Experimental study on the role and biomarker potential of CX3CR1 in osteoarthritis

Abstract

Background

Osteoarthritis (OA) is a chronic joint disorder marked by progressive degeneration of articular cartilage and the formation of secondary osteophytes. Despite extensive research, the underlying molecular mechanisms remain poorly understood. This study aimed to identify OA-associated genes and elucidate the molecular pathways implicated, with the goal of discovering reliable diagnostic biomarkers.

Methods

The microarray dataset was retrieved from the Gene Expression Omnibus (GEO) and analyzed using R software to identify the signature gene, CX3CR1. Differentially expressed genes (DEGs) correlated with CX3CR1 were subsequently subjected to Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), and immune infiltration analyses. A ceRNA regulatory network was also constructed. Vali-dation of CX3CR1 expression was conducted through qRT-PCR, Western blotting, and immunohistochemistry.

Results

CX3CR1 emerged as a candidate gene significantly associated with OA, exhibiting regulatory roles primarily in lipid metabolism-related and extra-cellular matrix-related biological processes and signaling cascades. The infiltration levels of immune cells, particularly activated mast cells, appeared to modulate OA progression. Both in vitro and in vivo experiments demonstrated elevated CX3CR1 expression in OA tissues relative to controls, with a robust positive correlation observed between CX3CR1 and MMP13 levels.

Conclusion

CX3CR1 represents a potential biomarker for OA diagnosis and therapeutic targeting, exerting its effects by modulating lipid metabolism, extracellular matrix dynamics, and immune cell infiltration.

1. Introduction

Osteoarthritis (OA), the most widespread chronic joint disorder, affects the entire joint and is primarily marked by structural alterations in hyaline articular cartilage, subchondral bone remodeling, and aberrant proliferation of synovial and vascular components within the joint space [1]. According to data from the Global Burden of Disease (GBD) study, OA prevalence rose by 13.25% between 1990 and 2019, imposing substantial healthcare demands and socio-economic pressures worldwide [2,3].

The pathogenesis of OA has been extensively investigated over recent decades. Skeletal development is markedly influenced by the Wnt signaling cascade, which is broadly categorized into β-catenin-dependent (canonical) and β-catenin-independent (non-canonical) branches [4]. Elevated β-catenin expression has been detected in articular surface chondrocytes from OA rats and human patients, with substantial evidence linking its overactivation to the initiation and progression of OA [5]. The NF-κB pathway, comprising a family of transcription factors, mediates articular cartilage degradation primarily by upregulating matrix metalloproteinases (MMPs), ADAMTS4, and ADAMTS5. Simultaneously, it triggers inflammatory responses and promotes chondrocyte apoptosis through the induction of PGE2, NOS, NO, and COX2, thereby intensifying joint degeneration [6]. Beyond mechanical overload, obesity accelerates OA progression via adipokine-mediated inflammatory pathways, which contribute to cartilage deterioration independent of biomechanical stressors [7].

However, the molecular mechanisms underlying OA and its associated biomarkers require further systematic validation. Gene microarray analysis and bioinformatics offer integrative approaches for investigating disease progression from multiple dimensions. These methodologies, coupled with in vitro experiments, contribute to a comprehensive delineation of disease-related molecular dynamics.

In this study, datasets GSE12021, GSE29746, GSE51588, GSE55457, and GSE1919 were retrieved from the GEO database to identify gene signatures implicated in OA progression. Functional enrichment analysis, immune cell infiltration profiling, and ceRNA network construction elucidated the involvement of these gene signatures in OA pathogenesis. In vitro assays served to validate candidate biomarkers and inform strategies for early intervention and therapeutic targeting.

2. Materials and methods

2.1. Data acquisition and processing

The datasets GSE12021, GSE29746, GSE51588, and GSE55457 from the GEO data-base (http://www.ncbi.nlm.nih.gov/geo/) were retrieved, followed by batch correction and normalization via the limma and sva packages in R. Overlapping genes across datasets were identified to construct an integrated gene expression matrix for downstream analyses. This combined dataset comprised 43 samples of normal cartilage tissue and 81 samples of OA cartilage tissue, categorized respectively as Control and Treat groups. GSE1919, containing 5 samples each of normal and OA cartilage tissue, served as an independent external validation dataset.

2.2. Identification of DEGs and signature genes

Eligible DEGs were identified across all datasets using the limma package with thresholds set at adj.P.Val < 0.05 and |Log2FC| > 1. Subsequent screening was conducted via lasso regression and support vector machine (SVM) algorithms, implemented through the glmnet and e1071 packages in R, respectively, yielding two distinct sets of candidate signature genes. Additionally, the RandomForest (RF) algorithm, applied through the ggplot2 package, evaluated DEG importance, and genes with RF scores exceeding 2 (rfGenes > 2) were selected. The intersection of gene sets derived from all three algorithms defined the final candidate signature genes, from which one was selected for downstream analysis.

2.3. Accuracy analysis of signature genes

Violin plots were generated via the ggpubr package to evaluate expression differences of intersecting signature genes between the control and OA groups, with statistical significance defined as p < 0.05. ROC curves constructed using the pROC package quantified the discriminative performance of these genes in distinguishing sample groups, where an area under the ROC curve (AUC) > 0.7 indicated acceptable classification reliability. Subsequently, differential expression and ROC analyses were conducted on the validation cohort to confirm the reproducibility and diagnostic capacity of the identified gene signatures.

2.4. Signature gene correlation analysis

Following identification of the disease-associated signature gene via the aforementioned screening strategy, control group samples were excluded from the integrated dataset using R software, retaining only those from the experimental group. Based on the median expression level of the signature gene, the experimental samples were stratified into high- and low-expression cohorts. Differentially expressed genes (DEGs) were subsequently identified by applying stringent criteria (adj.P.Val.Filter < 0.05 and Log2FCfilter > 1). The resulting DEGs and corresponding correlation data were compiled. Pearson correlation analysis was then conducted, with results visualized through the corrplot package.

2.5. Functional enrichment analysis

Following the identification of signature gene-associated candidates, DEGs were stratified into up-regulated (Log2FC > 0) and down-regulated (Log2FC < 0) subsets according to Log2FC thresholds. Functional enrichment profiling was then conducted using KEGG and GO analyses—including biological processes (BPs), cellular components (CCs), and molecular functions (MFs)—to delineate the specific functional landscapes linked to the signature gene-associated subsets.

2.6. Immune cell infiltration and correlation analysis

Immune cell infiltration across normal and OA samples was assessed using the CIBERSORT algorithm, which applies an inverse fold product method to estimate the relative abundance of 22 lymphocyte subtypes in bulk tissue transcriptomes [8]. Post-analysis, immune infiltration data were visualized through boxline plots generated via the corrplot and ggplot2 packages in R, with statistical significance defined as p < 0.05. Subsequently, Spearman correlation analysis was employed to evaluate associations between CX3CR1 expression and infiltrating immune cell populations, using cor$p.value < 0.05 as the significance threshold. The resulting correlations were also visualized to facilitate interpretation of immune cell–gene expression relationships.

2.7. Exploration ceRNA network of the signature gene

TargetScan [9], miRanda [10], miRDB [11] were jointly employed to predict regulatory interactions between the signature gene and associated miRNAs. Subsequently, lncRNAs capable of competitively binding miRNAs via MREs (microRNA response elements) were identified using the spongeScan database (http://spongescan.rc.ufl.edu). The ceRNA regulatory network, comprising the signature gene, miRNAs, and lncRNAs, was constructed and visualized using Cytoscape v3.10.1.

2.8. Extraction and culture of chondrocytes

Cartilage tissues were aseptically harvested from the knee joints of 3- to 7-day-old SD rats for chondrocyte isolation and in vitro culture. The collected cartilage was minced and subjected to 30 min of enzymatic digestion at 37 °C using trypsin (Solabio, China), followed by a 6-hour incubation in high-glucose DMEM (Gibco, USA) supplemented with type II collagenase (1 mg/mL; Gibco). Post-digestion, cell suspensions were centrifuged at 1500 rpm for 3 min, and the isolated chondrocytes were cultured at 37 °C with 5% CO2 in DMEM containing 10% FBS (Gibco, USA) and penicillin/streptomycin (Solarbio, China). Third-passage chondrocytes were used for subsequent assays. An in vitro OA model was established by exposing chondrocytes to 1 mg/mL lipopolysaccharide (LPS, Solarbio, China) for 24 h [12]. Experimental conditions were stratified into three groups: (1) untreated control; (2) OA group, receiving LPS stimulation (1 mg/mL); and (3) OA+AZD8797 group, pre-incubated with 2 mM AZD8797 for 1 h prior to LPS treatment [13]. All experimental protocols received ethical approval from the Medical Ethics Committee of the First Affiliated Hospital of Guangxi Medical University (Approval No: 2024-E012-01).

2.9. RNA extraction and qRT-PCR

Total RNA was isolated from chondrocytes (control, OA, and OA+AZD8797 groups) using the RNAeasy^™ Plus Animal RNA Isolation Kit with Spin Column (Beyotime, R0032, China). Subsequently, 1 μg of purified RNA was reverse-transcribed into cDNA using the PrimeScriptTM RT reagent kit with gDNA Eraser (Takara, China). mRNA expression levels of CX3CR1 and MMP13 in chondrocytes from each group were quantified via qRT-PCR. Gene-specific primers were designed based on GenBank sequences. Each 10 μL PCR reaction comprised 0.5 μL of forward and reverse primers, 1.5 μL of nuclease-free water, 2.5 μL of cDNA template, and 5 μL of SYBR Premix Ex Taq (Thermo Fisher Scientific, USA). Target mRNA expression was normalized to GAPDH using the 2 − ΔΔCt method, with expression levels compared to the control group. Each sample was analyzed in triplicate. The CX3CR1, MMP13 and GAPDH primer sequences are as follows: CX3CR1: Forward, 50-CGCAACTCGGAGGTCAACATC-3′0, Reverse, 50-AAGACAACAACCACCAAGAGGATG-3′0; MMP13: Forward, 50-ACAGTTGATAGACTCCGAGAAATGC-3′0, Reverse, 50-CACATCAGGCACTCCACATCTTG-3′0; GAPDH: Forward, 50-AAGTTCAACGGCACAGTCAAGG-3′0, Reverse, 50-GACATACTCAGCACCAGCATCAC-3′0.

2.10. Western blot

Protein extraction from the control, OA, and OA+AZD8797 groups was performed using RIPA lysis buffer supplemented with PMSF (BOSTER, China), a protease inhibitor mixture (HY-K0010, MCE), and 1% phosphatase inhibitor mixture (CW2383, CWBIO). Protein concentrations were quantified using the BCA assay (Beyotime, China), after which samples were mixed with loading buffer, denatured at 100 °C for 10 min, and subjected to SDS–PAGE on a 7.5% homemade polyacrylamide gel loaded with 10 μg total protein per lane. Separated proteins were transferred onto NC membranes, which were blocked with 5% skim milk for two hours at room temperature. Overnight incubation at 4 °C was conducted with primary antibodies against GAPDH (1:5000, Proteintech, China), CX3CR1 (1:2000, Proteintech, China), and MMP13 (1:2000, Proteintech, China). After three PBST washes, membranes were incubated for one hour at room temperature with Multi-rAb^™ HRP-Goat Antirabbit Recombinant Secondary Antibody (H + L) (1:5000, Proteintech, China). Protein signals were visualized using enhanced chemluminescence (ECL, Beyotime), and images were acquired using the Odyssey infrared imaging system.

2.11. Immunohistochemical analysis

Rat OA and control models were established using six male SD rats (200–220 g), allocated equally between groups. OA induction involved anterior cruciate ligament transection (ACLT) of the right knee under 2% sodium pentobarbital anesthesia [14]. After anesthesia, the knee joint was exposed via medial patellar incision following dehairing, disinfection, and sterile draping. Lateral patellar dislocation enabled full knee flexion, after which the anterior cruciate ligament (ACL) was transected. In contrast, the control group underwent sham operations involving joint capsule exposure and suturing without ligament disruption. After a 6-week post-operative period, rats without signs of infection or mortality were included; those developing complications were excluded. Knee joints were excised, surrounding tissues removed, and samples processed for histological assessment by immunohistochemistry. Tissues were fixed in 4% paraformaldehyde for 48 h, decalcified with EDTA (Boster), embedded in paraffin, and sectioned into 3 μm slices. Cartilage sections were treated with 3% hydrogen peroxide, blocked using 10% goat serum (Gibco), and incubated overnight at 4 °C with primary antibody CX3CR1 (1:400). After washing with PBS, sections were incubated for 2 h at room temperature with IgG secondary antibody (1:300, Thermo Fisher Scientific). Diaminobenzidine (DAB, Boster) was used for chromogenic detection, and sections were counterstained with hematoxylin. Visualization was conducted using an Olympus inverted microscope.

2.12. Statistical analyses

Bioinformatics analysis was performed using R (x64 4.3.1). Experimental data analysis was conducted with GraphPad Prism 8.0.2 and ImageJ, and results were reported as mean ± standard error. Group comparisons were evaluated using a two-tailed independent samples Student’s t test, with statistical significance defined at p < 0.05.

2.13. Statement

This study was approved by the Medical Ethics Committee of the First Affiliated Hospital of Guangxi Medical University (Approval No. 2024-E012-01). All experimental procedures adhered to the regulatory guidelines established by the Animal Research Ethics Committee of Guangxi Medical University. No human participants or human-derived tissue samples were involved in any part of the research.

3. Results

3.1. Identification of DEGs and signature genes

Using the limma package in R with filtering thresholds set at adj.P.Val.Filter <0.05 and Log2FCfilter >1, a total of 31 DEGs were identified in the merch dataset, comprising 12 up-regulated and 19 down-regulated genes. Visualization outputs included a heatmap and volcano plot of the DEGs (Figure 1A,B). Cross-validation of these 31 genes was conducted using three algorithms—lasso regression, SVM, and RF—within the R environment. The resulting Venn diagram (Figure 2E), integrating outcomes from each method (Figure 2A–D), revealed four convergent signature genes: NFIL3, COL8A1, CX3CR1, and FKBP5 (Table 1). Subsequent differential gene analyses indicated that NFIL3 was associated with 2 DEGs, COL8A1 with 9, and CX3CR1 with 29; no statistically significant DEGs were linked to FKBP5 (Table 2). Based on this comparative output, CX3CR1 was selected as the candidate signature gene for downstream investigation.

Figure 1.

Identification of DEGs. A: Heatmap illustrating DEGs derived from the differential analysis of datasets GSE12021, GSE29746, GSE51588, and GSE55457, with samples categorized into control and treatment groups. Red indicates high ex-pression levels; blue denotes low expression. B: Volcano plot depicting DEG distribution. Red marks upregulated genes, green indicates downregulated genes, and grey represents non-significant differences.

Figure 2.

Identification of signature gene. A: SVM analysis identified 14 genes corresponding to the highest classification accuracy (0.904). B: SVM analysis also revealed 14 genes at the point of lowest cross-validation error (0.0962). C: LASSO regression, optimized via the minimum log (lambda) criterion, selected 15 genes. D: Random Forest analysis yielded 8 genes ranked by importance scores (x axis) relative to gene names (y axis). E: Venn diagram summarizing the intersection of candidate genes obtained from LASSO, SVM, and RF analyses, highlighting the final signature genes.

Table 1.

Three algorithms for genetic screening results.

Algorithm	Genes
Lasso	NFIL3, COL8A1, CX3CR1, TLR7, FKBP5, PGLYRP1, PCDHB6, CLC, EREG, CRTAM, CRTAC1, COL11A1, ERAP2, COMP, ADH1A
SVM	CRTAM, NFIL3, CLC, COMP, CX3CR1, S100A12, IL1R2, ERAP2, FKBP5, XIST, DEFA4, DDX3Y, LRRC15, COL8A1
RF	NFIL3, COL8A1, CX3CR1, PCDHB6, FKBP5, PGLYRP1, TLR7, DDX3Y

Table 2.

The results of differential analysis for signature genes.

Signature genes	Associated up-regulated genes	Associated down-regulated genes
COL8A1	COL11A1, MMP13, LRRC15, COMP, PODNL1	GPC5, LGI1, IGJ, TNFRSF17
CX3CR1	LEP, WIF1, TSPAN8, NPY1R, SLC7A10, GPD1, ADIPOQ, FRZB, DNASE1L3, SSTR1, ADRB1, ACADL, TNMD, SGCG, ITIH5, FABP4, TCEAL2	WNT16, PPEF1, COL6A3, TUBB3, DKK3, HIST1H1A, MGC4294, POSTN, CA12, OGN, ASPN, CCL18
FKBP5	None	None
NFIL3	SAA1, C6	None

3.2. Accuracy analysis of signature gene

CX3CR1 expression was markedly elevated in OA samples compared to normal controls (p < 0.05), as demonstrated by the violin plot generated using ggpubr and the ROC curve analysis (Figure 3A). The AUC for CX3CR1 reached 0.806, surpassing the 0.7 threshold typically indicative of acceptable diagnostic performance. Validation cohort analyses, including violin plots and ROC curves (Figure 3B), yielded consistent results with the experimental group, supporting the reliability of CX3CR1 as a candidate gene for subsequent investigations based on the experimental cohort data.

Figure 3.

Accuracy analysis of signature gene. A: Violin plot demonstrating CX3CR1 expression levels in control versus treatment groups within the experimental dataset; ROC curve indicates classification performance using CX3CR1 (AUC = 0.806). ***p < 0.001. B: Violin plot showing CX3CR1 expression in the validation dataset; ROC curve confirms discriminative capability of CX3CR1 (AUC = 1.000). **p < 0.01.

3.3. Signature gene correlation analysis

Correlation analysis of CX3CR1 identified 29 associated DEGs, comprising 12 downregulated and 17 upregulated genes, as illustrated in the differential expression heatmap and volcano plot (Figure 4A,B). The bubble plot (Figure 4C) visualized the expression patterns and statistical associations among these DEGs, highlighting their interconnected regulatory dynamics with CX3CR1.

Figure 4.

Signature gene correlation analysis. A: Heatmap depicting the expression profiles of genes in samples stratified by CX3CR1 expression levels, with red and green indicating high and low expression groups, respectively. Up-regulation is marked in red, and down-regulation in blue. B: Volcano plot illustrating CX3CR1-associated DEGs: up-regulated genes are colored red, down-regulated green, and non-significant genes gray. C: Bubble plot showing the correlation between CX3CR1 and 29 associated DEGs; red bubbles indicate positive correlations, blue indicate negative correlations.

3.4. Functional enrichment analysis

In the GO enrichment analysis, the bar plot (Figure 5A) revealed significant enrichment (p < 0.05) of CX3CR1-associated up-regulated DEGs in BPs, primarily involving the modulation of adipocyte differentiation. Given the established in vivo association between disrupted lipid metabolism and OA pathogenesis, it is plausible that these DEGs contribute to disease progression via adipogenic pathways. In the category of CCs (Figure 5B), CX3CR1-associated down-regulated DEGs were predominantly enriched in the collagen-containing extracellular matrix (ECM), a key structural element of cartilage that supports chondrocyte proliferation and maturation. MF analysis further indicated that these down-regulated DEGs were closely linked to ECM structural constituents, particularly those conferring resistance to compressive forces. KEGG pathway analysis (Figure 6A) demonstrated that CX3CR1-associated up-regulated DEGs were enriched in lipid metabolism and hormone signaling pathways. Lipid metabolism is tightly regulated by hormonal signals [15]. and adipocytes actively produce various hormones and cytokines that interact with membrane-bound receptors to modulate physiological processes such as energy balance, inflammation, immune regulation, and cellular proliferation [16]. Multiple studies have reported that adipokines elevated in obesity correlate with OA development [17,18]. Among the enriched pathways, the PPAR signaling cascade, known for regulating genes associated with lipid metabolism, adipocyte differentiation, and glucose equilibrium, was also identified [19,20]. Furthermore, excessive activation of the Wnt signaling pathway, previously implicated in OA progression was observed [4]. CX3CR1-associated down-regulated DEGs were primarily enriched in immune-related pathways (Figure 6B), indicating a potential role for CX3CR1 and its downstream targets in modulating immune-mediated mechanisms underlying OA pathogenesis.

Figure 5.

GO functional enrichment analysis. A: Bar of GO enrichment analysis of CX3CR1-associated up-regulated genes. B: Bar of GO enrichment analysis of CX3CR1-associated down-regulated genes.

Figure 6.

KEGG functional enrichment analysis. A: Bar of KEGG enrichment analysis of CX3CR1-associated up-regulated genes. B: Bar plot of KEGG enrichment analysis of CX3CR1-associated down-regulated genes.

3.5. Immune cell infiltration and correlation analysis

In the immune infiltration analysis, the vapor bubble diagram (Figure 7A) illustrated the distribution of 22 immune cell types in both normal and OA sample groups. The boxplot (Figure 7B) identified three immune cell subtypes exhibiting statistically significant distributional differences between groups (p < 0.05): activated dendritic cells and resting mast cells were more prevalent in the normal group, whereas activated mast cells predominated in the OA group. Mast cells (MCs), tissue-resident immune cells, have been detected in elevated numbers in the synovial fluid of OA patients, alongside increased levels of MC-derived mediators such as histamine and trypsin [21]. These mediators—particularly IL-1 and TNF—are known to promote osteoclastogenesis and suppress osteoblast function, thereby contributing to dysregulated bone remodeling [22]. The observed elevation in activated MCs in OA samples implies their involvement in disease progression. The volcano plot (Figure 7C) demonstrated the correlation landscape among immune cell populations, where MCs displayed positive associations with eosinophils and monocytes, and negative associations with resting CD4 memory T cells and activated dendritic cells. Further, the lollipop plot (Figure 8A) highlighted four immune cell subsets significantly associated with CX3CR1 expression (p < 0.05). Specifically, CX3CR1 levels positively correlated with monocytes and eosinophils, and negatively correlated with resting CD4 memory T cells and activated dendritic cells (Figure 8B). These findings indicate that immune cell crosstalk, particularly involving CX3CR1, may modulate the immunopathological landscape of OA.

Figure 7.

Immune cell infiltration analysis. A: Stacked bar graphs comparing the relative abundance of 22 immune cell types in treatment and control groups. B: Violin plots representing the distribution of 22 immune cell subsets across control and treatment samples. Statistical significance was assessed using the Wilcoxon test (*p < 0.05). C. Heatmap of intercellular correlation among the 22 immune cell types; positive correlations are shaded red, negative correlations blue, with color intensity reflecting correlation strength.

Figure 8.

Immune correlation analysis. A: Lollipop plot depicting the correlation between CX3CR1 expression and immune cell subsets. Correlation coefficients >0 indicate positive associations; coefficients <0 indicate negative associations. Circle color reflects p value significance. B: Scatterplots based on Spearman’s rank correlation illustrating the relationship between CX3CR1 expression and the abundance of four selected immune cell types.

3.6. Exploration ceRNA network of the signature gene

Although lncRNAs do not encode proteins, they participate extensively in diverse cellular processes, including micro-regulatory functions, and have attracted growing attention in genetic research. The regulatory interactions between CX3CR1 and miRNAs were predicted using TargetScan, miRanda, and miRDB, identifying four miRNAs targeting CX3CR1 and 18 lncRNAs with shared MREs competing for miRNA binding (Figure 9). Based on these observations, CX3CR1 is hypothesized to influence OA pathogenesis through modulation of its own expression and that of as-sociated genes within the ceRNA network.

Figure 9.

The ceRNA network analysis revealed that CX3CR1 competitively binds to 4 miRNAs, which are respectively associated with 17 lncRNAs, blue for lncRNA, green for miRNA, and red for mRNA.

3.7. Validation of OA signature gene

qRT-PCR and Western blot analyses revealed a significant upregulation of CX3CR1 expression in the OA group relative to the normal group (p < 0.05). However, treatment with AZD8797 in the OA group led to a marked reduction in the expression levels of both CX3CR1 and MMP13 in inflammatory chondrocytes when compared to untreated OA samples (Figure 10A,B). Immunohistochemical analysis of knee joint tissues in SD rats further confirmed elevated CX3CR1 expression in the cartilage surface of OA-affected joints compared to normal counterparts (Figure 10C).

Figure 10.

Analysis and validation of the signature gene CX3CR1. A: Quantitative analysis showing mRNA expression levels of CX3CR1 and MMP13 in chondrocytes from control, LPS group and LPS + AZD8797 group. (***p < 0.001, *p < 0.05). B: Western blot images representing CX3CR1 and MMP13 protein expression and GAPDH as an internal reference in control, LPS group and LPS+ AZD8797 group. (**p < 0.01, *p < 0.05). C: Immunohistochemical staining of CX3CR1 in articular cartilage sections from control and OA groups and quantitative analysis of positive cell staining in immunohistochemistry of CX3CR1. The results of the quantitative analysis were obtained from three independent experiments (*p < 0.05).

4. Discussion

OA, the most common chronic joint disorder, remains among the few aging-associated diseases with limited therapeutic efficacy and no proven interventions capable of decelerating its progression [23]. Current treatment modalities offer minimal symptomatic relief and fail to alter the underlying disease trajectory. Addressing the challenge of mitigating disease advancement and improving patient outcomes necessitates intensified global efforts, particularly through comprehensive investigation into the pathogenesis and molecular pathways driving OA progression.

This study integrated data from public repositories and employed bioinformatics analysis to identify CX3CR1 as a signature gene significantly associated with OA progression. Experimental validation confirmed the correlation between CX3CR1 expression and OA pathophysiology. Up-regulated DEGs linked to CX3CR1 were predominantly enriched in adipose metabolism and adipokine-related pathways, implicating their involvement in OA development. Conversely, CX3CR1-associated down-regulated genes demonstrated a strong association with the collagen-rich extracellular matrix, a structural component vital to articular cartilage integrity, indicating potential regulatory effects on cartilage homeostasis. Furthermore, CX3CR1 expression modulated the in vivo abundance of MCs and related immune cell populations, thereby contributing to OA progression through immunological mechanisms.

CX3CR1, a monogamous chemokine receptor within the GPCR superfamily [24], serves as the exclusive receptor for the chemokine CX3CL1 (also known as fractalkine, FKN) and mediates pro-inflammatory leukocyte signaling [25]. Chemokines primarily regulate leukocyte chemotaxis and immune responses by establishing concentration gradients within tissues [26]. CX3CR1 is expressed by both infiltrating immune cells, such as monocytes [27], and tissue-resident populations, including macrophages [28]. The CX3CL1/CX3CR1 axis has been implicated in various pathological conditions, including atherosclerosis [29], neurological disorders [30],vasculitis [31], and cardiac dysfunction [32]. Previous research identified FKN as a chemoattractant for OA fibroblasts, which contributed to joint deterioration in OA through the secretion of inflammatory cytokines and matrix metalloproteinases (MMPs). In vitro results from western blot and qPCR analyses indicated a marked reduction in MMP13 expression in chondrocytes following CX3CR1 inhibition with AZD8797 in OA samples, highlighting a strong association between CX3CR1 activity and cartilage degeneration. Beyond its chemotactic role, FKN triggers OA fibroblast signaling via MAPK pathways including p38, JNK, ERK1/2, and Akt [33], while CX3CR1 activation promotes cartilage matrix breakdown by stimulating NF-κB-dependent transcription at the MMP-3 promoter, thereby enhancing the release of degradative enzymes such as MMPs, ADAMTS4, and ADAMTS5 during OA progression [34].

As previously discussed, β-catenin overactivation in articular chondrocytes contributes to both the initiation and progression of OA. Notably, CX3CR1 exhibits significantly elevated expression in OA cartilage compared to normal articular cartilage and modulates chondrocyte proliferation and apoptosis via the Wnt/β-catenin signaling cascade [35], This aligns with the KEGG enrichment results, where CX3CR1-associated up-regulated DEGs were predominantly mapped to the Wnt signaling pathway. Furthermore, immunohistochemical analysis of rat knee joints revealed markedly increased CX3CR1 expression in OA samples relative to controls.

The infrapatellar fat pad, as the primary adipose structure within the knee joint, serves as a major source of adipokine secretion. CX3CL1, an adipocyte-derived inflammatory chemokine, has been implicated in promoting monocyte adhesion to adipocytes. A strong positive correlation was observed between CX3CR1 expression and monocyte infiltration, indicating that elevated CX3CR1 may enhance monocyte accumulation within adipose tissue, contributing to localized inflammatory responses potentially associated with intra-articular inflammation. Notably, CX3CR1 expression is markedly upregulated in obesity and obesity-induced adipose inflammation [36], and the induction of chemokine expression constitutes a key mechanism driving early leukocyte recruitment and the onset of adipose tissue inflammation. Functional enrichment analysis revealed that CX3CR1-associated up-regulated DEGs were predominantly enriched in pathways linked to adipose metabolism, suggesting that CX3CR1 may modulate inflammatory signaling by regulating adipokine release. This mechanism appears to initiate intra-articular inflammation, subsequently promoting cartilage degradation and accelerating OA progression. However, further investigation is required to delineate the precise molecular pathways underlying these interactions.

Accumulating evidence indicates that intra-articular synovitis and immune dysregulation contribute substantially to the initiation and progression of OA, with immune cells and their cytokine networks implicated in disease etiology. Monocytes, macrophages, T-lymphocytes, and MCs represent the principal immune populations infiltrating OA-affected joints [37]. Immune infiltration analysis in this study identified a marked positive association between CX3CR1 and both monocytes and eosinophils. Upregulation of CX3CR1 has been shown to mediate monocyte recruitment, thereby promoting localized inflammatory responses. Although the role of eosinophils in OA remains undefined, elevated eosinophil levels have been observed in the subcutaneous adipose tissue of individuals with metabolic syndrome under pro-inflammatory conditions, and a distinct eosinophil subset has been reported in rheumatoid arthritis patients [38]. As key components of the innate immune surveillance system, MCs rapidly respond to endogenous stress signals and external pathogens. Upon activation by diverse stimuli, MCs release a spectrum of inflammatory mediators, including cytokines, chemokines, histamine, trypsin, and lipid-derived factors, which contribute to synovial inflammation, neovascularization, and bone degradation [37]. In this study, significantly elevated levels of activated MCs in OA samples, compared to normal controls, suggest a potential mechanism by which CX3CR1 may promote OA pathogenesis through modulation of immune cell infiltration within the joint microenvironment.

In this research, CX3CR1 was identified as a signature gene through bioinformatics analysis, followed by functional enrichment, immune cell infiltration profiling, and a series of in vitro and in vivo validations. The results indicate that CX3CR1 influences OA pathogenesis primarily through pathways regulating adipose metabolism and immune cell infiltration, particularly involving MCs. Consequently, CX3CR1 may serve as a potential biomarker for early OA diagnosis and clinical intervention. However, certain limitations remain. First, the datasets utilized are derived from publicly available platforms rather than directly acquired specimens. Second, this study does not investigate in detail the molecular mechanisms through which CX3CR1 contributes to disease progression; further exploration is required to delineate its downstream regulatory network. Although the findings have been substantiated by experimental validation, the development of more advanced models will be necessary to enhance mechanistic clarity and confirm diagnostic potential [36].

Funding Statement

This work received financial support from the Guangxi Natural Science Foundation (2023GXNSFAA026402), Guangxi Medical and Health Appropriate Technology Development and Extension and Application Project (GZSY22-62), Nanning Qingxiu District Science and Plan Project (2020018), Guangxi Science and Technology Base and Talent Special Project (GuikeAD19254003), and the Self-funded Project of the Health Department of Guangxi Zhuang Autonomous Region (Z2013039).

Ethics statement

Animal procedures were reviewed and approved by the Medical Ethics Committee of the First Affiliated Hospital of Guangxi Medical University (Approval No. 2024-E012-01). All experimental protocols adhered to the ARRIVE guidelines.

Disclosure statement

No commercial or financial relationships were identified during the study that could be interpreted as a potential conflict of interest.

Publisher’s note

This disclaimer indicates that all views expressed in this article are solely those of the author and do not necessarily reflect the position of the organization, publisher, editor, or reviewer with which he or she is affiliated. Any products evaluated in the article or any statements made by their manufacturers are not warranted or endorsed by the publisher.

Data availability statement

All datasets analyzed in this study were obtained from the publicly accessible Gene Expression Omnibus (GEO, https://www.ncbi.nlm.nih.gov/geo/). Experimental data generated during the current study are available from the corresponding author upon reasonable request.