Dried SH was acquired from Yeongcheon Hyundai Herbal Supplier (Yeongcheon, Korea) and confirmed by Professor Ki Hwan Bae, College of Pharmacy, Chungnam National University (Daejeon, Korea). All voucher specimens were deposited in an herbal bank at the KM Application Center, Korea Institute of Oriental Medicine (voucher number: E252). A total of 50.0 g of SH was extracted with 300 mL of a solvent consisting of 70% ethanol and 30% distilled water (DW) at 40 °C and 100 rpm in a shaking incubator for 24 h. The extract was filtered through a filter paper and concentrated in a rotary vacuum evaporator (Buchi, Tokyo, Japan) to remove residual ethanol. It was then freeze-dried to produce SHE powder and stored at −20 °C until use.

Dr. Younghoon Go of the Korea Institute of Oriental Medicine (Daegu, Korea) provided the HT22 cells used in this study. Cells were maintained in complete Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum (FBS) and 1% antibiotics at 37 °C in a 5% CO₂ atmosphere during the experiments. DMEM was purchased from Welgene (Gyeongsan, Korea), and FBS and the antibiotics were purchased from HyClone (Logan, UT, United States).

HT22 cells were incubated in 96-well culture plates (5 × 10³ cells/well) and then subsequently exposed to 5 mM glutamate with or without SHE (10, 20, or 30 μg/mL) or its six main components (isoorientin, orientin, vitexin, cynaroside, cosmosiin, and luteolin). After 24 h, Cell Counting Kit (CCK) solution (10 μL) was added to each well and incubated for 1 more hour. The viability of HT22 cells was determined by optical density at 450 nm using a microplate reader (Molecular Devices, San Jose, CA, United States). The CCK was purchased from Dojindo (Kumamoto, Japan).

Lactate dehydrogenase (LDH) release in the culture medium was analyzed using an LDH assay kit according to the manufacturer’s standard protocol. Cells were cultured in 96-well culture plates, pretreated with three concentrations of SHE or its six main components for 2 h, and then exposed to 5 mM glutamate. After 24 h, the cells were centrifuged, and a portion of the supernatant was collected and incubated with the reaction solution. After an additional 30 min of incubation, LDH activity was evaluated at the wavelength of 450 nm by a microplate reader.

2,7-Dichlorohydrofluorescein diacetate (H₂DCFDA) was used to detect intracellular ROS. HT22 cells were seeded at 5 × 10³ cells/well in 96-well culture plates. After 18 h, they were pretreated with SHE, cynaroside, or luteolin for 2 h, and then 5 mM glutamate was added to the wells. A total of 20 µM H₂DCFDA was added to each well for 30 m under dark conditions, and the stained cells were washed three times with cold phosphate-buffered saline (PBS). Fluorescence intensity was estimated using a microplate reader at an excitation wavelength of 485 nm and an emission wavelength of 525 nm, and fluorescence images were captured using a fluorescence microscope (Eclipse Ti, Nikon, Tokyo, Japan). H₂DCFDA was purchased from Invitrogen (Carlsbad, CA, United States).

Neuronal apoptosis of HT22 cells was measured and quantified using the FITC Annexin V Apoptosis Detection Kit I (BD Biosciences) according to the standard procedures. HT22 cells were seeded at 5 × 10⁵ cells/well in 6-well culture plates and pretreated with SHE for 2 h and then exposed to 5 mM glutamate for another 24 h. These cells were collected by trypsinization and washed with cold PBS twice. After washing, cells suspended in 500 μL binding buffer were stained with 5 μL propidium iodide (PI) and FITC-conjugated annexin V for 15 min at room temperature (RT) in the dark. These cells were then evaluated on a flow cytometer (FACSCalibur, BD Biosciences).

Mouse brain tissues and HT22 whole cells were lysed with radioimmunoprecipitation assay lysis buffer (Millipore, Burlington, MA, United States) containing protease and phosphatase inhibitors (Roche, Basel, Switzerland) and incubated on ice for 1 h before centrifugation to obtain protein lysates. The protein concentration of each sample was standardized using a BCA Kit (Thermo Fisher Scientific), and equal amounts of protein (20 µg) were separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis. The proteins were blotted onto polyvinylidene fluoride membranes and blocked with a solution containing 3% bovine serum albumin and Tween 20 in tris-buffered saline. The membranes were then incubated overnight at 4 °C with each primary antibody (1:1,000 dilution) and washed four times. The membranes were incubated with the appropriate secondary antibodies diluted at 1:5,000 for 1 h at RT and analyzed using a ChemiDoc™ Touch Imaging System (Bio-Rad, Hercules, CA, United States). Each protein band was normalized to its housekeeping protein and analyzed using ImageJ software (version 1.53e; National Institutes of Health, Bethesda, MD, United States). Table 1 details the primary and secondary antibodies used for the Western blot analysis.

Nitric oxide (NO) secretion was estimated by measuring the nitrite concentration in the BV2 cell culture medium using Griess reagent. BV2 cells cultured for 18 h in 96-well culture plates were treated with SHE (10, 50, or 100 μg/mL) and exposed to 100 ng/mL LPS for 1 h and later for 24 h. Then, the same amount of Griess reagent as the culture medium was added and incubated for 5 min at RT, and the absorbance was measured at 570 nm using a microplate reader.

Inflammatory cytokine levels in BV2 cell culture medium were measured using an ELISA Kit (Invitrogen, Carlsbad, CA, United States) according to the manufacturer’s protocol. BV2 cells cultured for 18 h in 24-well culture plates were pretreated with SHE for 1 h and stimulated with LPS for another 6 h. After that, the supernatant was obtained by removing the cell debris through centrifugation (15,000 g), and the cytokine levels were measured.

C57BL/6 mice (4 weeks old) were obtained from Samtako Bio Korea (Osan, Korea) and used for in vivo experiments. Male C57BL/6 mice were housed in a specific pathogen-free facility under controlled conditions (12-h light/dark cycle at 22 °C ± 2 °C), and food and water were provided ad libitum. All mice were allowed to acclimate to the new housing environment for 1 week before the experiment and randomly divided into four groups of eight mice each: normal (vehicle), SCO (SCO + vehicle), SHE 100 (SCO + SHE 100 mg/kg), and SHE 200 (SCO + SHE 200 mg/kg). SCO was acquired from Sigma-Aldrich (St. Louis, MO, United States) and used to induce cognitive impairment in mice. All mice were orally administered the vehicle or SHE once daily for 4 weeks, and 1 mg/kg of SCO was injected once daily to the mice in the SCO, SHE 100, and SHE 200 groups during the last week. All animal experiments were approved by the Animal Care and Use Committee of the Korea Institute of Oriental Medicine (reference number #D-22-045) and followed the relevant guidelines.

The water maze pool was divided equally into four quadrants, with a platform hidden approximately 1 cm below the water, as described previously (Morris, 1984). The Morris water maze (MWM) test was started on the 28th day after SHE administration to evaluate the spatial learning memory ability of the mice. Each trial lasted 60 s until the mouse found the hidden platform and climbed up on it. The mice underwent an acquisition trial for 6 days. The MWM test was analyzed by a video camera equipped with the SMART video tracking system (Panlab, Barcelona, Spain).

The passive avoidance (PA) test was performed to measure the working memory of the mice using the Shuttle Box Avoidance Basic Test Package (Med Associates Inc., Fairfax, VT, United States). This apparatus consists of two compartments connected to each other by a guillotine door. One of them is illuminated, and the other is a dark compartment with an electric bar mounted on the floor. On the training day, each mouse was placed in the illuminated compartment, and after a 30-s acclimation period, the guillotine door was opened, and when the mice moved to the dark chamber, a 0.2-mA electric foot shock (5 s) was delivered after a 2-s delay. The test procedure was the same in the test conducted 24 h later, but no electric shock was delivered when moving to the dark chamber. The latency to move to the dark compartment was recorded in all tests, up to a maximum of 180 s.

After the mice were sacrificed, brain tissue was collected and post-fixed in a 4% paraformaldehyde solution for 24 h. Specimens were dehydrated, immersed in paraffin, and then sliced into 5-µm-thick sections at RT for later use. Tissue sections were immersed in cresyl violet solution (0.5%) for 30 min, rinsed with DW, dehydrated with different ethanol concentrations, cleared using xylene, and mounted on glass coverslips (Chen et al., 2018). Images were visualized using a microscope.

Fresh fecal samples were collected from all the mice in each group, and the collected samples were placed in sterile tubes, immediately frozen in liquid nitrogen, and stored at −80 °C. DNA from each sample was extracted using the QIAamp PowerFecal DNA Isolation Kit (Qiagen), and the quality of the extracted DNA was assessed by electrophoresis and quantified using a Qubit 2.0 Fluorometer (Life Technologies, Carlsbad, CA, United States). 16S rRNA gene sequencing was then performed using the Illumina MiSeq platform at the KNU NGS Core Facility (Kyungpook National University, Daegu, Korea). Data were analyzed by QIIME 2 (version 2022.2) using DADA2 software (Bolyen et al., 2019; Callahan et al., 2016). The taxonomic classification of the representative sequences was determined through taxonomy using the q2-feature-classifier (sklearn) against the Greengenes database (version 13.8) (DeSantis et al., 2006; Pedregosa et al., 2011). Phylogenetic trees, both rooted and unrooted, were generated using the MAFFT, mask, and FastTree protocols (Katoh and Standley, 2013; Price et al., 2010). These trees were subsequently used for diversity analysis. The smallest sample contained 37,376 features. Further analyses were performed using the phyloseq (version 1.32.0) and vegan (version 2.5-6) R packages (McMurdie and Holmes, 2013).

All standard solutions contained methanol as the solvent to dissolve the compounds. Initially, standard stock solutions with 100 μg/mL concentration were prepared, followed by serial dilution with methanol to create a range of concentrations for the calibration curve. The SHE sample solution was prepared at 20 mg/mL concentration in methanol. Prior to analysis, all solutions were filtered using 0.45-μm regenerated cellulose membrane syringe filters (Sartorius, Germany). Table 2 details the compounds used for HPLC analysis.

HPLC analyses were carried out using a Dionex UltiMate 3000 system (Dionex Corp., Sunnyvale, CA, United States), which includes a binary pump, autosampler, column oven, and diode array UV/Vis detector (DAD). The operation of the system and data analysis were managed using Dionex Chromeleon software. Chromatographic separation was achieved using a Luna C18 column (250 mm × 4.6 mm, 5 μm, Phenomenex, Torrance, CA, United States). The column oven was maintained at 30 °C, and detection was performed at a UV wavelength of 320 nm. The mobile phase was composed of water (solvent A) and acetonitrile (solvent B), with gradient elution programmed as follows: a total of 0 min–60 min, with a gradient of 5% B (0 min–5 min), 5%–15% B (5 min–12 min), 15%–35% B (12 min–42 min), and 35%–80% B (42 min–60 min). The flow rate was set to 1.0 mL/min. After each run, the column was re-equilibrated to the initial solvent conditions for 10 min before the next sample injection.

Specificity was evaluated by comparing the chromatographic profiles of the standard with those of the sample extract. Linearity was determined by the correlation coefficient (r²) of the calibration curve for each compound at concentrations ranging from 15.625 μg/mL to 500 μg/mL. The regression equations were derived using the formula y = ax ± b, where y represents the peak area and x denotes the concentration of the sample.

The data values are expressed as the means with standard error of the mean from at least three independent replicate experiments. Statistical significance was measured through one-way analysis of variance and verified using Dunnett’s test after comparing SCO, glutamate, or LPS vs. each treated group. Statistical significance was determined as ^# p < 0.05 (vs. normal or control); ^* p < 0.05, ^** p < 0.01, and ^† p < 0.001 (vs. SCO, glutamate, or LPS); and ^‡ p < 0.05 (vs. glutamate + SHE 30).

We treated HT22 mouse hippocampal neurons with SHE at different concentrations to investigate its effect on cell viability in vitro. The CCK assay showed that SHE treatment alone had no effect on the viability of HT22 cells at up to 30 μg/mL concentration (Figure 1A). Therefore, in subsequent experiments, SHE was administered at doses of 30 μg/mL or less. Next, we measured the effect of SHE on glutamate-exposed neurotoxicity in HT22 cells. An approximately 63% decrease in cell viability was observed after exposure to glutamate (Figure 1B). However, SHE pretreatment effectively improved cell viability in a concentration-dependent manner. Upon pretreatment with 30 μg/mL SHE, cell viability improved to approximately 91% (Figure 1B). Treatment with glutamate increased the leakage of LDH, an indicator of apoptosis, by 225%, whereas pretreatment with SHE at any concentration (10 μg/mL–30 μg/mL) significantly reduced LDH leakage (Figure 1C). This protective effect of SHE against neurotoxicity was also observed against changes in cell morphology. Glutamate treatment induced nuclear condensation and cell shrinkage in HT22 cells, whereas pretreatment with SHE dose-dependently maintained the cell structure and inhibited morphological changes (Figure 1D).

We analyzed ROS accumulation in HT22 cells by fluorescence intensity with an H₂DCFDA probe to demonstrate the effect of SHE on neuronal oxidative stress caused by glutamate exposure. ROS accumulation notably increased in HT22 cells treated with glutamate. However, pretreatment with SHE significantly and concentration-dependently reduced glutamate-induced intracellular ROS production (Figure 1E).

To explore the activity of SHE on neuronal apoptosis exposed by glutamate treatment, flow cytometric analysis was performed using annexin V–PI double staining on damaged cells. Flow cytometry revealed that glutamate exposure for 24 h notably increased the apoptosis of HT22 cells. However, SHE pretreatment significantly attenuated glutamate-treated neuronal cell death concentration-dependently, and treatment at 30 μg/mL concentration maintained cell numbers similar to that of the non-treated control (Figure 2A).

As Nrf-2 activation promotes antioxidant enzyme expression and suppresses oxidative stress, exerting neuroprotective effects, we analyzed the effects of SHE on the nuclear translocation of Nrf-2 and the expression of antioxidant enzyme in HT22 cells. Figure 2B shows that SHE pretreatment increased the nuclear translocation of Nrf-2, effectively induced the expression of HO-1, and significantly restored the expression of NQO1 that was decreased by glutamate. Therefore, SHE induced an increase in antioxidant enzyme levels through the activation of Nrf-2, suggesting that the antioxidant properties contributed to the neuroprotective efficacy of SHE to some extent. Therefore, the activation of the Nrf-2 signaling pathway is thought to be involved in the neuroprotective effect of SHE.

We also analyzed the effects of SHE on ERK/CREB pathway activation and BDNF expression in HT22 cells treated with glutamate. Figure 2C shows that SHE effectively activated ERK and CREB by inducing phosphorylation compared to HT22 cells treated with glutamate alone, suggesting that it induced mature BDNF expression. To further evaluate the role of ERK in the neuroprotective effect of SHE, we evaluated the effect of combined treatment with SHE and MEK inhibitor U0126 (10 μM) on the ERK/CREB/BDNF pathway. As shown in Figure 2D, U0126 significantly reversed the SHE-induced increase in ERK/CREB phosphorylation and BDNF expression in glutamate-stimulated HT22 cells. These results suggest that SHE alleviates glutamate-induced neuronal apoptosis through the activation of the ERK/CREB/BDNF signaling pathway.

We first investigated the cytotoxicity of SHE against BV2 microglial cells using the CCK assay. As shown in Figure 3A, SHE treatment did not show cytotoxicity up to a concentration of 100 μg/mL. Therefore, subsequent experiments were performed at concentrations below 100 μg/mL. Next, we measured the production of neuroinflammatory factors to evaluate the anti-neuroinflammatory effect of SHE. Figure 3B shows that SHE treatment markedly inhibited NO secretion in a concentration-dependent manner after LPS stimulation. In addition, the level of inflammatory cytokines, including TNF-α, IL-6, and MCP-1, was effectively suppressed by SHE treatment (Figure 3C).

NF-κB is a key factor that plays an important role in regulating various pro-inflammatory mediators, such as NO and inflammatory cytokines (Roshak et al., 1996). Thus, we examined the effects of SHE on changes in the transcription factor (NF-κB) in LPS-stimulated BV2 cells. The Western blot analysis showed that SHE concentration-dependently inhibited p65 translocation into the nucleus (Figure 3D). In addition, as phosphorylation-activated MAPKs have a crucial role in regulating NF-κB activation, we additionally assessed the inhibitory effects of SHE treatment on the activation of the MAPK pathway. Our data revealed that SHE treatment strongly diminished phosphorylation of all the members of the MAPK family, including ERK, p38, and JNK (Figure 3E). These results suggest that SHE inhibits the expression of pro-inflammatory factors by regulating the NF-κB/MAPK signaling pathway in LPS-stimulated BV2 microglial cells.

We first determined the spatial memory function of mice using the MWM test to explore the effects of SHE on SCO-induced cognitive impairment. As shown in Figures 4A–C, the mice injected with SCO showed poor learning ability to find the hidden platform than the normal group. In contrast, the 100 mg/kg SHE-treated group showed a relatively quick and good learning ability to find the platform. From day 4 to day 6, the average escape latency and distance increased significantly in the SCO group, indicating that intraperitoneal administration of SCO reduced the spatial memory function of the mice. However, 100 mg/kg SHE administration strongly reduced both the latency (Figure 4A) and distance (Figure 4B) of escaping the water maze. On the other hand, administration of 200 mg/kg SHE showed a tendency to decrease both the escape latency and escape distance compared with SCO alone, but the effect was lower than that of 100 mg/kg and was not significant. In addition, the degree of recovery of the spatial memory function according to SHE administration by dose was also well demonstrated in the swimming trajectory on day 6 (Figure 4C).

Following the MWM test, a PA test was conducted to explore the influences of SCO and SHE administration on the working memory of mice. Little difference was found between each group in the training trial, but in the testing trial, the step-through latency of mice was significantly decreased by SCO administration (Figure 4D). However, the 100 mg/kg SHE group showed a significantly higher step-through latency than the SCO group, indicating that 100 mg/kg SHE administration improved the mice’s working memory (Figure 4D). In addition, the 200 mg/kg SHE group showed no improvement in working memory at all compared with the SCO group, suggesting that the efficacy of 100 mg/kg SHE was much better than that of 200 mg/kg, as shown in the MWM test (Figure 4D). Therefore, only the 100 mg/kg SHE group sample, which showed a lower dose but better cognitive improvement efficacy, was used in the subsequent in vivo analysis tests.

Histopathological changes in brain tissue were analyzed by Nissl staining with 0.5% cresyl violet solution to confirm neuronal damage in the mice brains caused by SCO injection and the effect of SHE administration thereon. Compared with that of normal mice, the hippocampal regions of mice treated with SCO alone appeared irregularly arranged and shrunken, indicating extensive damage, neuronal apoptosis, and the loss of large amounts of Nissl bodies (Figure 5A). Conversely, in the cortex and hippocampus of 100 mg/kg SHE-administered mice, the number of neurons with normal morphology was effectively restored (Figure 5A).

Next, hippocampal tissues were analyzed by immunoblotting to measure the signaling molecules associated with the alleviating effects of SHE on cognitive decline and neuronal damage in mice. The levels of ERK, CREB, PI3K, Akt, GSK-3β phosphorylation, and BDNF expression in mouse hippocampal tissue were significantly reduced by SCO injection. Conversely, the decreased ERK, CREB phosphorylation, and BDNF expression were strongly increased by 100 mg/kg SHE administration (Figure 5B). The activation levels of PI3K and its downstream signaling molecules Akt and GSK-3β, which were reduced by SCO, were also significantly restored by 100 mg/kg SHE administration (Figure 5C).

To examine the effect of SHE on the gut microbiome of SCO-induced cognitive impairment in mice, we performed a comprehensive analysis of microbial diversity and composition. Comparison of the alpha diversity of the gut microbiome of mice showed that the Shannon and evenness indices significantly increased following SCO administration. However, 100 mg/kg SHE administration restored microbial diversity to near normal levels (Figures 6A, B). Principal coordinates analysis revealed a distinct clustering of microbiome profiles between the SCO group and both the normal and 100 mg/kg SHE-administered groups. No significant clustering was observed between the 100 mg/kg SHE-administered and normal groups (Figure 6C). These findings demonstrate that 100 mg/kg SHE restored the microbiome composition altered by SCO administration. Relative abundance analysis indicated changes in microbial composition in the SCO-only group, including a decrease in Bacteroidetes and an increase in Firmicutes. Treatment with 100 mg/kg SHE significantly restored the balance of these bacterial phyla (Figures 6D–G). These results indicate that 100 mg/kg SHE administration not only mitigated cognitive impairment and neuronal damage but also positively modulated the gut microbiome in SCO-induced mice.

The primary components of SHE were flavonoids, which are well known for their diverse biological activities. The six key flavonoid compounds identified in the current analysis were evaluated further. The mobile phase used for the separation consisted of acetonitrile and water, which were optimized to achieve clear resolution of the components. A photodiode array detector was used specifically to monitor the absorbance at a wavelength of 320 nm, which is characteristic of flavonoid compounds. The identification of these flavonoids was based on a comparative analysis of their chromatographic retention times (tR) and UV absorption spectra with those of known standards (Figure 7). The retention times for the standards were as follows: isoorientin (compound 1, tR = 17.41 min), orientin (compound 2, tR = 18.12 min), vitexin (compound 3, tR = 20.04 min), cynaroside (compound 4, tR = 21.13 min), cosmosiin (compound 5, tR = 24.53 min), and luteolin (compound 6, tR = 31.78 min). Chromatographic separation achieved excellent selectivity for all six flavonoids, with no significant interferences observed within the 60-min analysis.

The contents of compounds 1–6 were determined based on calibration curves created by plotting the peak area against the concentration for each analyte using least-squares regression analysis. Each calibration equation was derived from five concentration levels, ranging from 15.625 μg/mL to 250 μg/mL for isoorientin (1), vitexin (3), cosmosiin (5), and luteolin (6), and from 31.25 μg/mL to 250 μg/mL for orientin (2) and cynaroside (4). The results indicated that the concentrations of the six compounds ranged from 4.94 mg/g to 26.73 mg/g (Table 3). This finding suggests that this validation method is suitable for detecting trace amounts in the extract.

The successful identification and quantification of these flavonoids established flavonoids as the major chemical components of SHE. Orientin and cynaroside were found at the highest concentrations among the components detected, indicating their prominence in the overall flavonoid profile of SHE. This analysis not only reinforced the significance of flavonoids in SHE but also highlighted the reliability of HPLC–DAD in the precise identification of bioactive components.

To explore the effects of the six main components of SHE on glutamate-induced neurotoxicity in HT22 cells, we first investigated the effects of each flavonoid on cell viability. Each flavonoid was treated at 1 μM–50 μM in cultured HT22 cells, and isoorientin, orientin, and vitexin did not show significant cytotoxicity at all concentrations (Figure 8A). In addition, cynaroside did not show cytotoxicity up to 10 μM, and cosmosiin and luteolin did not show cytotoxicity up to 5 μM (Figure 8A). Therefore, in further testing for the neuroprotective activity of each flavonoid, they were used only in the range where they did not show cytotoxicity to completely exclude potential cytotoxicity.

Next, we explored the effects of each flavonoid on glutamate-induced neurotoxicity in HT22 cells. The cell viability of HT22 was reduced by approximately 70% when exposed to glutamate, which was not recovered at all by pretreatment with isoorientin, orientin, vitexin, and cosmosiin (Figure 8B). However, pretreatment with cynaroside and luteolin effectively and concentration-dependently improved the cell viability (Figure 8B).

In addition, LDH leakage, one of the indicators of cell death, was sharply increased by glutamate treatment, which was slightly inhibited by pretreatment with isoorientin and orientin (Figure 8C). Vitexin and cosmosiin did not inhibit LDH leakage at all, whereas cynaroside and luteolin showed excellent inhibitory effects at relatively low concentrations (Figure 8C).

We analyzed the effects of cynaroside and luteolin, which showed neurotoxicity-inhibiting effects among the six flavonoids, on ROS accumulation in HT22 cells. As shown in Figures 8D, E, glutamate treatment alone dramatically increased intracellular ROS accumulation. However, this increase was significantly reduced by cynaroside and luteolin pretreatment, and excellent effects were observed at relatively low concentrations (Figures 8D, E).