2.1. Preparation of Total Water–Soluble Flavonoids

Total water–soluble flavonoids were separated from L. litseifolius (Hance) Chun (sweet tea) via ultrasonic-assisted water extraction according to our previous method [24]. Briefly, a fine powder of L. litseifolius (Hance) Chun (sweet tea) was obtained after pulverization and sieving and mixed with cold deionized water for the extraction of total water–soluble flavonoids at a liquid/solid ratio of 33:1 mL/g, a temperature of 61°C, a microwave power of 634 W, and a microwave time of 165 s. The extracts were filtered through a 0.45 μm filter with a vacuum pump after centrifugation at 5000 rpm for total water–soluble flavonoids. After content detection, total water–soluble flavonoids were concentrated to the desired concentration with a rotary evaporator for the following experiments.

2.2. Development of the GDM Mouse Model and Intervention

The animal experiments were approved by the Animal Ethical and Welfare Committee of Huaihua University (Huaihua: 2022 (A03028)). C57BL/6J mice were purchased from Changsha Tianqin Biotechnology Co., Ltd., and housed in a clean room with free access to food and water on a 12 h light/12 h dark cycle. The female mice (~9 to ~10 weeks of age and ~20 g of body weight) were mated with male mice (~11 to 14 weeks of age) at a 2:1 housing ratio to induce pregnancy. Vaginal plug detection was considered to indicate successful pregnancy, and the date of plug detection was defined as gestation day (GD) 1 [25]. On GD 1, pregnant mice were divided into three groups, namely, the normal control (NC) group (n = 10), GDM model (MD) group (n =10), and GDM model group treated with total water–soluble flavonoids from L. litseifolius (Hance) Chun (sweet tea) (LLHC) group (n = 10). As shown in Figure 1(a), pregnant mice in the MD and LLHC groups were fed a HFD, pregnant mice in the NC group were fed a normal diet beginning, and all the mice were weighed during pregnancy. On GDs 6–8, pregnant mice in the MD and LLHC groups were intraperitoneally injected with STZ (1% solution) at a dosage of 20 mg/kg/d (cat No. S8050, Solarbio Life Sciences, Beijing, China), and pregnant mice in the NC group were intraperitoneally injected with the same volume of normal saline. On GDs 10–16, pregnant mice in the MD and LLHC groups were intragastrically administered total water–soluble flavonoids at a dosage of 100 mg/kg/d. At GD 9 and GD 17 and at GD 10 and GD 18, the IPGTT and IPITT were carried out to monitor glucose homeostasis in all the mice. On GD 19, the experiment was terminated, and all the mice were anaesthetized and euthanized for anatomical observation and sample collection. The litter sizes of fetuses and corresponding placentas were counted and weighed. The weights of the heart, liver, spleen, lung, and kidney were also investigated to calculate the visceral indices.

2.3. IPGTT and IPITT

Glucose tolerance and insulin resistance were determined via the IPGTT and the IPITT, respectively. Half of the pregnant mice in each group were fasted and given free access to water for 12 h, and were then intraperitoneally injected with 20% glucose at a dose of 2 g/kg for the IPGTT (n =55). Another half of the pregnant mice in each group were fasted and given free access to water for 6 h and were then intraperitoneally injected with 7.5% U insulin at a dose of 0.75 U/kg for the IPITT (n =5). Blood samples from the tai vein were collected and directly detected via a glucometer (cat. No. GM501Air, Sinocare Inc., Changsha, China) at 0, 15, 30, 60, and 120 min after injection.

2.4. Histochemical Evaluation

The liver, an important organ of glucose metabolism, was collected for tissue sectioning, staining, and microexamination according to our previous study [25] and the literature [26]. Briefly, the livers were fixed in 4% PFA, dehydrated in a gradient series of ethanol, embedded in paraffin, sectioned into 5 μm thick tissue, and attached to the surface of microscope slides. After baking, deparaffinization, and rehydration, the sections were subjected to hematoxylin–eosin (H&E) staining or terminal deoxyribonucleotidyl transferase (TDT)–mediated dUTP–digoxigenin nick end labelling (TUNEL) staining, followed by dehydration with a gradient series of ethanol and clearing in xylene. Then, the mounted slides were visualized and imaged through a light microscope (BX53, Olympus), and the acquired photomicrographs were analysed via Image-Pro Plus 6.0 software (Media Cybernetics).

The liver sections were evaluated by 3 pathologists from Suzhou Cancer Cell Laboratory Co., Ltd. and Huaihua University. Five different fields of microscopic view were measured for the number of inflammatory cells, ballooning area, and steatosis area of each slice. The ratio of the average area was calculated using the pathological area divided by the total measured area.

2.5. Analysis of Network Pharmacology

According to the literature and our previous study [24], the main active ingredients of total water–soluble flavonoids used were as candidate components on the basis of three indicators: content ≥ 0.5%, oral bioavailability (OB) ≥ 30%, and drug likeness (DL) ≥ 1.5.

The canonical simplified molecular input line entry system (SMILES) of these active ingredients was obtained through a previous study and from the PubChem database (https://pubchem.ncbi. http://nlm.nih.gov/), which were uploaded to the Swiss Target Prediction (http://targetnet. http://scbdd.com/) to predict the targets of the components. The disease targets of GDM were obtained from the GeneCards (https://www.genecards.org/) and OMIM (https://omim.org/) databases. The targets with relevance scores greater than 10 were selected. The communicative targets between the active ingredients and GDM were screened through the Venny 2.1 platform (https://bioinfogp.cnb.csic.es/tools/venny/index.html) for further analysis.

The interaction network of active ingredients with disease targets was constructed via the Network Analyser of Cytoscape 3.10.1 software, in which the nodes indicated active ingredients or disease targets, and the edges indicated the interaction between any two. The protein interactions of communicative targets involved in the prevention and treatment of GDM were searched through the interaction database platform String 12.0 (https://cn.string-db.org/). The results of the String 12.0 analysis were also visualized with Cytoscape 3.10.1 software and were screened by clustering.

The molecular interactions between the active ingredients and core targets were confirmed by AutoDock and PyMOL software. Briefly, the 3D structures of the active ingredients and core targets were downloaded from PubChem and PDB databases and were transformed as pdbqt files, following by the removal of water molecules, solvent molecules, and ions by PyMOL software. The pdbqt files were imported into AutoDock software to deduce the binding model, binding atom, and minimum binding energy. A schematic diagram of the interaction between the active ingredients and core targets was constructed and visualized by AutoDock and PyMOL software.

The signalling pathways of the main anti-GDM components were analysed and visualized by Gene Ontology (GO) enrichment and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment in the Metascape platform (http://Metascape.org/gp/index.html) and the Cytoscape plugin ClueGO, with p < 0.05 and kappa scores ≥ 0.6.

2.6. Statistical Analysis

Statistical analyses were performed via one-way ANOVA with a post hoc multiple comparison (Bonferroni) test after normal distribution tests, which were performed via the Statistical Package for Social Science (SPSS, Version 19.0; SPSS Inc, Chicago, IL, United States). The final data are shown as the mean ± standard deviation (SD), with differences considered significant at p < 0.05 or p < 0.01.

3.1. Oral Administration of Total Water–Soluble Flavonoids Alleviated Glucose Intolerance and Insulin Resistance in GDM Patients

The increase in mouse body weight in the LLHC group was significantly lower than that in the MD group and tended to increase in the NC group (Figure 1(b)). These data suggested that the oral administration of total water–soluble flavonoids was beneficial for controlling excessive increases in body weight.

Before intervention with total water–soluble flavonoids, both glucose tolerance and insulin sensitivity were impaired in the LLHC and MD groups (Figure 1(c) and 1(d)) at GDs 9 and 10. However, they were regular in the NC group. These data demonstrated the successful development of the GDM mouse model.

Compared with those in the MD group, after intervention with total water–soluble flavonoids from GDs 10–16, both glucose tolerance and insulin sensitivity were partly rescued in the LLHC group, but the glucose tolerance and insulin sensitivity were lower than those in the NC group (Figure 1(e) and 1(f)) at GDs 17 and 18. These data revealed that total water–soluble flavonoids stabilized blood glucose when applied to GDM mice.

Compared with those in the NC group, the litter sizes of fetuses in both the MD and LLHC groups were significantly decreased. However, there were no significant differences in the litter size of foetuses between the MD and LLHC groups (Figure 1(g)).

Foetal and placental weights were also investigated for calculation of placental efficiency (foetal weight/placental weight), which revealed that the placental efficiency in the MD group was the highest among those in the NC, MD, and LLHC groups. These parameters were greater in the LLHC group than in the NC group (Figure 1(h)). These data indicated that high placental efficiency was induced when GDM occurred and was partly inhibited by the oral administration of total water–soluble flavonoids.

Routine blood tests revealed that the RBC, WBC, and PLT counts were not significantly different among the NC, MD, and LLHC groups (Figures 1(i), 1(j), and 1(k)). These data suggested that the oral administration of total water–soluble flavonoids did not induce systemic blood changes.

3.2. Oral Administration of Total Water–Soluble Flavonoids Inhibited Liver Cell Apoptosis to Improve Glucose Homeostasis in GDM Mice

The internal organs were weighed and compared. There were no significant differences in heart, spleen, lung, or kidney weights among the NC, MD, and LLHC groups (Figures 2(a), 2(c), 2(d), and 2(e)). However, liver weight was significantly greater in the MD group than in the NC group. There was no significant difference in liver weight between the LLHC and NC groups, but liver weight was significantly lower in the LLHC group than in the MD group. Moreover, the liver indices (liver weight/body weight) were significantly lower in the MD group than in the NC group. The liver indices of the LLHC- and NC-treated mice were not significantly different, but the liver indices of the LLHC-treated mice were significantly greater than those of the MD-treated mice (Figure 2(b) and 2(f)).

H&E staining of liver tissue sections from the LLHC group revealed that the number of inflammatory cells was significantly lower than that from the MD group but greater than that from the NC group (Figures 2(g) and 2(k)). There were many small holes, which are the indicators of adipose degeneration, in the liver sections from the MD group but few holes in those from the LLHC group and fewer small holes in those from the NC group. After detection and statistical analysis, the ratios of steatosis area and ballooning area in the liver sections from the LLHC group were significantly lower than those in the liver section from the MD group but greater than that from the NC group (Figures 2(h), 2(i), and 2(k)).

TUNEL staining of the liver tissue sections revealed that the apoptotic cells were red. The number of apoptotic cells was significantly greater in the liver cell sections from the MD group than in those from the NC group. The number of apoptotic cells was also significantly greater in the liver cell sections from the LLHC group than in those from the NC group but was significantly lower than that in the liver cell sections from the MD group. The ratio of apoptotic cells (the number of apoptotic cells/the number of normal cells) was greatest in the MD group and greatest in the LLHC group among the NC, MD, and LLHC groups (Figure 2(j) and 2(k)).

These data suggested that the oral administration of total water–soluble flavonoids partially inhibited the apoptosis of liver cells and rescued liver glucose metabolism and homeostasis.

3.3. Multiple Targets and Signalling Pathways Were Involved in the Process by Which Total Water–Soluble Flavonoids Protect Against GDM

To further investigate the potential mechanism of total water–soluble flavonoids against GDM, a total of 1204 GDM targets were identified from the GeneCards and OMIM databases. A total of 215 component targets were identified from the Swiss Target Prediction system. A total of 75 communicative targets associated with both GDM and total water–soluble flavonoids were identified (Figure 3(a)). All of these communicative targets were related to both the pharmacodynamic substances of total water–soluble flavonoids and the disease occurrence of GDM, including 85 nodes and 598 edges (Figure 3(b)). Analysis of protein interactions revealed 72 nodes and 292 edges with an average node degree of 15.9 and an average local clustering coefficient of 0.628 (Figure 3(c)). The communicative targets with the Top 20 scores (2.69e + 08), ranked by the maximal clique centrality (MCC) method, were selected for further analysis. The results revealed 20 nodes and 169 edges. These genes were AKT1, STAT3, EGFR, SRC, TNF, MMP9, PTGS2, HSP90AA1, ESR1, IGF1R, PARP1, GSK3B, MAPK14, MMP2, HRAS, AR, KDR, ESR2, IL2, and MMP1 in descending order (Figure 3(d)).

These eight targets, including INSR, IGF1R, EGFR, AKT1, GSK3β, MMP9, MMP2, and GAA, were considered closely associated with total water–soluble flavonoids against GDM and were selected to investigate their molecular interactions. Seven ingredients were used as ligands for molecular docking. The binding energies for molecular docking were recorded (Figure 4(a)). When the molecular docking binding energy is < 0, the two molecules have spontaneous binding ability; when the molecular docking binding energy is < −1.2 kcal/mol (−5.0 kJ/mol), the two molecules are well bound [27]. On the basis of these models, the best interaction sites and action sites between the core targets and active ingredients were recorded, including INSR with luteolin-7-O-β-glucoside (Figure 4(b)), IGF1R with quercetin (Figure 4(c)), EGFR with quercetin (Figure 4(d)), AKT1 with luteolin-7-O-β-glucoside (Figure 4(e)), GSK3β with quercetin (Figure 4(f)), MMP2 with luteolin (Figure 4(g)), MMP9 with luteolin (Figure 4(h)), and GAA with quercetin (Figure 4(i)). Therefore, quercetin, luteolin, and luteolin-7-O-β-glucoside had the greatest effects on the core targets of GDM.

3.4. Total Water–Soluble Flavonoids Improve Glucose Homeostasis in GDM Through Six Main Signalling Pathways Around the AKT1 Core Axis

All of these important targets were involved in 30 signalling pathways. These signalling pathways, including the apoptotic process signalling pathway, insulin receptor signalling pathway, steroid binding pathway, and insulin receptor substrate binding pathway, may be directly involved in GDM occurrence and outcome. These signalling pathways were considered upstream regulatory signals and included signal transduction, gene expression, protein kinase B signalling, the MAPK cascade, MAP kinase activity, phosphatidylinositol 3-kinase complex, and nitric oxide synthase regulation activity (Figure 5(a)). These signalling pathways were associated with 20 kinds of pathogenesis, in which endocrine resistance, the estrogen signalling pathway, the VEGF signalling pathway, and the AGE-RAGE signalling pathway in diabetic complications were directly or indirectly involved in GDM occurrence and outcome (Figure 5(b)).

As shown in Figure 5(c), the binding of Ins to INSR and the binding of IGF1 to IGFR act on IRS-1 to activate PI3K. Moreover, TNFα binds to TNFR to act on JNK1 to inhibit IRS-1, and IL2 binds to IL2R to act on STAT3 and SOS3 to inhibit IRS-1. There is a balance between Ins and TNF signalling and between IBS and IL2 signalling. EGF binding to EGFR also acts on PI3K, and E2 binding to the ER acts on SRC for PI3K activation. These signals of PI3K activation are transferred to AKT1 through the second messengers PIP3 and PKD1. AKT1 signalling blocks the inhibitory effect of GSK3β on GS, and activated GS enables glycogen synthesis to reduce the production of endogenous glucose and alleviate insulin resistance. AKT1 signalling also inhibits GLUT through blocking the inhibitory effects of AS160 and GAP on Rab5 to reduce the intake of exogenous glucose. Moreover, AKT1 signal stops the apoptotic signal in liver cells. All of the above signals benefit the cell cycle progression and survival of liver cells. After GSK3β is inhibited, NFATc1 dissociates from C/EBP and reacts with STAT to increase the transcription and translation of glycometabolism genes. AKT1 acts on CREB and TORC2 through CBP/p3000 to inhibit the transcription and translation of blood glucose-related genes. However, HSP90 acts on eNOS to promote NO production, which reduces liver cell fibrosis. In the process by which total water–soluble flavonoids improve glucose homeostasis, quercetin (acting on IGF1R, EGFR, GSK3B, and GAA), luteolin-7-O-β-glucoside (acting on INSR and AKT1), and luteolin (acting on MMP2 and MMP9) may act as medicinal substances.