Network-guided Evaluation of β-sitosterol in Inflammatory and Profibrotic Mesangial-Cell Models Relevant to Chronic Glomerulonephritis

Summary

This study integrates network pharmacology, validated molecular docking, and in vitro experiments to evaluate candidate compound-target relationships between Cordyceps sinensis constituents and chronic glomerulonephritis-related processes. β-Sitosterol was selected for compound-level validation and was found to attenuate inflammatory and profibrotic activation in mesangial-cell models.

Abstract

Chronic glomerulonephritis (CGN) is characterized by persistent inflammatory injury and progressive fibrotic remodeling, yet compound-target relationships underlying natural-product-based interventions remain incompletely defined. This study integrated network pharmacology, validated molecular docking, and in vitro experiments to evaluate candidate constituents of Cordyceps sinensis (C. sinensis) in CGN-related pathological processes. Candidate constituents of C. sinensis were screened, and the Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform-derived compound-associated targets were intersected with CGN-associated genes. Functional enrichment and protein-protein interaction analyses were used to prioritize biological processes and hub targets. PTGS2 was identified as a key overlapping target. Molecular docking was performed using a celecoxib-bound COX-2 structure. Re-docking of the co-crystallized celecoxib ligand reproduced the crystallographic pose with an RMSD of 0.876 Å, supporting the docking protocol. β-Sitosterol and linoleyl acetate showed predicted compatibility with the COX-2 docking region, with binding affinities of −7.2 and −7.5 kcal/mol, respectively. β-Sitosterol was selected for compound-level validation in HBZY-1 rat glomerular mesangial cells. At 0.5–10 µM, β-sitosterol did not markedly reduce cell viability. In Lipopolysaccharide (LPS)-stimulated cells, β-sitosterol reduced Tnf, Il6, and Ptgs2 expression, decreased prostaglandin E2 (PGE2) production, and reduced cyclooxygenase-2 (COX-2) protein abundance. In transforming growth factor β1 (TGF-β1)-treated cells, β-sitosterol reduced Col1a1 and Acta2 expression and decreased alpha-smooth muscle actin (α-SMA) protein abundance. These findings provide hypothesis-generating evidence that β-sitosterol modulates inflammatory and profibrotic activation in mesangial-cell models, but they do not establish direct COX-2 enzymatic inhibition or therapeutic efficacy in CGN.

Introduction

Chronic glomerulonephritis (CGN) comprises a heterogeneous group of glomerular disorders that can progress to chronic kidney disease and end-stage kidney disease¹. It is characterized by persistent glomerular injury, inflammatory activation, and progressive fibrotic remodeling^2,3. Although corticosteroids and immunosuppressive agents are used in selected patients, long-term disease control remains difficult in many cases, and treatment-related adverse effects are common^1,4,5. The progression of CGN involves interacting pathological processes, including immune activation, endothelial dysfunction, oxidative stress, and extracellular matrix accumulation^3,6,7. Therefore, identifying molecular nodes that connect inflammatory and profibrotic responses may help generate testable hypotheses for disease-relevant intervention strategies.

Cordyceps sinensis (C. sinensis) has been used in kidney-related clinical and experimental contexts and has been reported to exert anti-inflammatory, anti-apoptotic, and cytoprotective effects in several disease models^8,9,10. However, the molecular basis by which its candidate constituents may interact with glomerulonephritis-relevant targets remains insufficiently defined¹¹. Given the multi-component nature of C. sinensis and the mechanistic heterogeneity of CGN, network pharmacology provides a useful framework for prioritizing candidate compound-target-pathway relationships^12,13.

In this study, compound screening, target retrieval, disease-gene intersection analysis, functional enrichment, protein-protein interaction network analysis, molecular docking, and in vitro validation were integrated to explore potential molecular relationships between candidate constituents of C. sinensis and CGN-relevant pathological processes. Prostaglandin-endoperoxide synthase 2 (PTGS2) was prioritized as a hub target from the compound-disease network and was evaluated by molecular docking using a celecoxib-bound cyclooxygenase-2 (COX-2) structure with redocking validation. Based on the network results, docking analysis, and prior evidence for anti-inflammatory and anti-fibrotic activity, β-sitosterol was selected as a representative candidate compound for downstream validation^14,15,16. Lipopolysaccharide (LPS)-stimulated mesangial cells were used to assess inflammatory activation and the COX-2/prostaglandin E2 (PGE2)-associated response, while a transforming growth factor β1 (TGF-β1)-induced model was included to evaluate profibrotic marker expression.

Protocol

The network pharmacology and bioinformatics analyses were based on publicly available databases and did not involve human participants or animals. The in vitro experiments were conducted using an established cell line and did not require ethics approval according to institutional requirements.

Candidate constituent screening and target retrieval

Candidate constituents and their corresponding targets were retrieved from the Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform (TCMSP) using the search term “dongchongxiacao” on 13 July 2025. Compounds with predicted oral bioavailability of at least 30% and drug-likeness of at least 0.18 were retained. Seven candidate constituents met both screening criteria. The molecular weight, oral bioavailability, drug-likeness, and other available pharmacokinetic descriptors were recorded from TCMSP for descriptive comparison.

The compound-target information associated with the retained constituents was downloaded directly from TCMSP. Duplicate target entries were manually removed when constructing the nonredundant target set, while the original compound-target relationships were retained for subsequent network construction. Because TCMSP provided target names in their full-length form rather than standardized gene symbols, each target was searched against the UniProt database (accessed on 14 July 2025). The corresponding human entry was selected, and the target name was converted to its official gene symbol. The standardized compound-target dataset was used for disease-target intersection and compound-target network analysis.

Collection of glomerulonephritis-associated genes and target intersection

Disease-associated genes were retrieved from GeneCards, DisGeNET, PharmGKB, and the Therapeutic Target Database using the search term “glomerulonephritis” on 14 July 2025. For GeneCards, entries with relevance scores greater than the median score of 10 were retained. Eligible entries from the other databases were included according to their curated disease-target annotations. Gene names were standardized to official human gene symbols, and duplicate entries were removed. This procedure yielded 1,134 unique glomerulonephritis-associated genes. Because the search used the broader term “glomerulonephritis,” the resulting dataset was considered relevant to chronic glomerulonephritis-associated molecular processes rather than specific to a single histopathological subtype.

TCMSP-derived compound-associated targets were intersected with the disease-associated gene set using R. Twenty-three overlapping targets were identified. Venn diagrams were generated using the R package venn, and the compound-overlapping target network was constructed and visualized using Cytoscape. The relationships between candidate constituents and overlapping targets are summarized in Table 1.

Functional enrichment and protein-protein interaction analyses

Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were performed in R using clusterProfiler and org.Hs.eg.db. Gene Ontology terms were analyzed in the biological process, cellular component, and molecular function categories. P values were adjusted for multiple testing using the Benjamini-Hochberg method, and terms or pathways with adjusted P values below 0.05 were retained. Enriched terms and pathways were displayed using independently generated bar plots and dot plots.

The 23 overlapping targets were submitted to STRING, with Homo sapiens selected as the organism and a minimum interaction score greater than 0.4. The resulting protein-protein interaction network was imported into Cytoscape. Network topology was evaluated using CytoNCA based on degree, betweenness, closeness, eigenvector centrality, local average connectivity, and network centrality. Hub targets were defined as nodes for which the degree, betweenness centrality, closeness centrality, eigenvector centrality, local average connectivity, and network centrality values were all greater than the corresponding median values of the network. This analysis prioritized nine hub targets for subsequent interpretation.

Molecular docking and redocking validation

Molecular docking was performed using the celecoxib-bound cyclooxygenase-2 structure PDB ID 3LN1. Chain A and its associated heme group were retained as the receptor, while crystallographic water molecules and the co-crystallized celecoxib ligand were removed during receptor preparation. The native celecoxib conformation was separately extracted from chain A for redocking validation. Polar hydrogen atoms and Gasteiger charges were added using AutoDockTools, and the receptor was saved in PDBQT format. PDB ID 3LN1 is a 2.40 Å structure of murine COX-2 bound to celecoxib. The structures of β-sitosterol and linoleyl acetate were obtained from PubChem and prepared using Open Babel. Hydrogen atoms and Gasteiger partial charges were added, rotatable bonds were assigned, and each ligand was converted to PDBQT format. Celecoxib, β-sitosterol, and linoleyl acetate were docked using AutoDock Vina with the same receptor and search parameters.

The docking box was centered on the native celecoxib-binding region using the following coordinates: center_x = 30.9886 center_y = -22.2836 center_z = -16.5072 size_x = 26 Å size_y = 26 Å size_z = 26 Å. The exhaustiveness value was set to 16, the number of output modes to 10, and the energy range to 3 kcal/mol. The top-ranked pose based on the lowest predicted binding energy was retained for visualization. For protocol validation, native celecoxib was redocked under the same conditions, and the heavy-atom root mean square deviation between the crystallographic and redocked poses was calculated using PyMOL. A root mean square deviation below 2 Å was considered an acceptable reproduction of the reference pose. Residues located within 4 Å of each docked ligand were identified in PyMOL. Docking scores and poses were interpreted as estimates of structural compatibility and not as direct evidence of target binding or enzymatic inhibition.

Cell culture and compound preparation

HBZY-1 rat glomerular mesangial cells were cultured in high-glucose Dulbecco’s modified Eagle medium supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin at 37 °C under humidified conditions with 5% CO₂. Cells at passages 2–7 were used for all experiments, and mycoplasma-free status was verified before use. β-Sitosterol and celecoxib served as chemically defined compounds. β-Sitosterol was dissolved in absolute ethanol to generate a 10 mM stock solution (approximately 4.15 mg/mL). When required, brief sonication and warming to 50–60 °C were performed to achieve a visually clear solution. Celecoxib was dissolved in dimethyl sulfoxide to prepare a 10 mM stock solution (approximately 3.81 mg/mL). The stock solutions were aliquoted and stored at −20 °C. Before treatment, the stocks were diluted directly into prewarmed culture medium and mixed immediately.

Vehicle concentrations were matched within each experiment. In the cell viability assay, the final ethanol concentration was adjusted to 0.2% v/v in all groups because the highest β-sitosterol concentration tested was 20 µM. In the LPS-induced inflammatory experiment, the final concentrations of ethanol and dimethyl sulfoxide were adjusted to 0.1% and 0.01% v/v, respectively, in all groups. In the TGF-β1-induced profibrotic experiment, the final ethanol concentration was adjusted to 0.1% v/v in all groups. Equivalent volumes of the corresponding vehicles were added to the Control, LPS, and TGF-β1 groups.

Cell viability assay

Cells were plated in 96-well plates at a density of 5 × 10³ cells per well in 100 µL of complete medium and allowed to adhere for 24 h. To assess the concentration range of β-sitosterol, cells were incubated with 0.5, 1, 5, 10, or 20 µM β-sitosterol for 24 h. For the LPS-related cell viability assay, cells were pretreated with β-sitosterol at 1, 5, or 10 µM for 2 h, followed by exposure to 1 µg/mL lipopolysaccharide for an additional 24 h.

The treatment groups were Control, β-sitosterol at 0.5 µM, β-sitosterol at 1 µM, β-sitosterol at 5 µM, β-sitosterol at 10 µM, β-sitosterol at 20 µM, LPS, LPS plus β-sitosterol at 1 µM, LPS plus β-sitosterol at 5 µM, and LPS plus β-sitosterol at 10 µM. Cell Counting Kit-8 reagent was added at 10 µL per well, and the plates were incubated for 2 h. Absorbance was then recorded at 450 nm. Cell viability was determined after subtracting the cell-free blank and expressed relative to the Control group. Each condition included six technical wells in each of six independent biological experiments. The technical-well values from each experiment were averaged to generate one biological replicate value for statistical analysis.

LPS-induced inflammatory activation

For inflammatory activation experiments, HBZY-1 cells were plated in 6-well plates at a density of 2 × 10⁵ cells per well and treated after reaching approximately 70%–80% confluence. Cells were maintained in medium supplemented with 1% fetal bovine serum for 12 h before treatment. β-Sitosterol or celecoxib was administered 2 h before LPS stimulation. Subsequently, the cells were exposed to 1 µg/mL LPS for 24 h.

The experimental groups were Control, β-sitosterol at 10 µM, LPS, LPS plus β-sitosterol at 1 µM, LPS plus β-sitosterol at 5 µM, LPS plus β-sitosterol at 10 µM, and LPS plus celecoxib at 1 µM. Celecoxib was included as a functional positive control for inhibition of COX-2-dependent prostaglandin production. At the end of treatment, culture supernatants were collected for PGE2 measurement. Cells were collected separately for quantitative real-time PCR analysis of Tnf, Il6, and Ptgs2 and for western blot analysis of COX-2.

TGF-β1-induced profibrotic activation

A separate TGF-β1-induced model was used to assess profibrotic activation. HBZY-1 cells were plated in 6-well plates at a density of 2 × 10⁵ cells per well and maintained in medium supplemented with 1% fetal bovine serum for 12 h before treatment. Cells were pretreated with β-sitosterol for 2 h, followed by stimulation with 10 ng/mL recombinant TGF-β1 for 48 h.

The treatment groups included Control, β-sitosterol (10 µM), TGF-β1, TGF-β1 + β-sitosterol (5 µM), and TGF-β1 + β-sitosterol (10 µM). At the end of the treatment period, cells were harvested for quantitative real-time PCR analysis of Col1a1 and Acta2 and for western blot analysis of alpha-smooth muscle actin (α-SMA).

Quantitative real-time PCR

Total RNA was isolated from treated cells using a phenol-guanidinium-based RNA extraction reagent according to the manufacturer's protocol. RNA concentration and purity were determined using a spectrophotometer. Only samples with an A260/A280 ratio of 1.8–2.0 were used for reverse transcription. For each sample, 1 µg of total RNA was reverse transcribed into complementary DNA in a final reaction volume of 20 µL. The resulting complementary DNA was diluted fivefold with nuclease-free water before quantitative PCR. Quantitative PCR was carried out in a total volume of 20 µL containing 10 µL of 2× SYBR Green master mix, 0.4 µL each of 10 µM forward and reverse primers, 2 µL of diluted complementary DNA, and 7.2 µL of nuclease-free water. The amplification conditions consisted of an initial denaturation step at 95 °C for 30 s, followed by 40 cycles of denaturation at 95 °C for 5 s and annealing-extension at 60 °C for 30 s. A melting-curve analysis was subsequently performed from 65 °C to 95 °C to verify amplification specificity.

The measured genes were Tnf, Il6, Ptgs2, Col1a1, Acta2, and Gapdh. Gapdh served as the internal reference gene. Relative gene expression was determined using the 2^−ΔΔCt method, with the Control group used as the calibrator. Primer sequences are provided in Supplementary Table 1. Each sample was analyzed in technical triplicate, and six independent biological experiments were performed. The values obtained from the technical replicates were averaged before statistical analysis.

PGE2 measurement by enzyme-linked immunosorbent assay

Culture supernatants were collected after 24 h of LPS stimulation and centrifuged at 1,000 × g for 10 min at 4 °C to remove cell debris. PGE2 concentrations were determined using a commercially available enzyme-linked immunosorbent assay kit according to the manufacturer's instructions.

Standards and samples were loaded onto the assay plate in duplicate. After completion of the antibody-binding and color-development steps, absorbance was recorded at 450 nm using a microplate reader. PGE2 concentrations were determined from a four-parameter logistic standard curve and expressed as pg/mL. Supernatants from six independent cell experiments were analyzed. Duplicate measurements for each biological sample were averaged before statistical analysis.

Western blotting

Treated cells were lysed in radioimmunoprecipitation assay buffer supplemented with a protease inhibitor mixture. The lysates were incubated on ice for 30 min and centrifuged at 12,000 × g for 15 min at 4 °C. Protein concentrations in the supernatants were determined using a bicinchoninic acid protein assay. Equal amounts of protein (30 µg per lane) were mixed with loading buffer and denatured at 95 °C for 5 min. Proteins from the LPS experiments were resolved on 10% sodium dodecyl sulfate-polyacrylamide gels, whereas those from the TGF-β1 experiments were resolved on 12% gels. The separated proteins were subsequently transferred onto polyvinylidene fluoride membranes.

The membranes were blocked with 5% nonfat milk in Tris-buffered saline containing 0.1% Tween 20 for 1 h at room temperature and then incubated overnight at 4 °C with primary antibodies against COX-2 (1:1000), α-SMA (1:1000), and GAPDH (1:5000). COX-2 was analyzed in samples from the LPS experiment, whereas α-SMA was analyzed in samples from the TGF-β1 experiment. After washing, the membranes were incubated with horseradish peroxidase-conjugated secondary antibodies (1:5000) for 1 h at room temperature. Protein bands were visualized using an enhanced chemiluminescence substrate and captured with a chemiluminescence imaging system. The exposure times were approximately 10 s for COX-2, 10 s for α-SMA, and 2 s for GAPDH.

Band intensities were quantified using ImageJ. COX-2 and α-SMA signals were normalized to the corresponding GAPDH signal. Western blotting was performed using protein samples obtained from six independent cell culture and treatment experiments.

Statistical analysis

Statistical analyses were conducted using GraphPad Prism. Data are presented as the mean ± standard deviation. Six independent cell culture and treatment experiments were performed for each quantitative analysis, with each independent experiment representing one biological replicate. Technical replicate measurements obtained from the same biological experiment were averaged before statistical analysis and were not considered independent observations.

For the cell viability assay, each biological experiment included six technical wells per condition. Quantitative PCR reactions were performed in technical triplicate for each biological sample, and PGE2 measurements were performed in duplicate. Western blot quantification was based on six independently prepared protein samples. Thus, each quantitative bar contains six individual biological replicate values.

Differences among multiple groups were analyzed using one-way analysis of variance followed by Tukey’s multiple-comparison test. A two-sided P value < 0.05 was considered statistically significant. Individual biological replicate values were displayed together with the mean and standard deviation.

Results

Candidate Constituent Screening, Target Retrieval, and Target Intersection

The workflow for identifying candidate compound–disease overlapping targets is summarized in Figure 1A. Seven candidate compounds of C. sinensis met the predefined screening criteria of OB ≥30% and DL ≥0.18, including arachidonic acid, linoleyl acetate, β-sitosterol, peroxyergosterol, cerevisterol, cholesteryl palmitate, and CLR (Table 2). In parallel, 1,134 glomerulonephritis-associated genes were retrieved from GeneCards, DisGeNET, PharmGKB, and TTD, and their distribution across the four databases is shown in Figure 1B. Intersection analysis between the compound-associated target set and the disease-associated gene set identified 23 overlapping targets (Table 1, Figure 1C). The corresponding compound–overlapping target network is presented in Figure 1D, illustrating the relationships among representative candidate compounds, overlapping targets, and CGN. These overlapping targets included several genes related to inflammatory regulation, apoptosis, and vascular associated signaling, such as PTGS2, CASP3, MAPK1, PPARG, TNFRSF1A, and NOS3.

Functional Enrichment and Protein–Protein Interaction Analyses

To investigate the biological significance of the 23 overlapping targets, GO and KEGG pathway enrichment analyses were performed. The results revealed that these targets are significantly enriched in multiple inflammation- and metabolism-related pathways.

GO analysis classified the targets into the biological process (BP), cellular component (CC), and molecular function (MF) categories. The most significantly enriched BP terms included response to lipopolysaccharide, response to molecules of bacterial origin, and regulation of muscle system processes. At the cellular level, targets were enriched in membrane raft, caveola, and platelet alpha granule compartments, while MF terms such as oxidoreductase activity and nuclear receptor binding further implicated roles in oxidative stress and transcriptional regulation (Figure 2A).

KEGG analysis revealed that the lipid and atherosclerosis pathway (hsa05417) was the most significantly enriched (q-value <1 × 10⁻⁶), alongside pathways such as AGE–RAGE signaling in diabetic complications, TNF signaling, IL-17 signaling, and HIF-1 signaling (Figure 2B). Notably, the lipid and atherosclerosis pathway involved 10 out of the 23 targets, including CASP3, MAPK1, PPARG, NOS3, and TNFRSF1A.

Collectively, these enrichment results indicate that the overlapping targets are functionally associated with inflammatory, stress-response, and endothelial-related biological processes relevant to CGN.

Protein–Protein Interaction Network Analysis

A protein-protein interaction (PPI) network was constructed for the 23 overlapping targets using STRING, resulting in a network containing 23 nodes and 102 edges (Figure 3A). Topological analysis using CytoNCA identified nine highly connected hub targets: PTGS2, MAPK1, PPARG, CASP3, EGF, JUN, PECAM1, BCL2, and PRKCA (Figure 3B). Among these targets, PTGS2 showed prominent network centrality and was therefore selected for subsequent molecular docking analysis. These results highlight a subset of network-prioritized candidate regulators potentially relevant to the compound–disease intersection.

Molecular Docking and Redocking Validation

To further evaluate the structural plausibility of the PTGS2-associated candidate relationships identified by network analysis, molecular docking was performed using a celecoxib-bound COX-2 structure. The crystal structure PDB ID 3LN1 was selected because it contains celecoxib bound at the COX-2 active site, providing an experimentally defined ligand-binding pocket for docking validation. The co-crystallized celecoxib ligand was first extracted and re-docked into the same binding pocket. The redocked celecoxib pose closely reproduced the crystallographic pose, with a redocking root mean square deviation (RMSD) of 0.876 Å and a predicted binding affinity of -12.3 kcal/mol (Figure 4A). This result indicated that the docking protocol was able to reproduce the known ligand-binding pose under the selected docking conditions.

Using the same receptor, grid box, and docking parameters, β-sitosterol was docked into the celecoxib-defined COX-2 binding region. β-sitosterol showed a predicted binding affinity of -7.2 kcal/mol (Figure 4B). The docked pose was located within the predefined COX-2 docking region, and the residues surrounding the docked β-sitosterol pose included ASN567, ASP333, GLN178, GLN336, GLY340, HIS337, HIS80, PRO500, THR79, and TYR341.

Linoleyl acetate was also evaluated as a secondary computational candidate because it was retained in the candidate-constituent list and was linked to PTGS2 in the network analysis. Under the same docking conditions, linoleyl acetate showed a predicted binding affinity of -7.5 kcal/mol, which was slightly more favorable than that of β-sitosterol (Figure 4C). Its predicted pose was positioned close to the celecoxib reference ligand within the docking region. The residues surrounding linoleyl acetate included ALA502, ALA513, ARG106, ARG499, GLN178, GLY512, HIS75, ILE503, LEU338, LEU345, LEU517, MET508, PHE504, SER339, SER516, TRP373, TYR371, VAL102, VAL335, and VAL509.

Cell Viability, Inflammatory Activation, and Profibrotic Activation

β-sitosterol was selected for compound-level validation because it was identified as a PTGS2-associated candidate in the network analysis and had prior experimental support for anti-inflammatory and anti-fibrotic activity. Cell viability was first assessed to determine the concentration range suitable for subsequent experiments. Treatment with β-sitosterol at 0.5, 1, 5, and 10 µM did not markedly reduce cell viability compared with the control group, whereas 20 µM β-sitosterol caused a slight decrease. LPS exposure reduced cell viability, and cotreatment with β-sitosterol partially improved cell viability under LPS stimulation, with the most evident effect observed at 10 µM (Figure 5A). Based on these results, 1, 5, and 10 µM β-sitosterol were used for the following LPS-induced inflammatory activation experiments.

LPS stimulation markedly increased the mRNA expression of inflammatory genes in HBZY-1 cells. Compared with the control group, the LPS group showed higher expression levels of Tnf, Il6, and Ptgs2 (Figure 5B-D). β-sitosterol treatment reduced the LPS-induced upregulation of these genes in a concentration-related manner. The inhibitory trend was most apparent in the LPS plus 10 µM β-sitosterol group. Celecoxib, which was included as a functional COX-2 inhibitor control, also reduced inflammatory gene expression to some extent, although its effect was less pronounced than that observed with 10 µM β-sitosterol for several transcriptional markers (Figure 5B-D).

Because Ptgs2 encodes COX-2, and PGE2 is a major downstream prostaglandin product of COX-2 activity, PGE2 secretion was measured by ELISA. LPS treatment increased PGE2 concentration in the culture supernatant compared with the control group. β-sitosterol decreased LPS-induced PGE2 production in a concentration-related manner. Celecoxib also reduced PGE2 production and served as a positive control for functional inhibition of the COX-2/PGE2 axis (Figure 5E). Western blot analysis further showed that LPS increased COX-2 protein abundance. β-sitosterol treatment reduced COX-2 protein levels under LPS stimulation, with a stronger reduction observed at the higher concentration. In contrast, celecoxib reduced PGE2 production but showed a more limited effect on COX-2 protein abundance (Figure 5F).

To complement the LPS-induced inflammatory model, a TGF-β1-induced profibrotic activation model was used to assess whether β-sitosterol affected fibrotic marker expression in mesangial cells. TGF-β1 markedly increased the mRNA expression of Col1a1 and Acta2 compared with the control group (Figure 6A,B). β-sitosterol reduced the TGF-β1-induced increase in both markers, and the reduction was more evident at 10 µM than at 5 µM (Figure 6A,B). Western blot analysis showed a corresponding increase in α-SMA protein expression after TGF-β1 stimulation, whereas β-sitosterol treatment decreased α-SMA protein abundance in TGF-β1-treated cells (Figure 6C).

DATA AVAILABILITY:

The raw and processed data supporting this study have been deposited in Zenodo under DOI: https://zenodo.org/records/21649482. The full-scan images of the western bolt are provided in Supplementary File 1.

Figure 1. Workflow and network-based identification of candidate compound–disease overlapping targets of C. sinensis in glomerulonephritis. (A) Schematic workflow showing the identification of candidate compounds, retrieval and standardization of compound-associated targets, retrieval of glomerulonephritis-associated genes, intersection analysis, and construction of the compound–overlapping target network. (B) Venn diagram illustrating the distribution of glomerulonephritis-associated genes retrieved from GeneCards, DisGeNET, PharmGKB, and TTD. (C) Venn diagram showing the overlap between TCMSP-derived compound-associated targets and disease-associated genes, yielding 23 overlapping targets. (D) Compound–overlapping target network. The green diamond represents C. sinensis, blue hexagons indicate candidate compounds, yellow circles indicate overlapping targets, and the pink triangle represents CGN. Edges denote the corresponding compound–target or disease–target relationships. Please click here to view a larger version of this figure.

Figure 2. Functional enrichment analysis of the overlapping targets. (A) GO enrichment analysis of the 23 overlapping targets. Top BP, CC, and MF terms were identified. Bar length indicates gene count; color gradient represents q-value. (B) KEGG enrichment analysis of the 23 overlapping targets. Pathways are ranked by gene count and statistical significance. Please click here to view a larger version of this figure.

Figure 3. Protein–protein interaction network and hub target prioritization of the overlapping targets. (A) PPI network constructed from the 23 overlapping targets using STRING with an interaction score threshold >0.4. (B) The top nine hub targets are prioritized by CytoNCA-based network topology analysis. PTGS2 showed prominent network centrality and was selected for subsequent molecular docking analysis. Please click here to view a larger version of this figure.

Figure 4. Molecular docking and redocking validation of selected candidate compounds with COX-2. (A) Superposition of the crystallographic celecoxib pose and the top-ranked redocked pose in the celecoxib-bound murine COX-2 structure (PDB ID 3LN1). Redocking reproduced the experimental binding orientation. (B) Top-ranked docking pose of β-sitosterol within the celecoxib-defined COX-2 docking region. The crystallographic celecoxib pose is included as a spatial reference. (C) The top-ranked docking pose of linoleyl acetate within the same docking region. Please click here to view a larger version of this figure.

Figure 5. β-Sitosterol attenuates LPS-induced inflammatory activation and PGE2 production in HBZY-1 mesangial cells. (A) Cell viability was measured using the Cell Counting Kit-8 assay. Cells were treated with β-sitosterol at 0.5, 1, 5, 10, or 20 µM for 24 h. In the LPS-stimulated groups, cells were pretreated with β-sitosterol at 1, 5, or 10 µM for 2 h before exposure to 1 µg/mL LPS for 24 h. Cell viability was expressed relative to the Control group. (B–D) Quantitative real-time PCR analysis of Tnf (B), Il6 (C), and Ptgs2 (D) mRNA expression. Gene expression was normalized to Gapdh and expressed relative to the Control group. (E) PGE2 concentrations in culture supernatants measured by enzyme-linked immunosorbent assay. Celecoxib was included as a functional positive control for suppression of COX-2-dependent prostaglandin production. (F) Representative western blot images of COX-2 and GAPDH in the indicated treatment groups. Data are presented as mean ± SD from six independent biological experiments, with individual biological replicate values shown. Technical replicate values were averaged before analysis. Comparisons were performed using one-way ANOVA followed by Tukey’s multiple-comparison test. Significance is indicated by brackets in the individual panels: *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Please click here to view a larger version of this figure.

Figure 6. β-Sitosterol reduces TGF-β1-induced profibrotic marker expression in HBZY-1 mesangial cells. (A–B) Quantitative real-time PCR analysis of Col1a1 (A) and Acta2 (B) mRNA expression. Gene expression was normalized to Gapdh and expressed relative to the Control group. (C) Representative western blot images of α-SMA and GAPDH and in the indicated treatment groups. Data are presented as mean ± SD from six independent biological experiments, with individual biological replicate values shown. Technical replicate values were averaged before analysis. Comparisons were performed using one-way ANOVA followed by Tukey’s multiple-comparison test. Significance is indicated by brackets in the individual panels: *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Please click here to view a larger version of this figure.

Figure 7. Integrated interpretation of β-sitosterol-associated inflammatory and profibrotic responses in HBZY-1 cells. Network and docking analyses prioritized PTGS2/COX-2 within a broader context that included MAPK1, JUN, and PPARG. β-Sitosterol treatment was associated with reductions in LPS-induced Tnf, Il6, and Ptgs2 expression, COX-2 protein abundance, and PGE2 production, as well as reductions in TGF-β1-induced Col1a1, Acta2, and α-SMA expression. Please click here to view a larger version of this figure.

No.	Overlapping target	Candidate compound(s) associated with the target
1	PTGS1	β-sitosterol; linoleyl acetate; arachidonic acid
2	PTGS2	β-sitosterol; linoleyl acetate; arachidonic acid
3	CASP3	β-sitosterol; arachidonic acid
4	CASP8	β-sitosterol
5	BCL2	β-sitosterol
6	JUN	β-sitosterol
7	PON1	β-sitosterol
8	PRKCA	β-sitosterol
9	TGFB1I1	β-sitosterol
10	ADRB2	β-sitosterol
11	RXRA	linoleyl acetate; arachidonic acid
12	TNFRSF1A	arachidonic acid
13	TNFRSF1B	arachidonic acid
14	ALOX5	arachidonic acid
15	SELP	arachidonic acid
16	C1R	arachidonic acid
17	COL1A2	arachidonic acid
18	PPARG	arachidonic acid
19	PRKCB	arachidonic acid
20	NOS3	arachidonic acid
21	PECAM1	arachidonic acid
22	MAPK1	arachidonic acid
23	EGF	arachidonic acid

Table 1: Overlapping targets between TCMSP-derived compound-associated targets and glomerulonephritis-associated genes. Compound-target records were retrieved from TCMSP, manually deduplicated, and standardized to official human gene symbols using UniProt.

Molecule Name	MW	AlogP	OB (%)	Caco-2	BBB	DL	FASA	HL
Arachidonic acid	304.52	6.41	45.57	1.2	0.58	0.2	0.28	4.39
Linoleyl acetate	308.56	6.85	42.1	1.36	1.08	0.2	0.21	7.48
β-sitosterol	414.79	8.08	36.91	1.32	0.99	0.75	0.23	5.36
Peroxyergosterol	428.72	6.73	44.39	0.86	0.43	0.82	0.24	4.06
Cerevisterol	432.76	5.26	39.52	0.35	-0.29	0.77	0.22	5.08
Cholesteryl palmitate	625.19	14.35	31.05	1.45	0.68	0.45	0.18	7.93
Cholesterol (CLR)	386.73	7.38	37.87	1.43	1.13	0.68	0.2	4.52

Table 2: Candidate constituents of C. sinensis retrieved from TCMSP using oral bioavailability and drug-likeness screening criteria. Compounds with oral bioavailability of at least 30% and drug-likeness of at least 0.18 were retained. MW, molecular weight; ALogP, predicted octanol-water partition coefficient; OB, oral bioavailability; Caco-2, predicted Caco-2 permeability; BBB, blood-brain barrier penetration; DL, drug-likeness; FASA, fractional negative accessible surface area; HL, half-life.

Supplementary Table 1. Primer sequences used for quantitative real-time PCR in HBZY-1 rat glomerular mesangial cells. All sequences are presented in the 5′ to 3′ direction.Please click here to download this file.

Supplementary File 1. Full scan images for Western blot.Please click here to download this file.

Discussion

This study integrated network pharmacology, reference-ligand-validated molecular docking, and compound-level experiments to examine a PTGS2-associated hypothesis derived from candidate constituents of C. sinensis. Seven constituents met the predefined TCMSP screening criteria, and 23 targets overlapped with the glomerulonephritis-associated gene set. Protein interaction analysis prioritized nine central targets, among which PTGS2 was selected for docking and experimental assessment. Redocking of celecoxib reproduced its crystallographic pose, while β-sitosterol and linoleyl acetate showed moderate predicted compatibility with the COX-2 docking region. In HBZY-1 mesangial cells, β-sitosterol reduced LPS-induced inflammatory gene expression, COX-2 abundance, and PGE2 production, and reduced TGF-β1-induced expression of profibrotic markers. These findings connect computational prioritization with cellular phenotypes, but they support a candidate regulatory relationship rather than a defined therapeutic mechanism.

The enrichment results placed the overlapping targets within biological processes related to inflammatory stimulation, cellular stress, lipid regulation, vascular function, and tissue remodeling. The enrichment of responses to lipopolysaccharide and bacterial products was consistent with the inflammatory model used for experimental validation. TNF, IL-17, AGE-RAGE, and HIF-1 related pathways also provided a plausible functional context for glomerular inflammation and maladaptive remodeling^6,17,18,19. Nevertheless, pathway enrichment reflects statistical concentration of annotated genes and does not establish activation or inhibition of a pathway by β-sitosterol¹². The prominent lipid and atherosclerosis annotation should therefore be interpreted as shared inflammatory, apoptotic, and vascular biology rather than as evidence that atherosclerosis represents the principal disease mechanism in CGN. Similarly, TCMSP-based oral bioavailability, drug likeness, and target records were used to prioritize candidates.

PTGS2 emerged as a relevant node because it was present in the compound disease intersection, displayed high network centrality, and was linked experimentally to changes in Ptgs2 transcription, COX-2 protein abundance, and PGE2 production. This agreement across computational and cellular levels strengthens the rationale for examining the PTGS2-associated response. It does not, however, imply that PTGS2 is uniformly pathogenic in the kidney. COX-derived prostanoids participate in renal blood flow, salt handling, and adaptive homeostasis, and may exert protective or harmful effects according to cell type, receptor distribution, disease phase, and local concentration^20,21. In experimental nephritis, celecoxib reduced PGE2-related output, while PGE2 itself influenced prostaglandin receptor expression in mesangial cells²². These observations illustrate that the biological consequences of COX-2 and PGE2 extend beyond a simple linear proinflammatory pathway^20,23. Comparison with celecoxib helps distinguish target expression from enzyme function. Celecoxib produced the clearest reduction in PGE2, consistent with its role as a functional COX-2 inhibitor but exerted a more limited effect on Ptgs2 expression and COX-2 protein abundance²². β-Sitosterol showed a broader pattern that included lower Tnf, Il6, Ptgs2, COX-2, and PGE2. This difference does not indicate superior COX-2 inhibitory potency. Rather, it suggests that β-sitosterol may alter the inflammatory state in which PTGS2 is induced, whereas celecoxib primarily limits prostaglandin synthesis through inhibition of catalytic activity. Direct enzyme assays, target engagement studies, and PTGS2 perturbation would be required to determine whether β-sitosterol also interacts functionally with COX-2.

The network findings also suggest that PTGS2 should not be interpreted as an isolated molecular node. MAPK1, JUN, and PPARG were co-prioritized in the protein interaction network, placing prostanoid synthesis within a broader signaling and transcriptional context. In renal mesangial cells, sphingosine 1-phosphate increased COX-2 expression and PGE2 formation through S1P receptor 2-dependent p42/p44 MAPK signaling, providing direct evidence that MAPK activity can lie upstream of the COX-2/PGE2 response in this cell type²⁴. Because JUN is a principal component of the AP-1 transcription factor complex, AP-1 may provide one interface between MAPK activation and PTGS2 transcription. Consistent with this possibility, 15-deoxy-Δ12,14-prostaglandin J2 suppressed IL-1β-induced COX-2 expression together with AP-1 activity in cultured mesangial cells²⁵. In nonrenal cellular models, β-sitosterol reduced ERK and p38 activation, NF-κB signaling, COX-2 expression, and inflammatory cytokine production after LPS exposure^14,15. These observations provide a plausible context for the coordinated reductions in Tnf, Il6, Ptgs2, COX-2, and PGE2 observed in the present study. However, the cited β-sitosterol studies were conducted in microglial or endothelial cells rather than renal mesangial cells, and MAPK1 phosphorylation and JUN activity were not measured here. The present results, therefore, do not establish a MAPK1-dependent or JUN-dependent mechanism of PTGS2 regulation by β-sitosterol. PPARG may intersect with the PTGS2-centered network through lipid-sensitive transcriptional regulation and modulation of the profibrotic mesangial-cell phenotype rather than as a simple linear downstream effector of PTGS2. In glomerular mesangial cells, natural and synthetic PPARγ ligands inhibited TGF-β1-induced fibronectin expression, while separate work showed that PPARγ agonists reduced TGF-β-induced mesangial-cell activation, α-SMA expression, and collagen IV accumulation through attenuation of PKA and CREB signaling^26,27. More recent work in diabetic mice and high-glucose-stimulated glomerular mesangial cells placed NF-κB/COX-2 and PPARγ/UCP2 within parallel regulatory axes associated with renal inflammation and fibrosis, supporting potential functional convergence without demonstrating direct PTGS2-PPARG coupling²⁸. These findings provide a biologically plausible context for the reductions in Col1a1, Acta2, and α-SMA observed in the present TGF-β1 model. However, PPARG expression, transcriptional activity, and ligand engagement were not evaluated. The present data, therefore, cannot attribute the observed profibrotic response to PPARG activation or define a sequential MAPK1-PTGS2-PPARG signaling pathway. Importantly, MAPK1 and PPARG were prioritized at the level of the integrated compound-disease network and should not be interpreted as direct β-sitosterol targets based on the current analysis.

The docking analysis should be interpreted within the same evidential boundary. A redocking RMSD of 0.876 Å indicated that the selected docking protocol could reproduce the known celecoxib orientation in the 3LN1 structure. The substantially less favorable predicted affinities of β-sitosterol and linoleyl acetate compared with celecoxib supported spatial accommodation within the predefined region, but not equivalent binding strength or pharmacological action^29,30. The 0.3 kcal/mol difference between linoleyl acetate and β-sitosterol is too small to justify a biological ranking in isolation. β-Sitosterol was selected for downstream testing because its network association was accompanied by a broader experimental literature relevant to inflammation and fibrosis^14,15,16. Previous studies reported reductions in TNF-α, IL-6, and COX-2 in LPS-exposed microglial and endothelial cells, together with changes in MAPK and NF-κB signaling^14,15. Arachidonic acid requires a different interpretation because it is an endogenous COX substrate²¹. Its presence in the TCMSP-derived network does not indicate an anti-inflammatory inhibitory action and illustrates that database-based compound target relationships do not define directionality. The concentration-related reductions in inflammatory markers observed in HBZY-1 cells are consistent with these earlier cellular findings, while extending the evidence to a glomerular mesangial context^14,15. The TGF-β1 model addressed a separate aspect of the disease process³¹. β-Sitosterol reduced Col1a1 and Acta2 transcription and decreased α-SMA abundance after TGF-β1 stimulation. Previous work in human alveolar epithelial cells similarly found that β-sitosterol reduced TGF-β1-induced collagen, fibronectin, and α-SMA expression, although that study involved a different tissue and cellular process¹⁶. The present findings therefore support activity against a profibrotic phenotype in mesangial cells but do not establish inhibition of a specific TGF-β receptor or SMAD-dependent mechanism³¹.

Several limitations define the scope of interpretation. The disease gene set was generated using the broad term glomerulonephritis and is not specific to one clinical or histological CGN subtype. The computational analysis used human annotations, docking used a murine COX-2 structure, and experiments were performed in rat HBZY-1 cells. Orthologous relationships support comparison, but species-dependent differences in regulation and ligand recognition remain possible. A single immortalized mesangial cell line cannot represent interactions among mesangial cells, podocytes, endothelial cells, infiltrating immune cells, and the tubulointerstitial compartment. The study also did not evaluate direct binding, COX-2 enzymatic activity, PTGS2 dependence, primary human renal cells, in vivo efficacy, pharmacokinetics, or renal exposure. Finally, testing a defined β-sitosterol preparation resolves the earlier inconsistency between predicted constituents and an uncharacterized extract, but it does not establish that β-sitosterol accounts for the effects of whole C. sinensis or that other constituents are inactive.

Taken together, the network, docking, and cell-based findings support β-sitosterol as a representative candidate associated with reduced inflammatory and profibrotic activation in HBZY-1 cells. The results are consistent with modulation of a PTGS2/COX-2/PGE2-related response but do not establish direct COX-2 binding, PTGS2-dependent causality, or therapeutic efficacy in CGN. Further target-engagement, genetic perturbation, primary-cell, and in vivo studies are required (Figure 7).

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
Absolute ethanol	Sigma-Aldrich/Merck	E7023	200 proof, non-denatured molecular biology-grade ethanol; solvent for β-sitosterol stock.
AutoDock Vina	Scripps Research	Version 1.1.2	Molecular docking simulation
Biosafety cabinet	Thermo Fisher Scientific	1300 Series A2	Sterile cell culture operations
Celecoxib	MedChemExpress	HY-14398	Selective COX-2 inhibitor used as the positive control; 10 mM stock in DMSO.
Cell Counting Kit-8	Dojindo Molecular Technologies, Inc.	CK04	WST-8 colorimetric assay for cell viability; absorbance measured at 450 nm.
Chemiluminescence imaging system	Bio-Rad Laboratories	ChemiDoc MP	Western blot image acquisition
clusterProfiler R package	Bioconductor	Version 4.10.0	GO and KEGG enrichment analysis
CO2 cell culture incubator	Thermo Fisher Scientific	Forma Series II 3111	Cell culture incubation at 37 °C and 5% CO2
COX-2 (D5H5) Rabbit Monoclonal Antibody	Cell Signaling Technology	12282	Western blot validated for human, mouse, and rat COX-2; used at 1:1,000.
CytoNCA plugin	Cytoscape App Store	Version 2.1.6	Hub target topological analysis
Cytoscape software	Cytoscape Consortium	Version 3.8.0	Network visualization and analysis
Dimethyl sulfoxide	Sigma-Aldrich/Merck	D2650	Sterile-filtered BioReagent, ≥99.7%; solvent for celecoxib stock.
DisGeNET database	Integrative Biomedical Informatics Group	Not applicable	Disease-gene retrieval for glomerulonephritis
Dulbecco's Modified Eagle Medium (DMEM)	Gibco/Thermo Fisher Scientific	C11995500BT	Basal medium for HBZY-1 cell culture
Enhanced chemiluminescence substrate	Beyotime Biotechnology	P0018S	Western blot band visualization
Fetal bovine serum	Gibco/Thermo Fisher Scientific	10099-141C	Cell culture serum supplement
GAPDH (D16H11) Rabbit Monoclonal Antibody	Cell Signaling Technology	5174	Western blot loading control; reactive with human, mouse, and rat GAPDH; used at 1:5,000 after in-house optimization.
GeneCards database	Weizmann Institute of Science	NA	Retrieval of glomerulonephritis-associated genes
GraphPad Prism	GraphPad Software	Version 9.0.0.121	Statistical analysis of in vitro experimental data
HRP-conjugated goat anti-rabbit IgG	Proteintech Group	SA00001-2	Secondary antibody for rabbit primary antibodies
ImageJ software	National Institutes of Health	Version 1.54	Western blot densitometric analysis
Inverted phase-contrast microscope	Olympus	CKX53	Cell morphology observation
org.Hs.eg.db R package	Bioconductor	Version 3.18.0	Human gene annotation for enrichment analysis
Penicillin-Streptomycin solution	Gibco/Thermo Fisher Scientific	15140122	Antibiotic supplement for cell culture
PGE2 Parameter Assay Kit	R&D Systems, a Bio-Techne brand	KGE004B	Multi-species competitive ELISA for cell culture supernatants; assay range 39–2,500 pg/mL; colorimetric detection at 450 nm.
PharmGKB database	Pharmacogenomics Knowledgebase	NA	Disease-related target retrieval
PubChem database	National Center for Biotechnology Information	NA	Retrieval of ligand structures
PVDF membrane	Millipore/Merck	IPVH00010	Protein transfer membrane for western blotting
PyMOL molecular graphics system	Schrödinger, LLC	Version 2.6.2	Docking pose visualization
R software	R Foundation for Statistical Computing	Version 4.3.2	Target intersection and enrichment analysis
Rat glomerular mesangial cell line (HBZY-1)	Procell Life Science & Technology Co., Ltd.	CL-0117	In vitro mesangial cell validation model
RCSB Protein Data Bank	Research Collaboratory for Structural Bioinformatics	PDB ID: 3LN1	Retrieval of PTGS2/COX2 crystal structure
Recombinant human TGF-β1 protein	PeproTech, a Thermo Fisher Scientific brand	100-21-10UG	HEK293-derived recombinant human TGF-β1; profibrotic stimulation at 10 ng/mL.
SDS-PAGE electrophoresis system	Bio-Rad Laboratories	Mini-PROTEAN Tetra	Protein separation by SDS-PAGE
STRING database	ELIXIR/STRING Consortium	Version 12.0	Protein-protein interaction network construction
TCMSP database	Traditional Chinese Medicine Systems Pharmacology Database	NA	Screening of Cordyceps sinensis candidate compounds
Therapeutic Target Database	Innovative Drug Research and Bioinformatics Group	NA	Therapeutic target retrieval
α-SMA primary antibody	Proteintech Group	14395-1-AP	Detection of α-SMA protein expression
β-Sitosterol (purity >98%)	MedChemExpress	HY-N0171A	CAS 83-46-5; defined compound used for cell treatment; 10 mM stock in absolute ethanol.