Two novel phenolic glycosides from Saxifraga tangutica for NASH treatment: In vitro, network pharmacology, and molecular docking analysis

2.1. Phytochemical instruments and reagents

All the samples were analyzed on an Essentia LC–16 instrument (Shimadzu Enterprise Management Co., Ltd., Suzhou, China). Sample pretreatment and isolation were performed on a Newstyle® laboratory liquid chromatography system (Hanbon Science & Technology Co., Ltd., Huaian, Jiangsu, China). Electrospray ionization-mass spectrometry (ESI–MS) data were recorded with a Waters QDa electrospray ionization mass spectrometer from Waters Instruments Co., Milford, Massachusetts, USA. HR–ESI–MS spectra data were acquired on a Q Exactive Orbitrap device (Thermo Fisher Scientific Inc., Waltham, Massachusetts, USA). NMR data were collected with a Bruker Avance 600 MHz instrument (Bruker, Karlsruhe, Germany). The MCI GEL® CHP20P and Spherical C18 used in the sample pretreatment were purchased from Mitsubishi, Japan and SiliCycle, Canada, respectively. ReproSil–Pur C18 AQ analytical and preparative columns (4.6×250 mm, 20×250 mm, 5 μm) were purchased from Maisch Corporation (Germany). Click XION analytical and preparative columns (4.6×250 mm, 20×250 mm, 5 μm) were obtained from ACCHROM Corporation (Beijing, China).

The methanol (MeOH) and acetonitrile (ACN) used in this study were purchased from Kelon Chemical Reagent Factory (Chengdu, Sichuan, China). HPLC–grade H₂O was prepared with a Moore water purification system (Chongqing, China).

2.2. Sample extraction and isolation

The whole plant of S. tangutica was collected from Guoluo Tibetan Autonomous Prefecture (99°47′35″E, 34°33′37″N), Qinghai, China, and identified by Professor Lijuan Mei (Northwest Institute of Plateau Biology, Chinese Academy of Sciences). A voucher specimen (No. 0325734) was deposited at the Qinghai–Tibet Plateau Museum of Biology.

The dried and pulverized whole plants were weighed (500 g) and then extracted with 4 L of MeOH at room temperature. The extraction was repeated three times, each lasting 24 hrs. The combined 12 L MeOH extracts were concentrated under reduced pressure to approximately 300 mL, which was then poured into a container with 200 g of polyamide and dried (weight 265 g after drying, 13% yield). Twenty-two grams of the polyamide mixture was loaded into an empty chromatography tower (49×100 mm) and subjected to pretreatment via MCI GEL® CHP20P medium pressure liquid chromatography (MCI–MPLC). Combination of MeOH and water as a binary pump mobile phase. Chromatographic conditions: 0–120 min, 0–100% MeOH; 120–150 min, 100% MeOH. The flow rate was maintained at 57 mL/min. After the preparation process was repeated 12 times, a total of six fractions were obtained, and the target fraction (labeled Fr3) was concentrated under reduced pressure, yielding 2.66 g (4.09%).

Fr3 was completely dissolved in 30 mL of MeOH/H₂O (30:70 v/v) and passed through a 0.45 μM organic membrane before being processed by Spherical C18 medium-pressure liquid chromatography (MPLC). Combination of MeOH and water as a binary pump mobile phase. Chromatographic conditions: 0–120 min, 10–75% MeOH, with the flow rate set at 70 mL/min. After three cycles of preparation, four fractions were finally recovered, and the target fractions were labeled Fr31 and Fr34.

A total of 187.9 mg (7.06% yield) of Fr31 was dissolved in 1.5 mL of MeOH and filtered through a 0.45 μM organic membrane. Combination of ACN and water as a binary pump mobile phase. The chromatographic conditions were as follows: Click XION preparative column with a gradient elution procedure of 100–75% ACN over 60 min. The flow rate was set at 19 mL/min. The samples were loaded thrice and three fractions were finally recovered, labeled Fr311, Fr312, and Fr313. A Reprosil–Pur C18 AQ preparative column was used to further process Fr312. ACN (7%) and water (93%) as a binary pump mobile phase system was subjected to isocratic elution for 60 min. A total of 150 μL of Fr312 was prepared five times, and ultimately, two fractions were recovered.

A total of 715.2 mg (26.88% yield) of Fr34 was dissolved in 4 mL of MeOH and filtered through a 0.45 μM organic membrane. Water and ACN were used as the mobile phase system. The mixture was then subjected to gradient elution via a Click XION preparative column with ACN (100–65%) over 60 min, with a sample injection volume of 0.8 mL each time. The flow rate was set at 19 mL/min. Five fractions were finally recovered.

The samples were analyzed on a Reprosil–Pur C18 AQ analytical column under the following chromatographic conditions: 0–60 min, 5–28% ACN, and a constant steady flow rate of 1 mL/min. In this study, all chromatographic data were obtained at 254 nm.

2.3. Cell culture

The human normal hepatocyte cell line L02 was purchased from Procell Life Science & Technology Co., Ltd. (Wuhan, Hubei, China). The culture medium consisted of high-glucose DMEM (Procell, Wuhan, Hubei, China) containing 10% fetal bovine serum (Procell, Wuhan, Hubei, China) and 1% penicillin-streptomycin (Servicebio, Wuhan, Hubei, China). The cells were then incubated at 37°C with a humidity level of 5% CO₂. After treatment with the test compounds for 24 h, cell viability was determined with a CCK–8 kit (Vazyme, Nanjing, Jiangsu, China).

For the NASH cell model, L02 cells were seeded into 6–well plates at a density of 2×10⁵ cells per well. A fixed ratio (2:1) of oleic acid (Sigma, St. Louis, Missouri, USA) and palmitic acid (Sigma, St. Louis, Missouri, USA) was prepared as a 20 mM free fatty acid (FFA) solution for subsequent use. Approximately 24 h after the cells were passaged, the culture medium was removed, and fresh medium containing or not containing the drug, along with 600 μM FFA solution, was added to the 6–well plates. The cells were then incubated for an additional 24 h.

2.4. Physiological and biochemical assays

After the cellular NASH model described above was established, the cells were treated with β-D-galactoside-1,6-bis(3,4,5-trihydroxybenzoate) and -(-)-rhododendron-4’-β-D-glucoside for 24 h. After that, the total cholesterol (TC), total triglyceride (TG), alanine transaminase (ALT), and aspartate transaminase (AST) levels were measured in each experimental group via a kit (Beyotime, Shanghai, China; Abbkine, Wuhan, Hubei, China).

2.5. Screening novel phenolic glycoside targets related to NASH

The SDF files and SMILES codes for the small-molecule compounds were generated via ChemBio3D software. Their potential targets were predicted via two online websites: PharmMapper (https://lilab–ecust.cn/pharmmapper/index.html) and Super–PRED (https://prediction.charite.de/index.php). Homo sapiens was chosen as the target organism. All the predicted targets were ultimately integrated to establish separate target databases for β-D-galactoside-1,6-bis(3,4,5-trihydroxybenzoate) and -(-)-rhododendron-4’-β-D-glucoside.

Using “nonalcoholic steatohepatitis” as keywords, the following seven open–source databases were searched for targets related to NASH: GeneCards (https://www.genecards.org/), DisGeNET (https://www.disgenet.org/), DrugBank (https://go.drugbank.com/), CTD (https://ctdbase.org/), Therapeutic Target Database (https://db.idrblab.net/ttd/), OMIM (https://www.omim.org/), and PharmGKB (https://www.pharmgkb.org/). In addition, the mRNA expression profiles of NASH and normal samples (GSE126848) were searched in the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/). The differentially expressed genes (DEGs) were obtained after screening via the GEO2R package analysis, and the thresholds for the identification of DEGs were |logFC|> 1.5 and p<0.05. All targets were subsequently standardized and verified via the UniProt database (https://www.uniprot.org/). Homo sapiens as the species selection.

Through the online website VENNY (https://bioinfogp.cnb.csic.es/tools/venny/), common targets of small-molecule compounds and NASH were identified.

2.6. Screening of hub targets

The common targets were uploaded to the online STRING (https://string–db.org/) platform to construct a protein–protein interaction (PPI) network. Furthermore, Cytoscape v3.9.1 was utilized in the optimization and analysis of the PPI network, while “betweenness”, “closeness”, and “degree” in the CytoNCA plugin were used as metrics to evaluate the central network. The term “degree” refers to the number of connections between nodes in the network, which reflects the interaction information. Finally, the top six targets in terms of degree values were screened as the hub targets.

2.7. Molecular docking

The mol2 files of two novel phenolic glycosides (ligands) were generated and exported via ChemBio3D software. Based on Homo sapiens and high-resolution, the 3D structures of nine target proteins (receptors) were retrieved and downloaded from the RCSB Protein Data Bank (https://www.rcsb.org/). Using AutoDock tools, the ligands and receptors were prepared by removing water, adding hydrogen, and assigning charges. Furthermore, the AutoGrid program was used to configure the docking box (containing the entire protein), which specifies the 3D space for subsequent optimal conformation retrieval. After correctly generating the .gpf and .dpf files, AutoDock was run to simulate ligand–receptor binding, and finally, 100 complex conformations were generated. The optimal conformations were visualized via the PyMOL software.

2.8. Enrichment analysis

The online website DAVID (https://david.ncifcrf.gov/) was used to analyze the Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) annotations of the common targets. These data were further processed and then visualized via the bioinformatics (http://www.bioinformatics.com.cn/) platform. Biological processes (BP), cellular components (CC), and molecular functions (MF) comprised the GO analysis of this experiment.

2.9. Statistical analysis

All the assays were repeated thrice, and the results were expressed as mean ± standard error of mean (SEMs). The GraphPad Prism software was used for data analysis and visualization. One–way ANOVA was performed to determine the significance of the data. Typically, p<0.05 indicated a statistically significant difference.

3.1. Isolation and characterization of the compounds

Given the complexity of natural products, the combination of chromatographic methods with different separation selectivities is increasingly favored by scientists to study their chemical composition [20,21]. In this work, MCI GEL® CHP20P was used in the first pretreatment step of the S. tangutica crude extract, followed by Spherical C18 for further processing of the target fraction Fr3 recovered from MCI–MPLC. This strategy enables the preliminary fractionation of complex samples while removing some non-target components (e.g., pigments and biological macromolecules). The apparatus for MCI–MPLC, as well as the pretreatment chromatogram, has been shown in Figures 1(a) and (b). Figure 1(c) displays the Spherical C18 MPLC pretreatment chromatogram. Comparative analysis of the four fractions recovered by Spherical C18 MPLC and Fr3 revealed that there was no crossover between the samples, which basically covered the major peaks present in Fr3 (Figure 2a-e). For the target fractions Fr31 and Fr34, continued reverse–phase high-pressure preparation was unlikely to achieve high–purity compound separation because of poor peak resolution (Figure 2b and e). Given the excellent selection variability of the reversed–phase and hydrophilic columns, Click XIon was used to perform high–pressure preparations of these two fractions (Figure 2f-i), ultimately recovering eight fractions and identifying seven high–purity compounds through chromatographic analysis, labeled Fr311 (compound 1, 17.2 mg, 9.15% yield), Fr313 (compound 4, 12.3 mg, 6.54% yield), Fr341 (compound 5, 16.6 mg, 2.32% yield), Fr342 (compound 6, 4.5 mg, 0.62% yield), Fr343 (compound 7, 6.2 mg, 0.86% yield), Fr344 (compound 8, 17.2 mg, 2.4% yield), and Fr345 (compound 9, 25.2 mg, 3.52% yield). Further analysis of Fr312 on a Reprosil–Pur C18 AQ column revealed several visually separable peaks (Figure 3a). Consequently, Fr312 was subjected to two-dimensional high-pressure preparation via a Reprosil–Pur C18 AQ column (Figure 3b and c), which ultimately yielded two high–purity compounds, labeled Fr3121 (compound 2, 6.2 mg, 3.29% yield) and Fr3122 (compound 3, 3.1 mg, 1.64% yield). Figure S1 lists the reanalysis chromatograms of these nine compounds.

Close

Figure 1.

(a) The actual MCI GEL® CHP20P MPLC system and (b) S. tangutica crude sample pretreatment chromatogram. (c) Pretreatment chromatogram of Fr3 on Spherical C18 MPLC.

Export to PPT

Close

Figure 2.

Analytical chromatograms of (a) Fr3, (b) Fr31, (c) Fr32, (d) Fr33, and (e) Fr34 on a Reprosil–Pur C18 AQ column. (f) Comparative chromatograms of the analysis and (g) preparation of Fr31 on a Click XION column. (h) Comparative chromatograms of the analysis and (i) preparation of Fr34 on a Click XION column.

Export to PPT

Close

Figure 3.

Analytical chromatograms of (a) Fr312 on a Reprosil–Pur C18 AQ column. (b) Comparative chromatograms of the analysis and (c) preparation of Fr312 on a Click XION column.

Export to PPT

The spectral data of isolated compounds were compared with the literature, and their structures were finally determined: 1 is p-hydroxybenzoic acid-β-D-glucosyl ester [22], 2 is 3’-O-methyl-galloyl-β-D-glucose [23, 24], 3 is di-O-methylcrenatin [25], 4 is β-D-galactoside-1,6-bis(3,4,5-trihydroxybenzoate), 5 is syringic acid [26], 6 is caffeic acid [27], 7 is depside [28], 8 is 4-(4’-hydroxyphenyl)-2-butanone-4’-O-β-D-glucopyranoside [29], and 9 is -(-)-rhododendron-4’-β-D-glucoside. All of these compounds are the first reported in this plant; their structures have been shown in Figure 4, and detailed spectral data can be found in the supplementary material (Figure S2–44).

Close

Figure 4.

Chemical structures of all the isolated compounds.

Export to PPT

Based on the HR–ESI–MS data, the molecular formula of 4 was proposed to be C₂₀H₂₀O₁₄. The IR spectrum shows a broad and strong absorption band at approximately 3400 cm^-1, suggesting the presence of hydroxyl groups. Additionally, the peaks at 1600 cm^-1 (carbonyl groups and aromatic rings) and 1100 cm^-1 (glycosidic fragment) also presented clear absorption bands. [α]²⁵D = -19.40 implies that Fr3–1–3 has optical activity. In the ¹H–nuclear magnetic resonance (NMR) data, the downfield region (δ_H 6.90 and 6.86) indicates the presence of two aromatic rings that each exhibit meta-coupling. The seven proton signals at δ_H 5.80–3.70 suggest a galactoside; its configuration was determined to be β-D based on the coupling constants (8.37 Hz) of the sugar-terminal base proton signals and combined with the literature [30, 31]. In the ¹³C–NMR data, δ_C 164.9 and 164.3 revealed that the compound had two carbonyl carbon signals. The DEPT–135 spectra marked nine primary/tertiary carbons and only one secondary carbon. In the 2D spectrograms, the HSQC correlations of δ_H 6.90/δ_C 108.9 and δ_H 6.86/δ_C 108.7 match the hydrogen protons coupled between the two aromatic ring interstitials with the carbon atom signals, whereas the HSQC information of δ_H 5.80/δ_C 92.1 assigns the sugar-terminal base proton to C-1. The HMBC correlations of δ_C 164.9 with δ_H 6.90 and 5.80 identify the position of the carbonyl carbon (C-7ꞌ), whereas the HMBC correlations of H-2 (δ_H 4.98) with C-1 (δ_C 92.1) and C-3 (δ_C 74.0) were also observed. The ¹H–¹H COSY spectra further obscure the hydrogen proton information between neighboring positions. Finally, combined with other spectral information, 4 was determined to be a novel phenolic glycoside, named β-D-galactoside-1,6-bis(3,4,5-trihydroxybenzoate). The detailed spectrum information has been listed in Table 1.

The HR-ESI-MS data of 9 indicates that its molecular formula is C₂₂H₃₄O₁₂, and the IR data show the characteristic absorption bands of hydroxyl (∼3400 cm^-1) and glycoside fragments (∼1200 cm^-1). [α]²⁵D = -25.57 implies that 9 has optical activity. According to the ¹H-NMR data, the 4 proton signals at δ_H 7.15 and 6.99 indicate that the compound has a para-substituted benzene ring. The peaks at δ_H 2.66, 1.86, and 1.69 in the low-field region indicate a C–C single bond fragment, and δ_H 1.19 was shown to be a methyl signal. The signals at δ_H 4.32 (7.60 Hz) and δ_H 3.89 (7.98 Hz) correspond to the proton signals of the two sugar terminal groups with coupling constants implying that both sugars are in the β-D configuration [32]. Additionally, these peaks at δ_H 3.22–3.87 show 12 proton signals for two glucoses, which implies that 9 is a glycoside derivative of (-)-rhododendron. The DEPT–135 spectra revealed 16 primary/tertiary carbons and 4 secondary carbons. these secondary carbons correspond to a C–C single bond (C–3 and C–4) and the carbon atoms (C–6ꞌꞌ and C–6ꞌꞌꞌ) of two β-D glucose units [33]. In the 2D spectra, the HSQC correlation of δ_H 1.19 with δ_C 20.0, the HMBC correlation of δ_H 1.19 with δ_C 40.4, 75.1 identified the position of the methyl group. The HSQC correlations of δ_H 1.19/δ_C 20.0, δ_H 1.86, 1.69/δ_C 40.4, δ_H 2.66/δ_C 31.8, and δ_H 3.17/δ_C 75.1 assigned the protons at positions 1–4 to their respective carbon atoms. δ_C 102.3 showed an HMBC correlation with δ_H 3.17; The COSY correlations of δ_H 3.17/δ_H 4.32 and δ_H 2.66/δ_H 1.86 were also observed. Ultimately, after searching many studies, 9 was determined to be a novel (-)-rhododendrin derivative, named -(-)-rhododendron-4’-β-D-glucoside. The detailed spectrum information has been listed in Table 2.

3.2. Novel phenolic glycosides mitigate the damage caused by NASH

The anti-NASH activity of these two novel phenolic glycosides was evaluated, and the cell viability assay (Table S1) indicated that both β-D-galactoside-1,6-bis(3,4,5-trihydroxybenzoate) and -(-)-rhododendron-4’-β-D-glucoside exhibited no significant cytotoxicity at concentrations lower than 100 μM (Figure S40). Oil red O staining of the low–dose experimental group (5 μM) revealed that both β-D-galactoside-1,6-bis(3,4,5-trihydroxybenzoate) and -(-)-rhododendron-4’-β-D-glucoside were able to reduce intracellular lipid accumulation and attenuate cellular damage (Figure 5). Compared with the model group Table S2, these two novel phenolic glycosides significantly decreased the intracellular levels of TC, TG, ALT, and AST, with higher drug concentrations (20 μM) resulting in lower levels of these indicators. Notably, β-D-galactoside-1,6-bis(3,4,5-trihydroxybenzoate) was superior to -(-)-rhododendron-4’-β-D-glucoside in anti-NASH efficacy, which may be attributed to the position and number of substituents (phenolic hydroxyl and carbonyl groups).

Close

Figure 5.

Oil red O staining results for in the NASH model after 24 hrs of low–dose (5 μM) drug intervention. TG, TC, AST, and ALT in the low–, medium–, and high–dose groups (5 μM, 10 μM, and 20 μM). The data are presented as the means ± SEMs (n=3). *p<0.05, **p<0.01, ***p<0.001 compared with the Mod group. ###p<0.001 compared with the Con group

Export to PPT

Phenolics have long been a focus of research as biologically active secondary metabolites widely found in TTMs. Silymarin, caffeic acid, and quercetin are well–known natural products with abundant phenolic hydroxyl groups that provide sufficient hydrogen bond donors/acceptors required for hepatoprotective effects, but their poor aqueous solubility and bioavailability impede their deep utilization [34-36]. Glycosylation, a prevalent chemical modification of phenolics, markedly enhances aqueous solubility and can improve target specificity, thereby achieving both functional activity and bioavailability [37, 38]. In this study, both β-D-galactoside-1,6-bis(3,4,5-trihydroxybenzoate) and -(-)-rhododendron-4’-β-D-glucoside exhibited acceptable anti-NASH activity, and the presence of glycosidic fragments suggests that they possess superior aqueous solubility, which could be beneficial for the subsequent drug-like evaluation.

3.3. Hub targets of novel phenolic glycosides and NASH

A total of 7089 targets associated with NASH (Figure 6a), 897 targets related to β-D-galactoside-1,6-bis(3,4,5-trihydroxybenzoate), and 886 targets related to -(-)-rhododendron-4’-β-D-glucoside were identified. After the intersections were completed, 124 common targets were found for β-D-galactoside-1,6-bis(3,4,5-trihydroxybenzoate) in the regulation of NASH (Figure 6b), and another 124 common targets for the regulation of NASH by -(-)-rhododendron-4’-β-D-glucoside were identified (Figure 6c). The exported files of these common targets analyzed via the STRING online database were then reimported into Cytoscape for processing (Figure 6d and e). Ultimately, the hub targets of β-D-galactoside-1,6-bis(3,4,5-trihydroxybenzoate) regulated NASH were identified as HSP90AA1, TP53, TLR4, NFKB1, STAT1, and PRKACA (Figure 6f). The hub targets of -(-)-rhododendron-4’-β-D-glucoside regulated NASH were identified as NFE2L2, TP53, MYD88, PRKACA, NFKB1, and MTOR (Figure 6g).

Close

Figure 6.

(a) Schematic diagram of the number of NASH-related targets retrieved from 8 open-source databases. (b) Venn diagram depicting common targets between NASH and β-D-galactoside-1,6-bis(3,4,5-trihydroxybenzoate). (c) Venn diagram depicting common targets between NASH and -(-)-rhododendron-4’-β-D-glucoside. (d) Network topology diagram of common targets of β-D-galactoside-1,6-bis(3,4,5-trihydroxybenzoate) against NASH. (e) Network topology diagram of common targets of -(-)-rhododendron-4’-β-D-glucoside against NASH. (f) Six hub targets of β-D-galactoside-1,6-bis(3,4,5-trihydroxybenzoate) against NASH. (g) Six hub targets of -(-)-rhododendron-4’-β-D-glucoside against NASH.

Export to PPT

3.4. Molecular docking studies

Studies have shown that TLR4, NFKB1, NFE2L2, MYD88, HSP90AA1, and STAT1, which are related to inflammation and stress, and MTOR and PRKACA, which are related to lipid metabolism, are aberrantly expressed to varying degrees in the NASH process [39, 40]. The complex regulatory network and cascade effects highlight the challenges in NASH drug development. Through network pharmacology, these proteins serve as hub targets for the anti-NASH effects of β-D-galactoside-1,6-bis(3,4,5-trihydroxybenzoate) and -(-)-rhododendron-4’-β-D-glucoside. To further validate the results of network pharmacology, subsequent molecular docking studies are highly warranted.

Detailed docking information and binding free energy data have been presented in Table 3. The positions of the optimal docking results for each target protein and compound have been shown in Figure 7. The docking results of HSP90AA1 and NFE2L2 with their original ligands are shown in the supporting information as a positive control (Figure S45). Both compounds showed excellent binding energies (< -4 kcal/mol) with all the hub targets, and their binding mode was mainly hydrogen bonding. For β-D-galactoside-1,6-bis(3,4,5-trihydroxybenzoate), the ASP93, LEU103, TYR139, LEU103, GLY135, and LEU48 residues in the HSP90AA1 protein formed a cavity, which bound to this small compound through hydrogen bonds and contributed the best binding energy (-8.65 kcal/mol) in this study. The -(-)-rhododendron-4’-β-D-glucoside interacted with NFE2L2 through the VAL606, GLY367, VAL418, VAL465, ARG415, and VAL512 residues, which presented excellent binding energies (-7.88 kcal/mol). Although HSP90AA1 and NFE2L2 showed better binding energy with their original ligands, the two novel phenolic glycosides in this study established more hydrogen bonds with these two target proteins. Hydrogen bonding is an important noncovalent bonding interaction force for drug binding to target proteins, arising from lone-pair electrons on functional groups (e.g. hydroxyl, carbonyl) that engage in charge-transfer with complementary hydrogen donors or acceptors within the protein. This network of hydrogen bonds markedly enhances ligand–receptor affinity and specificity, thereby modulating diverse biological activities such as antioxidant and anti-inflammatory effects [41, 42]. Docking results implied that the abundant hydroxyl groups in these two novel phenolic glycosides increase their binding affinity to target proteins. The carbonyl group on β-D-galactoside-1,6-bis(3,4,5-trihydroxybenzoate) not only participates in hydrogen bonding but also enables more stable covalent interactions with amino acid residues via Michael addition reactions (theoretically). Moreover, the presence of a glycosidic structure not only establishes a highly stable ligand–receptor conformation through multipoint hydrogen bonding with key residues but also enhances aqueous solubility and membrane permeability, thereby improving pharmacokinetic profiles. These attributes provide a robust theoretical and experimental foundation for the future development of drug formulations, delivery systems and other innovative dosage strategies.

Close

Figure 7.

(a) 3D and 2D schematics of β-D-galactoside-1,6-bis(3,4,5-trihydroxybenzoate) and HSP90AA1 docking. (b) 3D and 2D schematics of β-(-)-rhododendron-4’-β-D-glucoside and NFE2L2 docking. (c) Schematic diagram of the docking results of β-D-galactoside-1,6-bis(3,4,5-trihydroxybenzoate) with NFKB1, (d) PRKACA, (e) STAT1, (f) TLR4, and (g) TP53. (h) Schematic diagram of the docking results of -(-)-rhododendron-4’-β-D-glucoside with MTOR, (i) MYD88, (j) NFKB1, (k) PRKACA, and (l) TP53.

Export to PPT

3.5. Enrichment analysis of two novel phenolic glycosides regulating NASH

The common targets were imported into the DAVID bioinformatics database for GO and KEGG analysis. As shown in Figure 8(a), the BP of β-D-galactoside-1,6-bis(3,4,5-trihydroxybenzoate) against NASH mainly includes the response to xenobiotic stimulus, inflammatory response, signal transduction, positive regulation of ERK1 and ERK2 cascade, and phosphorylation. The MF of β-D-galactoside-1,6-bis(3,4,5-trihydroxybenzoate) against NASH mainly includes identical protein binding, ATP binding, protein binding, and kinase activity. The CC of β-D-galactoside-1,6-bis(3,4,5-trihydroxybenzoate) against NASH mainly involves the extracellular exosome, cytosol, cytoplasm, and plasma membrane. As shown in Figure 8(b), the BP of -(-)-rhododendron-4’-β-D-glucoside against NASH mainly includes the response to hypoxia, inflammatory response, response to lipopolysaccharide, and positive regulation of transcription by RNA polymerase II. The MF of -(-)-rhododendron-4’-β-D-glucoside anti-NASH mainly includes protein binding, nuclear receptor activity, kinase activity, and ATP binding. The CC of -(-)-rhododendron-4’-β-D-glucoside anti-NASH mainly involves extracellular exosome, cytosol, cytoplasm, and nucleus.

Close

Figure 8.

Schematic diagram of the GO analyses of (a) β-D-galactoside-1,6-bis(3,4,5-trihydroxybenzoate) and (b) -(-)-rhododendron-4’-β-D-glucoside for treating NASH. Schematic diagram of KEGG analyses for the effects of (c) -(-)-rhododendron-4’-β-D-glucoside and (d) -(-)-rhododendron-4’-β-D-glucoside in treating NASH

Export to PPT

KEGG enrichment analysis revealed that these two novel phenolic glycosides affect the progression of NASH through additive and/or synergistic effects, with the cAMP signaling pathway, insulin signaling pathway, sphingolipid signaling pathway, and lipid and atherosclerosis signals emerging as particularly significant, which have been shown in several studies to have a direct correlation with the core pathological features of NASH [43-45]. Specifically, in the case of β-D-galactoside-1,6-bis(3,4,5-trihydroxybenzoate), a total of 127 signaling pathways were enriched, and the mechanism of its anti-NASH effect is closely related to the cAMP signaling pathway (Figure 8c). Abnormalities in the cAMP signaling pathway, a cyclic nucleotide signaling pathway involved in fatty acid β-oxidation, lead to an imbalance in glycolipid metabolism, which ultimately triggers a series of cascade reactions that contribute to the process of NASH [46]. β-D-galactoside-1,6-bis(3,4,5-trihydroxybenzoate) can regulate such a disorder, which in turn affects the aberrant expression of kinases (e.g., PKA) and inflammatory factors (e.g., IL6), thus realizing the efficacy of reducing NASH. For -(-)-rhododendron-4’-β-D-glucoside, KEGG analysis revealed a total of 104 enriched signaling pathways (Figure 8d). After screening, it was found that the anti-NASH effects of -(-)-rhododendron-4’-β-D-glucoside were closely related to the sphingolipid signaling pathway. This pathway inhibits the overexpression of inflammatory factors through the activation of specific sphingolipid kinases, which in turn cascade multiple downstream signaling pathways (e.g., P13K-AKT, p53, and NFκB), thereby reducing cell damage, slowing down or preventing the further progression of NASH to liver fibrosis and cirrhosis [47].

Overall, this study revealed the anti-NASH efficacy of two novel phenolic glycosides and uncovered their molecular mechanisms via network pharmacology. Their detailed pharmacodynamic comparisons have been listed in Table S3. This work has established a robust theoretical framework and technical platform for the future screening and optimization of more potent secondary metabolites from S. tangutica, which is expected to lead to the discovery of more innovative anti-NASH compounds with potential for clinical translation. In the case of these two novel phenolic glycosides as potential anti-NASH candidates, more rigorous in vivo pharmacodynamic, pharmacokinetic, and systemic safety evaluations will need to be conducted in future work to further exploit their medicinal potential.