Glycyrrhizic acid and its carrier-free micellar formulation: Unraveling the potential for enhanced oral prevention of hearing loss

Abstract

Hearing loss, a global health concern, significantly impacts patients with delayed language development, impaired neurocognitive function, and severe social problems. The main cause is the cochlear hair cell damage induced by oxidative stress and inflammation from ototoxic drugs, noise exposure or diabetes. Glycyrrhizic acid (GA), derived from edible herb licorice, is widely utilized in traditional Chinese medicine and clinical treatments for liver diseases. However, its potential in preventing hearing loss remains largely unexplored. Herein, we propose GA as a novel otoprotective agent and demonstrate its capability to prevent hearing loss. Our results show that GA effectively reduces oxidative stress and inflammation induced by cisplatin, aminoglycosides, or even noise and diabetes, thereby protecting cochlear hair cells. In hearing loss models, two commonly used administration methods were compared, with tympanic injection providing better protective effects than oral administration of GA. To enhance oral bioavailability, GA is employed as both the medicine and excipient, and formulated into micelles with curcumin, another extensively used bioactive compound. Interestingly, formulation parameters such as feeding ratio and temperature have little impact on micelle size but significantly affect the drug loading efficiency. The carrier-free strategy can achieve a high drug loading capacity and significantly increase the drug concentration in blood, offering improved preventive efficacy. Notably, the micelles also exhibit protection on kidneys and liver, and do not compromise the antitumor activity of cisplatin. Therefore, GA holds promise as an otoprotective candidate, with potential clinical applications for oral prophylaxis of hearing loss using the micellar formulation.

Graphical abstract

Highlights

• •GA effectively reduced oxidative stress, apoptosis, and inflammation.
• •GA showed good protection via intratympanic injection but limited oral efficacy.
• •The carrier-free micellar strategy enabled high loading capacity of GA.
• •The micelles significantly improved bioavailability and oral otoprotection.

1. Introduction

Hearing loss (HL) has emerged as a critical global health concern, which significantly impacts patients with delayed language development, impaired neurocognitive function, and social isolation (Collaborators, 2021). Approximately 1.5 billion people worldwide experience varying degrees of HL, which will rise in the coming years (Wang et al., 2024). The main cause is the cochlear hair cell damage induced by ototoxic agents-particularly chemotherapeutic drugs like cisplatin (CDDP) and aminoglycosides, noise, and chronic diseases such as diabetes (Qiao et al., 2024). However, current treatment options are limited and often rely on cochlear implants or hearing aids (Wolf et al., 2022). Systemic corticosteroid therapy, while used to reduce inflammation and edema in sudden sensorineural HL, carries notable side effects and remains controversial regarding its effectiveness and safety (Plontke et al., 2022). Recently, FDA approved sodium thiosulfate as the first drug to treat CDDP-induced HL, yet this compound chelates CDDP after administration, compromising the antitumor efficacy (Meijer et al., 2024). These limitations underscore an urgent need to develop novel therapeutics for HL.

Oxidative stress and inflammation in the inner ear have been considered as the mechanisms of HL (Paciello et al., 2024). CDDP, a first-line antitumor drug, can induce irreversible damage to inner ear by reactive oxygen species (ROS) and inflammation, and promote apoptosis to auditory hair cells (Chen et al., 2024). Similarly, aminoglycoside antibiotics such as gentamicin (GM), which is applied in bacterial infection treatments, may also lead to permanent hair cell damage through the overproduction of ROS and the activation of inflammation (Kim et al., 2023). Noise exposure and diabetes, also cause HL in similar ways. Diabetes-related HL has been attributed to vascular damage, including thickening of the stria vascularis basement membrane and subsequent ROS overproduction. Meanwhile, acoustic trauma induces excessive ROS generation through mechanical injury and excitotoxic stimulation of auditory neurons, both of which contribute to irreversible cochlear damage (Natarajan et al., 2023; Samocha-Bonet et al., 2021). Many compounds have been shown to reduce oxidative stress, alleviate inflammation, and prevent hair cell death. For instance, N-acetyl-L-cysteine (NAC), with the ability to produce antioxidant glutathione, can reduce the oxidative stress in noise- or CDDP-induced HL (Martinez-Banaclocha, 2022). The antioxidant ebselen, which can activate glutathione peroxidase, demonstrates preventing effects in noise-induced HL (Kil et al., 2022). These reports highlight the importance of addressing oxidative stress and inflammation in HL diseases. However, despite their anti-ototoxic effects, these chemicals may reduce the antitumor efficacy of CDDP and suffer from problems such as low oral bioavailability and high toxicity, limiting their further clinical applications.

Edible herbs have gained great interest due to their potent bioactivities (Liu et al., 2024). Licorice, recognized for both medicinal and edible properties, is listed as a top-grade herb in Shennong's Herbal Classics. Known for its diverse therapeutic effects, licorice possesses antibacterial, antiviral, antitumor and anti-inflammatory activities. It helps regulate immune function, protects the liver, combats arrhythmias, reduces blood lipids, prevents atherosclerosis, relieves coughs, and detoxifies the body. The 2020 edition of the Pharmacopoeia of the People's Republic of China also states that licorice can alleviate the toxicity and potency of drugs. Glycyrrhizic acid (GA), one of the most important active ingredients extracted from licorice, has demonstrated excellent antioxidant and anti-inflammatory effects. It is clinically applied in liver diseases including chronic hepatitis, liver fibrosis, and cirrhosis (Lee et al., 2024). Besides, GA shows therapeutic effects in other inflammation-related conditions such as dermatological diseases and colitis (Carbonell-Barrachina et al., 2003; Wu et al., 2022). Moreover, as a widely used medicine-food homology ingredient in traditional Chinese medicine (TCM), it demonstrates good safety. The US Food and Drug Administration (FDA) has also classified GA as a “generally recognized as safe” compound (Carbonell-Barrachina et al., 2003). Therefore, GA holds great promise to combat oxidative stress and inflammation for HL prevention. The employment of GA in HL treatment, especially ototoxicity, not only provides a new prevention strategy, but also expands the application of medicine-food homology herbs in modern medicine. Curcumin (Cur), a natural product from the rhizome of Curcuma longa, has attracted increasing attention due to its broad-spectrum biological activities, including anti-inflammatory, antioxidant, and anticancer effects (Wang et al., 2023; Zia et al., 2021). For instance, Cur is a potent antioxidant that can effectively scavenge free radicals, preventing oxidative damage to cells and protecting against diseases related to oxidative stress (He et al., 2015). These pharmacological properties make it a promising candidate for the treatment of ROS/inflammation-related diseases, such as hearing loss.

Herein, we employed GA as a novel otoprotective agent and evaluated the protective effects against drug (CDDP or GM)-induced ototoxicity, as well as HL caused by diabetes and noise. The involved efficacy in reducing oxidative stress, apoptosis, and inflammation was studied. To achieve efficient oral administration, a formulation strategy with improved bioavailability is needed. Notably, since GA is amphiphilic, we employed the concept of unification of medicines and excipients and innovatively formulated GA into micelles without adding additional carrier materials. Cur was introduced as another bioactive and promising component of the micelles to stabilize the formulation. The oral efficacy of the micelles towards the guinea pig HL models was evaluated. Finally, the bioavailability and biocompatibility were investigated.

2. Materials and methods

2.1. Materials

Glycyrrhizic acid ammonium salt (the ammonium salt of GA, >98 %) was purchased from Yuanye Bio-Technology Co., Ltd. (Shanghai, China). Curcumin (Cur, >98 %) was obtained from Macklin Biochemical Co., Ltd. (Shanghai, China). Cisplatin (CDDP, >99.5 %) and N-acetyl-L-cysteine (NAC, >99 %) were purchased from Aladdin (Shanghai, China). Acridine orange (AO) was purchased from Macklin Biochemical Co., Ltd. (Shanghai, China). Ifenprodil tartrate (IFD, >99.5 %) was purchased from Bide Pharmaceutical Co., Ltd. (Shanghai, China). Gentamicin sulfate (the sulfate of GM, ≥590 μg/mg) was supplied by Qiyun Biotechnology Co., Ltd. (Guangzhou, China). 2-[4-(Dimethylamino)styryl]-1-ethylpyridinium iodide (DASPEI, >99 %) and 3-aminobenzoic acid ethyl ester methanesulfonate (MS-222, >98 %) were purchased from Sigma-Aldrich (Saint Louis, USA). 2-(4-Amidinophenyl)-6-indolecarbamidine dihydrochloride (DAPI) and dichlorodihydrofluorescein diacetate (DCFH-DA) were provided by Beyotime Biotechnology (Shanghai, China). ELISA kits for TNF-α, IL-6, and IL-1β detecting were purchased from Jiancheng Bioengineering Institute (Nanjing, China). Acetonitrile and methanol (HPLC grade) were obtained from Aladdin Biochemical Technology Co., Ltd. (Shanghai, China). Other reagents were of analytical grade unless noted.

2.2. Zebrafish

Wild-type AB strain, Tg(ins:eGFP) and Tg(lyz:eGFP) transgenic zebrafish were purchased from China Zebrafish Resource Center (CZRC). Tg(Brn3c:mGFP) transgenic zebrafish were kindly provided by Professor Jing Lili from Shanghai Jiao Tong University.

All zebrafish were bred and maintained in our laboratory using a zebrafish housing system (Yishulihua Biotechnology Co., Ltd., Nanjing, China). The larvae used in this study were obtained from in-house natural mating of healthy adult zebrafish (male: female = 2:1) of the same strains. Fertilized embryos were incubated in embryo culture medium (1 mM MgSO₄, 120 μM KH₂PO₄, 74 μM Na₂HPO₄, 1 mM CaCl₂, 500 μM KCl, 15 μM NaCl, and 500 μM NaHCO₃) at 28.5 °C and cultured at a density of 250 fish per 500 mm² until 5 days post fertilization (5 dpf). Embryos were cleaned daily and provided with the new medium. Due to their early developmental stage, the sex of the larvae could not be determined.

2.3. Protective effect of GA on hair cell damage in zebrafish models

AB strain zebrafish larvae (5 dpf) were cultured in 48-well plates, with 5 larvae per well. The zebrafish were incubated with different concentrations of GA (1, 10, 50 μM) for 1 h, followed by modeling with GM (80 μM) for 1 h or CDDP (750 μM) for 4 h. The larvae were then stained with DASPEI (0.008 %) for 15 min. Finally, the zebrafish were washed three times with embryo culture medium, and anesthetized in MS-222 (0.01 %). The treated zebrafish were placed on slides and imaged using a fluorescence microscope. The average fluorescent area of the neuromasts was quantified using ImageProPlus6.0.

2.4. Quantitative RT-PCR

The total RNA of zebrafish larvae was extracted using RNA extraction kits (Servicebio). First-strand cDNA was synthesized using 10 μL of total RNA and detected by quantitative RT-PCR. Primer sequences were shown in Table S1. Amplification conditions were as follows: 95 °C for 30 s, followed by 40 cycles of 95 °C for 15 s and 60 °C for 30 s. Data were processed using the ΔΔCT method.

2.5. Antioxidant activity of GA in zebrafish models

Following the drug administration and modeling procedures described in Section 2.3, zebrafish larvae were treated, and the solutions were discarded. The larvae were then washed three times with embryo culture medium and stained with DCFH-DA (20 μM) for 1 h in the dark. After staining, the larvae were washed with embryo culture medium. The zebrafish larvae were placed on slides, and imaged using a fluorescence microscope. The average fluorescent area of the neuromasts was quantified using ImageProPlus6.0.

2.6. Anti-inflammatory activity of GA in zebrafish models

Tg(lyz:eGFP) transgenic zebrafish larvae were incubated with GA and modeled according to the procedures described in Section 2.3. After washing the larvae with embryo culture medium, the larvae were homogenized using a handheld electric homogenizer (MNT-991305, DEGUQMNT, Shanghai, China) on ice for 5 min. The mixture was then centrifuged at 3000 rpm for 15 min and the supernatant was collected. The concentrations of pro-inflammatory cytokines including TNF-α, IL-6, and IL-1β were measured using ELISA kits.

2.7. Animal

Seven-week-old male albino guinea pigs (250–300 g) were purchased from Suibei Experimental Animal Breeding Center in Baiyun District, Guangzhou. The selection of animal sex was consistent with our previous studies conducted in this laboratory(He et al., 2024). All animals were housed under standard laboratory conditions: temperature 22–24 °C, relative humidity 40–60 %, and a 12-hour light/dark cycle. Guinea pigs had free access to standard chow and water throughout the study. Prior to experimentation, all animals were acclimated for at least one week to the laboratory environment. All experimental protocols involving animals were approved by the Animal Ethics Committee of Guangdong Pharmaceutical University (Ethical approval number: CVF2017105), and were conducted in accordance with the ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines and the Guidelines for the Care and Use of Laboratory Animals.

2.8. Protective effect of GA in guinea pig models

Commonly used administration methods including intratympanic injections (i.t.) and oral administration (p.o.) were applied to study the effect of GA in drug-induced HL models. Guinea pigs were anesthetized with pentobarbital sodium (30 mg/kg), and auditory brainstem response (ABR) thresholds were measured at 4 kHz, 8 kHz, and click frequencies.

For GM model: Guinea pigs were randomly divided into four groups based on body weight (n = 3 per group): control group, GM group, GA (i.t.) group and GA (p.o.) group. For 10 consecutive days, animals of GA (p.o.) group were orally administered with GA (40 mg/kg) twice a day, GA (i.t.) group received intratympanic injections of GA (500 μg/kg) on day 0, 3, and 6, GM group received intraperitoneal injections of GM (150 mg/kg) once daily.

For CDDP model: Guinea pigs were randomly divided into four groups based on body weight (n = 3 per group): control group, CDDP group, GA (i.t.) group and GA (p.o.) group. For two consecutive days, animals were administered with GA (40 mg/kg) orally twice a day. On the third day, CDDP (12 mg/kg) was administered via intraperitoneal injection, followed by oral administration of the same dose of GA for three additional days. Animals of CDDP group were directly administered with CDDP (12 mg/kg) via intraperitoneal injection on the third day, GA (i.t.) group received an intratympanic injection of GA (500 μg/kg) both one day and 1 h before CDDP administration.

2.9. Preparation and characterization of GA micelles

GA micelles were prepared using a thin-film dispersion method. The total feeding mass, feeding ratio, temperature, solvent volume, and hydration time were designed as single-factor variables, while other factors were kept constant. The total feeding mass was set at 30, 60, 120 or 180 mg; the feeding ratio (GA:Cur) was 1:1, 3:1, 5:1 or 7:1; the temperature was 40, 50, 60 or 70 °C; the solvent volume was 10, 20, 30 or 40 mL; and the hydration time was 6, 12, 18 or 24 h. Three parallel experiments were conducted for each group, with drug loading efficiency (DLE) and particle size as the evaluation indicators. Briefly, GA and Cur were dissolved in ethanol at the specified mass ratio. The organic solvent was then removed under vacuum at the set temperature to obtain a thin film. Subsequently, a fixed volume of distilled water was added, and the mixture was sonicated for a predetermined duration. After centrifugation, the supernatant was filtered through a polycarbonate membrane with a pore size of 200 nm. Finally, the mixture was lyophilized to obtain GA micelles. For the optimized preparation, the feeding mass, feeding ratio, temperature and hydration time were determined at 60 mg, 5:1, 40 °C and 12 h, respectively.

The DLE was measured using an HPLC system (LC-20A, Shimadzu, Japan) equipped with a C₁₈ reverse-phase column (250 × 4.6 mm, 5 μm) at a temperature of 28 °C. A mobile phase consisting of 0.1 % formic acid in a water-methanol mixture (25:75, v/v) was used at a flow rate of 1 mL/min. The detection wavelength for GA and Cur was set at 245 and 425 nm, respectively. The calculation of drug loading efficiency was carried out as follows:

$$\mathit{Drug\; loading\; efficiency}\left(\%\right)=\frac{\mathit{encapsulated\; drug}}{\mathit{total\; drug\; introduced}}\times 100$$

A series of characterizations were performed on GA micelles. Full wavelength ultraviolet-visible (UV–vis) absorption spectra were recorded on a spectrophotometer (G6860A, Agilent Technologies, USA). Fluorescence emission spectra were measured from 450 to 750 nm (with a 5 nm slit width) under excitation at 420 nm using a fluorescence spectrophotometer (RF-5031pc, Shimadzu, Japan). Fourier-transform infrared (FT-IR) spectra were recorded in the range of 400–4000 cm⁻¹ on a spectrometer (Spectrum100, Shimadzu, Japan). The critical micelle concentration (CMC) of GA micelles was determined by fluorescence spectroscopy with Nile red (Sigma-Aldrich, USA), a hydrophobic fluorescent probe. The fluorescence intensity was monitored at an emission wavelength of 635 nm with an excitation of 530 nm, and different GA micelle concentrations were used to establish the characteristic inflection point corresponding to CMC. The stability of GA micelles was evaluated by monitoring the particle size and zeta potential of the samples during storage at 4 °C for 14 days. The method for determining the DLE was performed as previously described.

2.10. Protective effect of GA micelles in guinea pig models

Guinea pigs were randomly divided into five groups based on body weight (n = 3 per group): control group, CDDP/GM group, Cur group, GA group and GA micelles group. Animals were orally administered with Cur (8 mg/kg), GA (32 mg/kg) or GA micelles (40 mg/kg) according to the procedure described of CDDP/GM model in 2.8.

2.11. Data analysis

All experimental data were presented as means ± SD. The statistical analysis was performed using Graphpad Prism 8 software and the statistical significance was determined by one-way ANOVA. Results were considered statistically significant when *P < 0.05, **P < 0.01, ***P < 0.001.

3. Results

3.1. Protective effect of GA against hair cell damage in zebrafish

To construct hair cell damage models in zebrafish, two representative ototoxic drugs including GM and CDDP were employed. Drug concentrations and treatment protocols were chosen accordingly (Chen et al., 2022; Xiao et al., 2023). Cell nuclei and mitochondria in the four neuromasts (SO1, SO2, O1, and OC1) of the zebrafish were labeled using DAPI and DASPEI, respectively (Fig. S1A). As shown in Fig. S1B–E, the fluorescence of DAPI and DASPEI in zebrafish hair cells treated with 80, 160, and 320 μM GM decreased significantly, confirming the successful modeling of GM-induced hair cell damage. Similarly, zebrafish treated with CDDP exhibited dose-dependent hair cell damage (Fig. S1F–I). To establish the noise-induced damage model, zebrafish were exposed to noise with different frequencies, following the protocols described in previous studies(Han et al., 2022; Lara et al., 2022). Considering that underwater experiments present technical challenges in maintaining stable and measurable sound pressure levels (SPLs) due to factors such as tank resonance and particle motion, we chose frequency as a more controllable and biologically relevant variable to induce consistent auditory damage (Han et al., 2022). As shown in Fig. S2, the fluorescence area in the hair cells decreased after noise exposure, with a frequency-dependent manner of cell damage. Furthermore, diabetes, which is often characterized by pancreatic β-cells damage, insulin deficiency and hyperglycemia. Alloxan (AX) selectively destroys pancreatic β-cells and induces oxidative stress, while sucrose (Suc) provides a high-glucose environment that exacerbates oxidative stress and metabolic disturbances, resulting in diabetes, which subsequently leads to hair cell damage. In this study, we developed a zebrafish model of diabetes-induced hair cell injury by modifying previously published methods (Benchoula et al., 2019; Liang et al., 2024). As shown in Fig. S3, treatment with 1 mM AX and 1 % Suc resulted in a reduction in the DASPEI fluorescence by approximately 50 %, indicating the successful modeling of cell damage.

With these established hair cell damage models, the protective effect of GA was investigated (Fig. 1A). In GM- and CDDP-induced cell damage models, IFD and NAC were used as positive controls, respectively. The mitochondria were stained using DASPEI. As shown in Fig. 1B–C, the fluorescent areas of all GA-treated groups were larger than those of the GM group, among which the fluorescence of 10 μM GA-treated group was the highest. The preserved fluorescence of the mitochondria suggested that GA possessed a good protective effect on hair cell damage induced by GM. To further confirm this result, we performed dependent experiments with DAPI staining of nuclei and DASPEI staining of mitochondria. As shown in Fig. S4, the fluorescent areas of DAPI and DASPEI staining were larger than those of the GM group, meaning that GA effectively protected the hair cells. Besides, different concentrations of GA resulted in varying protective effects on CDDP-induced hair cell damage model, with 10 μM GA exhibiting the most significant protection (Fig. 1D–E). Nuclei and mitochondria staining dependence experiments were also conducted. As shown in Fig. S5, fluorescence in both nuclei and mitochondria staining increased after GA treatment compared to the CDDP group, indicating that GA could protect the hair cells from CDDP-induced damage. In addition, the protective effect of GA on noise- and diabetes-induced hair cell damage was evaluated. The DAPI area of the 25 μM GA-treated group was significantly larger, confirming that GA was able to reduce hair cells from noise damage (Fig. S6). In the diabetes-induced hair cell damage model, DAPI fluorescence was higher after treatment with 25 μM GA, suggesting that GA also prevented diabetes-induced hair cell damage (Fig. S7). Hyperglycemia can impair cochlear blood flow and promote ROS generation, leading to hair cell damage. The protective effect of GA may be related to its intrinsic antioxidant and anti-inflammatory properties.

Fig. 1.

Protective effect of GA against drug-induced hair cell damage in zebrafish. (A) Chemical structure of GA, and schematic representation of GA administration. (B) Fluorescent images of zebrafish hair cells following GA treatment in the GM-induced cell damage model. Green areas represented DASPEI staining of cell mitochondria. (C) Quantification of the mean fluorescence area of zebrafish hair cells shown in panel B. (D) Fluorescent images of zebrafish hair cells following GA treatment in the CDDP-induced cell damage model. (E) Quantification of the mean fluorescence area of zebrafish hair cells shown in panel D. Data were presented as mean ± SD, n = 5. **P < 0.01, ***P < 0.001 vs. GM or CDDP group. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

3.2. Antioxidant and anti-inflammatory activities of GA against GM- and CDDP-induced cell damage models in zebrafish

Since the anti-ototoxic effect of GA was associated with the generation of oxidative stress, the antioxidant activity of GA was studied (Fig. 2A). ROS production was indicated with DCFH-DA probe. As shown in Fig. 2B–C, the fluorescence intensity of the GA group decreased significantly, demonstrating that GA effectively inhibited ROS production. Similarly, in the CDDP-induced model, GA showed antioxidant effects on hair cell (Fig. 2D–E). We also measured the levels of gene expression associated with oxidative stress. As shown in Fig. 2F–G, GM treatment could upregulate the expression of CAT and SOD1 mRNA, but this negative effect was inhibited by GA. Similarly, GA treatment was able to reduce the expression of CAT and SOD1 mRNA in the CDDP-induced model, confirming the antioxidant activity of GA (Fig. 2H–I). The cell apoptosis was indicated with AO, and the neuromasts of zebrafish hair cells were located with DAPI. As shown in Fig. S8A, no significant AO signals were observed in the control group. In contrast, AO signals in the GM group suggested that GM induced apoptosis in zebrafish hair cells. However, the apoptosis could be reduced significantly by GA treatment. Similar results were observed in the CDDP model (Fig. S8B). Furthermore, the quantitative fluorescence in Fig. S8C–F indicated that GA could increase cell viability in both GM- and CDDP-treated models. The expression of the apoptosis-related gene CASP3 was also detected. As shown in Fig. S9, GA treatment downregulated the mRNA expression of CASP3, demonstrating that GA was able to reduce cell apoptosis.

Fig. 2.

The antioxidant and anti-inflammatory activity of GA in zebrafish. (A) Schematic illustration of the antioxidant and anti-apoptotic effects of GA. (B) Fluorescence images of zebrafish with GA treatment in the GM-induced hair cell damage model, stained with DCFH-DA (green). Scale bar: 250 μm. (C) Quantification of oxidative stress levels from panel B. (D) Fluorescence images of zebrafish with GA treatment in the CDDP-induced hair cell damage model. (E) Quantification of oxidative stress levels from panel D. (F–G) mRNA expression of CAT (F) or SOD1 (G) following GA treatment in the GM-induced hair cell damage model. (H–I) mRNA expression of CAT (F) or SOD1 (G) following GA treatment in the CDDP-induced hair cell damage model. (J) The accumulative neutrophils around the hair cells treated with CDDP. (K–M) Concentrations of pro-inflammatory cytokines, including TNF-α (K), IL-1β (L), and IL-6 (M) in the GM-induced hair cell damage model. (N) The accumulative neutrophils around the hair cells treated with GM. (O–Q) Concentrations of pro-inflammatory cytokines, including TNF-α (O), IL-1β (P), and IL-6 (Q) in the CDDP-induced hair cell damage model. Data were presented as mean ± SD, n = 5 for panels C, E, J, N, n = 3 for panels F–I, K–M, O–Q. *P < 0.05, **P < 0.01, ***P < 0.001 vs. GM or CDDP group. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

When damage occurs, neutrophils accumulate and release pro-inflammatory cytokines and chemokines, resulting in inflammation (Pei et al., 2016). Therefore, we investigated the anti-inflammatory activity of GA by evaluating the neutrophils accumulation and measuring the concentrations of pro-inflammatory cytokines. The SO1, SO2, O1, and OC1 neuromasts were located using DASPEI. As indicated by the strong GFP fluorescence in Fig. S10A, the neutrophil recruitment was significantly inhibited after GA treatment. A similar effect was observed in the CDDP model (Fig. S10B), where GA treatment reduced the accumulation of neutrophils at the site of hair cell damage. These results were further confirmed by the quantitative analysis of neutrophil (Fig. 2J and N). Additionally, the measurements of the pro-inflammatory cytokines in Fig. 2K–M and O–Q showed that GA treatment could reduce the concentrations of TNF-α, IL-1β, and IL-6 in both GM- and CDDP-induced damage models. These results suggested that GA could alleviate the inflammatory response induced by GM or CDDP, and thus provided protective effects on hair cells. Since both STAT3 and AKT1 are known to play important roles in inflammation and oxidative stress regulation (Butturini et al., 2020), the mRNA levels of STAT3 and AKT1 were evaluated in GM and CDDP models. As shown in RT-PCR results of Fig. S11, both the GM and CDDP drugs increased the mRNA levels of STAT3 and AKT1. However, pretreatment with GA significantly reduced the expression levels of STAT3 and AKT1, confirming that GA can regulate these proteins to counteract the ototoxic effects of GM and CDDP.

3.3. Protective effect of GA against CDDP/GM-induced HL in guinea pigs

GA was administered through intratympanic injection (i.t.) or oral administration (p.o.) as depicted in Fig. 3A–B. During the treatment process, both ABR thresholds and the integrity of outer hair cells (OHCs) were measured. As shown in Fig. 3C and Fig. S12A–D, the hearing threshold increased by approximately 70 dB in the GM group but only 25 dB in the GA group among all frequencies after intratympanic injection, suggesting effective prevention of HL. However, when GA was administered orally, the threshold shifts remained almost the same as those in the GM group, meaning that oral administration of GA didn't achieve the expected effect. In the CDDP-induced HL model, the results were similar to those observed in the GM model. Threshold shifts were significantly reduced in the GA (i.t.) group but remained almost unchanged in the GA (p.o.) group (Fig. 3D and Fig. S12 E–H). Fluorescent images showed that the OHCs in all cochlear turns of the GA (i.t.) group remained nearly intact, whereas an obvious loss of cells was observed in the GA (p.o.) group, indicating the good otoprotection by intratympanic GA administration (Fig. 3E–F). This was supported by the survival rates in Fig. 3G–H, where the GA (i.t.)-treated groups possessed approximately 95 % cell survival in both HL models.

Fig. 3.

Protective effect of GA against drug-induced HL in guinea pigs. Experimental procedures for GA treatment in the GM (A) and CDDP (B)-induced HL models. (C–D) Auditory brainstem response (ABR) threshold shift recorded after the administration of GA in the GM (C) and CDDP (D)-induced HL models. Qd, quaque die. GA was administrated in two ways including intratympanic injection (i.t.) and bis in die (bid) oral administration (p.o.). (E–F) Representative fluorescent images of outer hair cells (OHCs) with FITC-phalloidin staining in the GM (E) and CDDP (F)-induced HL models after GA treatment. Scale bar: 25 μm. (G–H) The survival rate of OHCs upon GA treatment in the GM (G) and CDDP (H)-induced HL models. Data were presented as mean ± SD, n = 3. *P < 0.05, **P < 0.01, ***P < 0.001 vs. GM or CDDP group.

3.4. Preparation and characterization of GA micelles

To overcome the low oral efficacy of GA, a micellar formulation strategy was proposed. As shown in Fig. 4A, GA was combined with Cur, a common ingredient in TCM, to form GA micelles using a thin-film dispersion method. The ability to form micelles was first confirmed by the CMC measurements. Fig. 4B revealed a CMC value of 0.387 mg/mL, indicating the formation of micelles. Key assembly parameters, including feeding mass, feed ratio (GA to Cur), temperature, solvent volume and hydration time, were then evaluated. Fig. 4C showed that the DLE of GA reached its maximum at 60 mg, while the DLE of Cur significantly decreased as the total mass increased. The DLE of Cur was also influenced by the feeding ratio of GA to Cur. As shown in Fig. 4D, the DLE of Cur increased dramatically with the ratio and reached the maximum at approximately 5:1, whereas the DLE of GA was maintained at a high level, suggesting the saturation of the encapsulation capacity. Additionally, the increase in temperature led to a decrease in DLE of both GA and Cur, likely due to the instability of the micelle structure at high temperature (Fig. 4E). In contrast, the volume of the solvent had minimal impact on the efficiency (Fig. 4F). However, prolonged hydration time slightly reduced DLE (Fig. 4G). Interestingly, these parameters had little impact on particle size, which only increased when the feeding mass exceeded 120 mg, indicating the well-control of the size during the assembly. Consequently, the optimized feeding mass, feeding ratio, temperature and hydration time could be determined at 60 mg, 5:1, 40 °C and 12 h, respectively. These conditions provided a foundation for the preparation of GA micelles with high DLE, controlled size and expected protective effect.

Fig. 4.

Preparation of GA micelles. (A) Schematic illustration of the preparation of GA micelles via a self-assembly technique. (B) Critical micelle concentration (CMC) of GA. Effects of total drug feeding mass (C), feeding ratio (D), temperature (E), hydration solvent volume (F), and hydration time (G) on the drug loading efficiency (DLE) and particle size of micelles. Data were presented as mean ± SD, n = 3.

With optimal conditions for the formulation described above, the GA micelles were obtained and characterized. DLS measurements in Fig. 5A revealed that the particle size of the obtained GA micelles was 70.11 ± 1.72 nm, with a zeta potential of −38.19 ± 3.50 mV showed in Fig. S13. TEM observations confirmed the spherical morphology of the micelles (Fig. 5B). Further characterization using FT-IR (Fig. S14) and UV–vis (Fig. 5C) implied the presence of GA and Cur. Fluorescence spectroscopy with the fluorescent peak at about 530 nm suggested the reserved fluorescent property of Cur in the micellar structure (Fig. 5D). Stability of GA micelles was analyzed by observing the changes in particle size and zeta potential. As shown in Fig. 5E and Fig. S15, the particle size and zeta potential remained almost unchanged during storage. Additionally, the drug loading content (DLC) of GA and Cur was 83.4 ± 0.5 % and 16.6 ± 0.4 %, respectively, suggesting a high drug loading capability of the formulation (Fig. 5F). These results demonstrated that the GA micelles were successfully prepared.

Fig. 5.

Characterization of GA micelles. (A) Hydrodynamic size distribution of GA micelles. (C) TEM image of GA micelles. Scar bar: 100 nm. (C) UV–vis spectrum of Cur, GA and GA micelles. (D) Fluorescence spectrum of Cur and GA micelles. (E) Particle size and polydispersity (PDI) changes of GA micelles during 14 days storage at 4 °C. (F) Drug loading content of GA micelles. Data were presented as mean ± SD, n = 3.

3.5. Molecular dynamics simulation of self-assembly

To study the self-assembly process of the GA and Cur molecules, molecular dynamics simulation was conducted. As observed in Fig. 6A, the structures of GA and Cur underwent continuous changes during the initial 20 ns. After 20 ns, the molecules gradually aggregated into clusters, which increased in size over time. By 60 ns, these structures stabilized into nanoclusters, indicating the completion of the self-assembly process. The final states of the structures were presented in Fig. 6B, revealing intermolecular hydrogen bonds and π-π stacking interactions. An average of 2.5 hydrogen bonds appeared during the self-assembly process, indicating the critical role of hydrogen bonds in forming micellar structures (Fig. 6C). Key parameters, including solvent accessible surface area (SASA), average root means square deviation (RMSD), and radius of gyration (Rg), were also detected. As shown in Fig. 6D, the SASA value was high at the beginning of the simulation, then decreased significantly and stabilized after 60 ns. This suggested that the interaction between Cur and GA molecules reduced their solvent exposure, leading to formation of stable structures. The increase and stabilization stages of RMSD indicated that the molecular structures deviated from the initial conformations and finally reached stable states after conformational rearrangement (Fig. 6E). The Rg values in Fig. 6F showed a gradual increase and stabilization over time, further demonstrating the formation of larger clusters. These results confirmed the molecular self-assembly ability to form micelles.

Fig. 6.

Molecular dynamics simulation of self-assembly. (A) Structural changes of GA and Cur molecules during 100 ns. (B) Molecular interaction of GA and Cur. Number of hydrogen bonds (C), solvent accessible surface area (D), average root means square deviation (E), and radius of gyration (F), and of the assemblies during the simulation time.

3.6. Enhanced oral bioavailability by GA micelles against CDDP/GM-induced HL in guinea pigs

The efficacy of the GA micelles via oral administration was investigated using the CDDP-induced guinea pig model (Fig. 7A). The results showed that after oral administration, the ABR threshold shift in the GA micelles group decreased significantly compared to the model group (Fig. 7B and Fig. S18A–E). In contrast, oral administration of Cur or GA free drugs did not show any protective effect. The integrity analysis of hair cells further confirmed the enhanced otoprotective ability (Fig. 7C and Fig. S16). We also conducted experiments in the GM-induced guinea pig model, and the efficacy was similar to that observed in the CDDP model (Figs. S17 and S18F–J). Oral administration of GA micelles provided more pronounced otoprotection compared to both the GM group and the groups treated with free drugs. Besides, as calculated in Table S2, the drug combination index was lower than 1, indicating that GA and Cur exhibited synergistic effects in the micellar formulation. Notably, the viability of HepG2 cells in the micelles/CDDP-treated groups was comparable to that of the CDDP group, indicating that the micelles did not influence the antitumor efficacy of CDDP (Fig. S19).

Fig. 7.

The protective effect of GA micelles against CDDP-induced HL in guinea pigs. (A) Experimental procedures for the treatment using GA micelles in CDDP-induced HL models. (B) ABR threshold shifts in guinea pigs after GA micelles administration. (C) Representative fluorescent images of hair cells with FITC-phalloidin staining after GA micelles administration. Scar bar: 25 μm. (D) Cumulative permeability of GA, Cur after administration of GA, Cur or GA micelles in the PAMPA model. (E) The drug concentration profiles in blood following oral administration of GA, Cur or GA micelles in guinea pigs. Data were presented as mean ± SD, n = 3. *P < 0.05, **P < 0.01, ***P < 0.001 vs. CDDP group.

The absorption and pharmacokinetics of GA-loaded micelles were investigated to assess whether the micellar formulation strategy enhanced the bioavailability of GA. The parallel artificial membrane permeability assay (PAMPA) model, using lecithin-coated polycarbonate membranes, was employed to evaluate drug diffusion (Guedes et al., 2022). The results demonstrated that the permeability of GA or Cur in the form of micelles was significantly increased compared to the free drugs (Fig. 7D). Furthermore, pharmacokinetic data indicated that the plasma concentrations of both GA and Cur were significantly increased in the GA micelles group compared to their respective free drug groups. As shown in Fig. 7E and Table 1, the Cmax values of GA and Cur in the micelle group reached 104.89 ± 10.46 ng/mL and 12.56 ± 10.88 ng/mL, respectively, representing 2.4-fold and 5.3-fold increases compared to the free drug groups. Similarly, the AUC_0–12h values were elevated to 560.3 ± 48.49 ng·mL⁻¹·h for GA and 66.32 ± 22.78 ng·mL⁻¹·h for Cur, indicating sustained systemic exposure.

Table 1.

The pharmacokinetic parameters after oral administration of free drugs or GA micelles in guinea pigs.

	GA	Cur	GA micelles
			GA	Cur
Cmax (ng/mL)	43.78 ± 12.75	2.38 ± 0.80	104.89 ± 10.46	12.56 ± 10.88
AUC0-12h (ng/mL·h)	301.8 ± 67.44	3.19 ± 1.63	560.3 ± 48.49	66.32 ± 22.78

3.7. Biocompatibility analysis

As shown in Fig. 8A, no significant hemoglobin release was observed from RBCs even at a high concentration of 400 μg/mL, indicating excellent biocompatibility of the micelles. H&E staining demonstrated that the structural and morphological integrity of both cochlear tissues and major organs remained intact without obvious signs of damage or inflammation (Fig. 8B and Fig. S20). Notably, damages to liver and kidneys were observed in the CDDP group but could be recovered after the treatment of GA micelles, suggesting strong protective effects on the liver and kidneys (Fig. 8C). Additionally, measurements of changes in body weight and serum biochemical index confirmed that GA micelles helped the recovery of body weight and improved the functions of liver and kidneys (Fig. 8D–G). Blood routine analysis also showed negligible systematic toxicity after the administration of GA micelles (Fig. S21). Overall, GA micelles did not cause obvious toxicity or organ damage, demonstrating good biocompatibility and a remarkable protective effect on CDDP-damaged liver and kidneys.

Fig. 8.

The biocompatibility of GA micelles. (A) Hemolysis rate of erythrocyte treated with different concentrations of GA micelles. (B) Hematoxylin-eosin (H&E) staining of stria vascularis (left) and spiral ganglion (right) in the inner ear of guinea pigs after GA micelles treatment. Scar bar: 100 μm. (C) H&E staining of liver and kidneys in CDDP-induced HL models with or without GA micelles treatment. Scar bar: 100 μm. (D) The body weight of guinea pigs before and after GA micelles treatment. (E–G) The serum biochemical index of aspartate aminotransferase (AST, E), urea nitrogen (BUN, F) and creatinine (Scr, G) before and after GA micelles treatment. Data were presented as mean ± SD, n = 3. *P < 0.05, **P < 0.01, ***P < 0.001 vs. CDDP group.

4. Discussion

In this study, we applied the concept of “medicine-food homology”, and demonstrated that GA significantly prevented ototoxicity of HL induced by ototoxic drugs, noise exposure, and diabetes. We successfully established zebrafish models of hair cell damage induced by GM, CDDP, diabetes, and noise exposure, following previously reported protocols with slight modifications. Remarkably, GA exhibited significant protective effects across all the tested damage models, with an optimal concentration showing the most pronounced effect. In the diabetes-induced HL model, hyperglycemia-related metabolic disturbances led to hair cell damage, while GA treatment effectively alleviated these pathological changes, likely through its intrinsic antioxidant and anti-inflammatory activities. Unlike the proportional relationship observed in hair cell damage models, GA's protective effect followed a non-linear pattern, which might be due to the complex interaction between GA and the underlying molecular mechanisms (Guo et al., 2024; Ma et al., 2024). We speculated that a threshold concentration of GA might be required to initiate protective effects, beyond which additional increases would not enhance the effect. Besides, the role of GA could involve multiple pathways that converged at certain dose thresholds. Such phenomena were observed in all models, likely due to the involvement of ROS generation and inflammation-related mechanisms in hair cell damage. Importantly, the protective effect of GA in GM and CDDP models was comparable to that of positive control treatments, demonstrating its broad-spectrum and potent protective properties.

The antioxidant and anti-inflammatory activities of GA were validated through experiments in zebrafish models. Different models have varying degrees of severity, and representative GM and CDDP-induced ototoxicity models were chosen for further investigation. Both GM and CDDP can increase the damage of mitochondria, leading to ROS overproduction. Our results showed that ROS levels, indicated by DCFH-DA and the expression of oxidative stress markers SOD1 and CAT, decreased significantly after GA administration. This indicated that GA could effectively reduce intracellular ROS. DAPI/AO staining showed that cell apoptosis was inhibited in GA-treated zebrafish compared to the untreated control group. The downregulation of CASP3 mRNA expression further supported the anti-apoptotic effects of GA. In addition, when hair cells are damaged by ototoxic substances, neutrophils can migrate to the neuromasts and produce pro-inflammatory cytokines, resulting in persistent inflammation (Hirose et al., 2017). In our experiments, GA was able to inhibit neutrophil accumulation and reduce the levels of pro-inflammatory cytokines, including TNF-α, IL-1β, and IL-6, supporting its anti-inflammatory activity. RT-PCR analysis revealed that GA treatment downregulated the expression of STAT3 and AKT1 mRNA. Both STAT3 and AKT1 are known to play pivotal roles in inflammation and oxidative stress regulation, where STAT3 mediates inflammatory responses and AKT1 is involved in cell survival and metabolism (Liu et al., 2023). Therefore, these findings demonstrated that GA could exert the anti-ototoxic effect by inhibiting ROS generation and inflammation.

GA also demonstrated efficacy in guinea pig HL models. ABR measurements and hair cell integrity analysis confirmed that GA conferred protection against GM and CDDP-induced ototoxicity when administered via intratympanic injection, consistent with the results of zebrafish model. However, the efficacy of GA varied with the administration route. Oral administration of GA showed negligible effects, which was likely due to its hydrolysis. GA is easily hydrolyzed in the digestive tract into glycyrrhetinic acid, which is poorly absorbed and primarily targets the liver, resulting in a reduced amount of active compounds entering the bloodstream (Yoshino et al., 2021). Moreover, compared to intratympanic injection which is widely used clinically for drug delivery to the inner ear and ensures enough local administration, oral administration of drugs might provide low bioavailability, further leading to insufficient concentration in the inner ear. Nevertheless, intratympanic injection is invasive and complex to perform, which causes discomfort or infection risk in patients. In contrast, while oral administration is more convenient and improves patient compliance, its therapeutic efficacy is often limited by poor absorption, metabolic degradation, and inefficient transport across biological barriers. These suggest that a more effective oral administration strategy is needed to improve GA's efficacy for HL prevention.

GA is amphiphilic and can form micelles that encapsulate lipophilic components, enhancing the drug stability and targeting specific tissues. Here, we innovatively employed GA with the united role of medicine and excipient and prepared GA micelles to improve its oral bioavailability. To stabilize the formulation of a micelle, the introduction of a lipophilic core is needed. Cur, another herb-derived bioactive compound, is commonly used as a colorant in the food industry and also has important pharmacological effects such as anti-oxidation and anti-inflammation. For more than a decade, previous studies have focused on its pharmacological activities and demonstrated its anti-ototoxic potential in HL (Chen et al., 2022; Fetoni et al., 2014; Fetoni et al., 2015; Yi et al., 2024). However, its lipophilicity and poor bioavailability remain major problems. In fact, GA and Cur are often combined in traditional formulations such as Jianghuang San and Shenxian Jiuqi Tang, which may enhance the therapeutic outcomes (Zeng et al., 2024). The combination of GA and Cur was selected based on their shared antioxidant and anti-inflammatory properties, historical compatibility in traditional Chinese prescriptions, and their ability to spontaneously co-assemble into stable micelles. As such, we used GA as the drug and the carrier of Cur to fabricate a novel kind of carrier-free micelles. The micelles were fabricated via thin-film dispersion, a widely used technique in nanomedicine. During the preparation process, key parameters including feeding mass, feed ratio, and temperature were found to influence the drug loading capability of the micelles, whereas the particle size remained relatively stable. This was attributed to the large loading capacity of the carrier-free nanoassemblies. Such assembly processes were further validated through molecular dynamics simulation, which revealed stable intermolecular interactions between GA and Cur, mainly involving hydrogen bonding and π-π stacking. These interactions facilitated the spontaneous formation of stable micellar structures without the need for excipients. With the optimized parameters, GA micelles with high DLC exceeding 80 % (GA) were successfully obtained. These micelles demonstrated excellent protective effects against CDDP/GM-induced ototoxicity in guinea pigs, even with oral administration. Importantly, GA and Cur exhibited a synergistic effect, likely due to the enhanced bioavailability of both drugs. In this study, GA and Cur alone were used as controls for GA micelles, with approximate doses set based on their loading ratios for experimental operability. Although slight deviations existed, the monotherapy groups exhibited significantly weaker protective effects compared to GA micelles, supporting the validity of the experimental conclusions.

The bioavailability measured by pharmacokinetics revealed that micelle formulation substantially enhanced the plasma concentrations and systemic exposure of GA and Cur, as reflected by increased C_max and AUC values compared to the free drugs. We hypothesize that there are two reasons. First, in the micelle formation, glycyrrhetinic acid was partially encapsulated inside the micelle, which could mask its liver-targeting properties and protect GA from hydrolysis. Second, the solubility of Cur has also been improved in the micellar formation. These structural advantages collectively contribute to the enhanced pharmacokinetic performance observed. However, the specific mechanism needs to be further investigated. Importantly, unlike the FDA-approved otoprotective agent sodium thiosulfate, GA micelles did not affect the antitumor activity of CDDP as revealed by our results. Furthermore, the micelles were able to reduce the damage to liver and kidneys, a main side effect of CDDP. In conclusion, the micellar formulation could be a promising strategy for oral delivery of GA, offering effective prevention of HL while maintaining the antitumor activity of CDDP.

5. Conclusion

In summary, we employed GA as an otoprotective agent to prevent HL induced by various factors, including ototoxic drugs (GM and CDDP), noise, and diabetes. Our results demonstrated that GA effectively protected hair cells from damage in all studied zebrafish models. Experiments revealed that GA reduced ROS and pro-inflammatory factors induced by GM or CDDP. PCR validation further confirmed that the expressions of oxidative stress and inflammation-associated mRNA STAT3 and AKT1 were downregulated after GA pretreatment. In the guinea pig model, GA also displayed protective effects against ototoxicity, but the efficacy varied according to the administration route. To improve oral efficacy, we utilized GA's united role as both the active pharmaceutical ingredient and excipient, and formulated it into micelles. Optimization of the assembly conditions demonstrated that the feeding ratio (GA to Cur) and temperature could affect the drug loading efficiency. Molecular dynamics simulation further validated the self-assembly capability of the micelles. The obtained micellar formulation of GA effectively enhanced the bioavailability and improved its anti-ototoxic effects in CDDP-induced HL. Moreover, the micelles exhibited protective effects on the kidneys and liver. Cell experiments also revealed that they did not influence the antitumor efficacy of CDDP. These results suggested that GA could serve as a promising otoprotective agent and offer great opportunities for oral prevention of HL with micellar formulation.

CRediT authorship contribution statement

Huaan Li: Writing – review & editing, Writing – original draft, Methodology, Funding acquisition, Data curation. Bohan Lin: Writing – original draft, Methodology, Data curation. Shuangwu Wang: Methodology, Investigation. Ying Zhang: Methodology, Investigation. Zeming Zhou: Methodology, Investigation. Dingsheng Wen: Investigation, Formal analysis. Zhifeng Zhang: Methodology, Investigation. Xiaohua Feng: Methodology, Investigation. Lu Wen: Writing – review & editing, Supervision, Formal analysis. Jun He: Writing – review & editing, Supervision, Formal analysis. Gang Chen: Writing – review & editing, Supervision, Funding acquisition.

Ethics statement

All experimental procedures involving animals were reviewed and approved by the Animal Ethics Committee of Guangdong Pharmaceutical University (Ethical approval number: CVF2017105). All animal studies were conducted in accordance with the ARRIVE guidelines (Animal Research: Reporting of In Vivo Experiments) and the Guidelines for the Care and Use of Laboratory Animals.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

Supplementary material

Data availability

Data will be made available on request.