Discovery of Apigenin as a Multi-Target Therapeutic Drug for Osteoarthritis Complicated With Type 2 Diabetes Mellitus: Based on Network Pharmacology and in Vitro Experimental Validation

Abstract

Background

Osteoarthritis (OA) frequently coexists with type 2 diabetes mellitus (T2DM), and both diseases share common pathological mechanisms, particularly chronic inflammation and metabolic dysregulation. However, the shared molecular networks and potential multi-target therapeutic agents remain poorly understood.

Methods

Gene expression datasets related to OA and T2DM were obtained from the GEO database. Common differentially expressed genes (DEGs) were identified and analyzed using Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses. A protein–protein interaction (PPI) network was constructed to identify hub genes. Molecular docking was subsequently performed to screen potential natural compounds targeting these hub genes. Experimental validation was conducted in an IL-1β plus high-glucose-induced SW1353 chondrocyte comorbidity model using qRT-PCR, Western blotting, CCK-8, TUNEL, and ELISA assays.

Results

A total of 99 shared DEGs were identified between OA and T2DM, mainly enriched in inflammation- and metabolism-related pathways, including TNF, PI3K–Akt, NF-κB, and AGE–RAGE signaling pathways. Thirteen hub genes were screened, among which CX3CR1, IRF8, and NCF1 were identified as key targets. Molecular docking demonstrated that the natural flavonoid apigenin exhibited strong binding affinity toward these proteins, particularly CX3CR1 (binding energy: −9.3 kcal/mol). In vitro experiments showed that CX3CR1, IRF8, and NCF1 mRNA expression levels were significantly elevated in the OA–T2DM model (P < 0.01). Apigenin treatment dose-dependently inhibited NF-κB activation, improved cell viability, reduced apoptosis, and decreased IL-6 and TNF-α secretion (all P < 0.01).

Conclusions

This study combined bioinformatics analysis, molecular docking, and experimental validation to elucidate the shared inflammatory molecular mechanisms underlying OA and T2DM. The findings provide the first evidence that apigenin alleviates OA–T2DM-associated cellular injury through a multi-target anti-inflammatory mechanism, supporting its potential as a therapeutic candidate for OA complicated with T2DM.

Keywords

apigenin osteoarthritis type 2 Diabetes Mellitus network pharmacology molecular docking NF-κB signaling pathway

1. Introduction

Osteoarthritis (OA) and Type 2 Diabetes Mellitus (T2DM) represent two chronic conditions that are widely prevalent globally, especially among older adults.¹ OA is a condition characterized by the deterioration of articular cartilage, inflammation of the synovial membrane, and changes in the subchondral bone, leading to primary symptoms such as joint pain, stiffness, and impaired function, which significantly affect the quality of life of those affected.² In contrast, T2DM is a metabolic condition defined by insulin resistance and dysfunction of pancreatic β-cell, resulting in irregular blood glucose levels and a variety of complications, including cardiovascular issues, kidney disease, and nerve damage.³ Recent epidemiological research has highlighted a notable comorbidity between OA and T2DM.^4,5 There is a considerable increase in the occurrence of OA among individuals with T2DM, while those suffering from OA also show a heightened risk for developing T2DM, indicating a potential shared pathophysiological mechanism connecting both disorders.⁶

Despite differences in clinical manifestations and affected tissues, OA and T2DM share multiple pathological mechanisms,Chronic low-grade inflammation is a pervasive physiological condition that can subtly yet significantly affect various aspects of health. This type of inflammation is characterized by a prolonged immune response, which can lead to an array of health complications over time. In addition to inflammation, oxidative stress plays a crucial role in cellular damage, resulting from an imbalance between free radicals and antioxidants in the body. This oxidative imbalance further exacerbates the risk of chronic diseases.Moreover, the accumulation of advanced glycation end products (AGEs) contributes to the deterioration of health by promoting inflammation and oxidative stress. AGEs are harmful compounds that form when proteins or fats combine with sugars in the bloodstream, and their buildup can interfere with normal metabolic processes. Lastly, abnormalities in the insulin signaling pathway remain a critical factor in the development of various metabolic disorders.^7,8 Insulin resistance, which arises when cells no longer respond effectively to insulin, can lead to severe health implications, including type 2 diabetes and cardiovascular diseases. Collectively, these mechanisms underscore the interconnectedness of chronic inflammation, oxidative stress, AGE accumulation, and insulin signaling in their impact on metabolic health and disease progression.For instance, Cytokines involved in inflammation, including tumor necrosis factor-alpha (TNF-α), interleukin-6 (IL-6), and interleukin-1 beta (IL-1β), are crucial in the degradation of cartilage in OA and in contributing to insulin resistance in T2DM.^9,10 Furthermore, AGEs, through binding to their receptor (RAGE), not only promote the development of diabetic complications but also accelerate chondrocyte apoptosis and extracellular matrix degradation, exacerbating the OA process.¹¹ Although these shared mechanisms have been initially elucidated, the specific molecular interactions, shared gene networks, and key signaling pathways between OA and T2DM remain unclear, limiting the development of therapeutic strategies targeting both conditions.

As bioinformatics undergoes swift progression,employing genome-wide expression datasets generated at high speed and volume to screen for disease-related differentially expressed genes (DEGs) has become a crucial approach for uncovering disease molecular mechanisms.¹²By integrating transcriptomic data of OA and T2DM from public databases, shared DEGs can be systematically identified, followed by functional enrichment analysis to reveal their involvement in biological processes and signaling pathways. Further,constructing a protein-protein interaction (PPI) network enables the identification of key genes (hub genes) at the core of the network, which are likely to exert a dominant influence on the common pathogenesis of both diseases.Building on the identification of key targets, molecular docking technology provides an efficient and cost-effective computational tool for screening potential therapeutic compounds.¹³ By modeling the spatial binding events between low-molecular-weight ligands and their cognate proteins,binding affinity and interaction modes can be predicted, facilitating the selection of promising lead compounds. Subsequently, validating the efficacy of these compounds in disease models through in vitro experiments is a crucial step in advancing their clinical translation.^14,15

At present, investigations have addressed the molecular underpinnings of OA or T2DM separately, whereas integrated insights into their co-occurring pathobiology remain sparse.¹⁶ Systematic screening of shared key genes between OA and T2DM and the exploration of multi-target therapeutic strategies hold significant scientific and clinical value.¹⁷ Therefore, this study aims to systematically screen for shared DEGs between OA and T2DM, identify key signaling pathways and core targets, and screen for small molecule compounds that efficiently bind to these targets through molecular docking, using an integrated research strategy of bioinformatics, computational simulation, and experimental validation. Subsequently, an in vitro model of OA complicated with T2DM was established using human chondrocytes to validate the expression changes of key targets and the therapeutic potential of the lead compound. This study seeks to elucidate the molecular basis underlying the comorbidity of OA and T2DM, providing new targets and experimental evidence for the development of common therapeutic strategies for both conditions and concurrently supplying methodological guidance for comorbidity investigations.

2. Materials and Methods

2.1. Bioinformatics Analysis

2.1.1. Data Acquisition, Preprocessing, and Selection Criteria

Gene expression microarray datasets for osteoarthritis (OA) and type 2 diabetes mellitus (T2DM) were downloaded from the NCBI Gene Expression Omnibus (GEO) database.For OA, the dataset with accession number GSE98918 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE98918) was selected. This dataset is based on the GPL20844 platform (Agilent-072364 SurePrint G3 Human GE v3 8x60K Microarray) and includes 24 human meniscus samples (12 from OA patients and 12 from non-OA controls). Inclusion criteria for OA samples were: (1) diagnosis of knee osteoarthritis according to the American College of Rheumatology criteria; (2) undergoing total knee arthroplasty. Non-OA control samples were obtained from patients undergoing arthroscopic partial meniscectomy due to meniscus tear, with no clinical or arthroscopic evidence of OA (Kellgren-Lawrence grade 0).For T2DM, the dataset with accession number GSE76896 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE76896) was selected. This dataset is based on the GPL570 platform (Affymetrix Human Genome U133 Plus 2.0 Array) and includes islet transcriptome profiles from organ donors and pancreatectomised patients. After data preprocessing and quality control, a total of 232 samples were included in the present analysis, comprising 174 non-diabetic controls and 58 diabetic cases (including type 2 diabetes, impaired glucose tolerance, and type 3c diabetes). The diagnosis of T2DM was based on clinical history, treatment with glucose-lowering drugs, and absence of anti-GAD65 autoantibodies. Exclusion criteria were: (1) organ donor samples with no history of diabetes but with blood fructosamine >285 μmol/L or glucose >11.1 mmol/L; (2) samples with insulin levels <1 standard deviation from the within-group mean.To ensure data comparability and minimize batch effects, all raw expression data were background-adjusted, quantile-normalized, and probe-summarized using the affy package (v1.74.0) in R v4.2.0. Batch correction was performed using the ComBat algorithm in the sva package (v3.44.0) when necessary. Differential expression analysis was conducted using the limma package (v3.52.4).

Although the OA dataset (GSE98918) was derived from meniscus tissue and the T2DM dataset (GSE76896) originated from pancreatic islets, the primary aim of this study was not to identify tissue-specific alterations but rather to explore the shared systemic molecular mechanisms underlying OA-T2DM comorbidity. Increasing evidence suggests that chronic low-grade inflammation, oxidative stress, metabolic dysregulation, and AGE–RAGE signaling are common pathological features linking these two diseases across different tissues. Therefore, datasets were selected based on the following criteria: (1) clearly defined clinical diagnosis; (2) sufficient sample size; (3) availability of raw data for standardized preprocessing; and (4) representation of disease-relevant pathological processes. To improve comparability and reduce inter-platform and inter-tissue variability, independent DEG analysis, quantile normalization,and batch effect correction using the ComBat algorithm were performed before intersection analysis.

2.1.2. Differential Expression Gene Screening

The limma algorithm was applied to the normalized dataset to identify genes with significant expression changes.Expression shifts within the OA and T2DM datasets were quantified independently.with |log2(Fold Change)| > 1 and an calibrated P-value (adj. P-value) < 0.05 serving as thresholds to screen differentially expressed genes.(DEGs).The threshold of |log2(Fold Change)| > 1 and adjusted P-value < 0.05 was selected based on widely accepted transcriptomic analysis standards to balance sensitivity and specificity in identifying biologically meaningful differentially expressed genes while minimizing false-positive findings.Differential expression results were visualized using volcano plots and heatmaps to illustrate the distribution of significantly upregulated and downregulated genes.

2.1.3. Screening and Functional Enrichment Analysis of Shared DEGs

We employed Venn analysis to intersect the differentially expressed genes from OA and T2DM datasets, thereby identifying the shared DEGs.The analysis of functional enrichment for the shared differentially expressed genes (DEGs) was conducted utilizing the clusterProfiler package, focusing on Gene Ontology (GO) categories including biological processes, cellular components, and molecular functions, in addition to pathways from the Kyoto Encyclopedia of Genes and Genomes (KEGG).Statistical significance was defined asP-value < 0.05 and Q-value < 0.1.Bar and bubble charts were generated to illustrate the enrichment outcomes.

2.1.4. Construction of Protein-Protein Interaction Network and Identification of Core Genes

Shared DEGs were uploaded to STRING v11.5 to generate a PPI map, retaining edges whose confidence exceeded 0.7. After the network was transferred into Cytoscape 3.9.1, the CytoHubba extension evaluated node centrality with four metrics—Maximal Clique Centrality (MCC), Degree Centrality (Degree), Maximum Neighborhood Component (MNC), and Edge Percolated Component (EPC). Genes consistently ranked at the top across all four algorithms were consolidated, yielding 13 hub genes for downstream analyses.To improve the robustness and reliability of hub gene identification, four independent topological algorithms (MCC, Degree, MNC, and EPC) were simultaneously applied. Only genes consistently ranked among the top candidates across multiple algorithms were retained for subsequent analyses, thereby reducing potential bias caused by relying on a single network centrality metric.

2.2. Molecular Docking

The two-dimensional (2D) representations of small molecule ligands were sourced from the PubChem database, accessible at https://pubchem.ncbi.nlm.nih.gov/. These 2D structures were subsequently transformed into three-dimensional (3D) models utilizing ChemOffice software. The final 3D representations were then saved in the mol2 file format for further analysis and application.Protein targets were chosen from the RCSB PDB database (https://www.rcsb.org/) by utilizing crystal structures with high resolution.Solvent molecules and phosphate groups were eliminated from the proteins with PyMOL 2.6, and the resulting structures were stored as PDB files.Both macromolecular and small molecule structureswere handled through AutoDock 1.5.6,including adding hydrogens to proteins and small molecules, and determining torsion forces for small molecules. The docking box coordinates were precisely defined to ensure accurate molecular docking analysis. This process utilized AutoDockVina software, which is specifically designed to investigate the interactions between proteins and ligands. Following the docking procedure, the most favorable conformation of the protein-ligand complexes was identified by systematically comparing their respective docking scores, allowing for the selection of the most promising interactions for further study.Two- and three-dimensional interaction maps between the ligands and critical residues were generated in Discovery Studio 2019 and PyMOL 2.6 for graphical interpretation of the docking output.As a rule, affinities below −5.0 kcal/mol denote robust ligand engagement, whereas values under −7.0 kcal/mol reflect exceptionally tight binding.It should be noted that molecular docking provides only preliminary computational predictions of ligand–target interactions based on static protein conformations. The docking results do not fully reflect dynamic intracellular environments or protein conformational flexibility. Therefore, the predicted binding affinities require further validation using molecular dynamics simulations and experimental binding assays such as surface plasmon resonance (SPR) or isothermal titration calorimetry (ITC).

2.3. Experimental Validation

2.3.1. Cell Culture and Disease Model Establishment

The human chondrocyte cell line (SW1353) (Catalog No.: HTB-94, American Type Culture Collection, Manassas, VA, USA) was cultured in DMEM/F12 medium (Catalog No.: 11330032, Thermo Fisher Scientific, New York, NY, USA) supplemented with 10% fetal bovine serum (Catalog No.: 10099-141, Thermo Fisher Scientific, New York, NY, USA) and 1% penicillin-streptomycin solution (Catalog No.: 15140-122, Thermo Fisher Scientific, New York, NY, USA) at 37°C in a 5% CO₂ atmosphere. To simulate disease conditions, the following experimental groups were established: Control group (normal glucose concentration, 5.5 mM); OA model group (stimulated with 10 ng/mL human recombinant interleukin-1β (IL-1β) (Catalog No.: 200-01B, PeproTech Inc., New Jersey, USA) for 24 hours); T2DM model group (cultured in medium containing high glucose concentration, 30 mM D-(+)-glucose (Catalog No.: G7021, Sigma-Aldrich Corporation, Missouri, USA)); OA+T2DM comorbidity model group (treated with both IL-1β (10 ng/mL) and high glucose (30 mM)); Treatment groups (comorbidity model supplemented with different concentrations of the lead compound (apigenin, Catalog No.: A3145, Sigma-Aldrich Corporation, Missouri, USA)).

2.3.2. Validation of Key Gene Expression (qRT-PCR)

TRIzol reagent was employed to recover the complete RNA complement from the cells.(Catalog No.: 15596026, Thermo Fisher Scientific, Massachusetts,USA).The assessment of RNA purity and concentration was conducted by means of a NanoDrop microvolume spectrophotometer.(Model No.: ND-2000, Thermo Fisher Scientific, Massachusetts, USA). The synthesis of cDNA was performed using the PrimeScript RT Reverse Transcription Kit.(Catalog No.: RR047A, TaKaRa Bio Inc., Shiga, Japan). The execution of real-time quantitative PCR was performed with TB Green Premix Ex Taq II.(Catalog No.: RR820A, TaKaRa Bio Inc., Shiga, Japan).PCR amplification and detection were carried out on a QuantStudio5 platform (Model No.: QuantStudio 5, Thermo Fisher Scientific, California, USA). The thermal cycling protocol comprised an initial denaturation phase at 95°C for 30 sec, succeeded by 40 amplification cycles, each involving a 5-sec denaturation at 95°C and a combined annealing step at 60°C for 30 seconds.We employed GAPDH as the housekeeping gene for normalization and determined relative mRNA abundance by applying the 2^-ΔΔCt analytical method.

2.3.3. Protein Level and Pathway Validation (Western Blotting)

Cellular protein extraction was performed utilizing RIPA lysis buffer supplemented with protease inhibitors (Catalog No.: P0013B, Beyotime Institute of Biotechnology, Shanghai, China). Protein quantification was carried out employing the BCA Protein Assay Kit. (Catalog No.: P0012, Beyotime Institute of Biotechnology, Shanghai, China). Equivalent quantities of protein were resolved by SDS-PAGE (12%) and subsequently electroblotted onto PVDF membranes.(Catalog No.: IPVH00010, Merck KGaA, Darmstadt, Germany). We blocked the membranes with 5% skimmed milk(Catalog No.: 9999S, Cell Signaling Technology, Inc., Massachusetts, USA)followed by incubation with primary antibodies(anti-PTPRC, Catalog No.: 13017-1-AP, Proteintech Group, Inc., Illinois, USA; anti-p65, Catalog No.: 8242S; anti-p-p65, Catalog No.: 3033S, Cell Signaling Technology, Inc., Massachusetts, USA). for 16 hours at 4°C under constant rotation.Following three TBST washes (10 min each), the membranes were exposed to horseradish peroxidase-conjugated goat anti-rabbit secondary antibody(1:5000 dilution,Catalog No.: 7074S, Cell Signaling Technology, Inc., Massachusetts, USA)for 60 minutes under ambient temperature conditions.Protein signals were detected via enhanced chemiluminescence (ECL) substrate incubation(Catalog No.: WBKLS0100, Merck KGaA, Millipore brand, Darmstadt, Germany), followed by quantitative densitometric analysis of band intensities using ImageJ.

2.3.4. Cell Function Assays

We evaluated cellular viability employing the CCK-8 reagent (Catalog No.: CK04, Dojindo Molecular Technologies, Inc., Kumamoto, Japan). Cell suspensions were aliquoted into 96-well microplates at a density of 96 cells(Catalog No.: 3599, Corning Incorporated, New York, NY, USA). Post-treatment, we introduced 10 μL of CCK-8 solution per well, followed by a 2-hour incubation period under standard culture conditions.The absorbance readings were acquired at 450 nm wavelength using a Multiskan GO microplate spectrophotometer(Model No.: Multiskan GO, Thermo Fisher Scientific, Massachusetts, USA). Apoptotic cells were identified employing the Tunel assay kit(Catalog No.: 11684817910, Roche Diagnostics GmbH, Basel, Switzerland) to analyze apoptosis rates in each group. Secretion levels of inflammatory cytokines were measured using commercially available human TNF-α (Catalog No.: 88-7346-22) and human IL-6 (Catalog No.: 88-7066-22) ELISA kits (Thermo Fisher Scientific, Massachusetts, USA), Measuring the secretion levels of inflammatory cytokines.

2.4. Statistical Analysis

All experiments were independently repeated at least three times, and the data are presented as mean ± standard deviation (SD). Statistical analyses were performed using GraphPad Prism software (version 10.1.2, GraphPad Software, San Diego, CA, USA) and R software (version 4.2.0).Before statistical comparison, data normality was assessed using the Shapiro–Wilk test, and homogeneity of variance was evaluated using Levene’s test. For comparisons between two groups, an unpaired Student’s t-test was applied. For comparisons among multiple groups, one-way analysis of variance (ANOVA) followed by Tukey’s post hoc test was used.Differential expression analysis of microarray datasets was performed using the limma package in R, with |log₂(fold change)| > 1 and adjusted P-value (Benjamini–Hochberg correction) < 0.05 considered as the threshold for significant differentially expressed genes (DEGs).Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were conducted using the clusterProfiler package, with P-value < 0.05 and q-value < 0.1 considered statistically significant.For molecular docking analysis, binding energies were compared among candidate compounds. Binding energy values < −5.0 kcal/mol were considered to indicate favorable binding activity, whereas values < −7.0 kcal/mol indicated strong binding affinity.A two-sided P-value < 0.05 was considered statistically significant, and P < 0.01 was considered highly statistically significant.

3. Results

3.1. Screening of Shared Differentially Expressed Genes (DEGs) Between Osteoarthritis and Type 2 Diabetes Mellitus

To comprehensively clarify the common molecular pathogenic mechanisms underlying osteoarthritis (OA) and type 2 diabetes mellitus (T2DM),This study initially conducted a differential analysis of gene expression data retrieved from OA and T2DM sourced from the GEO database.Results of principal component analysis (PCA) illustrated that,within the OA dataset,sample points of the control group and the treatment group showed a distinct tendency toward separation(Figure 1A),indicating that the disease state of OA substantially changed the expression of articular cartilage tissue-related genes. Similarly, in the T2DM dataset,Samples from the control and treatment groups exhibited obvious distribution discrepancies on the PCA plot,particularly along the direction of the first principal component (PC1) (Figure 1B), confirming the impact of metabolic disturbances such as hyperglycemia or insulin resistance on gene expression.Based on these findings,were filtered and identified.The heatmap (Figure 1C) and volcano plot (Figure 1E) associated with OA visually presented the gene sets that were significantly upregulated and downregulated in OA.It is probable that these genes are extensively engaged in pathological cascades such as inflammatory responses and cartilage matrix degradation.Likewise, the heatmap (Figure 1D) and volcano plot (Figure 1F) related to T2DM revealed genes that underwent significant changes in T2DM, which may be closely associated with abnormal glucose and lipid metabolism and dysregulated insulin signal transduction.

Figure 1.

Screening and analysis of DEGs in osteoarthritis and type 2 Diabetes Mellitus. (A) Principal component analysis (PCA) plot of gene expression data from osteoarthritis (OA), illustrating the distribution of samples from the control group (blue) and the treatment group (orange). (B) PCA plot of gene expression data from type 2 Diabetes Mellitus (T2DM), illustrating the distribution differences between samples from the control group (blue) and the treatment group (orange). (C) Heatmap of differentially expressed genes in OA, displaying changes in gene expression levels between the control group and the treatment group (red indicates upregulation, blue indicates downregulation). (D) Heatmap of differentially expressed genes in T2DM, presenting gene expression patterns and clustering analysis results under large-sample conditions. (E) The volcano plot depicting differentially expressed genes in OA reveals the magnitude of alterations in gene expression and statistical significance, and significantly upregulated and downregulated genes are labeled therein. (F) The volcano plot depicting T2DM differentially expressed genes highlights genes with significantly altered expression

3.2. Functional Enrichment Analysis of Shared DEGs Reveals Key Biological Pathways

By taking the Venn intersection of the differentially expressed gene (DEG) sets from OA and T2DM, 99 shared differentially expressed genes were successfully identified (Figure 2A).We successfully pinpointed 99 shared differentially expressed genes. Of these, 1,954 DEGs were specific to T2DM, with 895 unique to OA.These 99 shared genes, serving as a molecular bridge connecting the two diseases, suggest the presence of common molecular drivers in the pathological progression of OA and T2DM.To obtain a more profound understanding of the biological relevance of the 99 shared DEGs.We further performed systematic Gene Ontology (GO) functional annotation and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses.Results of GO enrichment analysis indicated that, in OA (Figure 2B), the shared DEGs were notably concentrated in biological processes such as “extracellular matrix organization” and “cellular response to cytokine stimulus,” and their molecular functions centered on “cytokine activity” and “receptor ligand activity.”In T2DM (Figure 2D), these genes were also prominently concentrated in biological processes including “extracellular matrix synthesis” and “inflammatory response cascade,”indicating that extracellular matrix remodeling and chronic inflammatory responses are core pathological foundations shared by OA and T2DM.KEGG pathway analysis further confirmed these findings at the signaling pathway level. In both OA (Figure 2C) and T2DM (Figure 2E), the shared DEGs significantly accumulated in pathways including the TNF, NF-κB, PI3K-Akt, and AGE-RAGE (in diabetic complications) signaling pathways, as well as cytokine-cytokine receptor interaction.These pathways form a complex network that jointly regulates inflammation, metabolism, cell survival, and stress responses, clearly outlining a multi-pathway synergistic interaction map for the comorbidity of OA and T2DM at the molecular level.

Figure 2.

Functional enrichment analysis of shared differentially expressed genes. (A) Venn diagram displaying the quantity of common DEGs between T2DM and OA,as well as the number of genes unique to each disease. (B) Bar chart of GO functional annotation and enrichment for OA differentially expressed genes, presenting significantly enriched entries in biological processes, cellular components, and molecular functions. (C) KEGG pathway enrichment bubble plot for OA DEGs, showing significantly enriched signaling pathways. (D) Bar chart of GO functional enrichment analysis for differentially expressed genes in T2DM. (E) Bubble plot of KEGG pathway enrichment for differentially expressed genes in T2DM

3.3. Construction of PPI Network and Identification of Core Hub Genes

To precisely locate the regulatory factors at the core of the comorbid protein network among the 99 shared differentially expressed genes (DEGs), this study submitted these genes to the STRING database for protein-protein interaction (PPI) network construction.To ensure the high reliability and validity of the obtained results.we employed the CytoHubba plugin in Cytoscape software, utilizing four topological algorithms—Maximal Clique Centrality (MCC), Degree Centrality (Degree), Maximum Neighborhood Component (MNC), and Edge Percolated Component (EPC)—to score the network nodes. The intersection of the top genes detected by the four algorithms was obtained using a Venn diagram,we ultimately screened out 13 core hub genes (Figure 3A). These genes exhibited high comprehensive scores in the importance score heatmap constructed using the five algorithms (Figure 3B).The final PPI network diagram (Figure 3C) visually demonstrated dense interactions among the proteins encoded by these 13 core genes, which collectively formed a tightly interconnected network module. Genes within this module, such as PTPRC involved in immune regulation, ITGB3 mediating cell adhesion, and SLC2A1 responsible for glucose transport, likely play pivotal roles as intermediaries in the interaction network between OA and T2DM. These genes represent the most promising molecular targets for subsequent targeted drug screening.

Figure 3.

Construction of protein-protein interaction network and identification of core hub genes. (A) Venn diagram showing the intersection of core genes selected by four algorithms, with an intersection of 13 core genes. (B) Heatmap displaying the scoring matrix of the 13 core genes under five algorithms. (C) PPI network diagram displaying the interactive associations among proteins encoded by the 13 core genes

3.4. Molecular Docking Identifies Natural Compounds With High-Affinity Binding to Core Targets

To identify potential therapeutic compounds targeting the aforementioned hub genes, we conducted molecular docking studies on three key targets: CX3CR1, IRF8, and NCF1. The aim was to explore the interactive relationships as well as binding affinities between the compounds and pathway-related target molecules. A variety of small molecules, including natural flavonoids such as apigenin, luteolin, quercetin, and the positive control drug rosiglitazone, were screened from a natural product library. The docking results showed that the natural flavonoid apigenin exhibited strong binding potential to all three targets (Table 1). As shown in Figure 4A, apigenin demonstrated an exceptionally low binding free energy of -9.3 kcal/mol with CX3CR1, and binding energies of -8.0 kcal/mol with both IRF8 and NCF1, indicating its multi-target inhibitory potential.As shown in Figure 4B, luteolin exhibited binding energies of -8.0 kcal/mol and -8.5 kcal/mol with IRF8 and NCF1, respectively. As shown in Figure 4C, the positive control drug rosiglitazone demonstrated binding energies of -8.6 kcal/mol with CX3CR1 and -8.7 kcal/mol with NCF1. As shown in Figure 4D, quercetin exhibited a binding energy of -8.4 kcal/mol with NCF1.The heatmap (Figure 4E) displays the molecular docking binding energy matrix of different compounds against different target genes. Collectively, apigenin showed the strongest binding affinity to CX3CR1 (-9.3 kcal/mol) and exhibited favorable binding activity to all three targets. Given its outstanding overall performance, apigenin was selected as the lead compound for subsequent in vitro experimental validation.

Table 1.

Molecular Docking Binding Energies (kcal/mol) of Candidate Compounds With Core Targets

Compound	CX3CR1	IRF8	NCF1
Apigenin	-9.3	-8.0	-8.0
Luteolin	-	-8.0	-8.5
Quercetin	-	-	-8.4
Rosiglitazone	-8.6	-	-8.7

Figure 4.

Molecular docking screening of natural compounds against core targets. (A) Molecular docking results of apigenin with CX3CR1, IRF8, and NCF1. (B) Molecular docking results of luteolin with IRF8 and NCF1. (C) Molecular docking results of rosiglitazone with CX3CR1 and NCF1. (D) Molecular docking results of quercetin with NCF1. (E) Heatmap of molecular docking binding energies (kcal/mol) of different compounds (apigenin, luteolin, quercetin, rosiglitazone) against different targets (CX3CR1, IRF8, NCF1). The color gradient ranges from white to deep red, representing binding energies from low to high

3.5. In Vitro Validation of Core Gene Expression and Therapeutic Potential of Apigenin

3.5.1. Upregulation of Core Genes in a Cellular Model of OA Complicated With T2DM

To validate the results of bioinformatics analysis, we established a human chondrocyte model of OA complicated with T2DM induced by a combination of IL-1β and high glucose. qRT-PCR results demonstrated that, in contrast to the normal control group,hub genes CX3CR1, IRF8, and NCF1 had notably higher mRNA expression in the model group (p < 0.01), matching the bioinformatics-predicted upregulation pattern. (Figure 5).

Figure 5.

Upregulation of core genes in a cellular model of OA complicated with T2DM. The mRNA expression levels of the hub genes CX3CR1, IRF8, and NCF1 in the OA + T2DM combined cellular model. Data are illustrated as mean ± standard deviation. *p < 0.05, **p < 0.01, ***p < 0.001 vs. the control group

3.5.2. Apigenin Effectively Inhibits Inflammatory Pathways and Improves Cellular Function

After administering apigenin at varying concentrations (5,10,20 μM) in the combined model, a series of experiments demonstrated its significant therapeutic effects.Western Blot analysis indicated that apigenin could remarkably and concentration-dependently diminished the phosphorylation level of the key protein p65 in the NF-κB pathway (p < 0.05, **p < 0.01) (Figure 6A). Figure 6B shows that the CCK-8 assay demonstrated that 20 μM apigenin could effectively restore the reduced cell viability caused by IL-1β and high glucose (p < 0.05).As illustrated in Figure 6C, the Tunel assay demonstrated that the model group exhibited a significantly higher apoptosis rate,while apigenin treatment could concentration dependently reverse this trend and significantly reduce the cell apoptosis rate (p < 0.05). As depicted in Figure 6C, ELISA results revealed that apigenin treatment exerted a significant inhibitory effect on the secretion of key inflammatory cytokines IL-6 and TNF-α in the cell supernatant(p < 0.05, **p < 0.01).

Figure 6.

Apigenin effectively inhibits inflammatory pathways and improves cellular function. (A) He regulatory effect of apigenin on NF-κB signaling pathway activation. (B) Cell viability detection via the CCK-8 method. (C) Detection of the cell apoptosis rate using the Tunel method. (D) Quantification of inflammatory cytokines IL-6 and TNF-αin the cell culture supernatant by means of the ELISA method. Data are presented as mean ± standard deviation. *p < 0.05, **p < 0.01 vs. the model group

4. Discussion

This study comprehensively elucidated the shared molecular mechanisms of comorbidity between osteoarthritis (OA) and type 2 diabetes mellitus (T2DM) via the combination of network pharmacology, molecular docking, and in vitro experimental verification. It was found that the natural flavonoid apigenin exhibits multi-target therapeutic potential by targeting key hub genes such as CX3CR1, IRF8, and NCF1. This research not only presents a novel perspective for elucidating the crosstalk between OA and T2DM but also establishes a theoretical foundation for the development of multi-target therapeutic regimens for these comorbidities.¹⁸Although the GEO datasets used in this study were derived from different tissue sources (meniscus tissue for OA and pancreatic islets for T2DM), this design was intentional and aligned with the objective of identifying shared systemic molecular mechanisms rather than tissue-specific gene signatures. OA and T2DM are increasingly recognized as systemic metabolic-inflammatory diseases characterized by common pathological features such as chronic inflammation, oxidative stress, and abnormal glucose metabolism. Therefore, cross-tissue transcriptomic integration provides valuable insight into the molecular crosstalk underlying disease comorbidity. Nevertheless, tissue heterogeneity may still introduce biological bias, and future studies should further validate these findings using cartilage tissue, synovium, and metabolic target organs.

The preliminary bioinformatics analysis in this study successfully constructed a molecular bridge connecting OA and T2DM. The 99 shared differentially expressed genes (DEGs) we identified and the significantly enriched signaling pathways they are involved in strongly confirm that chronic inflammation is the core pathological basis driving the co-progression of the two diseases.This finding is highly consistent with the study by Walrabenstein et al,¹⁹ which clearly points out that metabolic inflammation is the key link connecting OA and T2DM. It is particularly noteworthy that the results of the KEGG analysis echo the role of the AGE-RAGE signaling pathway in diabetic complications as reported by Rosa et al. However, for the first time in the context of OA and T2DM comorbidity, this study systematically identifies this pathway, together with classical inflammatory pathways (TNF and NF-κB), as shared core pathological pathways.^20-22Subsequently, by constructing a PPI network, we precisely located the core hub genes regulating this comorbid network. The 13 identified core genes, such as PTPRC, ITGB3, and SLC2A1, are known to have important functions in immune regulation, cell adhesion, and glucose transport.^22-27 Particular attention should be paid to the subsequently in-depth studied CX3CR1, IRF8, and NCF1.^28-30 As the receptor for Fractalkine, the association of CX3CR1 with arthritis has been confirmed by the study of Lee et al, who observed that the CX3CL1/CX3CR1 signaling axis undergoes significant upregulation in OA synovitis.^31,32Meanwhile, the role of this pathway in diabetic neuropathy is also widely recognized.³³ In this study, we identified and validated the hub status of CX3CR1 in the comorbid network of OA and T2DM, providing a new molecular explanation for why these two diseases often co-occur.³⁴ Similarly, IRF8 is a key regulator of macrophage M1 polarization,³⁵ which is in good agreement with the primary role of M1 macrophages in the OA inflammatory milieu documented by Zhang et al.³⁶ Our work extends the role of IRF8 from the simple OA inflammatory environment to the broader context of T2DM comorbidity. As a key mediator of oxidative stress, the role of NCF1 is in line with the established oxidative stress theories in diabetes and OA. However, through network analysis, this study positions NCF1 as a core node connecting the oxidative stress states of the two diseases.^34,37 Our analysis indicates that these genes occupy central positions in the comorbid network and may constitute a synergistic amplification network of inflammation and oxidative stress that jointly drives disease progression.

The most crucial finding is that through molecular docking prediction and subsequent in vitro experimental validation, the natural compound apigenin can efficiently bind to the above-mentioned core targets and exhibit significant therapeutic effects. Molecular docking results show that apigenin has strong binding affinity to CX3CR1, IRF8, and NCF1, especially with a binding energy as low as -9.3 kcal/mol to CX3CR1, indicating extremely high binding affinity and stability.^38-40This finding is of great significance because although the anti-inflammatory and antioxidant effects of apigenin have been reported in multiple studies, such as Meyer et al's report on its ability to improve metabolic syndrome⁴¹ and Ali et al's discovery of its protective effect on cartilage,⁴² its direct molecular targets, especially its multi-target characteristics in the context of OA and T2DM comorbidity, have remained unclear until now.⁴³ Our study fills this gap. This computational prediction was validated in subsequent cell experiments. In the IL-1β and high glucose-induced OA combined with T2DM chondrocyte model, we confirmed that apigenin can concentration-dependently inhibit NF-κB pathway activation, enhance cell viability, reduce cell apoptosis, and diminish IL-6 and TNF-α(key inflammatory cytokines) secretion.The IL-1β plus high-glucose-induced SW1353 cell model used in this study is a widely accepted in vitro approach for simulating the inflammatory and metabolic microenvironment of OA complicated with T2DM. IL-1β represents a major pro-inflammatory cytokine involved in cartilage degradation and synovial inflammation in OA, whereas high glucose reflects the diabetic metabolic stress associated with T2DM. Their combined stimulation better mimics the pathological coexistence of chronic inflammation and metabolic dysfunction observed clinically in OA patients with diabetes. Nevertheless, this model remains a simplified system and cannot fully reproduce the complexity of in vivo conditions, including immune cell interactions, mechanical loading, and systemic metabolic regulation.Mechanistically, our results suggest that apigenin may exert its anti-inflammatory effects through suppression of the CX3CR1/NF-κB signaling axis. Among the identified hub genes, CX3CR1 showed the strongest binding affinity with apigenin in molecular docking analysis, indicating that it may serve as a key upstream target. Previous studies have demonstrated that CX3CR1 signaling participates in inflammatory cell recruitment and activation of downstream NF-κB signaling. Therefore, it is plausible that apigenin interferes with CX3CR1-mediated inflammatory signaling, thereby reducing p65 phosphorylation, suppressing IL-6 and TNF-αsecretion, and ultimately alleviating chondrocyte apoptosis and inflammatory injury. However, direct mechanistic validation through gene knockdown, overexpression, or specific inhibitor studies was not performed in the present study and warrants further investigation.These results are consistent with the NF-κB inhibitory effect observed by Wang et al when using apigenin to treat rheumatoid arthritis,⁴⁴ but we have extended the application of this mechanism to the more challenging environment of OA and T2DM comorbidity. These results form a complete causal chain: apigenin interferes with the inflammation signal transduction mediated by targets such as CX3CR1 by directly acting on them, thereby inhibiting the downstream NF-κB core inflammatory pathway. Ultimately, it alleviates the inflammatory damage, metabolic stress, and apoptosis of chondrocytes at the functional level, effectively intervening in the disease phenotype.As a natural product extensively distributed in vegetables and fruits, apigenin has been documented to possess diverse pharmacological activities, including anti-inflammatory, antioxidant, and anti-diabetic properties.^45,46 Collectively, the results of this study demonstrate that it may be a promising multi-target natural remedy for addressing OA and T2DM comorbidity.Compared with traditional single-target drugs, apigenin’s synchronous action on multiple pivotal nodes within the disease network offers the potential to more effectively reestablish the network’s equilibrium,potentially producing synergistic therapeutic effects and reducing the risk of drug resistance caused by signal pathway compensation. This provides an extremely attractive new strategy for solving the treatment challenges of complex comorbidities.⁴⁷

Several limitations of this study should be acknowledged. First, the GEO datasets used for bioinformatics analysis were derived from different tissue sources (meniscus tissue for OA and pancreatic islets for T2DM), which may introduce biological heterogeneity despite standardized normalization and intersection analysis. Although this design aimed to identify shared systemic molecular mechanisms rather than tissue-specific alterations, further validation using cartilage tissue, synovium, and metabolic target organs is still necessary. Second, hub gene identification was primarily based on network topology analysis and bioinformatics prediction, which may not fully reflect causal biological importance in vivo. Functional studies involving gene knockdown, overexpression, and pathway-specific intervention are needed to further validate the biological roles of CX3CR1, IRF8, and NCF1. Third, molecular docking was performed using static protein structures and does not account for protein conformational flexibility, intracellular dynamics, or real binding kinetics. Therefore, the predicted ligand–target interactions require further validation using molecular dynamics simulations as well as experimental binding assays such as surface plasmon resonance (SPR) and isothermal titration calorimetry (ITC). Fourth, the in vitro validation was conducted using the SW1353 chondrocyte cell line rather than primary human chondrocytes, which may limit physiological relevance. Moreover, the IL-1β plus high-glucose model represents only a simplified simulation of OA complicated with T2DM and cannot fully reproduce the complexity of in vivo conditions, including immune cell interactions, mechanical loading, and systemic metabolic regulation. In addition, no animal experiments or clinical validation were performed in this study, which limits the direct translational significance of the findings. Future studies should incorporate primary chondrocytes, animal models, and multicenter clinical samples to further confirm the therapeutic efficacy of apigenin. Importantly, before clinical application, the pharmacokinetic characteristics, bioavailability, optimal therapeutic dosage, long-term safety, and potential synergistic effects of apigenin in combination with existing anti-inflammatory or anti-diabetic therapies should be systematically evaluated. These efforts will help promote the translational development of apigenin as a multi-target therapeutic strategy for OA complicated with T2DM.

5. Conclusion

In conclusion, this study successfully employed a research strategy from “computational prediction” to “experimental validation,” revealing the shared inflammation-related molecular network and key hub genes of OA and T2DM. It also proposed for the first time that apigenin can effectively alleviate the pathological progression of the combined diseases at the cellular level through a multi-target action mechanism. Our work advances the understanding of the comorbid mechanisms linking OA and T2DM, and in doing so, establishes a solid theoretical foundation and provides experimental data support for the development of apigenin as a novel, safe, and effective multi-target therapeutic drug.

Footnotes

Acknowledgements

The authors would like to thank the GEO, PubChem, and RCSB PDB databases for providing open access data. We are also deeply grateful to Professor Chunlei Liu for his valuable technical guidance and support throughout this study.

ORCID iD

Saifei Meng

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Guangdong Provincial Basic and Applied Basic Research Foundation(Grant No. 2022A1515220177). The High-level Talent Introduction Scientific Research Initiation Project, Qingyuan Hospital Affiliated to Guangzhou Medical University (Qingyuan People’s Hospital) (Grant No.15001019001294).

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

The data in this study are derived from PubChem (https://pubchem.ncbi.nlm.nih.gov/, chemical-related data) and RCSB PDB (, structure-related data). Both databases are publicly accessible, with data complying with their sharing policies and no additional restrictions on access and use.

References

Elmaleh-Sachs

Schwartz

Bramante

Nicklas

Gudzune

Jay

. Obesity Management in Adults: A Review. JAMA. 2023;330(20):2000-2015. doi:10.1001/jama.2023.19897.

Knights

Redding

Maerz

. Inflammation in osteoarthritis: the latest progress and ongoing challenges. Curr Opin Rheumatol. 2023;35(2):128-134. doi:10.1097/BOR.0000000000000923.

Młynarska

Czarnik

Dzieża

, et al. Type 2 Diabetes Mellitus: New Pathogenetic Mechanisms, Treatment and the Most Important Complications. Int J Mol Sci. 2025;26(3):1094. doi:10.3390/ijms26031094.

Song

. Identification of the shared genes in type 2 diabetes mellitus and osteoarthritis and the role of quercetin. J Cell Mol Med. 2024;28(4):e18127. doi:10.1111/jcmm.18127.

Mamadapur

Gaidhane

Padhi

, et al. Burden of rheumatic diseases among people with diabetes: A systematic review and meta-analysis. Narra J. 2024;4(3):e863. doi:10.52225/narra.v4i3.863.

Rydberg

Zimmerman

M Nilsson

Gottsäter

B Dahlin

. [The diabetic hand - a forgotten complication]. Lakartidningen. 2024;121:23207.

Vaishya

Misra

Patralekh

Kalra

Vaish

Migliorini

. Association of type 2 diabetes and osteoarthritis: an umbrella review of systematic reviews and meta-analyses. Eur J Orthop Surg Traumatol Orthop Traumatol. 2025;35(1):111. doi:10.1007/s00590-025-04231-6.

Tchetina

Markova

Sharapova

. Insulin Resistance in Osteoarthritis: Similar Mechanisms to Type 2 Diabetes Mellitus. J Nutr Metab. 2020;2020:4143802. doi:10.1155/2020/4143802.

Wei

Umar

, et al. Risk of metabolic abnormalities in osteoarthritis: a new perspective to understand its pathological mechanisms. Bone Res. 2023;11(1):63. doi:10.1038/s41413-023-00301-9.

10.

Zhang

Liu

. Integrated bioinformatic analysis of the molecular mechanisms between type 2 diabetes mellitus and osteoarthritis. Medicine (Baltimore). 2024;103(35):e39469. doi:10.1097/MD.0000000000039469.

11.

Hsieh

Peng

Shi

Huang

Chang

. The Potential Dual Roles of Metallothionein-1/-2 in Diabetic Osteoarthritis. J Cell Physiol. 2025;240(6):e70056. doi:10.1002/jcp.70056.

12.

Zanfardino

Franzese

Geraci

. DeClUt: Decluttering differentially expressed genes through clustering of their expression profiles. Comput Methods Programs Biomed. 2024;254:108258. doi:10.1016/j.cmpb.2024.108258.

13.

Yin

Duo

, et al. Unlocking biological insights from differentially expressed genes: Concepts, methods, and future perspectives. J Adv Res. 2025;76:135-157. doi:10.1016/j.jare.2024.12.004.

14.

Hutchinson

Barratt

Johnston

, et al. Biomarkers to predict, prevent, and treat persistent pain: omics. Pain. 2025;166(11S):S103-S105. doi:10.1097/j.pain.0000000000003673.

15.

Lan

Gao

, et al. Elucidation of the molecular mechanism of type 2 diabetes mellitus affecting the progression of nonalcoholic steatohepatitis using bioinformatics and network pharmacology: A review. Medicine (Baltimore). 2024;103(37):e39731. doi:10.1097/MD.0000000000039731.

16.

Blüher

. An overview of obesity-related complications: The epidemiological evidence linking body weight and other markers of obesity to adverse health outcomes. Diabetes Obes Metab. 2025;27(Suppl 2):3-19. doi:10.1111/dom.16263.

17.

Rosati

Palmieri

Brunelli

, et al. Differential gene expression analysis pipelines and bioinformatic tools for the identification of specific biomarkers: A review. Comput Struct Biotechnol J. 2024;23:1154-1168. doi:10.1016/j.csbj.2024.02.018.

18.

Meurot

Jacques

Martin

, et al.

Targeting the GLP-1/GLP-1R axis to treat osteoarthritis: A new opportunity?

J Orthop Transl. 2022;32:121-129. doi:10.1016/j.jot.2022.02.001.

19.

Walrabenstein

Wagenaar

van de Put

, et al. A multidisciplinary lifestyle program for metabolic syndrome-associated osteoarthritis: the “Plants for Joints” randomized controlled trial. Osteoarthritis Cartilage. 2023;31(11):1491-1500. doi:10.1016/j.joca.2023.05.014.

20.

Kreitmaier

Katsoula

Zeggini

. Insights from multi-omics integration in complex disease primary tissues. Trends Genet TIG. 2023;39(1):46-58. doi:10.1016/j.tig.2022.08.005.

21.

Engin

. Reappraisal of Adipose Tissue Inflammation in Obesity. Adv Exp Med Biol. 2024;1460:297-327. doi:10.1007/978-3-031-63657-8_10.

22.

Clyne

. Endothelial response to glucose: dysfunction, metabolism, and transport. Biochem Soc Trans. 2021;49(1):313-325. doi:10.1042/BST20200611.

23.

Yue

Wang

Zhou

. PTPRC functions as a prognosis biomarker in the tumor microenvironment of cutaneous melanoma. Sci Rep. 2023;13(1):20617. doi:10.1038/s41598-023-46794-6.

24.

Huang

Pan

, et al. A novel duplication mutation of SLC2A1 gene causing glucose transporter 1 deficiency syndrome. Gene. 2024;928:148762. doi:10.1016/j.gene.2024.148762.

25.

Luo

Feng

Luo

Tian

. ITGB3 is a Novel Prognostic Biomarker and Correlates with Aberrant Methylation and Tumor Immunity in Clear Cell Renal Cell Carcinoma. Comb Chem High Throughput Screen. 2025;28(13):2233-2248. doi:10.2174/0113862073311689240730112355.

26.

Shaikh

Beck

Alwarawrah

MacIver

. Emerging mechanisms of obesity-associated immune dysfunction. Nat Rev Endocrinol. 2024;20(3):136-148. doi:10.1038/s41574-023-00932-2.

27.

Slieker

Donnelly

Akalestou

, et al. Identification of biomarkers for glycaemic deterioration in type 2 diabetes. Nat Commun. 2023;14(1):2533. doi:10.1038/s41467-023-38148-7.

28.

Subbarayan

Joly-Amado

Bickford

Nash

. CX3CL1/CX3CR1 signaling targets for the treatment of neurodegenerative diseases. Pharmacol Ther. 2022;231:107989. doi:10.1016/j.pharmthera.2021.107989.

29.

Moorman

Reategui

Poschel

Liu

. IRF8: Mechanism of Action and Health Implications. Cells. 2022;11(17):2630. doi:10.3390/cells11172630.

30.

Zhang

Wax

Huang

Petersen

. Meta-Analysis and Systematic Review of the Association between a Hypoactive NCF1 Variant and Various Autoimmune Diseases. Antioxidants. 2022;11(8):1589. doi:10.3390/antiox11081589.

31.

Wojdasiewicz

Poniatowski

Kotela

Deszczyński

Kotela

Szukiewicz

. The chemokine CX3CL1 (fractalkine) and its receptor CX3CR1: occurrence and potential role in osteoarthritis. Arch Immunol Ther Exp (Warsz). 2014;62(5):395-403. doi:10.1007/s00005-014-0275-0.

32.

Lin

Pan

. The Causal Effects Between Circulating Inflammatory Proteins and Osteoarthritis: A Mendelian Randomization and Transcriptomic Analysis. J Pain Res. 2025;18:3383-3402. doi:10.2147/JPR.S523677.

33.

Sambataro

Sambado

Colardo

, et al.

Neuroinflammation and osteomyelitis in adults with Type 2 diabetes mellitus and peripheral neuropathy without and with foot lesions. What comes first?

J Diabetes Complications. 2025;39(11):109176. doi:10.1016/j.jdiacomp.2025.109176.

34.

O’Neill

Brault

Stasia

Knaus

. Genetic disorders coupled to ROS deficiency. Redox Biol. 2015;6:135-156. doi:10.1016/j.redox.2015.07.009.

35.

Guo

Yang

Peng

Yao

. Inhibition of IRF8 Negatively Regulates Macrophage Function and Impairs Cutaneous Wound Healing. Inflammation. 2017;40(1):68-78. doi:10.1007/s10753-016-0454-8.

36.

Zhang

Lin

Zeng

, et al. Synovial macrophage M1 polarisation exacerbates experimental osteoarthritis partially through R-spondin-2. Ann Rheum Dis. 2018;77(10):1524-1534. doi:10.1136/annrheumdis-2018-213450.

37.

van Dalen

SCM

Kruisbergen

NNL

Walgreen

, et al. The role of NOX2-derived reactive oxygen species in collagenase-induced osteoarthritis. Osteoarthritis Cartilage. 2018;26(12):1722-1732. doi:10.1016/j.joca.2018.08.014.

38.

Murali

Nair

Radhakrishnan

Presanna

Nath

. Evaluation of the Nimbamrithadhi Panchathiktha Kashayam against SARS CoV-2 based on Network Pharmacology and Molecular Docking analysis. Comb Chem High Throughput Screen. 2023;26(2):436-447. doi:10.2174/1386207325666220519112846.

39.

Lee

Yang

. The Role of Fractalkine in Diabetic Retinopathy: Pathophysiology and Clinical Implications. Int J Mol Sci. 2025;26(1):378. doi:10.3390/ijms26010378.

40.

Yan

, et al. [Effects and active substances of ethanol extract from Dendrobium officinale on metabolic hypertensive rats induced by comprehensive dietary]. Zhongguo Zhong Yao Za Zhi Zhongguo Zhongyao Zazhi China J Chin Mater Medica. 2019;44(22):4896-4904. doi:10.19540/j.cnki.cjcmm.20190610.401.

41.

Meyer

Bolarinwa

Wolfram

Linseisen

. Bioavailability of apigenin from apiin-rich parsley in humans. Ann Nutr Metab. 2006;50(3):167-172. doi:10.1159/000090736.

42.

Ali

Okla

Abuelreich

, et al. Apigenin and Rutaecarpine reduce the burden of cellular senescence in bone marrow stromal stem cells. Front Endocrinol. 2024;15:1360054. doi:10.3389/fendo.2024.1360054.

43.

Zhang

Jiang

, et al. Targeting Senescence with Apigenin Improves Chemotherapeutic Efficacy and Ameliorates Age-Related Conditions in Mice. Adv Sci. 2025;12(20):e2412950. doi:10.1002/advs.202412950.

44.

Wang

Zhong

Wang

, et al. [Activity of Codonopsis canescens against rheumatoid arthritis based on TLRs/MAPKs/NF-κB signaling pathways and its mechanism]. Zhongguo Zhong Yao Za Zhi Zhongguo Zhongyao Zazhi China J Chin Mater Medica. 2022;47(22):6164-6174. doi:10.19540/j.cnki.cjcmm.20220727.401.

45.

Morakul

Teeranachaideekul

Limwikrant

Junyaprasert

. Dissolution and antioxidant potential of apigenin self nanoemulsifying drug delivery system (SNEDDS) for oral delivery. Sci Rep. 2024;14(1):8851. doi:10.1038/s41598-024-59617-z.

46.

Tang

Wang

, et al. Apigenin Ameliorates H2O2-Induced Oxidative Damage in Melanocytes through Nuclear Factor-E2-Related Factor 2 (Nrf2) and Phosphatidylinositol 3-Kinase (PI3K)/Protein Kinase B (Akt)/Mammalian Target of Rapamycin (mTOR) Pathways and Reducing the Generation of Reactive Oxygen Species (ROS) in Zebrafish. Pharmaceuticals. 2024;17(10):1302. doi:10.3390/ph17101302.

47.

Ivanov

Popov Aleksandrov

Božunović

, et al. Bioactive potential of Tripleurospermum inodorum with detailed insight into anti-inflammatory activity through in vitro, in vivo evaluations and network pharmacology. Inflammopharmacology. 2025;33(6):3363-3376. doi:10.1007/s10787-025-01769-z.