Puerarin (>98%) was purchased from Jiangsu Yongjian Pharmaceutical Technology Co., Ltd. (Jiangsu, China); fluoxetine hydrochloride (Prozac) was purchased from Eli Lilly and Company; simvastatin was obtained from Dalian Meilun Biotech Co., Ltd. (Dalian, China); rat ELISA kits of IL-6 and TNF-α were purchased from PeproTech (Rocky Hill, United States); the IL-10 rat ELISA kit was purchased from eBioscience (San Diego, United States); the BCA Protein Assay Kit was purchased from Pierce (Rochford,United States); Goat Anti-Mouse IgG H&L (Alexa Fluor® 647) antibodies were purchased from Abcam (Cambridge, MA); anti-TLR4, anti–claudin-1, anti-occludin, anti–β-actin, and HRP-conjugated Goat Anti-Rabbit IgG antibodies were purchased from Servicebio (Wuhan, China); the UNIQ-10 column TRIzol total RNA extraction kit was obtained from Sangon Biological Engineering Technology & Services Co., Ltd. (Shanghai, China). The FastStart Universal SYBR Green Master (ROX) kit was purchased from Roche (Mannheim, Germany); the calcium-dependent cytosolic phospholipases A2 (cPLA2) assay kit and prostaglandin E₂ (PGE₂) Express EIA Monoclonal Kit were purchased from Cayman Chemical (Minneapolis, United States). The COX-2 inhibitor screening kit was obtained from Beyotime Biotechnology (Shanghai, China).

Male Sprague–Dawley (SD) rats (160–180 g) and a high-fat diet containing 4.6 kcal/g (49% fat, 20% protein, and 31% carbohydrate) were purchased from Qingdao Daren Fucheng Animal Husbandry Co., Ltd. [SCXK: (Lu) 2019 0003, Qingdao, China]. The rats were housed in a ventilated, temperature-controlled (22–24°C), and standardized sterile animal room at Jining Medical University. All of the rats were adapted for 1 week with free access to food and water. All animal experiments in the present study were performed according to the National Institutes of Health guidelines for animal care and approved by the Ethics Committee of Jining Medical University (Approval No. 2019-YX-04).

The rats were randomly divided into 10 groups with 10 rats per group, including 1) the normal control group, 2) HFD group, 3) CUMS group, 4) HFD/CUMS group, 5) HFD/CUMS + fluoxetine (10 mg/kg) group, 6) HFD/CUMS + simvastatin (10 mg/kg) group, 7) HFD/CUMS + fluoxetine + simvastatin group, (8–10) HFD/CUMS + puerarin (30, 60, and 120 mg/kg) groups. As shown in Figure 1B, the rats received a HFD except those which were fed with a normal diet in the control and CUMS groups for 11 weeks. All rats received 1% (w/v) sugar water training since the third day of the seventh week for 4 days: two bottles of 1% sucrose per cage for 24 h; one bottle of water and one bottle of 1% sugar water for another 24 h; and two bottles of water for another two days. Since the eighth week, the rats in the CUMS and HFD/CUMS groups were exposed to one random stressor between 9:00 and 11:00 a.m. per day for 4 weeks. The stressors included noises for 3 h (60 dB), tail clamp for 2 min, swimming in cold water (10°C) for 5 min, inversion of light/dark cycle for 24 h, food deprivation for 24 h, water deprivation for 24 h, and level shaking for 5 min (1 time/s). Since the 11th week, the rats in various groups were given sterile 0.9% saline, fluoxetine, simvastatin, fluoxetine + simvastatin, puerarin (30, 60, or 120 mg/kg), respectively, once per day for 7 days. Fluoxetine, simvastatin, and puerarin were dissolved in sterile 0.9% saline solution and stored at 4°C for up to 24 h.

The open-field test was used to explore the behavior of rats. The open-field apparatus was a 100 × 100 × 40 cm black box equipped with a video tracking system (Ethovision, Netherlands). Each rat was placed into the center of the box and was observed for 6 min. The rearing numbers, time resting, time activity, and total distance traveled during the last 4 min were recorded and statistically analyzed.

The body weight of all rats was recorded at days 0, 7, 14, 21, 28, 35, 42, 49, 56, 63, 70, and 77. The 1% sucrose preference test was carried out at days 50, 57, 64, 71, and 78 to evaluate the anhedonic-like state of rats (Pothion et al., 2004). Before the test, all of the rats were deprived of food and water for 24 h and then fed with one bottle of water and one bottle of 1% sucrose solution for 1 h (8:00–9:00 a.m.). Water and sucrose solution consumption was obtained by weighing the bottles before and after the test. The sucrose preference rate was calculated according to the following formula: Sucrose consumption  ( % ) = Sucrose intake Sucrose water intake + Water intake ×100 .

After the behavior test, blood of anesthetized rats was collected into heparinized anticoagulant tubes and centrifuged at 3,000 rpm for 10 min. Plasma was then obtained and stored at −20°C until use. Simultaneously, the hippocampal tissues of rats were quickly harvested, frozen in lipid nitrogen immediately, and stored at −80°C. The levels of IL-6, TNF-α, and IL-10 in plasma and hippocampal tissues were detected by commercial ELISA kits, according to the manufacturer’s instructions. The results of hippocampal tissues were corrected as picogram per milligram protein.

After anesthetization, the rats were killed, and 2 cm of small intestines was dissected and fixed in 10% formalin solution for hematoxylin and eosin (H&E) staining. The histological and pathological changes of each slice were observed by using an optical microscope at a magnification of 100 ✕.

The small intestines and hippocampal tissues (∼50 mg) were quickly harvested, frozen in liquid nitrogen immediately, and stored at −80°C. The protein expression of TLR4, claudin-1, and occludin was determined by Western blotting. Small intestines and hippocampal tissues were homogenized using lysis buffer at a ratio of 1:5 (mg/μl) and quantified by the BCA protein assay kit (Thermo, United States). Samples containing an equal amount of the protein (40 μg) were subjected to 10% SDS-PAGE electrophoresis and blotted on the PVDF membrane. After blocking with 5% bovine serum albumin (BSA) in phosphate-buffered saline-Tween 20 (PBST) for 2 h at room temperature, the PVDF membrane was incubated with respective primary antibodies at 4°C overnight. The dilution of anti-TLR4, anti–claudin-1, anti-occludin, and anti–β-actin antibody was 1:1000, 1:500, 1:1000, and 1:1500, respectively. After washing with PBST, the membrane was incubated with horseradish peroxidase–conjugated secondary antibody (1:3000) for 2 h at room temperature. Then, the membranes were washed five times using PBST and incubated with the ECL substrate. Finally, the membrane was scanned into a computer, and densitometry was quantified using ImageJ software.

Total RNA of hippocampal tissues was extracted using the Sangon UNIQ-10 column TRIzol total RNA extraction kit and reverse-transcribed using the ImProm-II Reverse Transcription System cDNA synthesis kit, according to the manufacturer’s instructions. Primer sequences used for TLR4, occludin-1, claudin, and β-actin are shown in Table 1. The real-time RT-PCR reactions were performed using the SYBR Green system on an ABI 7500 PCR System (Thermo Fisher, MA, United States). MRNA expression was normalized to β-actin mRNA levels. Relative expression was determined relative to the normal control using the 2^−∆∆C_T method (Livak and Schmittgen, 2001).

After dissection, the rat brain tissues were immediately fixed in 10% (v/v) neutral formalin solution. Then, the tissues were embedded in paraffin and cut to 5 μm thick. The slides were deparaffinized and rehydrated before immunofluorescence staining using the anti-TLR4 antibody overnight at 4°C. After washing with phosphate-buffered saline (PBS, ph = 7.4), TLR4-positive cell detection was enhanced by using a Goat Anti-Mouse IgG H&L (Alexa Fluor® 647) antibody (Abcam, Cambridge, MA) in the dark. Moreover, the nuclei were stained with 4′,6′-diamidino-2-phenylindole (DAPI, 2 μg/ml). The stained slides were observed on a CRI fluorescence imaging system (Maestro2, CRI, United States) and analyzed using ImageJ software.

Rat hippocampal tissues (∼80 mg) were transferred into 2-ml centrifuge tubes. A total of 600 μl chloroform methanol mixed solution (2:1, precooled at −20°C) was added and vortexed for 30 s. After transferring the samples to ice for 40 min, 190 μl H₂O was added and vortexed for 30 s. After transferring the samples on ice for 10 min, the supernatant was centrifuged at 12,000 rpm for 5 min at room temperature. Then, ∼300 μl of the lower layer was transferred into a new contribute tube. This cycle was repeated once, and the samples were concentrated and dried in vacuum. Finally, the samples were dissolved with 200 μl isopropanol, and the supernatant was filtered through a 0.22-μm membrane to obtain the prepared samples for LC-MS.

Separation and chromatography of the samples on an Agilent 1290 infinity series UHPLC (Agilent Corporation, CA, United States) coupled to a Triple TOF® 6600 mass spectrometer system (AB SCIEX, United States)were performed. In brief, chromatographic separation was performed on a Phenomenex Kinetex C18 column (100 × 2.1 mm, 1.7 μm) maintained at 55°C. The temperature of the autosampler was 8°C. Gradient elution of the analytes was carried out with acetonitrile: water = 40:60 (10 mM ammonium formate) (A) and isopropanol: acetonitrile = 90:10 (10 mM ammonium formate) (B) at a flow rate of 0.30 ml/min. The injection volume was 2 μl in the positive mode and 6 μl in the negative mode. An increasing linear gradient of solvent B (v/v) was used as follows: 0–1.5 min, 40% B; 1.5–10.5 min, 40–85% B; 10.5–14 min, 85% B; 14–14.1 min, 85–100% B; 14.1–15 min, 100% B; 15–15.2 min, 100%–40% B; and 15.2–18 min, 40% B.

The mass spectrometry parameters were used as follows: ion source gas 1 (GS1), 60 psi; ion source gas 2 (GS2), 60 psi; curtain gas (CUR), 30 psi; temperature, 600°C; and ion spray voltage floating (ISVF), 5,000 V or −4,500 V in positive or negative modes; declustering potential (DP) 100 V. The data-dependent acquisition (DDA) method was used for MS/MS acquisition. Each acquisition cycle consists of one rapid TOF MS survey scan (200 ms) followed by the consecutive acquisition of 11 product ion scans (50 ms each). For the TOF MS survey scan, the mass range is from 200 to 2,000 Da, and collision energy (CE) is set as 10 V. For the product ion scan, the mass ranges are from 100 to 2,000 Da, and collision energy (CE) is set as 45 ± 25 V. Dynamic background subtraction was applied. Dynamic exclusion was implemented to remove some unnecessary information in the MS/MS spectra.

LipidView software (v4.2) was used to annotate the obtained raw data and then peak identification, filtration, and alignment were performed. The data were normalized by MetaboAnalyst 5.0 (https://www.metaboanalyst.ca/). Principal component analysis (PCA) and partial least square analysis (PLS-DA) were carried out. Differentially altered lipid metabolites were screened based on variable importance in projection (VIP) values. Metabolites with VIP >1 and p (corr) < 0.05 were further evaluated with an independent sample t-test. Moreover, the false discovery rate (FDR) was also used to evaluate the significance of differences in each metabolite. An FDR less than 0.05 was considered as significant.

The binding ability and binding mode of puerarin to cPLA2 and COX-2 were investigated through molecular docking studies. The crystallography structure of cPLA2 [PDB ID: 1DB5 (Schevitz et al., 1995)] and COX-2 [PDB ID: 3QH0 (Liang, 2011)] was retrieved from the PDB database and preprocessed in BIOVIA Discovery Studio 2017. Puerarin was then docked to cPLA2 and COX-2 with the CDOCKER module in Discovery Studio 2017. The binding site sphere radius was set to 10 Å for COX-2 and 9 Å for cPLA2. Twenty docked poses were generated for each docking simulation. The top-ranked poses were selected for further analysis. 3D figures were generated by Discovery Studio 2017. 2D diagrams were generated by LigPlot v.1.0 (Wallace et al., 1995).

For the enzyme activity detection of cPLA2, small intestine and hippocampal tissues were perfused with PBS containing 0.16 mg/ml heparin to remove red blood cells and clots. Then, the tissues (50 mg) were homogenized in 1 ml pre-cold buffer (50 mM HEPES, pH 7.4, containing 1 mM EDTA). After being centrifuged, the supernatant was analyzed according to the instructions of the commercial kit (Cayman Chemical, Ann Arbor, MI, United States). The samples were diluted at 1:5 and 1:10, respectively, for hippocampal tissues and small intestines. The absorbance was recorded at 414 nm using a plate reader. To detect the enzyme activity of COX-2, the COX-2 inhibitor screening kit (Beyotime Biotechnology, China) was used. The hippocampal tissues and small intestines were homogenized using lysis buffer at a ratio of 1:5 (mg/μl) and quantified by the BCA protein assay kit (Thermo, United States). The enzyme activity of COX-2 of the samples was analyzed following the instructions of the commercial kit and corrected as Relative Fluorescence Unit (RFU) per gram protein.

Production of PGE₂ in the hippocampus and small intestine tissues was detected by enzyme-linked immunosorbent assay (ELISA). The hippocampal tissues and small intestines were homogenized using lysis buffer at a ratio of 1:8 (mg/μl) and quantified by the BCA protein assay kit (Thermo, United States). The concentration of PGE₂ was detected according to the instruction of the commercial PGE₂ EIA monoclonal kit (Minneapolis, United States). The results were corrected as picogram per milligram protein.

All data were represented as mean ± S.E.M. N indicated the number of animal samples in each group. Statistical analyses were performed with Origin Pro 2021 software by Student’s t-test (between two groups) or one-way analysis of variance (ANOVA, among multiple groups). The p-value less than 0.05 was considered statistical significant.

In order to investigate the protective effect of puerarin on depression, an HFD/CUMS-induced depression-like rat model was used. As shown in Figures 2A,B, in comparison with the normal control group, the body weight of rats in the HFD group significantly increased from the 56^th day, while significant weight loss was observed in the CUMS-alone group at the 77^th day (p < 0.01). However, no significant difference was found among other groups because an HFD and CUMS have opposite effects on weight gain. Anhedonia, the main symptom of depressive disorder, was detected by the 1% sucrose preference test. Compared with the normal control group, rats in the HFD, CUMS, and HFD/CUMS groups showed significantly less sucrose consumption (Figure 2C, p < 0.01). The drug treatment groups, including simvastatin, fluoxetine, simvastatin + fluoxetine, and puerarin (30, 60, and 120 mg/kg), showed significant higher consumption of sucrose solution than the HFD/CUMS group (p < 0.01). In the open-field test (Figures 2D–F), HFD, CUMS, and HFD/CUMS stimulation prolonged time resting, while decreasing time activity and number. of rearing and total distance (p < 0.01), showing behaviors indicative of depression. In comparison with the HFD/CUMS group, simvastatin, fluoxetine, simvastatin + fluoxetine, and puerarin (30, 60, and 120 mg/kg) treatments significantly alleviated the depression-like behaviors (p < 0.01).

To further investigate the protective role of puerarin on HFD/CUMS-induced inflammation in the peripheral and CNS, the protein levels of IL-6, TNF-α, and IL-10 in plasma and hippocampal tissues were detected. The results in Figure 3 demonstrated that chronic stimulation of HFD, CUMS, and HFD/CUMS significantly induced inflammatory damage by enhancing the protein level of IL-6 and TNF-α and reducing the expression of IL-10 in the hippocampus and plasma (p < 0.01). However, in comparison with the HFD/CUMS group, simvastatin + fluoxetine or puerarin treatment antagonized the abnormal protein expression of IL-6, TNF-α, and IL-10 in hippocampal tissues and plasma (p < 0.01 or p < 0.05). Moreover, a dose-dependent manner was observed in puerarin-treated (30, 60, and 120 mg/kg) groups.

As shown in Figure 4A, in the normal control group, the small intestinal mucosal villi were arranged regularly, and no obvious damage was observed. The small intestinal mucosa of the rats in the HFD, CUMS, and HFD/CUMS groups were severely damaged, including the following characteristics: 1) the thickness from the serosa to muscularis was less; 2) the villi were broad and short and considerably swollen, broken, or necrotic; and 3) the arrangement of the lamina propria cells was completely disordered. Long-term stimulation of CUMS showed that the serosal layer was loosely arranged or fell off from the muscle layer. Among which, the morphological structure of the small intestine mucosa of rats in the HFD/CUMS group was the most severely damaged. Compared with the HFD/CUMS group, the changes of the small intestine mucosa were improved. Moreover, the simvastatin + fluoxetine-treated group and puerarin (120 mg/kg)-treated groups showed that the villi were basically intact, arranged tightly, and the structure of each layer was relatively clear and complete.

The pathological damage of the small intestine may be related to the increase in intestinal permeability mediated by abnormal expression of TLR4. Subsequently, the mRNA and protein expressions of TLR4, occludin, and claudin-1 were determined by real-time RT-PCR and Western blotting, respectively. As displayed in Figures 4B,E, HFD, CUMS, and HFD/CUMS stimulations significantly increased the mRNA and protein concentrations of TLR4 in the small intestine (p < 0.01). However, simvastatin, fluoxetine, simvastatin + fluoxetine, and puerarin (30, 60, or 120 mg/kg) treatments significantly downregulated the abnormal TLR4 levels (p < 0.01). Moreover, rats in the HFD, CUMS, and HFD/CUMS groups increased intestinal permeability by reducing the mRNA and protein expression of occludin (Figures 4C,F) and claudin-1 (Figures 4D,G, p < 0.01 or p < 0.05). After the rats received simvastatin, fluoxetine, simvastatin + fluoxetine, or puerarin (30, 60, and 120 mg/kg), the mRNA expressions of occludin and claudin-1 were considerably upregulated (p < 0.01 or p < 0.05). Similarly, the protein expression of occludin and claudin-1 was significantly enhanced by simvastatin + fluoxetine or puerarin (30, 60, and 120 mg/kg) (p < 0.01 or p < 0.05). These results indicated that the intestinal permeability was restored.

The TLR4 levels in the prefrontal cortex were identified as red, and the nucleus was stained in blue. As displayed in Table 2 and Figure 5A, the number of TLR4-positive cells in the HFD, CUMS, and HFD/CUMS groups was more than that of the normal control group (p < 0.01). Moreover, the intensity of red fluorescence in the HFD/CUMS group was stronger than that of the other groups. In comparison with the HFD/CUMS group, the rats that received fluoxetine, simvastatin, simvastatin + fluoxetine, or puerarin (30, 60, and 120 mg/kg) treatments showed that the number of TLR4-positive cells was reduced to 70.35, 37.78, 30.36, 65.20, 29.64, and 17.02% (p < 0.01). Moreover, we detected the expression of hippocampal TLR4. As shown in Figure 5B, HFD, CUMS, and HFD/CUMS stimulations significantly enhanced the expression of TLR4 (p < 0.01 or p < 0.05). However, rats in the simvastatin, simvastatin + fluoxetine, or puerarin (30, 60, and 120 mg/kg) groups displayed a remarkably decreased TLR4 level (p < 0.01 or p < 0.05).

Lipid metabolites are important signaling molecules in cells. The effect of puerarin on lipid metabolism is shown in Figure 6. PLS-DA score plots showed a complete separation between the control and HFD/CUMS groups and HFD/CUMS and PUE groups, indicating that there were different lipid metabolites among the three groups. The top 15 most different metabolites were LacCer (d19:0/12:1), PC (O-22:2/14:1), TG (16:0/22:6/18:1), PC (14:0/18:4), Cer (14:0/21:0), PC (16:1/22:2), PE (P-16:0/18:2), PC (P-18:0/18:1), PC (22:5/38:0), Sph (d16:0), DG (16:1/20:3/20:0), PE (16:1/22:6), lyso-PE (20:0/0:0), PC (16:0/18:3), and TG (16:0/18:0/18:0). Among the differently altered metabolites, more than 65% were phospholipids. Therefore, based on the phospholipid metabolism abnormalities, the group heatmap is shown in Figure 6D. In comparison with the HFD/CUMS group, the puerarin-modified phospholipid biomarkers were identified as PC (15:1/20:1), PE (15:1/16:1), and PI (18:2/20:1).

The docking results revealed that puerarin exhibited moderate binding affinity to cPLA2 and weak binding affinity to COX-2 (Figure 7A). The ligand bonded to cPLA2 and COX-2 mainly through hydrogen bonding and hydrophobic interactions. As illustrated in Figure 7B, puerarin formed several hydrogen bonds with Ser530, Tyr385, and Phe518 of COX-2, as well as a number of hydrophobic interactions with Val349, Leu352, and Met522. Puerarin formed hydrogen bonds with Gly29, Cys44, Asp48, and His47 of cPLA2 and hydrophobic interactions with Val30, Tyr21, His6, and Ala17.

To confirm the results of molecular docking, the effect of puerarin on enzyme activities of cPLA2 and COX-2 was evaluated. As shown in Figures 7C–E, HFD, CUMS, and HFD/CUMS stimulation significantly increased the enzyme activities of cPLA2 and COX-2, resulting in enhanced PGE₂ concentration in hippocampal and small intestinal tissues (p < 0.01). In comparison with the HFD/CUMS group, simvastatin, simvastatin + fluoxetine, and puerarin (30, 60, and 120 mg/kg) treatments significantly decreased the enzyme activity of cPLA2 (p < 0.05 or p < 0.01). Moreover, all drug-treated groups, including fluoxetine, simvastatin, simvastatin + fluoxetine, and puerarin (30, 60, and 120 mg/kg), showed weaker enzyme activities of COX-2 and a lower PGE₂ level than those in the HFD/CUMS group (p < 0.01).