In our previous research, UHPLC-Q-Exactive Orbitrap HRMS analysis was conducted on TGD (11). By comparing and analyzing the chemical composition of each herb detected in the viscera and serum of rats after oral administration of TGD, the effective chemical components of each TCM that worked in the formula were explored. Epimedium is an important component of TGD. A total of 250 mg Epimedium sample (Sichuan Neo-Green Pharmaceutical Technology Development Co., Ltd.) was taken and placed in a 2ml centrifuge tube. Thereafter, 20% methanol (4 ml) was added for centrifugation (4˚C, 10,000 x g, 15 min). Subsequently, 200 µl supernatant was taken as the experimental sample. Chromatographic separation was implemented through a Waters ACQUITY UPLC BEH C18 column (2.1x100 mm, 1.7 µm; Waters Corporation) at a column temperature of 40˚C. The mobile phase included two solvents; mobile phase A consisted of methanol whereas mobile phase B comprised a 0.1% formic acid aqueous solution. Elution was completed according to the following schedule: 4% A from 0 to 4.0 min, 4-12% A at 4.0-10.0 min, 12-70% A at 10.0-30.0 min, 70-95% A at 30.0-35.0 min, 95% A at 35.0-38.0 min, and 4% A at 42.0-45.0 min. The flow rate was 0.3 ml/min and the injection volume was 2 µl. The resolution was 70,000 full width at half maximum and the cumulative time was 300 msec. Data collection and analysis of the results of UHPLC-Q-Exactive Orbitrap HRMS analysis (Thermo Fisher Scientific, Inc.) were performed using Xcalibur 4.1 software (Thermo Fisher Scientific, Inc.).

Drug targets were retrieved using the key word ‘Epimedium’ in the TCM systems pharmacology (TCMSP) (https://www.91tcmsp.com/), SwissTarget (http://swisstargetprediction.ch/), PharmMapper (http://lilab-ecust.cn/pharmmapper/) and HERB databases (http://herb.ac.cn/), utilizing the criteria oral bioavailability ≥30% and drug similarity ≥0.18. Subsequently, with ‘Alzheimer's disease’ being used as the key word, drug targets were searched in the GeneCards (https://www.genecards.org/), DisGeNet (https://disgenet.com) and Online Mendelian Inheritance In Man (OMIM) databases (https://www.omim.org/), and Venn diagrams (http://bioinformatics.psb.ugent.be/webtools/Venn/) were used to screen and generate overlapping targets between diseases and drugs. The ‘drug disease target’ was visualized with Cytoscape 3.8.2 (https://cytoscape.org/download.html), while common targets of drugs were imported in STRING database (https://string-db.org/) for protein-protein interaction (PPI) analysis. In the PPI network, proteins were denoted by nodes, whereas protein interactions were signified by edges. Nodes of various colors and sizes represented diverse degree values, and larger nodes and darker red colors declared greater degree values and more important targets. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were performed on the top 20 core targets in Metascape database (https://metascape.org/gp/index.html). Subsequently, the exported clustering network was sorted in accordance with the P-value using R 4.1.2 (https://www.r-project.org/) and a bubble chart was generated.

Epimedin C was supplied by Chengdu Biopurify Phytochemicals Ltd. (CAS no. 110642-44-9), and its purity was detected by HPLC-Diode Array Detection to be 95-99%. In addition, 17β-estradiol (17β-E₂) was provided by Sigma-Aldrich; Merck KGaA (cat. no. E2758). All samples were stored in the laboratory at 4˚C for future use.

PC12 cells were obtained from The Cell Bank of Type Culture Collection of The Chinese Academy of Sciences and were cultured in the Dulbecco's Modified Eagle medium (cat. no. C11995500BT; Gibco; Thermo Fisher Scientific Inc.) with 10% FBS (cat. no. 10099-141C; Gibco; Thermo Fisher Scientific Inc.) at 37˚C and 5% CO₂. The compound 17β-E₂ can be produced in the brain and it participates in regulating the reproductive axis (32). In addition, it can act on synaptic plasticity, and improve neural pathways and neurodegenerative diseases (32-34). Therefore, the present study used 17β-E₂ in the positive control group. PC12 cells were classified into the normal control, H₂O₂ model, 17β-E₂ and Epimedin C (1, 5 and 10 µM) groups. A concentration of 150 µM H₂O₂ (25) (cat. no. 7722-84-1; Sigma-Aldrich; Merck KGaA) was applied for 4 h at 37˚C and 5% CO₂ to induce OS within PC12 cells in the model and treatment groups. A total of 24 h before H₂O₂ treatment, Epimedin C (1, 5 and 10 µM) was introduced into the TCM groups, whereas 17β-E₂ (1 nM) was added into the positive control group for 24 h at 37˚C and 5% CO₂. To determine whether the JNK pathway affected Epimedin C-induced Nrf2 activation, PC12 cells from the model and treatment groups were also treated with the JNK agonist anisomycin (5 µM; cat. no. S7409; Selleck Chemicals). Following 4 h of H₂O₂ treatment, the culture medium was removed, and the cells were further incubated with anisomycin for 24 h at 37˚C and 5% CO₂.

The cultured cells were digested, centrifuge at 1,200 x g for 5 min at 25˚C and counted. Subsequently, cells (2x10⁴/well) were seeded into a 96-well plate and were cultured for 12 h under 37˚C, 5% CO₂ and 90% humidity conditions. The cells were then treated with different concentrations of Epimedin C (0-160 µM) for 24 h at 37˚C and 5% CO₂. Subsequently, cell viability was assessed according to the instructions of the Cell Counting Kit-8 (cat. no. GK3607; Gen-View Scientific Inc.) and the optimal drug administration concentration was screened. In addition, according to the manufacturer's instructions, a lactate dehydrogenase (LDH) detection kit (cat. no. C0016; Beyotime Institute of Biotechnology) was used to assess the cytotoxicity of Epimedin C at different concentrations (0-160 µM) to evaluate their safety. Six replicates were set for each sample and the absorbance was measured at 450 nm for the CCK-8 assay and at 490 nm for the LDH assay using a microplate reader (BioTek; Agilent Technologies, Inc.).

The ROS assay kit (cat. no. S0033; Beyotime Institute of Biotechnology) was adopted to detect cellular ROS levels. PBS was introduced to dilute DCFH-DA to 10 µmol/l. After removing the culture medium, the diluted DCFH-DA at 10 µM was added and the cells were incubated for 20 min at 37˚C. The cells were then rinsed three times with PBS to thoroughly remove the DCFH-DA that had not entered the cells. A microplate reader (BioTek; Agilent Technologies, Inc.) was adopted for testing ROS levels. The fluorescence was measured at an excitation wavelength of 485 nm and an emission wavelength of 525 nm.

The lipid peroxidation level in PC12 cells was evaluated using the MDA assay kit (cat. no. S0131S; Beyotime Institute of Biotechnology). PC12 cells (1x10⁴/well) were collected and lysed using RIPA lysis buffer (cat. no. P0013K; Beyotime Institute of Biotechnology). The cell lysate was then centrifuged at 12,000 x g for 5 min at 4˚C to remove insoluble debris. Finally, the absorbance of the resulting supernatant was measured at 532 nm using a microplate reader. The MDA concentration in the samples was calculated based on a standard curve prepared concurrently.

The PC12 cells were scraped and digested into a cell suspension, and were immediately fixed with 2.5% glutaraldehyde in 0.1 M phosphate buffer (pH 7.4) for 24 h at 4˚C. Following primary fixation, the cells were post-fixed in 1% OsO₄ tetroxide for 2 h at 4˚C, then dehydrated through a graded series of ethanol (50, 70, 90 and 100%). The dehydrated cells were embedded in epoxy resin. Ultrathin sections were cut using an ultramicrotome at a thickness of 70-90 nm and collected on copper grids. The sections were then stained with 1% uranyl acetate at room temperature for 10 min, and the samples were observed using a Talos L120C transmission electron microscope (Thermo Fisher Scientific, Inc.) operated at an accelerating voltage of 80-120 kV. Digital images were acquired for analysis.

After intervention with different groups of drugs, PC12 cells were immersed in 4% paraformaldehyde for 30 min at 4˚C, after which the cells were rinsed three times with PBS and permeabilized using 0.1% Triton X-100 for 3 min at 25˚C. Subsequently, the TUNEL reaction solution (cat. no. C1170S; Beyotime Institute of Biotechnology) was added to the cells and incubated for 1 h in the dark at 37˚C. The cells were then rinsed with PBS and later subjected to nuclear counterstaining with DAPI (cat. no. C1006; Beyotime Institute of Biotechnology) at room temperature for 5 min. The number of TUNEL-positive cells from five random regions in three individual samples was quantified using a fluorescence microscope (Ti-E; Nikon Corporation), and TUNEL-positive cell rate (%) was calculated as: TUNEL-positive cell area/total cell area x100.

Flow cytometry was applied for the quantitative analysis of apoptosis and mitochondrial membrane potential (MMP) in PC12 cells, using the CytExpert software (version 2.5; Beckman Coulter, Inc.) for analysis. The Annexin V-FITC/PI cell apoptosis detection kit (cat. no. 556547; BD Biosciences) was utilized for detecting cell apoptosis. Briefly, cells (2x10⁵/well) were harvested and resuspended in 400 µl binding buffer. Thereafter, Annexin V-FITC and PI (5 µl each) were added and mixed, and the cells were incubated in the dark for 15 min at room temperature. Afterwards, flow cytometry (CytoFLEX SRT; Beckman Coulter Inc.) was used to analyze early and late apoptotic cell proportion.

The level of MMP can reflect the status of apoptosis. Briefly, the cells (2x10⁵/well) were cultured in a 6-well plate and rinsed once with PBS, Subsequently, 1 ml JC-1 (cat. no. C2006; Beyotime Institute of Biotechnology) staining solution was introduced, mixed sufficiently with the media and incubated at 37˚C in an incubator for 20 min. Afterwards, the supernatant was discarded, and the sample was rinsed twice with JC-1 staining buffer (1X). An Altra flow cytometer (CytoFLEX SRT; Beckman Coulter Inc.) was used to verify the changes in MMP in each group of cells. The JC-1 aggregates (healthy mitochondria) were detected in the PE channel (575 nm emission), while the JC-1 monomers (depolarized mitochondria) were detected in the FITC channel (530 nm emission). The ratio of the geometric mean fluorescence intensity (PE/FITC) was calculated to quantify the MMP.

After treatment, PC12 cells were subjected to lysis with RIPA buffer that contained 1% PMSF (cat. no. ST506; Beyotime Institute of Biotechnology). The BCA protein detection kit (cat. no. P0011; Beyotime Institute of Biotechnology) was used to measure protein concentration. Subsequently, PBS and 4X loading buffer were added according to protein concentration, mixed evenly and boiled at 95˚C for 10 min. Proteins (20 µg/lane) were separated by SDS-PAGE on 10% gels and were then transferred to polyvinylidene fluoride membranes (cat. no. ISEQ00010; Merck KGaA). After blocking for 1 h at room temperature in 5% BSA (Beyotime Institute of Biotechnology), the membranes were subjected to primary antibody incubation overnight at 4˚C using antibodies against: GAPDH (1:1,000; cat. no. 5174; CST Biological Reagents Co., Ltd.), JNK (1:1,000; cat. no. 9252; CST Biological Reagents Co., Ltd.), phosphorylated (p)-JNK (1:1,000; cat. no. 4668; CST Biological Reagents Co., Ltd.), HO-1 (1:1,000; cat. no. 43966; CST Biological Reagents Co., Ltd.), Bcl-2 (1:1,000; cat. no. 2870; CST Biological Reagents Co., Ltd.), Bax (1:1,000; cat. no. 14796; CST Biological Reagents Co., Ltd) and Nrf2 (1:500; cat. no. ab137550; Abcam). The next day, the membranes were washed six times with TBS-0.01% Tween and were incubated with a HRP-conjugated Goat Anti-Rabbit IgG (H+L) secondary antibody (1:10,000; cat. no. 33101ES60; Shanghai Yeasen Biotechnology Co., Ltd) at room temperature for 1 h. After incubation, the membranes were washed six times with TBST. Subsequently, the membranes were visualized using Omni-ECL (Epizyme; Ipsen Pharma). The gel imaging system (Chemiscope6300; Clinx Science Instruments Co., Ltd.) was used for imaging and semi-quantification was performed using ImageJ (version 1.6.0; National Institutes of Health).

Data are presented as the mean ± SD. Unpaired Student's t-tests were used for pairwise comparisons between two groups. For comparisons across more than two groups, the homogeneity of variances was first assessed using Levene's test. Following a significant Levene's test (P<0.05) which indicated heterogeneity of variances, a Welch's ANOVA was conducted in place of the standard one-way ANOVA. Post hoc comparisons against the control or H₂O₂ group were then performed using Dunnett's T3 test to account for the unequal variances. In cases where Levene's test was not significant (P>0.05), indicating homogenous variances, a standard one-way ANOVA was performed, followed by Bonferroni's post hoc test for all comparisons against the control or H₂O₂ group. GraphPad Prism version 9.3 (Dotmatics) was employed for data analysis. P<0.05 was considered to indicate a statistically significant difference.

Comparing the active ingredients in TGD that can enter the bloodstream, as identified by UHPLC-Q-Exactive Orbitrap HRMS in our previous study (11), a total of 20 active ingredients in Epimedium that could enter the rat bloodstream and exert their effects were identified, including Epimedin C, Epimidin B and others (Table I). The total ion chromatograms of active ingredients within Epimedium compound are displayed in Fig. 1.

To compensate for the lack of recent updates in drug databases, which may result in incomplete inclusion of components, network pharmacological analysis (using AD as an example) was conducted in conjunction with MS. Using ‘Epimedium’ as the key word, and oral bioavailability ≥30% and drug similarity ≥0.18 as the screening criteria in TCMSP, SwissTarget, PharmMapper and HERB databases, a total of 23 components of Epimedium were obtained, and 379 drug targets were retrieved. Using ‘Alzheimer's disease’ as the key word, a total of 1,165 drug targets were obtained from the GeneCards, DisGeNet and OMIM databases after deduplication. Based on the Venn diagram, 108 common targets of Epimedium and AD were identified (Fig. 2).

H₂O₂ stimulation can cause OS damage, leading to apoptosis or necrosis in PC12 cells (35,36). Except for the control group, all the other groups of PC12 cells were exposed to 150 µM H₂O₂ treatment for 4 h for modeling OS (25). Before treatment with H₂O₂, these cells were pretreated with 0-160 µM Epimedin C (Fig. 3A) for 24 h to evaluate the protective effects of Epimedin C on PC12 cells and to identify its optimal concentration range for improving OS status (Fig. 3B). The results showed that 0-80 µM Epimedin C pretreatment in PC12 cells did not have an effect on the viability of PC12 cells compared with the control. Among these concentrations, Epimedin C at concentrations of 1, 5 and 10 µM had an improved cell survival rate than other concentrations; however, this was not statistically significant. Therefore, the concentrations of 1, 5 and 10 µM were used for subsequent experiments. Relative to the OS model group, after 24 h of intervention with 17β-E2 and Epimedin C, the viability of PC12 cells was significantly improved (Fig. 3C). Among them, the cell survival rate significantly rose in a dose-dependent manner in response to pretreatment with 1, 5 and 10 µM Epimedin C. LDH analysis was applied to detect the toxicity of different treatments on PC12 cells in each group; according to the results, relative to the control group, the H₂O₂-induced group showed a significant increase in LDH release (Fig. 3D). Alternatively, the cells pretreated with Epimedin C and 17β-E₂ displayed a significant decrease in LDH secretion and cytotoxicity compared with those in the H₂O₂-induced group. These comprehensive results indicated that intervention with 10 µM Epimedin C had the best therapeutic efficacy in PC12 cell survival and the least toxic side effects.

MDA and ROS are involved in producing OS free radicals and are key indicators for measuring oxidative and antioxidant capacity (37,38). The results of the present study revealed that relative to the control group, MDA and ROS contents in PC12 cells induced by H₂O₂ were significantly increased (Fig. 3E and F). In cells that had been pretreated with 17β-E2 or 1, 5 and 10 µM Epimedin C, the MDA and ROS contents were reduced to varying degrees compared with those in the H₂O₂ group, indicating that Epimedin C is beneficial for alleviating the OS status of PC12 cells. Notably, 10 µM Epimedin C was the most effective concentration.

OS can induce cell apoptosis associated with increased ROS production and decreased MMP levels (39). Therefore, the cells were stained with Annexin V-FITC and PI, and the apoptosis rate was measured using flow cytometry (Fig. 4A and B). The present results showed that, after treatment with 150 µM H₂O₂ for 4 h, the PC12 cell apoptosis rate was significantly increased. By contrast, intervention with Epimedin C resulted in a decline in apoptosis rate compared with that in the H₂O₂ group, suggesting that Epimedin C can inhibit cell apoptosis.

A reduction in MMP represents a hallmark event during early cell apoptosis, which is evident by the transition of JC-1 from red fluorescence to green fluorescence (40). To further evaluate the apoptosis status of each group of cells, the MMP loss within H₂O₂-treated PC12 cells was observed through JC-1 staining. As shown in Fig. 4C and D, the MMP of H₂O₂-treated PC12 cells was decreased, whereas pretreatment with Epimedin C (1, 5 and 10 µM) resulted in a dose-dependent improvement in MMP, with varying degrees of upregulation.

TUNEL staining can be conducted to detect apoptotic cells with large amounts of DNA degradation during the late stage of apoptosis (41). Therefore, TUNEL staining was performed on each group of cells in the current study (Fig. 5A). The results of the TUNEL assay revealed that relative to the control group, following H₂O₂ induction modeling, the apoptosis of PC12 cells in the model group (level of red fluorescence) was significantly increased (Fig. 5B). Relative to the model group, after drug intervention, the apoptosis rates of PC12 cells in the 17β-E₂ and Epimedin C groups were significantly improved. The degree of apoptosis was ameliorated to varying degrees in response to different concentrations of Epimedin C, with the largest improvement observed in the 10 µM Epimedin C group.

Within the present study ultrastructural changes of PC12 cells in each group was observed using TEM (Fig. 6). Under physiological conditions, the mitochondria are present with prominent cristae and intact membranes (42). The control group cells had clear nuclear membranes and uniform chromatin. After H₂O₂ induction, marked mitochondrial damage was observed in PC12 cells, evident as mitochondrial membrane rupture and chromatin condensation, accompanied by vacuolization. As aforementioned, 10 µM Epimedin C had the largest effect on improving H₂O₂-mediated OS damage to PC12 cells. Accordingly, TEM was performed on PC12 cells treated with 10 µM Epimedin C to detect ultrastructural changes. The results indicated that treatment with 10 µM Epimedin C could inhibit mitochondrial damage and alleviate mitochondrial swelling in PC12 cells, indicating that 10 µM Epimedin C can improve oxidative damage and inhibit cell apoptosis.

The ‘drug-disease-target’ network was visualized with Cytoscape 3.8.2 software, and common targets between drugs (Epimedium) and diseases (AD) were imported in STRING database for PPI analysis. The PPI results included 110 nodes and 2,102 edges (Fig. 7A and B). Thereafter, the top 20 core genes of Epimedium for treating AD were screened using R language (R 4.1.2), including BCL2, APP, JUN, CASP3, ESR1, IL1B and TNF. The top 20 core targets underwent GO analysis in Metascape database. As a result, the shared targets were involved in various GO biological processes (Fig. 7D), including ‘positive regulation of apoptotic process’, ‘regulation of apoptotic signaling pathway’, ‘regulation of inflammatory response’ and ‘cellular response to nitrogen compound’. Moreover, the GO molecular functional results (Fig. 7E) showed that the shared targets involved in ‘DNA-binding transcription factor binding’, ‘cytokine receptor binding’, ‘protease binding’ and ‘protein kinase binding’. While GO cellular component results (Fig. 7F) indicated that the shared targets were closely related to ‘platelet alpha granule lumen’, ‘neuronal cell body’, ‘transcription regulator complex’ and ‘receptor complex’. Additionally, as revealed by KEGG enrichment analysis, the shared targets were enriched in pathways such as ‘MAPK signaling pathway’, ‘endocrine resistance’, cell clearance and ‘apoptosis’ (Fig. 7C).

JNK belongs to the MAPK family and has a crucial effect on cell proliferation, differentiation, apoptosis, immune response and embryonic development (43). From the network pharmacological analysis, KEGG enrichment analysis showed that the pathway through which Epimedium improves AD was significantly enriched with the ‘MAPK signaling pathway’. In addition, the PPI network revealed that the top 20 core targets included APP, JUN and BCL2. Notably, JUN (also known as C-JUN) is involved in the JNK pathway (44), and activated JUN promotes the expression of various pro-apoptotic proteins (45). The Nrf2 pathway is an important regulatory factor for cellular antioxidant response (46), which is involved in triggering the expression of antioxidant and cell protective genes, and enhancing cellular antioxidant capacity (47); therefore, activation of Nrf2 can reduce neuronal damage caused by OS and inflammation (48). The free Nrf2 then transfers to the nucleus and binds to antioxidant-related elements, leading to the expression of various antioxidant and detoxifying genes, such as HO-1(49). Research has shown that MAPKs are closely interrelated with Nrf2 nuclear translocation in eliminating OS (50).

Therefore, to study the neuroprotective role mediated by Epimedin C, WB was conducted for verifying p-JNK, Nrf2 and HO-1 protein levels. Relative to the control group, H₂O₂ increased p-JNK levels but decreased Nrf2 levels, whereas Epimedin C downregulated p-JNK expression, and upregulated Nrf2 and HO-1 levels induced by H₂O₂ (Fig. 8A-D). Furthermore, it was observed that compared with in the H₂O₂ model group, Epimedin C decreased Bax expression while increasing Bcl-2 expression (Fig. 8A and E).

To confirm that Epimedin C mediated protection via the JNK pathway, the present study used a JNK agonist (anisomycin) for supplementary validation. The results revealed that p-JNK was activated and significantly upregulated in PC12 cells co-cultured with JNK agonist (Fig. 8F and H). By contrast, after treatment with Epimedin C and anisomycin, p-JNK expression decreased compared with following treatment with anisomycin alone, whereas Nrf2 and HO-1 levels were significantly increased (Fig. 8F-I).