Naringin Alleviates Knee Osteoarthritis by Targeting TNF-α and PTGS2: An Integrated Network Pharmacology, Molecular Simulation, and Experimental Validation Study

Abstract

Knee osteoarthritis (KOA) is a chronic degenerative joint disorder driven largely by persistent inflammation and progressive cartilage damage. Naringin, a bioactive flavonoid abundant in citrus fruits, has shown potential anti-inflammatory effects; however, its molecular mechanisms in KOA remain unclear. In this study, an integrated approach combining network pharmacology, molecular docking, molecular dynamics (MD) simulations, and in vitro experiments was employed to investigate the anti-inflammatory effects of naringin in KOA. Network pharmacology analysis identified 59 potential KOA-related targets of naringin, among which TNF, PTGS2, TP53, CASP3, and PPARG were recognized as core targets. Functional enrichment indicated these targets were primarily associated with inflammation- and apoptosis-related pathways, especially the TNF and IL-17 signaling pathways. Molecular docking and MD simulations revealed strong binding affinity and stable interactions between naringin and the key inflammatory mediators TNF-α and PTGS2. In an IL-1β-stimulated C28/I2 human chondrocyte model, naringin dose-dependently improved cell viability and significantly suppressed TNF-α and PTGS2 expression at both mRNA and protein levels. These findings provide mechanistic evidence that naringin alleviates KOA-associated chondrocyte inflammation by modulating key inflammatory mediators, supporting its potential as an anti-inflammatory therapeutic candidate for KOA.

1. Introduction

Knee osteoarthritis (KOA) is a chronic degenerative joint disease characterized by progressive articular cartilage degradation, synovial inflammation, subchondral bone remodeling, and joint dysfunction, and it represents one of the leading causes of disability and reduced quality of life among middle-aged and elderly populations [1,2]. With the rapid aging of the global population, the prevalence of KOA has increased markedly in recent decades, posing a serious public health challenge worldwide [3,4]. Accumulating evidence indicates that the initiation and progression of KOA involve multiple pathological processes, including inflammatory responses, oxidative stress, apoptosis, autophagy dysregulation, and extracellular matrix metabolic imbalance. These processes are orchestrated through complex molecular mechanisms characterized by multi-target and multi-pathway interactions [5,6,7,8].

Currently, the clinical management of KOA mainly relies on nonsteroidal anti-inflammatory drugs (NSAIDs), analgesics, intra-articular injections, and surgical interventions. Although these treatments can alleviate pain and improve joint function to some extent, long-term use is often associated with adverse effects such as gastrointestinal, cardiovascular, hepatic, and renal toxicity. Moreover, there is still a lack of ideal disease-modifying drugs capable of effectively delaying or reversing cartilage degeneration [9,10]. Therefore, the development of safe and effective therapeutic strategies with disease-modifying potential has become a major focus of both basic and clinical KOA research.

Naringin, a natural flavonoid compound widely distributed in citrus fruits, exhibits a broad spectrum of biological activities, including anti-inflammatory, antioxidant, anti-apoptotic, and immunomodulatory effects [11]. Previous studies have demonstrated that naringin exerts beneficial pharmacological effects in cardiovascular diseases, metabolic disorders, and bone metabolism–related diseases [12,13,14,15]. In recent years, increasing evidence suggests that naringin may exert protective effects against osteoarthritis by suppressing the release of inflammatory mediators, modulating oxidative stress, and maintaining chondrocyte homeostasis [16,17,18]. However, systematic investigations into the molecular mechanisms underlying naringin-mediated intervention in KOA remain limited, and its key therapeutic targets and core signaling pathways have yet to be comprehensively elucidated.

Importantly, recent studies have shifted the flavonoid research paradigm from descriptive anti-inflammatory effects toward a mechanistic dissection of stress-responsive signaling networks, particularly those governing chondrocyte fate decisions. Glycosylated flavonoids, such as naringin, have been shown to modulate TNF- and IL-17-driven inflammatory cascades, transcriptional activation, and apoptosis-associated pathways in various degenerative and inflammatory disease models [11,19,20]. Given the central roles of TNF-α and prostaglandin-endoperoxide synthase 2 (PTGS2) in amplifying inflammatory signaling, promoting matrix catabolism, and driving chondrocyte dysfunction during KOA progression [21], focusing on this inflammatory axis may provide a more precise and biologically relevant framework for dissecting the chondroprotective mechanisms of naringin.

From a translational perspective, increasing attention has been paid to factors that critically influence the in vivo efficacy of flavonoids, including bioavailability, metabolic stability, and formulation strategies. As a glycosylated flavonoid, naringin undergoes intestinal metabolism and biotransformation, potentially limiting its systemic exposure despite robust in vitro activity. Recent studies have highlighted that formulation optimization and delivery strategies can substantially improve the pharmacokinetic behavior and therapeutic performance of flavonoids in inflammatory and degenerative disease models [22,23]. In this context, a clear understanding of the core molecular targets and signaling pathways regulated by naringin is a necessary prerequisite for its rational development as a disease-modifying or adjunct therapeutic agent for KOA.

Network pharmacology, an emerging research paradigm integrating systems biology, bioinformatics, and pharmacology, enables the systematic exploration of complex interactions among drugs, targets, pathways, and diseases at a network level. This approach is particularly well-suited for elucidating the multi-target mechanisms of natural products and active compounds derived from traditional medicines [24]. Meanwhile, molecular docking and molecular dynamics (MD) simulation techniques allow for the prediction of binding modes, affinities, stability, and dynamic behaviors between active compounds and target proteins at the molecular level, providing critical theoretical support for mechanistic studies [25,26,27,28]. The integration of network pharmacology, molecular docking, MD simulation, and experimental validation has thus become a widely accepted and powerful strategy in contemporary natural product research [29].

Based on this background, the present study focused on naringin as the investigational compound and systematically applied network pharmacology to identify its potential therapeutic targets and key signaling pathways involved in KOA treatment. Molecular docking and MD simulations were subsequently performed to evaluate the binding affinity and structural stability of naringin with the core target proteins. Furthermore, the computational predictions were validated through a combination of in vitro and in vivo experiments, thereby comprehensively elucidating the molecular mechanisms underlying naringin’s protective effects against KOA, as illustrated in the study workflow (Figure 1). This study aims to provide a solid scientific basis for the application of naringin in the prevention and treatment of KOA and to offer novel insights and theoretical support for the multi-target therapeutic mechanisms of natural bioactive compounds in KOA management.

2. Results

2.1. Screening of Potential Targets of Naringin for Osteoarthritis

A total of 385 potential naringin targets were predicted using the TCMSP, SwissTargetPrediction, and PharmMapper databases. After removing duplicates, 377 unique targets were obtained. Osteoarthritis-related transcriptomic data were downloaded from the GEO database (GSE8077), and differential expression analysis was performed between osteoarthritic and normal control samples. Differentially expressed genes (DEGs) were identified using the criteria of |log₂Fold Change| > 1 and p < 0.05, resulting in 384 DEGs, including 247 upregulated and 137 downregulated genes (Figure 2A,B). The |log₂Fold Change| > 1 cutoff ensures that only genes with at least a twofold expression difference are selected, highlighting those with substantial transcriptional alterations likely to be biologically relevant in OA. The p < 0.05 threshold provides statistical rigor, reducing the inclusion of genes with random or marginal expression changes.

Meanwhile, 3367 OA-related targets were retrieved from GeneCards, OMIM, DisGeNET, DrugBank, and TTD databases. After merging all sources and removing duplicates, a total of 3373 targets were obtained. By intersecting the naringin targets with the OA-related targets, 59 potential key targets of naringin for OA treatment were identified (Figure 2C and Supplementary Table S1). This intersection strategy enriches for targets that are both differentially expressed in OA and theoretically modulated by naringin, thereby prioritizing biologically meaningful targets while avoiding inflation of the network with marginal or nonspecific genes. A “compound–target–disease” network of these 59 intersecting targets was subsequently constructed using Cytoscape 3.9.1 (Figure 3 and Supplementary Table S2).

2.2. Protein–Protein Interaction Network and Core Target Screening

The intersecting targets were imported into the STRING database to construct a PPI network, which was visualized in Cytoscape. The PPI network comprised 59 nodes and 327 interaction edges (Figure 4A and Supplementary Table S3). Core targets were screened using the CytoHubba plugin based on the Degree algorithm. TNF, TP53, CASP3, ESR1, PPARG, MMP2, and PTGS2 were identified as highly connected hub genes in the network (Figure 4B).

2.3. Functional Annotation and KEGG Pathway Enrichment Analysis

GO functional annotation of the intersecting targets revealed that, at the biological process (BP) level, they were primarily enriched in responses to hypoxia, estradiol, hormonal stimulus, and environmental stimuli. At the cellular component (CC) level, targets were mainly located in the extracellular matrix, endoplasmic reticulum lumen, and chromosomal regions. At the molecular function (MF) level, targets were mainly involved in protein binding, receptor binding, and transcription factor activity regulation (Figure 5 and Supplementary Table S4). KEGG pathway enrichment analysis showed that the intersecting targets were significantly enriched in apoptosis, IL-17 signaling, TNF signaling, cancer pathways, p53 signaling, lipid metabolism, atherosclerosis, and cellular senescence pathways (Figure 6 and Supplementary Table S5). Figure 7A,B depict the top 50 enriched pathways, which can be categorized into four classes: Metabolism (lipid metabolism), Environmental Information Processing (signal transduction), Cellular Processes (cell growth and death), Organismal Systems (immune system, nervous system, endocrine system, circulatory system), and Human Diseases (cancer overview, infectious disease: viral, cancer: specific types, neurodegenerative diseases, drug resistance: antineoplastic).

2.4. Molecular Docking Results

Core targets identified by network pharmacology (CASP3, ESR1, MMP2, PTGS2, TNF, and TP53) were selected for molecular docking with naringin. The results showed that naringin could stably bind to the active pockets of all target proteins, with binding energies lower than −5.0 kcal/mol. Among them, PTGS2 (binding energy: −9.8 kcal/mol) and TNF (binding energy: −8.7 kcal/mol) exhibited the lowest binding energies and the most stable binding conformations (Figure 8 and

Table 1. Molecular docking results of key active ingredients and key target proteins.

Active Ingredient	Binding Energy (kcal/mol)
	TNF	TP53	CASP3	ESR1	MMP2	PTGS2
Naringin	−8.7	−8.0	−8.2	−8.2	−8.5	−9.8

).

2.5. Molecular Dynamics Simulation Results

To further investigate the stability of protein–ligand interactions, the TNF-α–naringin and PTGS2–naringin complexes with the most favorable binding energies were selected for 100 ns MD simulations, and their structural stability was evaluated from multiple perspectives. In terms of overall conformational stability, the root-mean-square deviation (RMSD) values of both complexes rapidly converged during the early stage of the simulation and remained stable thereafter (Figure 9A), indicating that the complexes reached a stable equilibrium state. The structural compactness of the target proteins was assessed by the radius of gyration (Rg). As shown in Figure 9B, the Rg values of both systems fluctuated within a narrow range throughout the simulation, confirming that naringin binding did not induce protein unfolding or structural loosening. Regarding the stability of the interaction interface, solvent-accessible surface area (SASA) analysis demonstrated that the buried surface area of naringin within the binding pocket remained nearly constant during the simulation (Figure 9C), suggesting a tight and stable binding interface. Hydrogen bond analysis further revealed that both complexes formed and maintained stable hydrogen-bonding networks (Figure 9D), providing direct evidence for high-affinity protein–ligand interactions. At the level of local residue dynamics and energetic features, root-mean-square fluctuation (RMSF) analysis showed that, except for the flexible terminal regions, residue fluctuations of the target proteins were effectively restrained in both complexes (Figure 9E), indicating preserved structural rigidity. Consistently, Gibbs free energy landscape analysis revealed well-defined and highly concentrated low-energy basins for both systems (Figure 9F), thermodynamically confirming the high conformational stability and binding advantage of the TNF-α–naringin and PTGS2–naringin complexes. To further elucidate the energetic determinants of ligand binding, molecular mechanics Poisson–Boltzmann surface area (MM-PBSA) analysis was performed. Per-residue energy decomposition demonstrated that the overall binding affinity was predominantly governed by a limited number of key residues (Figure 9G). Among these, GLN104 in TNF-α and LEU357 in PTGS2 exhibited notable negative energy contributions, indicating their favorable roles in stabilizing ligand binding. Global binding free energy component analysis revealed that molecular mechanics energy, particularly van der Waals interaction energy, served as the major driving force for complex formation, whereas Poisson–Boltzmann solvation energy exerted an unfavorable contribution to binding (Figure 9H). Comparative analysis indicated that the TNF-α–naringin complex exhibited stronger overall binding affinity than the PTGS2–naringin complex, which was consistent with the residue-level and energy decomposition results.

2.6. Effects of Naringin and Celecoxib on C28/I2 Cell Viability

To determine the safe working concentrations of naringin and the positive control drug celecoxib, the effects of various concentrations of each drug on C28/I2 cell viability were first assessed. CCK-8 assay results (Figure 10A,B) indicated that naringin had no significant cytotoxicity within the 0–40 μM range, while celecoxib showed negligible toxicity at 0–20 μM. Cell viability decreased significantly when naringin concentration reached 80 μM or higher, and celecoxib concentration reached 40 μM or higher (p < 0.05). Based on these results, 10, 20, and 40 μM were selected as low, medium, and high concentrations for naringin, and 20 μM was used for celecoxib in subsequent experiments.

To further evaluate the protective effects under pathological conditions, an IL-1β-induced chondrocyte inflammation model was established. Compared with the control group, treatment with 10 ng/mL IL-1β for 24 h significantly reduced cell viability (p < 0.0001), confirming the successful establishment of the inflammation model. In this model, co-treatment with naringin at 10, 20, or 40 μM or celecoxib at 20 μM significantly reversed IL-1β-induced cell viability reduction in a dose-dependent manner (Figure 10C). Notably, the high-dose naringin group (40 μM) exhibited comparable protective effects to the celecoxib group (20 μM) with no statistically significant difference (p > 0.05), suggesting that high-dose naringin achieved a protective efficacy similar to that of a clinically established anti-inflammatory drug.

2.7. Effects of Naringin and Celecoxib on Inflammatory Cytokine Secretion

ELISA results (Figure 11) showed that IL-1β stimulation markedly increased TNF-α, IL-6, and PGE2 levels in the culture supernatants (p < 0.0001). Naringin treatment dose-dependently suppressed the secretion of these cytokines. Compared with the model group, high-dose naringin (40 μM) significantly reduced TNF-α, IL-6, and PGE2 levels (p < 0.0001). Similarly, the positive control celecoxib group (20 μM) demonstrated significant inhibition, with no significant difference compared to high-dose naringin (p > 0.05), further confirming the anti-inflammatory potential of naringin.

2.8. Effects of Naringin and Celecoxib on TNF-α and PTGS2 mRNA Expression

qRT-PCR analysis revealed that IL-1β stimulation significantly upregulated the mRNA expression of TNF-α and PTGS2 (p < 0.0001). Treatment with naringin at all concentrations significantly downregulated the expression of these genes in a dose-dependent manner (Figure 12). Notably, medium- and high-dose naringin slightly outperformed celecoxib in suppressing PTGS2 expression, although the difference was not statistically significant (p > 0.05). The celecoxib group also significantly inhibited TNF-α and PTGS2 transcription (p < 0.0001), consistent with the predicted target mechanisms from network pharmacology.

2.9. Effects of Naringin and Celecoxib on TNF-α and PTGS2 Protein Expression

To further validate the network pharmacology predictions and transcriptional results at the protein level, Western blot analysis was performed. As shown in Figure 13A, IL-1β stimulation significantly increased TNF-α and PTGS2 protein levels. Intervention with different concentrations of naringin or celecoxib markedly suppressed the expression of both proteins. Quantitative analysis of band intensity (Figure 13B,C) indicated a dose-dependent inhibitory effect of naringin. Although the inhibitory effects of medium- and high-dose naringin on TNF-α and PTGS2 were numerically slightly lower than those of the celecoxib group (20 μM), the differences were not statistically significant (p > 0.05). These results further confirmed at the protein level that naringin could achieve anti-inflammatory efficacy comparable to the clinically established anti-inflammatory drug celecoxib.

3. Discussion

To provide an integrated overview of the proposed molecular mechanism, the interactions among naringin, key inflammatory targets, enriched signaling pathways, and observed cellular outcomes are schematically summarized in Figure 14. In this study, we employed an integrated research strategy combining network pharmacology, molecular docking, molecular dynamics simulations, and in vitro experimental validation to systematically elucidate the potential molecular mechanisms of naringin in the treatment of knee osteoarthritis (KOA). Network pharmacology analysis revealed a significant overlap between naringin-related targets and KOA-associated targets. Subsequent protein–protein interaction (PPI) network topology analysis identified several key nodes, including TNF, PTGS2, TP53, CASP3, and PPARG, suggesting that naringin may influence multiple biological processes related to inflammation, apoptosis, and metabolic regulation. Although TP53 and CASP3 exhibited relatively high degree values within the PPI network, hub gene prioritization in the present study was not based solely on topological metrics. Instead, biological relevance to the inflammatory microenvironment of KOA chondrocytes was considered a critical complementary criterion. TNF and PTGS2 function as upstream inflammatory amplifiers that directly regulate cytokine production, prostaglandin synthesis, and extracellular matrix catabolism, thereby acting as disease-driving nodes in the context of KOA pathogenesis. In contrast, TP53 and CASP3 are canonical stress- and apoptosis-associated regulators that are broadly activated across diverse cellular injury contexts and are more likely to represent downstream cellular responses rather than primary inflammatory triggers in chondrocytes. Accordingly, TNF and PTGS2 were prioritized for experimental validation to more precisely capture the inflammation-centered chondroprotective mechanism of naringin.

KEGG pathway enrichment analysis indicated that the potential targets of naringin were primarily involved in the TNF signaling pathway, IL-17 signaling pathway, apoptosis, and lipid metabolism-related pathways. Among these, the TNF signaling pathway plays a central role in amplifying inflammation and promoting cartilage matrix degradation in KOA [30,31,32]. TNF-α induces chondrocytes to produce various inflammatory mediators and matrix metalloproteinases, accelerating the degradation of type II collagen and proteoglycans, making it a critical driver of KOA progression [33,34]. PTGS2 (COX-2), as a rate-limiting enzyme in the inflammatory cascade, facilitates the synthesis of prostaglandin E2 (PGE2), further amplifying inflammation and aggravating cartilage damage [35,36]. Thus, TNF-α and PTGS2 are considered key therapeutic targets for anti-inflammatory treatment in KOA.

Molecular docking results demonstrated that naringin could stably bind to the active pockets of TNF-α and PTGS2 with low binding energies, indicating a strong theoretical binding affinity. Further MD simulations confirmed that the naringin–target protein complexes remained structurally stable over the simulation period, with minimal RMSD fluctuations and steady free energy changes, suggesting that the interactions between naringin and TNF-α or PTGS2 are stable under dynamic conditions. These findings provide molecular-level evidence supporting the anti-inflammatory potential of naringin.

In vitro validation using an IL-1β-induced C28/I2 human chondrocyte inflammation model, which mimics the inflammatory microenvironment of KOA, demonstrated that naringin at non-cytotoxic concentrations significantly improved IL-1β-induced reductions in cell viability and dose-dependently downregulated TNF-α and PTGS2 mRNA and protein expression. Notably, the anti-inflammatory effects of medium- and high-dose naringin were comparable to those of the positive control, celecoxib, highlighting its potential therapeutic value. These results are highly consistent with predictions from network pharmacology and molecular simulations, experimentally confirming that naringin may alleviate chondrocyte inflammatory damage by targeting TNF-α and PTGS2.

Overall, these findings suggest that naringin may directly target key molecules, such as TNF-α and PTGS2, thereby suppressing the overactivation of inflammatory signaling pathways, reducing inflammatory damage to chondrocytes, and slowing KOA progression. This “multi-target–multi-pathway” regulatory mode aligns with the pharmacological characteristics of natural flavonoids and corresponds well with the complex pathological mechanisms of KOA.

Furthermore, the study indicates that naringin may play a role in regulating chondrocyte apoptosis and autophagy. The high connectivity of CASP3 and TP53 in the PPI network, along with the significant enrichment of apoptosis-related pathways in KEGG analysis, suggests that naringin could mitigate chondrocyte death by modulating apoptotic pathways. This provides a foundation for future research into the potential chondroprotective and disease-modifying effects of naringin. Notably, although CASP3 and TP53 emerged as high-connectivity nodes in the PPI network and apoptosis-related pathways were significantly enriched in KEGG analysis, the present experimental validation was intentionally focused on inflammatory readouts. This design choice was based on the fact that TNF-α– and PTGS2-mediated inflammatory signaling represents an early and dominant pathological axis in IL-1β–stimulated chondrocytes, whereas apoptotic activation typically occurs as a secondary consequence of sustained inflammatory stress. Therefore, CASP3 and TP53 are more likely to function as downstream effectors that reflect cell-fate decisions rather than as primary regulatory targets in the acute inflammatory model employed in this study.

It should be emphasized that although high-dose naringin exhibited anti-inflammatory efficacy comparable to celecoxib in the IL-1β–stimulated chondrocyte model, the underlying pharmacological mechanisms of these two agents are fundamentally distinct. Celecoxib acts as a selective PTGS2 (COX-2) inhibitor, directly suppressing prostaglandin synthesis through enzymatic inhibition. In contrast, naringin does not function as a classical enzyme inhibitor but exerts regulatory effects at the transcriptional and signaling network levels by simultaneously modulating multiple upstream inflammatory mediators, including TNF-α, PTGS2, and related signaling pathways.

The strengths of this study lie in the comprehensive combination of computational predictions and experimental validation, providing multi-level evidence from network analysis and molecular interactions to cellular functions. However, several limitations exist. First, experiments were conducted only in C28/I2 cells, which may not fully recapitulate the complexity of the in vivo joint microenvironment. Second, the pharmacokinetics and long-term in vivo efficacy of naringin have not been evaluated. Third, network pharmacology and molecular simulation results depend on database information and may be incomplete or biased. Notably, while our experimental validation focused on the suppression of TNF-α and PTGS2, CASP3 and TP53 also emerged as high-connectivity nodes in the network analysis. This suggests that naringin may exert broader protective effects on chondrocyte survival under prolonged or severe stress conditions, which were not directly assessed in the current study. Future investigations incorporating apoptosis- and survival-oriented readouts, as well as animal models and clinical samples, are warranted to further explore the long-term efficacy and molecular mechanisms of naringin in chondroprotection, inflammation regulation, and joint function improvement.

4. Materials and Methods

4.1. Materials and Reagents

The human chondrocyte cell line C28/I2 (ATCC; Manassas, VA, USA), naringin (Shanghai YuanYe Bio-Technology Co., Ltd., Cat. No.: S24000; Shanghai, China), celecoxib (MedChemExpress, Cat. No.: HY-14398; Monmouth Junction, NJ, USA), DMEM/F12 medium (Gibco, Cat. No.: 11330032; Grand Island, NY, USA), fetal bovine serum (Gibco, Cat. No.: 10099141C; Grand Island, NY, USA), penicillin-streptomycin solution (Gibco, Cat. No.: 15140122; Grand Island, NY, USA), recombinant human interleukin-1β (PeproTech, Cat. No.: 200-01B; Rocky Hill, NJ, USA), CCK-8 assay kit (Dojindo, Cat. No.: CK04; Kumamoto, Japan), TRIzol reagent (Thermo Fisher Scientific, Cat. No.: 15596026; Waltham, MA, USA), ReverTra Ace qPCR RT Kit (Toyobo, Cat. No.: FSQ-101; Osaka, Japan), PowerUp SYBR Green Master Mix (Applied Biosystems, Cat. No.: A25742; Foster City, CA, USA), RIPA lysis buffer (Beyotime Biotechnology, Cat. No.: P0013B; Shanghai, China), BCA assay kit (Beyotime Biotechnology, Cat. No.: P0012S; Shanghai, China), ELISA kits for PGE2, TNF-α, and IL-6 (Shanghai Enzyme-linked Biotechnology Co., Ltd., Cat. Nos.: ml002859, ml002095, ml002293; Shanghai, China), PTGS2 antibody (Abcam, Cat. No.: ab52915; Cambridge, UK), TNF-α antibody (Abcam, Cat. No.: ab41037; Cambridge, UK), HRP-conjugated goat anti-rabbit secondary antibody (Beijing Zhongshan Jinqiao Biotechnology Co., Ltd., Cat. No.: ZB-2301; Beijing, China), HRP-conjugated goat anti-mouse secondary antibody (Beijing Zhongshan Jinqiao Biotechnology Co., Ltd., Cat. No.: ZB-2305; Beijing, China), Thermo Scientific Heracell™ 240i CO₂ incubator (Thermo Fisher Scientific; Waltham, MA, USA), Suzhou Purification SW-CJ-2FD laminar flow hood (Suzhou Purification; Suzhou, China), Olympus CKX41 inverted phase-contrast microscope (Olympus; Tokyo, Japan), BioTek Synergy H1 microplate reader (BioTek; Winooski, VT, USA), Applied Biosystems QuantStudio 5 qPCR system (Applied Biosystems; Foster City, CA, USA), Bio-Rad Mini-PROTEAN Tetra electrophoresis system (Bio-Rad; Hercules, CA, USA), Trans-Blot Turbo transfer system (Bio-Rad; Hercules, CA, USA), and Tanon-5200Multi chemiluminescence imaging system (Tanon; Shanghai, China) were used in this study.

4.2. Network Pharmacology Analysis

Potential targets of naringin were predicted using the TCMSP, SwissTargetPrediction, and PharmMapper databases. The obtained targets were deduplicated, standardized, and converted to Gene Symbols, with the species limited to Homo sapiens. KOA-related targets were retrieved from GeneCards, OMIM, DisGeNET, DrugBank, and TTD databases using the keywords “knee osteoarthritis” or “osteoarthritis.” The intersection of naringin targets and disease targets was analyzed using Venn diagrams to identify potential therapeutic targets and construct a “herb–component–target–disease” network. The intersecting targets were then imported into the STRING database to construct a protein–protein interaction (PPI) network, which was visualized in Cytoscape. Core targets were screened based on topological parameters such as Degree using the CytoHubba plugin. Gene Ontology (GO) functional annotation and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis were further performed on the intersecting targets using the Metascape platform.

4.3. Molecular Docking

Core targets identified by network pharmacology were selected as receptors. Three-dimensional structures of the target proteins were obtained from the Protein Data Bank (PDB; http://www.rcsb.org/), with preference for high-resolution structures containing native ligands. The 3D structure of naringin was downloaded from PubChem and energy-minimized using Chem3D. Receptor proteins were processed with AutoDock Tools 1.5.7 by removing water molecules, adding hydrogen atoms, and assigning charges. The docking grid was set to cover the active sites. Molecular docking was performed using AutoDock Vina to evaluate the binding affinity between naringin and core targets. Binding modes and key amino acid interactions were visualized in PyMOL 2.4.

4.4. Molecular Dynamics Simulation

The “naringin–core target protein” complex with the lowest binding energy and the most favorable binding mode from molecular docking was selected for MD simulation using GROMACS 2023. The CHARMM force field was applied, and naringin parameters were generated using CGenFF. The system was solvated in TIP3P water and neutralized with appropriate Na⁺/Cl⁻ ions. After energy minimization, equilibration was performed under NVT and NPT ensembles. MD simulations were subsequently run for 100 ns (5 million steps, 2 fs per step) at 300 K and 1 atm. Trajectory analysis was conducted using GROMACS tools. Root-mean-square deviation (RMSD), interaction fractions, and Gibbs free energy were calculated. To further assess binding energetics, MM-PBSA calculations were performed on equilibrated MD trajectories using the g_mmpbsa tool. The total binding free energy was decomposed into van der Waals, electrostatic, polar solvation, and nonpolar solvation contributions. Additionally, per-residue energy decomposition was performed to identify key residues contributing to ligand stabilization.

4.5. Cell Culture and Inflammatory Model

C28/I2 cells were cultured in DMEM/F12 complete medium supplemented with 10% FBS and 1% penicillin–streptomycin at 37 °C in a humidified incubator with 5% CO₂. Cells in the logarithmic growth phase were used for experiments. To simulate the inflammatory microenvironment of KOA, an in vitro inflammation model was established. Cells were seeded in appropriate plates and cultured until 70–80% confluence, then serum-starved for 12 h. Experimental groups were treated with complete medium containing 10 ng/mL IL-1β for 24 h to induce chondrocyte inflammation and degeneration.

4.6. Experimental Grouping

Cells were divided into six groups: Control (untreated), Model (10 ng/mL IL-1β for 24 h), Naringin low-dose (Nar-L, IL-1β + low concentration naringin for 24 h), Naringin medium-dose (Nar-M, IL-1β + medium concentration naringin for 24 h), Naringin high-dose (Nar-H, IL-1β + high concentration naringin for 24 h), and Positive Control (PC, IL-1β + celecoxib for 24 h).

4.7. Cell Viability Assay

C28/I2 cells were seeded at 5 × 10³ cells/well in 96-well plates. After attachment, cells were treated as follows: (a) naringin or celecoxib at 0, 5, 10, 20, 40, 80, and 160 μM for 24 h; (b) 10 ng/mL IL-1β for 24 h; (c) IL-1β combined with different concentrations of naringin or celecoxib for 24 h. After treatment, 10 μL of CCK-8 solution was added to each well and incubated for 2 h. Absorbance was measured at 450 nm using a microplate reader, and cell viability was calculated. Non-toxic and effective concentrations were selected for subsequent experiments.

4.8. ELISA for Inflammatory Cytokines

After treatment, culture supernatants from each group were collected and centrifuged at 1000× g for 10 min at 4 °C to remove cell debris. The concentrations of TNF-α, IL-6, and PGE2 in the supernatants were measured using the corresponding ELISA kits according to the manufacturers’ protocols. Absorbance was recorded at 450 nm, and cytokine levels (pg/mL) were calculated based on standard curves.

4.9. Quantitative Real-Time PCR

Total RNA was extracted using TRIzol reagent, and RNA concentration and purity were determined. Reverse transcription was performed to synthesize cDNA. qRT-PCR was conducted using SYBR Green Master Mix with β-actin as an internal control to quantify TNF-α and PTGS2 mRNA levels. Relative gene expression was calculated using the 2^−ΔΔCT method. Primer sequences are listed in

Table 2. Primer Sequences.

Gene		Sequence
TNF-α	Forward	GTGAGGAGCACGTAGTCGG
	Reverse	ACGGCATGGATCTCAAAGACA
PTGS2	Forward	CGAGGTGTATGTATGAGTGT
	Reverse	AGTGGGTAAGTATGTAGTGC
β-actin	Forward	TCACCATGGATGATGATATCGC
	Reverse	ATAGGAATCCTTCTGACCCATGC

.

4.10. Western Blot Analysis

Total protein was extracted with RIPA lysis buffer and quantified using the BCA method. Equal amounts of protein were separated by SDS-PAGE and transferred to PVDF membranes. Membranes were blocked with 5% nonfat milk and incubated overnight at 4 °C with primary antibodies against TNF-α and PTGS2. After washing with TBST, membranes were incubated with HRP-conjugated secondary antibodies for 1 h at room temperature. Protein bands were visualized using ECL, and densitometric analysis was performed using ImageJ 2. β-actin was used as an internal reference to calculate relative protein expression.

4.11. Statistical Analysis

All experiments were independently repeated at least three times. Data are expressed as mean ± standard deviation (mean ± SD). Statistical analysis was performed using GraphPad Prism 8.0. Differences among multiple groups were analyzed by one-way ANOVA, and pairwise comparisons were conducted using Tukey’s post hoc test. p < 0.05 was considered statistically significant.

5. Conclusions

In summary, this study employed a combined strategy of network pharmacology, molecular docking, MD simulation, and in vitro validation to systematically elucidate the potential molecular mechanisms of naringin in the treatment of KOA. Our results suggest that naringin may target key inflammatory molecules such as TNF-α and PTGS2, suppress the overactivation of inflammatory signaling pathways including TNF and IL-17, and thereby alleviate IL-1β-induced chondrocyte inflammatory damage and improve cell viability. Collectively, these findings indicate that naringin functions as a natural lead compound with therapeutic potential against KOA inflammation, providing a theoretical and experimental foundation for its further development as a safe, multi-target natural therapeutic for the prevention and treatment of osteoarthritis.