Flores et al. 2023; Lephatsi et al. 2023; Vikhar Danish Ahmad et al. 2024).

Animals and drugs

DBA/2 J mice (male, postnatal day 21–22) were obtained from Beijing HFK Bioscience Co., Ltd. (Beijing, China). Mice were raised under a 12-h light/dark cycle and were given ad libitum access to food and water. Two compounds, quercetin (Q4951) and baicalin (572,667) were purchased from Merck Corporation (Darmstadt, Germany). The animal experiments were carried out in accordance with the guidelines for the Care and Use of Laboratory Animals (NIH publication no. 85–23, revised 1985). The animal study was reviewed and approved by the Animal Ethical and Welfare Committee of Hebei University (IACUC-2021009SM).

Auditory brainstem responses (ABR) testing

Drug administration and experiment groups

To explore the effects of flavonoids on ARHL in vivo, quercetin, and baicalin were representatively assessed in the current study. Quercetin with different dosages ranging from 15 to 100 mg/kg (Sagit et al. 2015; Hirose et al. 2016; Gundogdu et al. 2019) and baicalin with different dosages ranging from 15 to 30 mg/kg(Kang et al. 2010; Yang et al. 2011) have been evaluated in the treatment of ototoxicity and noise-induced hearing loss. At the beginning of the current study, DBA/2 J mice (postnatal day 21–22) without serous otitis media and tympanic membrane perforation underwent ABR testing as a baseline, after the mice were subsequently divided randomly into three groups: (1) saline group, (2) quercetin group, and (3) baicalin group. After baseline evaluation of ABR, in the first month, the mice from quercetin group and baicalin group were intraperitoneally administrated with quercetin (15 mg/kg, dissolved in saline with 1% DMSO) and baicalin (15 mg/kg, dissolved in saline with 1% DMSO) once a day for a month, respectively, while the mice from control group were administrated with same volume of saline (1% DMSO) once a day for 1 month. After 1 month of continuous dosing, the auditory brainstem responses (ABRs) of these mice (postanal day 51–53) were assessed. However, substantial improvement in hearing ability was undetectable, so the dosages of quercetin and baicalin were both increased in the following month. In the second month, the mice from the quercetin group and baicalin group were continuously intraperitoneally administrated with quercetin (50 mg/kg, dissolved in saline with 2% DMSO) and baicalin (30 mg/kg, dissolved in saline with 2% DMSO) once a day for a month, respectively; the mice from control group were administrated with same volume of saline (2% DMSO) once daily for 1 month. Finally, the ABRs of these mice (postanal day 81–83) were assessed.

ABRs recording and analysis

Mice were anesthetized by intraperitoneal injection of esketamine (80 mg/kg) and xylazine (25 mg/kg) (Medchem, New Jersey, USA) and were placed on an isothermal pad (Homeothermic Monitoring System, Harvard Apparatus) to maintain body temperature at 37 °C. ABR recordings were performed in a sound-isolation booth (background noise, 30 ~ 35 dB) using the Tucker-Davis Technologies RZ6 workstation. Three subdermal needle electrodes were respectively inserted into the vertex (active electrode), right mastoid (reference electrode), and left shoulder (ground electrode). The impedances of both the input channel and the reference ranged from 0.5 to 0.9 kΩ. In a free-field setup, sound stimuli were delivered through an MF1 speaker positioned 10 cm in front of the ears. ABRs were evoked by a series of short tone pips at frequencies of 4, 5.6, 8, 11.3, 16, 22.6, and 32 kHz, and sound levels were decreased from 90 to 0 dB SPL by 5 dB steps. Waveforms were averaged 400 trials for each stimulus. Thresholds were defined as the minimal sound levels that can elicit ABR waves. Latencies of ABR wave were defined as the time when the stimulus begins at the peak of the wave, and amplitudes of ABR wave were calculated by averaging the ΔV of both sides of the peak.

Tissue preparation

After completing ABR measurements, mice were euthanized, and their cochleae were dissected. During the dissection process, we verified the presence of otitis media. Only cochleae from mice without otitis media were used for the subsequent experiments. The cochleae were fixed with 4% PFA, perfused through a small opening at the apex made by a needle, oval window and round window, and were left in PFA overnight. After fixation, cochleae were decalcified in phosphate-buffered saline (PBS) containing 10% EDTA at room temperature for 7 days, and subsequently used for subsequent specific experiments. The left cochleae and right cochleae of these mice were utilized for immunofluorescent staining and Hematoxylin and Eosin (HE) staining, respectively. For immunofluorescence staining, the cochlear sensory epithelium was dissected. After removing the stria vascularis, Reissner’s membrane, and tectorial membrane, it was cut into three turns for immunofluorescent staining. For HE staining, cochleae underwent gradient dehydration of ethanol and then embedded with paraffin. Paraffin-embedded samples were cut into 4-µm thick sections through the modiolus using a Leica Cryostat. Slices containing 3–4 Corti’s organs were stained with HE.

Immunofluorescence staining

The cochlear sensory epithelium was permeabilized in PBS with 1% Triton X-100 for 30 min and blocked with 10% BSA for 60 min at room temperature. The tissues were then incubated with the primary antibodies at 4 °C overnight: polyclonal rabbit anti-Myosin VIIα (25–6790, Proteus Bioscience, USA) diluted at 1:300 and mouse (IgG1) anti-CtBP2 (612,044, BD Pharmingen) diluted at 1:300. After being washed three times with PBS, the tissues were incubated in the secondary antibodies for 2 h at room temperature in darkness: Alexa Fluor 555-conjugated donkey anti-rabbit IgG (A-31572) diluted at 1:1000 and Alexa Fluor 488-conjugated goat anti-mouse IgG1 (A-21126) (Thermo Fisher Scientific, USA) diluted at 1:1000. After being washed, the sensory epithelium was mounted on a slide rinsed with ProLong™ Gold anti-fade Mountant (P36984, Thermo Fisher Scientific, USA). Images were taken using a laser scanning confocal microscope (FV3000, Olympus). Hair cells were captured under a 20 × objective, and carboxyl-terminal-binding protein 2 (CtBP2) of IHCs were imaged (z-stack mode) with a × 63 objective. Imaris software (Bitplane AG, Zurich, Switzerland) was used to quantify CtBP2 immunofluorescence puncta on the surface of each IHC.

Hematoxylin and Eosin (H&E) staining

Paraffin-embedded cochlear sections were deparaffinized in xylene and dehydrated in ethanol. Then, the sections underwent orderly incubation in hematoxylin and eosin for 3–4 min. After washing with tap water, the sections were dehydrated in gradient alcohol, cleared in xylene, and later coverslipped. The images of whole cochlear sections were captured under a × 40 objective. The morphology of stria vascularis (SV) and the density of spiral ganglion neurons (SGNs) were quantified in the cross-sectional areas of each turn. The cross-sectional area of SV and Rosenthal’s canal was quantified by using the CaseViewer software. The density of SGNS from apical to basal turns was calculated by dividing the number of neurons by the cross-sectional area of the corresponding Rosenthal’s canal (cells per square millimeters).

Quantitative real-time PCR (qPCR)

The total RNA of cochleae was extracted using Aiso Plus (Takara, Dalian, China). Reverse transcription and quantitative PCR (qPCR) were done using PrimeScript™ RT reagent Kit with gDNA Eraser (Takara) and TB Green® Premix Ex Taq™ Kit (Takara). qPCR was performed using the CFX96 Real-Time PCR system (Bio-Rad Laboratories, Hercules, CA, USA). Primer sequences are listed in Table S14. Relative expression levels were calculated by the 2^−ΔΔCT method.

Statistical analysis

Data were presented as mean ± SEM and analyzed using the GraphPad Prism software (GraphPad Software Inc., San Diego, USA). Two-tail student’s t-test and analysis of variance (ANOVA) followed by Bonferroni’s post-hoc comparisons were used for difference analysis. P < 0.05 was considered to be statistically significant.

Results

Shared targets between common compounds of three TCM formulas and ARHL

All active chemical ingredients from three TCM formulas (ELZC, TQEL, and EL pill) were collected from the TCMSP database. A total of 134 compounds co-existed in the three formulas were selected. Due to the ADME characteristics of TCM drugs in routine oral administration, a subset of 11 common compounds (sitosterol, quercetin, baicalin, campesterol, stigmasterol, diop, mairin (betulinic acid), kaempferol, mandenol, ethyl oleate, and leucanthoside (swertiajaponin)) passed the OB and DL filters and was chosen for further study (Table 1). The SMILES structures of these 11 common compounds were imported into the Prediction Target of the Swiss Target Prediction network database, yielding 276 unique putative targets (Table S2). Meanwhile, a total of 3350 ARHL-related targets were collected from GeneCard and OMIM databases. Venn analysis revealed 145 shared targets between common compounds and ARHL (Fig. 2). These 145 overlapping genes were considered potential key targets of TCM for treating ARHL and were further explored.

Table 1.

The common active compounds from three Chinese medicine formulas

MOL ID	Molecule name	Molecular weight	OB (%)	DL	Herb
MOL000359	Sitosterol	414.79	36.91	0.75	RRP, GRER, SRAO, Li, COS, CM, Al, ArR, AuR, CR
MOL000098	Quercetin	302.25	46.43	0.28	GF, Li, CM, RB, Al
MOL002776	Baicalin	446.39	40.12	0.75	SR, RB, Al
MOL000493	Campesterol	400.76	37.58	0.71	SR, RD
MOL000449	Stigmasterol	412.77	43.83	0.76	RRP, SR, CA, GF, ASR, COS, RD, RB, ArR, AuR
MOL002879	Diop	390.62	43.59	0.39	SR, COS
MOL000211	Mairin	456.78	55.38	0.78	Li, CMAR
MOL000422	Kaempferol	286.25	41.88	0.24	CTER, GF, Li, RAA, CM, RB
MOL001494	Mandenol	308.56	42	0.19	GF, COS
MOL002883	Ethyl oleate	310.58	32.4	0.19	GF, COS
MOL003137	Leucanthoside	462.44	32.12	0.78	GRER, COS

ID, MW, OB, and DL data are from the TCMSP website (http://tcmspw.com/tcmsp.php). The targets were obtained from Swiss Target Prediction website, and the number of targets corresponding to compounds after intersection with disease targets was obtained

RRP Rehmanniae radix praeparata, GRER Gentianae radix et rhizoma, SRAO Scutellariae radix Alisma orientale, Li licorice, COS Cornus officinalis sieb, CM Cortex Moutan, Al aloe, ArR Arisaematis rhizoma, AuR Aucklandiae radix, CR Citrus reticulata, GF Gardeniae fructus, RB Radix Bupleuri, SR Scutellariae radix, RD Rhizoma Dioscorea, CA Caulis Akebiae, ASR Angelicae sinensis radix, CMAR Cortex Moutan Aucklandiae Radix, RAA rhizoma Anemones altaicae

Fig. 2.

Venn diagram illustrating the overlap of targets. The Venn map demonstrates that 145 targets were shared between 11 common active compounds from three TCM formulas and ARHL

To elucidate the complex interactions between the 11 common compounds and the 145 shared targets, we constructed an interacting network of these compounds and targets. The compound-target network consisted of 174 nodes (including 11 compound nodes and 145 compound target nodes) and 485 edges (Fig. 3). The target number of each compound was listed in Table S3. A total of 10 compounds including sitosterol, quercetin, baicalin, campesterol, stigmasterol, diop, mairin, kaempferol, mandenol, and ethyl oleate, interacted with more than two targets. Among them, two flavonoids (quercetin, kaempferol) exhibited the most potential targets, while baicalin also targeted more than ten potential targets (Table S3). Notably, one target of Leucanthoside (another flavonoid) revealed through SwissTarget Prediction did not overlap with ARHL-related targets. We found that 88 of the 145 putative targets were shared between two or more compounds, indicating that these compounds act on some of the same biological processes.

Fig. 3.

Compound-target network visualization. The network diagram exhibits 174 nodes and 485 edges. Blue nodes represent ARHL, and orange nodes represent TCM. Brown nodes represent compounds, and green targets denote proteins co-acting with TCM and disease. Node color gradation corresponds to the degree value

Protein–protein interaction network of shared targets

To elucidate the correlation, the 145 shared targets were input into the STRING database to construct a protein–protein interaction (PPI) network. The PPI network revealed the interaction among the shared targets, with nodes representing proteins and edges representing interactions between them. The median values for degree centrality (DC), betweenness centrality (BC), and closeness centrality (CC) of all nodes of compounds were calculated (10, 0.005, and 0.3248, respectively). According to the network topology properties, 23 targets were excluded due to no interaction with other targets or the formation of an independent cluster. The remaining 122 core targets (confidence coefficient > 0.7) were predicted as the core targets of TCM formulas action on ARHL (Fig. 4A, Table S4, Table S5). Among them, SRC, AKT1, MAPK1, EGFR, PIK3R1, RXRA, TNF, ESR1, PTPN11, and AR were the top 10 nodes in terms of degree value (Fig. 4A, Table 2, Table S4). Next, we used MCODE to screen the top three sub-networks with high scores from the network of PPI. A total of three tightly connected network clusters were identified with MCODE scores ≥ 3.5 and nodes ≥ 6. Sub-network 1 included 18 nodes and 114 edges (Fig. 4B), sub-network 2 included 7 nodes and 38 edges (Fig. 4C), and sub-network 3 included 12 nodes and 40 edges (Fig. 4D).

Fig. 4.

Protein–protein interaction network of the target proteins. A Construction of the PPI network consisting of 122 nodes and 832 edges. Node size and color gradation correspond to the degree value. Orange nodes represent key targets, while green nodes represent other targets. Twenty-three nodes not connected to the network have been removed (PDE10A, GHRM2, RORA, HRH4, CSNK2A1, ACP1, MTNR1A, ABCC1, ADRA2A, HRH3, C5AR1, TYR, GABBR1, PDE6D, ALOX5AP, XDH, ASAH1, GABRG2, GABRB2, GABRB3, GABRA1, G6PD, GCK). The top three sub-networks with high scores are identified via MOCDE cluster analysis; B sub-network 1 comprises 18 nodes and 114 edges; C sub-network 2 comprises 7 nodes and 38 edges; D sub-network 3 comprises 12 nodes and 40 edges

Table 2.

Top ten key targets acting on ARHL

Number	Target	Protein names	Betweenness centrality	Closeness centrality	Degree
No. 1	SRC	Proto-oncogene tyrosine-protein kinase Src	0.337	0.487	68
No. 2	AKT1	RAC-alpha serine/threonine-protein kinase	0.247	0.476	62
No. 3	MAPK1	Mitogen-activated protein kinase 1	0.272	0.449	58
No. 4	EGFR	Epidermal growth factor receptor	0.214	0.468	58
No. 5	PIK3R1	Phosphatidylinositol 3-kinase regulatory subunit alpha	0.076	0.418	50
No. 6	RXRA	Retinoic acid receptor RXR-alpha	0.127	0.423	38
No. 7	TNF	Tumor necrosis factor	0.189	0.427	38
No. 8	ESR1	Estrogen receptor	0.077	0.407	36
No. 9	PTPN11	Tyrosine-protein phosphatase non-receptor type 11	0.014	0.392	34
No. 10	AR	Androgen receptor	0.140	0.399	34

Protein names are obtained from the Uniprot website (https://www.uniprot.org/). The normalized betweenness centrality values are divided by (n − 1)(n − 2)/2, where n is the number of nodes in the PPI network

Enrichment analysis of gene ontology and KEGG pathway for shared targets

To understand the function and the underlying significance of the TCM formulas’s therapeutic effect, the above 122 core targets were entered into the DAVID platform for GO term and KEGG pathways enrichment analysis. The details of GO and KEGG are listed in Table S6. The top 5 enriched GO terms showed biological processes mainly concentrated on positive regulation of MAP kinase activity, positive regulation of transcription from RNA polymerase II promoter, negative regulation of the apoptotic process, response to xenobiotic stimulus, and protein autophosphorylation. The cellular component is mainly composed of receptor complex, membrane raft, macromolecular complex, plasma membrane, and cytoplasm. The molecular functions were related to ligand-activated sequence-specific DNA binding, enzyme binding, protein serine/threonine/tyrosine kinase activity, zinc ion binding, and sequence-specific DNA binding (Fig. 5A). Additionally, the top ten KEGG enrichment pathways predominantly involved the PI3K-Akt pathway (consisting of targets of PDGFRB, GSK3B, FLT1, FLT3, INSR, PRKCA, PIK3R1, EGFR, IL2, PIK3CG, IGF1R, RXRA, CDK6, KIT, CDK2, MDM2, KDR, AKT1, MAPK1, JAK2, MET), MAPK signaling pathway (consisting of targets of PDGFRB, FLT1, FLT3, INSR, PRKCA, TNF, EGFR, IGF1R, RPS6KA3, KIT, KDR, AKT1, MAPK1, MAPT, MET), and neurodegeneration pathway (consisting of targets of GSK3B, APP, NOS2, SIGMAR1, PSEN2, PRKCA, PSEN1, PTGS2, TNF, GRIN2B, CDK5, NOX4, MAPK1, MAPT, CDK5R1) (Fig. 5B).

Fig. 5.

Gene ontology and KEGG pathway enrichment analysis of core targets working on ARHL. A GO-enriched terms in the biological processes (BP), cellular components (CC), and molecular functions (MF) categories, with the top 5 terms highlighted for each category (p < 0.05); B KEGG pathway analysis highlighting the top 10 enriched pathways (p < 0.05). GO and KEGG analysis of the sub-network are shown in C–H. GO analysis of sub-network 1 (C), sub-network 2 (E), sub-network 3 (G), KEGG pathway analysis of sub-network 1 (D), sub-network 2 (F), and sub-network 3 (H). Purple nodes represent relatively important targets. Data sourced from the DAVA database, with a heatmap generated using http://www.bioinformatics.com.cn

Furthermore, GO and KEGG analyses of the three clusters were also performed. Cluster 1 mainly concentrated on the biological process of protein autophosphorylation; molecular function on the enzyme binding, membrane receptor protein tyrosine kinase activity, and growth factor receptor binding. The KEGG enrichment analysis provided support for the pharmacological effects of cluster 1 being associated with the Ras signaling pathway, EGFR tyrosine kinase inhibitor resistance, and focal adhesion (Fig. 5C and D; Table S7). Cluster 2 was enriched in the thyroid hormone signaling pathway (consisting of targets of RXRA, THRB, THRA) (Fig. 5E and F; Table S8), while Cluster 3 was mainly enriched in the PI3K-Akt signaling pathway (consisting of targets of GSK3B, FLT1, CDK6, KIT, CDK2, PIK3R1, JAK2, IL2, PIK3CG) (Fig. 5G and H; Table S9).

Molecular docking reveals the interaction mode between compounds and targets

To elucidate the binding activity between the 11 common active compounds of three formulas and ten core target proteins with high degree values (SRC, AKT1, MAPK1, EGFR, PIK3R1, RXRA, TNF, ESR1, PTPN11, and AR), molecular docking analyses were conducted. The LibDock scores and interaction modes of the molecular docking between compounds and targets are depicted in the heatmap (Fig. 6) and Tables S10–11. Notably, 87 out of 110 docking scores exceeded 50 (ranging from 58 to 179), indicating a stable combination involving hydrogen bonds, hydrophobic interactions, and electrostatic forces between these compounds and core targets.

Fig. 6.

Heatmap of LibDock scores for eleven common active compounds and the core targets proteins. The x-axis represents the bioactive compounds; the y-axis represents the core targets. A darker color indicates a higher value, indicating better binding affinity between compounds and target proteins. The details of molecular docking are presented in Tables S10 and S11. AKT1, RAC-alpha serine/threonine-protein kinase; SRC, proto-oncogene tyrosine-protein kinase Src; MAPK1, mitogen-activated protein kinase 1; EGFR, epidermal growth factor receptor; PIK3R1, phosphatidylinositol 3-kinase regulatory subunit alpha; RXRA (NR2B), retinoic acid receptor RXR-alpha; TNFα, tumor necrosis factor-α; ESR1, estrogen receptor; PTPN11, tyrosine-protein phosphatase non-receptor type 11; AR, androgen receptor

Furthermore, to validate the reliability of binding affinity, specific ligands for three targets (Akt, TNFα, RXRA) were employed as positive controls. IP4, a specific agonist of protein kinase Akt, binds specifically to the PH domain and activates Akt in the cytosol (Jo et al. 2012). The IP4-Akt docking model revealed that IP4 forms hydrogen bonds with seven critical residues (Lys14, Gly16, Glu17, Tyr18, Ile19, Arg23, Arg25) within the active pocket of Akt, yielding a Libdock score of 71.2103. Similarly, the 11 compounds also formed hydrogen bonds and engaged in hydrophobic interactions with different key residues of Akt, with Libdock scores approximating or exceeding that of IP4 (ranging from 58.0256 to 99.476) (Fig. 7A–B, Table S10). BMS649, a specific agonist of RXRA (retinoid X receptor, alpha) (Egea et al. 2002), was used as another positive control. Docking results demonstrated that BMS649 forms hydrogen bond with three key residues (Arg316, Leu326, Ala327) and engages in hydrophobic interactions with 12 residues (Ile268, Ala271, Ala272, Leu309, Ile310, Phe313, Ile324, Ile345, Phe346, Val349, Ile428, Cys432), yielding a Libdock score of 127.495. Furthermore, six compounds (baicalin, ethyl oleate, kaempferol, mairin, mandenol, quercetin) exhibited hydrogen bonding and hydrophobic interaction patterns with several key residues similar to those of BMS649, with Libdock scores ranging from 91.5324 to 122.918 (Fig. 8A–B, Table S10). UCB-6876, an antagonist of TNFα, binds to the asymmetrical crystal form of the TNF trimer (O’Connell et al. 2019). Docking analysis revealed that UCB-6876 forms hydrogen bonds with two residues (Gly121, Tyr151) and engage in hydrophobic interactions with six other residues (Leu57, Tyr59, Gly121, Ile155, Tyr119) of TNFα, resulting in a Libdock score of 83.8009. Remarkably, eight compounds (baicalin, campesterol, ethyl oleate, kaempferol, mairin, mandenol, quercetin, sitosterol) demonstrated hydrogen bonding and hydrophobic interaction profiles similar to those of UCB-6876, with higher Libdock scores ranging from 96.9731 to 178.942 (Fig. 9A–B, Table S10).

Fig. 7.

Molecular docking of flavonoids with AKT1. (A) PH domain and hydrophobic pocket of AKT1. (B–F) Interactions of IP4, baicalin, kaempferol, quercetin, and leucanthoside with AKT1 exhibit stable binding interactions. The key residues involved in the binding are highlighted. The details of molecular docking and Libdock scores are presented in Table S10

Fig. 8.

Molecular docking of flavonoids with RXRA. A The ligand binding domain of RXRA (Synonym: NR2B1). B–E Interactions of BMS-649 (Synonym, SR11237), baicalin, kaempferol, and quercetin with RXRA in the same binding pocket, with high binding affinity. The key residues involved in binding are highlighted. Leucanthoside cannot interact with this binding pocket of RXRA, data is not shown. The details of molecular docking and Libdock scores are presented in Table S10

Fig. 9.

Molecular docking of flavonoids with TNFα. A Central hydrophobic pocket within TNFα homotrimer (chain A, B, C). B–E Interactions of UCB-6876, baicalin, kaempferol, and quercetin with TNFα in the same binding pocket, with high binding affinity. The key residues involved in binding are highlighted. Leucanthoside cannot interact with this binding pocket of TNFα, data not shown. The details of molecular docking and Libdock scores are presented in Table S10

Among the 11 compounds, three flavonoids (quercetin, kaempferol, baicalin) had interaction with almost all ten core targets, and another flavonoid, leucanthoside interacted with six core targets, although no putative targets were found via network pharmacology. The docking details between four flavonoids and the three core targets (Akt, RXRA, TNFα) were also present, showing stable docking similar to positive ligands (Figs. 7C–F, 8C–E, and 9C–E). The results strongly indicated that these active compounds exhibited a high biological affinity for several core targets.

Molecular docking reveals the interaction mode between flavonoids and potassium channels

Age-related hearing impairment involves the downregulation of the BKCa and Kv7.4 potassium channels in the cochlea and auditory pathway, prompting recent therapeutic interest in pharmacologically activating these two potassium channels (Peixoto Pinheiro et al. 2021, 2022). Flavonoids derived from medicinal herbs have been identified as potent activators of BKCa and Kv7.4 channel activators (Lee et al. 2018; Redford and Abbott 2020). Molecular docking was employed here to investigate the stable interaction and activation potential of the flavonoids discovered in the current study with these channels. CTBIC, a potent activator of the BKCa channel, binds to a pocket between two adjacent α-subunits in the outer vestibule of the channel, thereby potentiating its activity (Lee et al. 2012). Our findings demonstrated that CTBIC forms hydrogen bonds with Trp275, Thr298, Thr298, and Arg301 while engaging in hydrophobic interactions with seven other residues (Trp246, Trp275, Val278, Leu302, Val305) within the binding pocket of the BKCa channel, with a Libdock score of 97.10 (Fig. 10A–C, Tables S12, S13). Furthermore, four flavonoids (quercetin, baicalin, kaempferol, leucanthoside) exhibit the ability to form hydrogen bonds and hydrophobic interactions with key residues of the BKCa channel, although it is somewhat inconsistent with those observed in the CTBIC-BKCa complex. Notably, the Libdock scores of the four flavonoid-BKCa complexes (75.8555, 107.097, 90.2986, and 97.4712, respectively) closely resemble that of the CTBIC-BKCa complex (Fig. 10D–G, Table S12, S13).

Fig. 10.

Molecular docking of four flavonoids with BKCa and Kv7.4. A Macroscopic diagram of BKCa channel. B CTBIC, a potent activator of BKCa, binds to the hydrophobic pocket and activates BKCa channel. C–G Docking results of CTBIC, baicalin, kaempferol, and quercetin, leucanthoside with BKCa channels. H Macroscopic diagram of Kv7.4 potassium channel. I ML213, a potent activator of Kv7.4, binds to the hydrophobic pocket and activates the Kv7.4 channel. J–N Docking results of ML213, baicalin, kaempferol, and quercetin, leucanthoside with Kv7.4 channels. The details of molecular docking and Libdock scores are presented in Tables S12 and S13

On the other hand, ML213, a specific activator of the Kv7.4 channel, interacts stably with key residues within a hydrophobic pocket formed by the S4-S5 linker, S5, and S60 from the neighboring subunit at the middle of the membrane (Zheng et al. 2022). Our analysis of the ML213-Kv7.4 complex revealed hydrogen bonds with Trp242, Leu305 and Ser309, along with hydrophobic interactions with several key residues (Trp242, Leu305, Leu306, Phe310, Phe311, Pro314), resulting in a Libdock score of 75.7005 (Fig. 10H–J, Table S12, S13), consistent with previous studies (Kim et al. 2015; Zheng et al. 2022). Notably, the four flavonoids exhibit similar hydrogen bonding and hydrophobic interaction patterns with key residues of the Kv7.4 channel, displaying higher Libdock scores (106.855, 93.7003, 108.367, and 80.2967, respectively) compared to the ML213-Kv7.4 complex (Fig. 10K–N, Table S12, S13). These molecular docking results support the notion of stable associations between the four flavonoids and potassium channels, suggesting their pharmacological activation potential and potential contribution to mitigating auditory senescence.

Experimental validation of flavonoids in mitigating auditory senescence in DBA/2 J mice

Flavonoid treatment delays ABR wave I amplitude decrease

Network pharmacology and molecular docking analyses identified eleven major active compounds from three TCM formulas against ARHL and their potential regulatory targets. To verify the protective effects of these compounds, quercetin, and baicalin were selected due to their significant interactions with ARHL-related targets and their high binding affinity with target proteins, as demonstrated by the molecular docking results. Therefore, two flavonoids (quercetin, baicalin) were investigated for their ability to mitigate auditory senescence in DBA/2 J mice, which exhibit early-onset progressive hearing loss (Willott et al. 1984; Johnson et al. 2008). The details of the experimental design of auditory measurements and drug administration are depicted in Fig. 11.

Fig. 11.

Experimental scheme to investigate the effects of quercetin and baicalin on ARHL in DBA/2 J mice. After ABR recordings were obtained from DBA/2 J mice (P21-P22) as the baseline, they were randomly divided into three groups. Group 1 served as a control with saline treatment, and Groups 2 and 3 were treated with quercetin and baicalin, respectively. One month after drug administration, ABR recordings were performed. Two months after drug administration, ABR recordings were first completed, and then cochleae were taken from these mice for experimental verification

Initially, auditory brainstem responses (ABRs) were recorded in mice from three groups (saline, quercetin, baicalin) at various frequencies and time points to assess the impact of these compounds on hearing phenotype. Compared with baseline (P21-23), mice treated with saline for 1 month exhibited significant increases in thresholds at 22.6 kHz (from 81.55 ± 0.7064 to 87.78 ± 1.690 dB, p < 0.05) and 16 kHz (from 60.34 ± 2.162 to 80.56 ± 2.115, p < 0.0001), indicative progressive hearing loss in DBA/2 J mice. However, 1 month of treatment with quercetin or baicalin at 15 mg/kg failed to attenuate the uptrend in hearing thresholds (Fig. 12A). Additionally, neither quercetin nor baicalin treatment exhibited significant effects on the amplitudes and latencies of wave I across the frequency spectrum (from 5.6 to 11.3 kHz) (Fig. 12B–G). These results suggest that 1 month of low-dose (15 mg/kg) quercetin or baicalin treatment does not mitigate age-related auditory senescence of DBA/2 J mice, likely due to inadequate drug concentration within the cochlea, attributable to the relatively impermeable internal environment of the inner ear. Consequently, mice were administered with relatively higher doses of quercetin (50 mg/Kg) or baicalin (30 mg/Kg) for an additional month. The results showed there was no significant change in the hearing thresholds across the frequency spectrum (from 4 to 11.3 kHz) of mice treated with quercetin (50 mg/Kg) or baicalin (30 mg/Kg), compared to mice treated with saline (Fig. 13A). Quercetin treatment did not impact the latencies and amplitudes of wave I across the frequency spectrum (from 4 to 8 kHz), compared with saline treatment (Fig. 13B–G). However, baicalin treatment resulted in larger wave I amplitudes at low frequencies (4 kHz and 5.6 kHz) compared to saline treatment (Fig. 13C and E). These findings suggest that baicalin treatment may modestly delay the decrease in wave I amplitude at low frequencies with age.

Fig. 12.

ABR threshold and wave I analysis in mice treated with flavonoid for 1 month. A This plot shows ABR thresholds for two treatment groups compared to the control group (saline) 1-month post-administration. A two-way ANOVA revealed a main effect of frequency (F (6, 169) = 266.4, p < 0.0001). Post-hoc analyses indicated no significant threshold changes with quercetin or baicalin compared to saline. Arrows indicate thresholds have exceeded 90 dB in some mice at high frequencies; thus, actual values at 32 kHz and 22.6 kHz may be higher than shown. N values at high frequencies: saline (32 kHz, 7/9; 22.6 kHz, 8/9), quercetin (32 kHz, 7/10; 22.6 kHz, 10/10), and baicalin (32 kHz, 5/10; 22.6 kHz, 8/10). B Two-way ANOVA showed the main effects of treatment (F (2, 123) = 7.492, p = 0.0009) and intensity (F (4, 123) = 72.75, p < 0.0001). Post-hoc analyses revealed prolonged latency in the baicalin group at 80 dB compared to saline. C The amplitude at 5.6 kHz showed no significant changes with quercetin or baicalin compared to saline. The values represent the amplitude at a frequency of 5.6 kHz. Two-way ANOVA indicated the main effects of treatment (F (2, 118) = 4.431, p = 0.0139) and intensity (F (4, 118) = 19.12, p < 0.0001). D Latency at 8 kHz showed no significant differences with treatment. Two-way ANOVA revealed a main effect of intensity (F (4, 116) = 10.66, p < 0.0001). E Amplitude at 8 kHz showed a significant increase at 80 to 90 dB with baicalin administration. Two-way ANOVA indicated the main effects of treatment (F (2, 115) = 10.55, p = 0.0005) and intensity (F (4, 115) = 9.42, p < 0.0001). F Latency at 11.3 kHz showed no significant changes with treatment. Two-way ANOVA revealed the main effects of treatment (F (2, 136) = 5.101, p = 0.0073) and intensity (F (6, 136) = 26.25, p < 0.0001). G Amplitude at 11.3 kHz increased significantly at 70 and 90 dB with baicalin administration. Two-way ANOVA indicated a main effect of treatment (F (2, 85) = 7.176, p = 0.0013, n = 8–10)

Fig. 13.

A ABR threshold and Wave I analysis in mice treated with flavonoid for 2 months. The plot shows ABR thresholds for two treatment groups compared to the control group (saline) 2 months post-administration. Two-way ANOVA revealed the main effects of frequency (F (3, 85) = 7.741, p = 0.0001) and treatment (F (2, 85) = 2.495, p = 0.0885). Post-hoc analyses showed no significant threshold changes with quercetin or baicalin. Arrows indicate actual thresholds at 11.3 kHz are higher than shown. N values: saline (11.3 kHz, 8/9), quercetin (11.3 kHz, 8/9), and baicalin (11.3 kHz, 5/9). B Latency at 4 kHz showed a main effect of intensity (F (2, 48) = 22.22, p < 0.0001), with no significant changes due to treatment. C Amplitude at 4 kHz showed the main effects of intensity (F (2, 42) = 10.86, p = 0.0002) and treatment (F (2, 42) = 7.613, p = 0.0015). No significant changes relative to saline with quercetin or baicalin. D Latency at 5.6 kHz showed a main effect of intensity (F (3, 69) = 29.54, p < 0.0001), with no significant changes due to treatment. E Amplitude at 5.6 kHz showed a significant increase with baicalin. Two-way ANOVA indicated main effects of intensity (F (3, 69) = 11.75, p < 0.0001) and treatment (F (2, 69) = 15.96, p < 0.0001). F Latency at 8 kHz showed main effects of intensity (F (4, 71) = 20.67, p < 0.0001) and treatment (F (2, 71) = 4.736, p = 0.0117), with no significant changes due to treatment. G Amplitude at 8 kHz showed no main effect of treatment. Post-hoc analyses revealed no significant changes with quercetin or baicalin. The saline group experienced severe hearing loss at 4 kHz, limiting data analysis. Data are presented as mean ± SEM and evaluated with two-way ANOVA followed by Bonferroni post-test (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001; n = 4–8)

Baicalin treatment preserves ribbon synapses integrity in IHCs

To explore the potential morphological changes in the cochlea associated with improved hearing function following baicalin treatment, the cochleae of mice subjected to ABR experiments were collected and analyzed. H&E staining revealed no significant differences in the areas of the stria vascularis or the density of spiral ganglion neurons (SGNs) across cochlear turns between mice treated with quercetin or baicalin and control treated with saline (Fig. 14A–D), ruling out these structures as the source of differences in wave I amplitude. As shown in Fig. 15, immunofluorescent staining revealed substantial loss of outer hair cells (OHCs), particularly in the basal and middle turns, and inner hair cells (IHCs) in the basal turn of the cochlea in control mice (Fig. 15A), consistent with previous reports of cochlear morphological degeneration and hearing loss in aging DBA/2 J mice. Quercetin or baicalin treatment did not significantly rescue the OHCs and IHCs loss across all cochlear turns (Fig. 15B).

Fig. 14.

Stria Vascularis morphology and SGNs count in mice treated with flavonoid. A Representative H&E staining images of the stria vascularis in the apical, middle, and basal turns of cochleae from three groups (scale bar, 50 µm). B No visible atrophy was detected. The sectional area of stria vascularis in the apical, middle, and, basal turns showed no significant differences among all the three groups (all p > 0.05, two-way ANOVA followed by Bonferroni post-test). C Representative H&E staining images of SGNs in the apical, middle, and basal turns of the cochlea from three groups (scale bar, 50 µm). D SGN density (number of SGNs/area of Rosenthal’s canal) showed no significant difference among the three groups. Two-way ANOVA showed a main effect of treatment (F (2, 21) = 4.925, p = 0.0176) and a main effect of the cochlear region (F (2, 21) = 23.36, p < 0.0001). Post-hoc analyses showed that compared with the control, SGN density showed no significant difference after quercetin and baicalin administration (all p > 0.05 at, two-way ANOVA followed by Bonferroni post-test, n = 4 ~ 5 cochleae in each group)

Fig. 15.

Quantitative analysis of hair cells and ribbon synapses in mice treated with flavonoids. A Representative immunofluorescent images of cochlear OHCs and IHCs from DBA/2 J mice from three groups, captured at the specific regions (apical, middle, and basal turns) of cochleae (scale bar, 20 µm). B Quantitative analysis of the number of OHCs and IHCs. For OHCs, a two-way ANOVA of histograms showed the main effect of treatment (F (2, 25) = 224.2, p < 0.0001). For IHCs, a two-way ANOVA of histograms also showed a main effect of treatment (F (2, 25) = 52.21, p < 0.0001). Post-hoc analyses showed that the numbers of OHCs and IHCs in all three specific regions exhibited no significant differences among the three groups (all p > 0.05). C Representative images of IHC synapses from the apical, middle, and basal cochlea of DBA/2 J mice stained with presynaptic CtBP2 (green puncta) clustered at the basolateral pole of IHCs. The IHC nuclei were also labeled due to the nuclear expression of CtBP2 (scale bar, 10 µm). D Quantitative analysis of the number of CtBP2 puncta per IHC. A two-way ANOVA showed a main effect of treatment (F (2, 21) = 30.60, p < 0.0001). Post-hoc analyses showed that the numbers of CtBP2 puncta per IHC in middle and basal turns exhibited no significant differences among the three groups. In the apical turn, the number of CtBP2 puncta per IHC in the baicalin group was significantly larger than that in the control group. Data are presented as mean ± SEM, n = 4 ~ 5 cochleae in each group; *p < 0.05, compared with control group

Then we quantified the ribbon synapses between IHCs and auditory nerve fibers, since wave I amplitude is primarily determined by SGN firing. Each IHC’s basal pole is surrounded by approximately a dozen ribbon synapses, identifiable as closely juxtaposed pairs of fluorescent puncta representing presynaptic CtBP2 (a major component of the synaptic ribbon that facilitates glutamate release via exocytosis) and postsynaptic glutamate receptor 2 (Glowatzki and Fuchs 2002; Nouvian et al. 2006). The number of presynaptic ribbons (CtBP2) was quantified to assess IHC synapses. Compared to control mice, mice treated with quercetin exhibited no significant difference in the number of IHC synapses across all turns of the cochlea, but mice treated with baicalin showed a significantly greater number of IHC synapses in the apical turn of cochlea (Fig. 15C–D), corroborating the larger ABR wave I amplitudes at low frequency observed in baicalin-treated mice. These results support the conclusion that baicalin treatment partially preserves low-frequency sound transmission by reducing the loss of ribbon synapses per IHC and the decline in SGN firing in aging DBA/2 J mice.

Baicalin treatment modulates the expression levels of multiple targets

Molecular docking analysis revealed that baicalin interacts stably with multiple potential targets involved in inflammation and various signaling pathways (Fig. 6). These target proteins are implicated in regulating auditory pathophysiology, including chronic inflammation in the peripheral and the central auditory system contributing to ARHL development (Menardo et al. 2012; Verschuur et al. 2012; Nash et al. 2014; Watson et al. 2017). To investigate whether these target proteins and associated proteins are involved in the improved auditory function observed in baicalin-treated DBA/2 J mice, their expression levels were quantified. Quantitative PCR experiments revealed significantly higher expression levels of RXRA, MAPK1 (ERK2), SRC, and ESR1 (estrogen receptor α) in the cochlea of baicalin-treated mice, along with lower expression levels of PIK3CB_p110β, PTPN11, and AR (androgen receptor), compared to control mice. Moreover, the expression levels of inflammation-related genes including NLRP3, TNFα, IL1β, and IL6 were markedly reduced in baicalin-treated mice compared to control mice (Fig. 16). These findings suggest baicalin may delay the decline in auditory function in DBA/2 J mice by modulating multiple target proteins, though the molecular mechanisms warrant further exploration.

Fig. 16.

Expression analysis of target proteins involved in the otoprotective effectiveness of baicalin. A The mRNA expression levels of four inflammation-related molecules (TNFα, NLRP3, IL1β, and IL6) in the cochleae of control mice and baicalin-treated mice. B The mRNA expression levels of eleven targets (RXRA, EGFR, PIK3CB_p85β, PIK3CB_p110β, AKT1, mTOR, MAPK1, SRC, PTPN11, ESR1, and AR) implicated in the otoprotective effects of baicalin. Student’s two-tailed t-test was used for statistical tests. Data are presented as the mean ± SEM, *p < 0.05; **p < 0.01; ****p < 0.0001, compared with control group

Discussion

Aging is a gradual and irreversible biological process, characterized by time-dependent functional declines in all organ systems and significant increases in the risks of various aging-related diseases. To date, aging is believed to be driven by twelve hallmarks (Lopez-Otin et al. 2013, 2023). Given the multiplicity of potential pathogenesis, exploring therapeutic strategies to decelerate aging and treat aging-related diseases, presents an arduous challenge. Age-related hearing loss (ARHL) is a typical and multifaceted age-related disease characterized by progressive and irreversible degenerative processes. ARHL is attributed to impaired function and consequent cellular loss both in the peripheral and central auditory pathways (Lee 2013). The complex and heterogeneous pathophysiology underlying ARHL is not fully understood so far. Although there are still no specific and effective medications approved for ARHL treatment, strategies such as antioxidants, caloric restriction, anti-inflammatory agents, anti-apoptotic compounds, and modulation of membrane ion channels/receptors have been considered promising (Bednar et al. 2015; Tavanai and Mohammadkhani 2017; Bowl and Dawson 2019; Wang and Puel 2020; Rybalko et al. 2021). Due to its extensive history in preventing and treating age-related disorders, traditional Chinese medicine (TCM) including Chinese medicine formulas and traditional Chinese herbs, is considered as a potential resource for preventing and treating ARHL. Several formulas (such as ELZC, Jian-Er, and Yi-Qi Cong-Ming decoction), herbs (Ginkgo biloba), and compounds (such as resveratrol, ursolic acid, and resveratrol) extracted from herbs have been shown promise in partially delaying the loss of hearing capability in humans and animals, as well as in rescuing hearing loss-related pathological changes in the cochlea, probably due to their multi-component and/or multi-target properties (Hu et al. 2023; Yan et al. 2023). Of course, in the practice of TCM treatment for hearing-related diseases, there remain several challenges to address. For example, what are the potential molecular mechanisms of TCM or herbal medicine to protect hearing? What are the key active ingredients in formulas and herbs, their targets, and mechanisms, could they serve as monotherapies for hearing loss treatment?

The present study has identified eleven common compounds (quercetin, baicalin, kaempferol, sitosterol, stigmasterol, campesterol, diop, mairin (betulinic acid), mandenol, ethyl oleate, and leucanthoside (swertiajaponin)) from three TCM formulas (ELZC, TQEL, and EL) that are cataloged in the Chinese Pharmacopoeia and utilized in clinical practice for the treatment of ARHL. These eleven key bioactive compounds are presumed to form the foundation of the therapeutic efficacy of the three TCM formulas in ARHL treatment. Notably, three of these compounds (quercetin, baicalin, kaempferol) have been associated with mitigating drug-induced ototoxicity and noise-induced hearing loss. Quercetin, a naturally non-toxic flavonoid, for instance, has been demonstrated to rescue ABR threshold shift and alleviate histopathological cochlear lesions induced by gentamicin and cisplatin in rodent models (Sagit et al. 2015; Gundogdu et al. 2019). Quercetin has also shown protective effects on auditory function in noise-induced hearing loss in rodents, suppressing ABR threshold shifts, preserving outer hair cells, and reducing inflammation and oxidative stress (Hirose et al. 2016; Goodarzi et al. 2023). Baicalin and its aglycon baicalein extracted from Scutellaria baicalensis have also exhibited protective effects on auditory function in noise-induced hearing loss in rodent models (Kang et al. 2010; Rodriguez et al. 2017). Kaempferol, a natural flavonoid, has been found to inhibit cisplatin-induced apoptosis in HEI-OC1 cells by upregulating the expression of heme oxygenase and catalytic subunit of glutamate-cysteine ligase (Gao et al. 2010). Furthermore, kaempferol treatment can ameliorate noise-induced alterations in oxidative status and improve neurobehavioral performance in rodents exposed to noise-induced stress (Akefe et al. 2020). Although the protective effects on auditory function of the other eight compounds remain unclear, some of them have been associated with other diseases, attributed to their unique pharmacological properties. As summarized in a recent review(Bakrim et al. 2022), stigmasterol, an unsaturated phytosterol possesses multiple pharmacological effects such as antioxidant, anti-inflammatory, anti-osteoarthritis, anticancer, anti-diabetic, and neuroprotective properties. Mairin, also known as Betulinic acid, a pentacyclic triterpene compound, has been verified to be able to modulate a number of essential mediators in the inflammatory process (Oliveira-Costa et al. 2022). Furthermore, it also can induce mitochondrial apoptotic pathways in differentiated PC12 cells through reactive oxygen species (ROS)(Wang et al. 2017). The anti-inflammatory and anti-oxidant activities of these compounds might contribute to the three TCM formulas in the treatment of ARHL. The potential targets of action for ARHL of all the eleven compounds in the auditory system were determined through network analysis and molecular docking in the present study.

The Venn diagram in Fig. 2 illustrates that among eleven common compounds and ARHL, there are 145 shared targets identified. With the exception of leucanthoside, each of the other ten compounds had more than two potential targets, showing the multi-component and multi-target properties of TCM. Furthermore, based on protein–protein interaction (PPI) network analysis, 123 targets were identified as potential core targets of TCM for ARHL treatment. Among them, SRC, AKT1, MAPK1, EGFR, PIK3R1, RXRA, TNFα, ESR1, PTPN11, and AR emerge as the top 10 nodes with the highest degree values, indicating their pivotal roles as hub targets. Molecular docking analysis confirmed the strong biological affinity between the eleven active compounds and these ten hub targets. Particularly noteworthy is the stable interaction observed between three flavonoids (quercetin, baicalin, kaempferol) and almost all ten hub targets, suggesting their potential significance in ARHL treatment.

Furthermore, through GO function and KEGG pathway enrichment analysis, we found that the 123 shared targets are primarily involved in the PI3K/AKT pathway, MAPK signaling pathway, Ras signaling pathway, neurodegeneration pathway, and hormone signaling pathway and are therefore clearly associated with hearing loss. The PI3K-Akt signaling pathway, known for its role in cellular responses to external stimuli, regulates various cellular processes. Dysfunction of this pathway is implicated in metabolic, cardiovascular, and neurological disorders (Fruman et al. 2017; Matsuda et al. 2019; Goyal et al. 2023). Akt, a pivotal effector downstream of PI3K and the core node of the PI3K-Akt pathway, regulates cellular processes by modulating multiple targets through its serine/threonine kinase activity. The detection of activated Akt in several inner ear structures underscores its indispensable role in inner ear physiology (Hess et al. 2006). While all three Akt isoforms (Akt1, Akt2, Akt3) are expressed in the cochlea and contribute to normal hearing, Akt2 and Akt3 specifically play roles in hair cell survival in response to aminoglycoside ototoxic stress (Brand et al. 2015). Emerging evidence confirms that activation of the PI3K-Akt pathway protects cochlear hair cells and spiral ganglion neurons from gentamicin and cisplatin ototoxicity (Chung et al. 2006; Heinrich et al. 2015; Kucharava et al. 2019; Liu et al. 2021b; He et al. 2022). Pharmacological activation of the PI3K-Akt pathway has been suggested as a promising treatment for noise-induced hearing loss, as it can reduce apoptosis and restore cellular homeostasis (Haake et al. 2009, Kurioka et al. 2014; Chen et al. 2015). Importantly, attenuation of the PIP3/Akt cascade disturbs cellular homeostasis, shifting the balance towards cell death, and thereby contributing to age-related hearing loss (Sha et al. 2010). Based on existing research, pharmacological activation of PI3K-Akt signaling emerges as a potential therapeutic strategy for ARHL, possibly delaying the irreversible loss of sensory HCs and degeneration of SGNs in the aging cochlea. Our study revealed that eleven common compounds from three TCM formulas exhibit stable binding to the active pocket of Akt1, with a binding mode similar to that of IP4, a specific Akt agonist (Jo et al. 2012). Notably, among the eleven compounds, seven compounds (baicalin, leucanthoside, kaempferol, diop, ethyl oleate, mairin, mandenol) demonstrated higher binding affinity with Akt than IP4, suggesting their potential as potent activators of Akt. Furthermore, our study also uncovered molecular interactions between these eleven compounds and PIK3R1 (also known as p85α), a regulatory subunit of PI3K, although the precise modulatory action remains to be elucidated. These findings suggest that the activation of the PI3K/Akt signaling pathway may be implicated in the therapeutic mechanism of TCM treatment for ARHL.

Mitogen-activated protein kinases (MAPKs) are Serine-threonine protein kinases known for regulating cellular responses to extracellular stimuli. Classical MAPKs such as ERK1/2/5, JNK1/2/3, and p38 isoforms (MAPK11/12/13/14) play pivotal roles in mediating tissue growth/survival and stress responses, respectively (Kyriakis and Avruch 2012). ERK1/2 (also called MAPK2/1) activation is commonly thought to promote cell proliferation and survival. Direct evidence in the cochlea shows that inhibiting the activation of ERK1/2 in the cochlea leads to the loss of OHCs and is accompanied by enhanced gentamicin-induced toxicity (Battaglia et al. 2003). Moreover, activation of ERK1/2 in cochlear supporting cells has been shown to maintain OHC survival (Bell and Oberholtzer 2010; Hayashi et al. 2013). However, increasing numbers of studies indicate a dual role of the ERK1/2 pathway in the cochlea. Gentamicin-induced apoptosis of auditory cells involves ERK1/2 pathway activation, leading to ERK translocation to the nucleus, transcriptional upregulation of the proapoptotic factor Hrk, and initiation of apoptosis (Kalinec et al. 2005). Activation of the ERK1/2 pathway is also implicated in cisplatin- and noise-induced hearing loss, thus, inhibition of this pathway in response to cochlear cell damage represents a valuable strategy for hearing loss protection and treatment (Ingersoll et al. 2024). The JNK and p38 signaling pathways are mainly stimulated by environmental stresses and contribute to cell apoptosis (Davis 2000). JNK pathway activation is associated with neomycin- and noise-induced hair cell loss, its inhibitors such as CEP-1347 and D-JNKI-1 can prevent the death of both cochlear hair cells and neurons, offering protection against acoustic and ototoxic trauma, and preserve hearing function (Pirvola et al. 2000, Wang et al. 2003). Inhibition of p38 activation has been linked to protection against aminoglycoside and noise exposure (Wei et al. 2005, Jamesdaniel et al. 2011, Bas et al. 2012, Maeda et al. 2013). In the context of ARHL, there is compelling evidence indicating the activation of JNK and p38 signaling linked to oxidative stress in the cochlea of aging CBA mice (Sha et al. 2009). These findings support that therapies targeting the inhibition of MAPK pathways hold promise as therapeutic strategies for ototoxic drug- and noise-induced hearing loss, as well as ARHL. In our study, four flavonoid compounds (baicalin, quercetin kaempferol, leucanthoside) from TCM formulas may modulate the MAPK signaling pathway by binding to and regulating the phosphorylation or expression of MAPK proteins (e.g., ERK2/MAPK1), thereby contributing to ARHL treatment.

Chronic inflammation plays a critical role in the development of age-related diseases, including ARHL. Among the cytokines expressed in the cochlea, TNFα stands out as one of the most important due to its pro-inflammatory properties. It promotes a robust inflammatory response and activates downstream apoptotic signaling pathways that can lead to the death of auditory hair cells (Keithley et al. 2008; Watson et al. 2017). Blocking TNFα expression and associated signaling cascades such as ERK, p38, and JNK signaling may attenuate inflammatory and pro-death responses in the cochlea, thus potentially rescuing hearing loss (Haake et al. 2009, Dhukhwa et al. 2019). As discussed earlier, chemical compounds derived from TCM herbs possess diverse medicinal properties, including anti-inflammatory effects. These compounds can interact with various molecules involved in inflammatory pathways, thereby decreasing the activity of cytokines, chemokines, and inflammatory enzymes. In our study, network pharmacological analysis revealed that baicalin, a typical flavonoid known for anti-inflammatory properties, targets TNFα. Molecular docking analysis demonstrated a higher binding affinity of baicalin with TNFα compared to UCB-6876, a specific TNFα antagonist, suggesting its potential as a potent TNFα inhibitor. Moreover, seven other compounds (kaempferol, quercetin, campesterol, ethyl oleate, mairin, mandenol, and sitosterol) also showed stable binding with TNFα, exhibiting a similar interaction mode to that of UCB-6876. This suggests their potential pharmacological property in combating ARHL-associated inflammation and apoptosis in the auditory system.

To date, a growing body of research has revealed the intricate interaction between hormonal fluctuations and the onset and progression of ARHL. Various hormones such as sex hormones, thyroid hormones, glucocorticoids, aldosterone, growth hormones, and insulin have been implicated in mediating antiapoptotic effects on cells crucial to the structure of the cochlea and auditory pathway, and modulating the functionality of hair cells and the electrolyte balance within the endo- and perilymph (Frisina et al. 2021). Despite the potential benefits, the prospect of long-term hormone treatment, especially in the context of ARHL, is currently deemed impractical due to the associated systemic side effects and the lack of feasible local application options. Consequently, the focus shifts towards the modulation of hormone receptors and related molecules as a more viable strategy. Our study revealed that the regulation of hormone-mediated signaling pathways may be implicated in the treatment of ARHL, as evidenced by GO and KEGG analysis. Significantly, PPI analysis of core targets highlighted estrogen receptor α (ESR1) and androgen receptor (AR) as two of the top ten nodes in terms of degree value. There is a common physiological phenomenon that females have better hearing than males of the same age, which suggests the opposite role of two sex hormone-related pathways in the regulation of hearing (Guimaraes et al. 2004; Henry 2004). Previous studies have revealed that ESRα (ESR1) and ESRβ (ESR2) expressed in cochlea help determine auditory sound processing response parameters such as ABR threshold and the amplitude and latency of wave I (Williamson et al. 2019). Age-related reduction in the two types of ESR and a decrease in estrogen production along with the downregulation of estrogen-related pathways contribute to perceptual processing deficits of ARHL(Frisina et al. 2021). A recent study reported that inhibition of AR by flutamide (a non-steroidal antiandrogen drug) protects against cochlear injuries in kanamycin-induced hearing loss rats (Chun et al. 2021). Among 11 common compounds derived from three TCM formulas, nine compounds (sitosterol, quercetin, baicalin, campesterol, stigmasterol, diop, kaempferol, mandenol, ethyl oleate) were identified to stably bind to ESR1, while four compounds (quercetin, kaempferol, mandenol, ethyl oleate) exhibited stably binding to AR. These findings suggest that these compounds hold promise for modulating the functions of ESR1 and AR, potentially offering therapeutic benefits for ARHL.

Potassium channels play a pivotal role in maintaining the proper functioning of the auditory system. Among these channels, certain potassium channels have emerged as promising therapeutic targets against ARHL (Peixoto Pinheiro et al. 2021). Kv7.4, encoded by KCNQ4, is an M-type K⁺ channel expressed in the inner ear that plays an indispensable role in auditory function, contributing to potassium recycling and homeostasis maintenance. Mutations in KCNQ4 lead to DFNA2, an autosomal dominant form of progressive hearing loss in humans (Kharkovets et al. 2000), which is predominantly caused by a slow degeneration of OHCs resulting from chronic depolarization (Kharkovets et al. 2006). Numerous subsequent studies have confirmed the correlation between impaired function or reduced activity of Kv7.4 and ARHL, establishing it as a clinically promising drug target for ARHL treatment (Rim et al. 2021). A recent report stated that pharmacological activation of this potassium channel using ACOU085 (a potent agonist of the Kv7.4) protects hearing function and promotes OHC survival in a senescence-accelerated mouse-prone 8 (SAMP8) mouse model (Peixoto Pinheiro et al. 2022). Natural compounds and their derivatives from herbs are well-known sources of potassium channel modulators, especially flavonoids (Richter-Laskowska et al. 2023). Quercetin, a typical flavonoid with a wide range of biological activities, has been shown to potentiate outward K⁺ currents mediated by Kv7 channels exogenously in oocytes (Redford and Abbott 2020). Our study demonstrated that quercetin stably binds to the activation pocket of Kv7.4 channels, exhibiting similar hydrogen bonding and hydrophobic interaction patterns, and a higher Libdock score compared to ML213, a potent activator of Kv7 channels (Zheng et al. 2022). This result provides molecular details of the interaction between quercetin and Kv7 channels, elucidating its potential to increase Kv7 currents. Additionally, three other flavonoids (baicalin, kaempferol, leucanthoside) also exhibited stable binding to Kv7.4, with a similar interaction mode to ML213. These findings suggest potential roles for these flavonoids as Kv7 agonists, which thereby contribute to mitigating auditory senescence.

On the other hand, large-conductance Ca²⁺-activated K⁺ (BK) channels expressed in cochlear hair cells mediate fast-activating outward K⁺ conductance, predominantly contributing to high-frequency hearing in mammals (Pyott and Duncan 2016; Latorre et al. 2017). Previous studies have revealed that germline knockout of BKα (encoded by KCNMA1), the pore-forming subunit of the BK channel in mice, has no effect on hearing thresholds but increases resistance to noise-induced hearing loss (Pyott et al. 2007). BK channels in IHCs are crucial for temporal fine structure coding of sound and signal detection in a noisy environment (Kurt et al. 2012). Interestingly, a recent study reported that the BK current in IHCs is largely reduced in early-onset hearing loss mice with age (Jeng et al. 2021). These lines of evidence hint at the potential importance of the BK channel in ARHL. Recently, the impact of flavonoids on the function of BK channels has been extensively studied. At least a dozen flavonoids have been shown to ameliorate various pathologies through modulating the activity of BK channels (Richter-Laskowska et al. 2023). For example, quercetin activates plasma-membrane BK channel by increasing channel open probability in cultured human endothelial cells, human bladder cancer cells, and mouse ileal myocytes (Kuhlmann et al. 2005; Kim et al. 2011; Melnyk et al. 2019). Additionally, quercetin activates mitochondrial BK channels at ten times lower concentration compared to plasma membrane BK channels (Kampa et al. 2021). Kaempferol increases the open probability of BK channels in human umbilical vein endothelial cells, resulting in membrane hyperpolarization (Xu et al. 2008). Through molecular docking analysis, the present study provided molecular details of the interaction between the two flavonoids (quercetin and kaempferol) and the pore-forming subunit of the BK channel, which serves as the cornerstone of their pharmacological activity. Moreover, two other flavonoids (baicalin and leucanthoside) both can bind to the activation pocket of BK channels with high affinity, displaying a similar interaction mode to CTBIC (a potent activator of BK channels), which suggests the possibility of these two flavonoid compounds as novel BK channel activators. Although the role of the BK channel in ARHL development needs further exploration, these findings suggest potential roles for the four flavonoids as BK channel agonists, and possibly contributing to mitigating auditory senescence. Certainly, further exploration is needed into how flavonoids modulate the biophysical properties of the two potassium channels. In short, targeting the crucial channels within the organ of Corti could be a promising therapeutic option for ARHL, as the survival of cochlear hair cells is dependent on the functional K⁺ recycling circuit.

Then, the present study assessed the protective effects of two flavonoids (quercetin, baicalin) on ARHL in DBA/2 J mice, which serve as a valuable model for studying age-related progressive hair cell death and hearing loss (Willott et al. 1984; Johnson et al. 2008). DBA/2 J mice carry recessive alleles that lead to the development of an early-onset hearing loss that starts at as early as 3 weeks of age, with the loss of cochlear HCs and SGNs being the primary pathological features during the progression of hearing loss (Noben-Trauth et al. 2003; Shin et al. 2010). Our findings revealed that compared to baseline (3 weeks of age), older mice (7 and 11 weeks of age) exhibited a gradual deterioration in hearing function, characterized by age-related shifts in ABR thresholds and a progressive loss of cochlear hair cells from high to low frequencies, consistent with prior research. Previous studies have highlighted the therapeutic potential of quercetin in age-related diseases (Deepika and Maurya 2022), and the present study also explored its regulatory targets and associated signaling pathways in ARHL. However, our investigation showed that daily systematic administration of quercetin for 2 months failed to significantly attenuate the age-related elevation of ABR thresholds or rescue the loss of cochlear hair cells in aging DBA/2 J mice. This unexpected outcome is likely attributed to the ineffective drug concentration within the cochlea resulting from the systematic administration dosage utilized in this study (daily 15–50 mg/kg). As is well-known, the cochlea represents a relatively enclosed microenvironment due to the presence of the blood–brain and blood-labyrinth barriers, which limits effective drug penetration into the inner ear. This speculation finds support in a recent study demonstrating that high-concentration systematic administration of quercetin (daily 100 mg/kg) could protect against noise-induced oxidative damage in the rat cochlea (Goodarzi et al. 2023). On the other hand, our results indicated that systematic administration of baicalin (daily 15–30 mg/kg) for 2 months delayed the decline in ABR wave I amplitude at low frequencies and rescued the loss of ribbon synapses in apical IHCs, although without significant effects on ABR thresholds shifts or hair cell loss in aging DBA/2 J mice. These results suggest that baicalin treatment may preserve low-frequency sound transmission by mitigating the loss of ribbon synapses between IHCs and auditory nerve fibers, as well as attenuating the decline in SGN firing in aging DBA/2 J mice. Previous studies have shown that baicalin and its aglycone baicalein possess multifaceted therapeutic properties, attributed to their abilities to scavenge reactive oxygen species and regulate various signaling pathways implicated in apoptosis, inflammation, and autophagy (Hu et al. 2022). Our study also elucidated the molecular interactions between baicalin and these signaling molecules. Notably, at the molecular level, baicalin treatment significantly downregulated the expression levels of several inflammatory-related factors (NLRP3, TNFα, IL1β, and IL6) in the cochlea of DBA/2 J mice, providing compelling evidence for its anti-inflammatory properties in attenuating age-related cochlear damage and hearing loss. Another noteworthy finding is the significant upregulation of ESR1 and the downregulation of AR in the cochlea of DBA/2 J mice following baicalin treatment. As discussed above, estrogen supplementation and enhancement of ESR expression, as well as pharmacological inhibition of AR, have been shown to protect against hearing loss. The result suggests that the bidirectionally regulation of ESR1 and AR expression may contribute to the otoprotective effects of baicalin. Considering the related results in this study, our findings suggest that baicalin may exert therapeutic effects in ARHL through its multi-targets and multi-pathways mechanisms, involving anti-inflammation, activation of KCNQ4 and BK potassium channels, and modulation of sex hormone-related pathways.

Conclusion and perspectives

In summary, this study has elucidated eleven shared bioactive compounds of three TCM formulas, which may confer therapeutic benefits in ARHL by modulating potential regulatory targets involved in several key signaling pathways. Notably, these targets/pathways include PI3K-Akt, MAPK, hormone-related signaling pathways, and potassium channels. These results offer valuable insights into the shared mechanisms underlying the efficacy of the three TCM formulas in mitigating ARHL. Moreover, baicalin emerges as a promising therapeutic candidate for ARHL due to its multi-targets and multi-pathways mechanisms, involving anti-inflammatory properties, activation of Kv7.4 and BK potassium channels, and modulation of sex hormone-related pathways. These findings may benefit further exploration in the quest for novel substances and therapeutic avenues for the treatment of ARHL. However, developing highly effective TCM drugs for ARHL treatment with minimal side effects requires great effort. Some areas for improvement include: identifying the active blood-enriching ingredients of TCM formulas, including metabolites, using liquid chromatography-mass spectrometry; applying quantitative transcriptome and proteome analyses to solidify and expand the disease-related target database; identifying the targets and pharmacological properties of individual chemical components to distinguish inhibitory effects from activated effects on targets; increasing the bioavailability of chemical components through chemical structure modification; determining dosage regimens and developing innovative delivery systems in light of specific circumstances. By addressing these aspects, future research will improve the efficacy and safety of TCM-based treatments for ARHL.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

ARHL

Age-related hearing loss

TCM

Traditional Chinese medicine

ELZC

Er-Long-Zuo-Ci

TQEL

Tong-Qiao-Er-Long

EL

Er-Long

TCMSP

Traditional Chinese Medicine Systems Pharmacology Database and Analysis platform

OMIM

Online Mendelian inheritance in man

OB

Oral bioavailability

DL

Drug similarity

PPI

Protein-protein interaction

GO

Gene ontology

KEGG

Kyoto encyclopedia of genes and genomes

ABRs

Auditory brainstem responses

IHCs

Inner hair cells

OHCs

Outer hair cells

SGNs

Spiral ganglion neurons

SV

Stria vascularis

CTBP2

Carboxyl-terminal-binding protein 2

Authors contributions

Y.Z.: conceptualization, data curation, funding acquisition, project administration, writing-original draft, writing-review&editing. W.Y.S.: investigation, methodology, software, visualization. Q.Z.: formal analysis, methodology, software, validation, visualization. H.W.G: investigation, methodology, software, visualization. Y.X.Y. : methodology, software. Z.Y.T: writing-review & editing. N.L: funding acquisition. H.J.W.: resource, supervision. Y.H.J: project administration, supervision, writing-review&editing. All authors reviewed the manuscript. The authors declare that all data were generated in-house and that no paper mill was used.

Funding

This research was supported by the Natural Science Foundation of Hebei Province (H2022201059, H2022201050), the Science and Technology Research Project of Universities in Hebei Province (QN2024198), and the Medical Science Foundation of Hebei University (2022A04).

Data availability

The original contributions presented in the study are included in the article/supplementary material. Further inquiries can be directed to the corresponding author.

Declarations

Competing interests

The authors declare no competing interests.