A total of 65 specific pathogen-free (SPF) grade female Sprague-Dawley (SD) rats, aged 3 weeks and weighing between 52 and 66 g, were purchased and housed at the Animal Experiment Center of China Three Gorges University (Animal Certificate No. 42010200008980, No. 42010200009387). The rats had free access to water and food and were maintained under a 12-hour light/dark cycle. After a one-week acclimation period without intervention, the rats were used to establish a PCOS model. The study involving experimental animals have been conducted in accordance with the ARRIVE guidelines, the U.K. Animals (Scientific Procedures) Act, 1986 and associated guidelines, EU Directive 2010/63/EU for animal experiments, and the National Research Council's Guide for the Care and Use of Laboratory Animals. Ethical approval for this research was granted by the Animal Welfare Ethics Committee of China Three Gorges University (Approval Number: 2020080X).

Establishment of PCOS Rat Model: The model (PCOS) group of rats received continuous subcutaneous injections on the back of the neck of DHEA (0.06 mg/(g*d); Aladdin Industrial Corporation, Shanghai) and sesame oil (0.1 ml/rat; Shanghai Macklin Biochemical Technology Co., Ltd.), once daily for 24 consecutive days, along with a high-fat diet (consisting of 56% basic feed, 12% lard, 2% soybean oil, 5% egg yolk powder, 20% sugar, 3% milk powder, 1.5% cholesterol, 0.5% sodium deoxycholate; Baituo Experimental Feed Business Department, Hongshan District, Wuhan). The control (Control) group received the same dose of sesame oil by subcutaneous injection on the back of the neck, once daily for 24 consecutive days, and were given a regular diet. Daily vaginal smears were used to monitor the estrous cycle, and hematoxylin and eosin (HE) staining was used to observe morphological changes in the ovaries. Fasting blood glucose, insulin, and sex hormone levels were assessed to evaluate the success of the PCOS model establishment.

After successful modeling, the rats were further divided into three groups: the model (PCOS) group, the model plus metformin (PCOS + Met) group, and the model plus QGW (PCOS + QGW) group. That is, the PCOS group received intragastric injection of physiological saline, the PCOS + Met group received intragastric injection of metformin hydrochloride (500 mg/kg/d, Merck Pharmaceuticals), which the dose of metformin is the same as the dose used in PCOS model in rats in other reported studies (24), and the PCOS + QGW group received intragastric administration of QGW prescription, which originates from the Qing Dynasty's “Yi Fang Ji Jie”. Containing 12 g of Pinellia ternata (Thunb.) Makino; Batch Number 240101, 8 g of Atractylodes lancea (Thunb.) DC.; Batch Number 241001, 16 g of Cyperus rotundus L.; Batch Number 240402, 12 g of Poria cocos (Schw.) Wolf.; Batch Number 240801, 8 g of Massa medicata fermentata; Batch Number 231001, 8 g of Citrus reticulata Blanco; Batch Number 231001), and 8 g of Conioselinum anthriscoides ‘Chuanxiong’. Batch Number 240301; these herbal granules were purchased from China National Medicines Corporation Zhonglian Pharmaceutical Co., Ltd. Prepared according to the aforementioned ratio, a single dose of QGW granules weighs 12.66 g, and a 70 kg adult takes this amount daily. To translate the daily human dose to a rat's dose, Rat dose = 6.3 × Human dose/Human weight × Rat weight. The low dose of QGW = 6.3 × (12.66 g/70 kg) = 1.1394 g/kg, with the doses in the ratio of 1:2:4 for low, medium, and high doses, respectively. Based on the results of preliminary experiments, both the medium dose and high dose of QGW were able to significantly reduce fasting insulin and Home-IR levels. We selected the high dosage for intragastric administration, with equal volumes for each group, administered once daily, divided into two administrations in the morning and evening, for 10 consecutive days.

The rats were anesthetized with an intraperitoneal injection of 3% pentobarbital sodium (100 mg/kg), and blood was collected from the abdominal aorta. The blood was then centrifuged (1,500 rpm/min) for 15 min to collect the serum, which was stored at −80 °C for subsequent analysis of fasting blood glucose, fasting insulin, follicle-stimulating hormone (FSH), luteinizing hormone (LH), and testosterone. The ovaries and uterus were removed, and the left ovary and a portion of the uterus were fixed with 4% paraformaldehyde, followed by dehydration, embedding, and sectioning for subsequent HE staining and immunohistochemistry. A portion of the uterus was fixed with electron microscopy fluid (G1102, Saiwei'er) for scanning electron microscopy experiments. The remaining fresh ovaries and uterus were stored at −80 °C for gene and protein expression analysis. After completing the experiments, rats were euthanized by cervical dislocation following re-anesthesia using the same method.

Rat Hoxa11 RNA interference fragments (si-Hoxa11) were synthesized by GenePharma (Shanghai Jema Company, China), with the sense strand sequence GCCGUCUCGUCCAAUUUCUTT and the antisense strand sequence AGAAAUUGGACGAGACGGCTT. Fifteen PCOS model rats were selected, anesthetized, and a mixture of animal transfection reagent (Beijing InGen Biotechnology Co., Ltd.) and siRNA was injected into the uterine cavity. Among them, five rats were injected with si-NC on both sides of the uterine cavity, and the other ten were injected with si-Hoxa11 on both sides. They were then divided into three groups: si-NC control (PCOS + si-NC) group: intervened with physiological saline; si-Hoxa11 interference (PCOS + si-Hoxa11) group: intervened with physiological saline; si-Hoxa11 interference followed by QGW intervention (PCOS + si-Hoxa11 + QGW) group: intervened with QGW. Each group received equal volumes, once daily for 7 consecutive days. The rats were euthanized after being anesthetized with an intraperitoneal injection of 3% pentobarbital sodium, and the ovaries and uterus were removed and stored at −80 °C for gene and protein expression analysis.

The stages of the rat estrous cycle were determined based on vaginal cytology. Vaginal smears were collected between 9 and 10 a.m. each day. Vaginal epithelial cells were collected by flushing with physiological saline, fixed on slides, and stained with HE. The stages of the rat estrous cycle are divided into: Proestrus: mainly nucleated epithelial cells; Estrus: almost all anucleated cornified cells, which may be clustered; Metestrus: a large number of leukocytes and a few anucleated cornified cells; Diestrus: leukocytes, nucleated cells, and anucleated cornified cells coexist, with leukocytes predominating.

The ovarian tissue was fixed with 4% paraformaldehyde for 24 h, dehydrated with different concentrations of alcohol, embedded in paraffin, and sectioned. The largest cross-section of the ovary was taken and stained with HE to count the number of cystic follicles and the number of granulosa cell layers.

ELISA kits (Jiangsu Enzyme Mark Biotechnology Co., Ltd.) were used to detect fasting blood glucose fasting insulin, FSH, LH, and testosterone levels in the serum of rats in each group. All steps of the ELISA kit were strictly followed according to the instructions, and the absorbance was measured at a wavelength of 450 nanometers using an enzyme-linked immunoassay reader (Thermo, China) to calculate the serum hormone content.

Total RNA from the uterus was extracted using the Trizol method. A cDNA synthesis chain kit (Hunan Aikorui Bioengineering Co., Ltd.) was used to remove DNA from RNA and reverse transcribe RNA into cDNA. A qRT-PCR kit (Hunan Aikorui Bioengineering Co., Ltd.) was employed for real-time fluorescence quantitative PCR (qRT-PCR). The qRT-PCR results were calculated using the ΔΔCt method, and the expression of target gene mRNA was expressed using the 2^-ΔΔCt method. Gapdh was used as the reference standard, and the relevant primers are listed in Table 1.

Total proteins were extracted from uterine tissue using RIPA lysis buffer (G2002-100ML, Servicebio) and phosphatase inhibitor (G2007-1ML, Servicebio). Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was performed with varying concentrations based on the molecular weight of the proteins. Proteins were separated by electrophoresis and transferred onto polyvinylidene fluoride (PVDF) membranes (G6015-0.45, Servicebio). The PVDF membranes were immersed in a primary antibody solution and incubated overnight at 4 °C, with primary antibodies including β-actin (1:1250, AC026, ABclonal), Hoxa10 (1:1250, A8550, ABclonal), Hoxa11 (1:1250, A2976, ABclonal), Glut4 (1:1250, A7637, ABclonal), Igf1 (1:1250, A24744, ABclonal), Insulin-like growth factor binding protein-1(Igfbp1)(1:1250, A11109, ABclonal), Il6 (1:1250, A21264, ABclonal), Itgβ3 (1:1250, A19073, ABclonal), Lif (1:1250, A1288, ABclonal). After incubation with the secondary antibody (1:4000, AS014, ABclonal) at room temperature, signals were detected on a ChemiScope5300 (Shanghai Qinxiang Science) using ECL reagent (G2161-200ML, Servicebio), and the grayscale analysis was performed using Image J.

The uterus was fixed with 4% paraformaldehyde, dehydrated, embedded, and sectioned. Sections were deparaffinized with xylene and soaked in alcohol of varying concentrations. Antigen retrieval was performed using citrate retrieval solution (pH 6.0). Sections were soaked in a 3% hydrogen peroxide solution for 20 min at room temperature to block endogenous peroxidase activity, and then covered and sealed with 10% goat serum for 30 min. Primary antibodies included progesterone receptor(PR) (25871-1-AP, PTG) and estrogen receptor(ER) (18309-1-AP, PTG). Sections were incubated with the antibodies overnight at 4 °C, followed by incubation with the secondary antibody (PN0046, PinoFly Bio) for 1 h at room temperature. The sections were developed with DAB, counterstained with hematoxylin, dehydrated, and mounted. Under a microscope, six random fields were selected from each section, and the average optical density was calculated using Image.

The uterus was fixed with electron microscopy fluid at room temperature for 2 h and then transferred to 4 °C for storage. After fixation, it was post-fixed with 0.1M phosphate buffer PB (pH 7.4, G0002, Saiwei'er) and osmium tetroxide (18456, Ted Pella Inc). It was dehydrated with ethanol of varying concentrations, dried using a critical point dryer (K850, Quorum), and then subjected to sample conduction treatment with an ion sputter coater (MC1000, HITACHI). The samples were observed and imaged under a scanning electron microscope (SU8100, HITACHI). Three random fields of view (10.0 μm) were selected for each sample, and then mature pinocytic protrusions were manually counted using Image J software under blind conditions.

Retrieve drugs from the TCMSP database (http://tcmspw.com/tcmsp.php), including Pinellia ternata (Thunb.), Atractylodes lancea (Thunb.), Cyperus rotundus, Poria cocos, Tangerine peel, Ligusticum chuanxiong Hort, and Massa Medicata Fermentata. The screening criteria for effective components of traditional Chinese medicine are oral bioavailability (OB) ≥30% and drug-likeness (DL) ≥0.18. The active components of Massa Medicata Fermentata are obtained through reference literature with PMIDs: 38299285 and 36527339. Potential target genes of the drugs are obtained from the TCMSP, DrugBank, and Swiss TargetPrediction databases (http://www.swisstargetprediction.ch/). Using |log2FC| ≥ 1 and corrected p < 0.05 as thresholds, download the GSE137354 dataset from the GEO database (platform GPL17586), which includes gene expression profiles of endometrium from 3 PCOS patients and 3 normal controls. Perform differential gene analysis of GSE137354 using R language limma with ggplot2 (3.3.6) and plot a volcano plot of differential genes (17). Convert drug target proteins to target gene IDs to obtain target genes, denoted as Qi Gong; then, denote the differential genes (DEGs) of the GSE137354 dataset as PCOS with Endometrial Disorder (PCOS-EM), and create a Venn diagram of the intersection to identify common targets. Import the intersecting target genes into the STRING database (18), select the species as “Homo sapiens,” and perform PPI analysis on the intersecting target genes. Further use the Degree algorithm in the cytoHubba plugin of Cytoscape to screen the intersecting target genes to obtain Hub genes. Search for corresponding active ingredients and drugs for the genes. To explore the biological pathways in which hub genes are involved in the disease, use the R package clusterprofile version 3.18.1 to perform single-gene pathway enrichment analysis on key genes, with a threshold of p-value < 0.05 and FDR < 0.25 as the criteria for selecting enriched pathways (19).

Download the 3D structure of gene proteins from UniProt (https://www.uniprot.org/) and the 3D structure of active ingredients from PubChem (https://pubchem.ncbi.nlm.nih.gov/). Next, preprocess the protein receptor molecules: use AutoDock to remove water (20), ligands, cofactors, and ions that are not needed for docking from the predicted protein structures, add hydrogen, and finally obtain the 3D structure of the protein for subsequent molecular docking. Preprocess the ligand small molecules: first, convert the downloaded 2D structure SDF files to pdb format using Open Babel (21), then use the pybel module to convert the 2D structure of the small molecules to 3D structure; next, use AutoDock to add hydrogen and set it as a ligand. First, perform blind docking using AutoDock Vina (reason: blind docking is performed when there is no information about the catalytic site/binding site in the protein;), where the preprocessed 3D structures of the ligand and receptor molecules and the configuration file for all docking parameters are input. Finally, use PyMOL to visualize the results with the lowest binding energy (22).

To ascertain the impact of Qigong Pill on the endometrial receptivity in PCOS, a network pharmacology analysis was conducted. Based on databases such as TCMSP/DrugBank and Swiss TargetPrediction, a total of 103 active components and 1527 protein targets were retrieved for QGW (Supplementary Table S1). Based on the GSE137354 dataset, 1,690 DEGs were identified in PCOS patients, of which 1,084 were upregulated and 606 were downregulated, as shown in the volcano plot (Figure 1A). The intersection of 161 target genes corresponding to the drug and 1,690 disease genes resulted in 15 potential therapeutic targets for drug-disease (Figure 1B). A PPI analysis was performed on the intersecting target genes, and a PPI network diagram of the 15 intersecting genes was constructed (Figure 1C). After screening the intersecting target genes and applying the Degree algorithm, five Hub genes were ultimately obtained: HOXA10, HOXA11, IGFBP1, CRP, and MPO (Figure 1D). Single-gene GO analysis indicated that HOXA10 and HOXA11 are involved in uterine development and embryonic development, while IGFBP1 is involved in the insulin receptor signaling pathway and positive regulation of cell growth. CRP is associated with innate immunity, inflammatory response, and acute inflammatory response. MPO is related to mechanical injury, fungal defense, and oxidative stress (Supplementary Figure S1).

The estrous cycle in the Control group showed regular changes, with one estrous cycle every 4 days, while the PCOS group had no cyclic changes and was arrested in the metestrus phase (Figure 2A). Serum testosterone, fasting insulin, and Homeostasis Model Assessment of Insulin Resistance (HOMA-IR) in the PCOS group were all increased compared to the Control group, with P < 0.01, which was statistically significant (Table 2). The number of cystic follicles in the PCOS group was increased compared to the Control group, and the granulosa cell layer was reduced, as seen in Figures 2B,C.

After treatment with QGW (PCOS + QGW group), the estrous cycle in PCOS rats became more regular compared to the PCOS group (Figure 2A), and LH and serum testosterone (T) were significantly reduced, both with statistical significance (P < 0.01) (Table 2). After treatment with QGW, the number of cystic follicles in the ovaries of PCOS rats decreased (P < 0.01), and the number of granulosa cell layers increased (Figures 2B,C). PCOS model rats treated with metformin (PCOS + Met group) showed a decrease in LH/FSH and testosterone levels compared to the PCOS group, both with statistical significance (P < 0.01). In addition, both QGW and metformin were able to reduce the FSH level in PCOS rats (Table 2). Metformin was also able to reduce the number of cystic follicles and increase the number of granulosa cell layers in PCOS model rats (P < 0.01) (Figures 2B,C).

After treatment with QGW (PCOS + QGW group), there was a significant reduction in fasting insulin, fasting blood glucose, and HOMA-IR in PCOS rats compared to the model group (PCOS group), with P < 0.01 indicating statistical significance. QGW demonstrated a similar effect to metformin, both being able to reduce fasting insulin, fasting blood glucose, and HOMA-IR in the PCOS model mice (Table 2). However, the reduction in fasting blood glucose levels by QGW was close to that of the control group, whereas metformin reduced blood glucose levels below those of the normal control group.

Comparison of endometrial receptivity markers between the PCOS group and the control group revealed that the expression levels of Hoxa10, Hoxa11, Glut4, Igf1, Itgβ3, and Lif were decreased, while the expression levels of Il-6 and Igfbp1 were increased as shown by qRT-PCR and immunoblotting results (Figures 3A–R). The number of pinopodes in the endometrium of rats in the PCOS group was significantly reduced compared to the control group (Figures 4E,F). After treatment with QGW (PCOS + QGW group) or metformin (PCOS + Met group), compared to the PCOS group, qRT-PCR and immunoblotting results showed increased expression of Hoxa10, Hoxa11, Glut4, Igf1, Itgβ3, and Lif, and decreased expression levels of Il-6 and Igfbp1.However, QGW showed a more pronounced effect in reducing the expression levels of Il-6 (Figures 3A–R). Immunohistochemical results revealed an increase in PR expression and a decrease in ER expression in the PCOS + QGW group compared to the control group. Metformin treatment also showed similar therapeutic effects (Figures 4A–D). Scanning electron microscopy results indicated that, compared to the PCOS group, both metformin and QGW treatments increased the number of pinopodes in the endometrium of PCOS model rats (Figures 4E,F).

To ascertain if the therapeutic effects of QGW are facilitated through the Hoxa11-Itgβ3 signaling pathway in PCOS rats, we utilized si-Hoxa11 to selectively knock down Hoxa11 in the endometrium of the PCOS model rats, creating the PCOS + si-Hoxa11 group for subsequent comparative analysis. Compared to the PCOS + si-NC group, qRT-PCR and immunoblotting analysis results showed that the RNA and protein expression levels of Hoxa11 in the PCOS + si-Hoxa11 group were both reduced, indicating successful knockdown of the Hoxa11 gene [Figures 5(1)B, G]. Compared to the PCOS + si-NC group, after knocking down Hoxa11 (PCOS + si-HOXA11 group), the RNA and protein expression levels of Hoxa10, Itgβ3, and Igf1 were significantly reduced, while the RNA and protein expression levels of Igfbp1 were significantly increased, with P < 0.01 indicating statistical significance. There were no significant changes in the RNA and protein expression of Glut4 in both groups, with P > 0.05, indicating no statistical significance. Compared to the PCOS + si-Hoxa11 group, the PCOS + si-Hoxa11 + QGW group (after treatment with QGW) showed increased RNA and protein expression of Hoxa10, Hoxa11, Itgβ3, Glut4, and Igf1, and decreased RNA and protein expression levels of Igfbp1, with P < 0.001 indicating statistical significance. In contrast to the PCOS + si-NC group, the PCOS + si-Hoxa11 + QGW group exhibited enhanced RNA and protein expression for Hoxa10, Hoxa11, Glut4, and Igf1, with P < 0.001 signifying statistical significance (Figure 5). Meanwhile, the protein expression levels of Igfbp1 and Itgβ3 remained unaltered, with P > 0.05 suggesting no statistical significance (Figures 5G–N). When compared to the PCOS + si-NC group, the PCOS + si-Hoxa11 + QGW group displayed no significant alterations in the RNA expression of Itgβ3 (P > 0.05), yet there was a marked decrease in the RNA expression of Igfbp1, with P < 0.001 denoting significant biological statistical significance (Figures 5A–F).

QGW has identified a total of 103 active components, among which wogonin, luteolin, lauric acid, methyl palmitate, quercetin, and gamma-aminobutyric acid were found to act on the genes HOXA10, HOXA11, IGFBP1, and IL-6 (Supplementary Table S2). HOXA10, HOXA11, and IGFBP1 were docked with wogonin, while IL-6 was docked with luteolin. The results of molecular docking are represented by the binding energy, which indicates the binding activity between the receptor and ligand. A binding energy of less than 0 KJ/mol indicates that the ligand and receptor have the ability to spontaneously bind. The smaller the binding energy, the higher the affinity between the receptor and ligand, and the greater the likelihood of interaction. A binding energy of less than −1.2 kcal/mol (−5 KJ/mol, 1 Kcal = 4.184 KJ) indicates high binding activity, while a binding energy of less than −1.7 Kcal/mol (−7.0 KJ/mol) indicates very strong binding activity. The molecular docking results (Table 3; Figure 6) show that there is a certain degree of interactive force between the targets and their corresponding active ingredients, forming multiple hydrogen bonds, which are important forces in the interaction. In summary, the docking binding free energy of less than −5 indicates that HOXA10, HOXA11, IGFBP1, and IL-6 have good binding capabilities with their corresponding active ingredients.