Next Article in Journal
Feasibility and Preliminary Efficacy of Wearable Focal Vibration Therapy on Gait and Mobility in People with Multiple Sclerosis: A Pilot Study
Previous Article in Journal
Keratinocyte Capability for Smooth Sheet Formation on a Step Pattern Substrate
Previous Article in Special Issue
Exploring Sarcopenic Obesity in the Cancer Setting: Insights from the National Health and Nutrition Examination Survey on Prognosis and Predictors Using Machine Learning
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Bioengineering
Volume 12
Issue 9
10.3390/bioengineering12090930
bioengineering-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Mendelian Randomization and Transcriptome Analyses Reveal Important Roles for CEBPB and CX3CR1 in Osteoarthritis

by
Hui Gao
1,†,
Xinling Gan
1,†,
Jing He
1,2,* and
Chengqi He
1,2,*
1
Rehabilitation Medicine Center and Institute of Rehabilitation Medicine, West China Hospital, Sichuan University, Chengdu 610000, China
2
Key Laboratory of Rehabilitation Medicine in Sichuan Province, Chengdu 610000, China
*
Authors to whom correspondence should be addressed.
†
These authors contributed equally to this work.
Bioengineering 2025, 12(9), 930; https://doi.org/10.3390/bioengineering12090930
Submission received: 2 July 2025 / Revised: 8 August 2025 / Accepted: 26 August 2025 / Published: 29 August 2025
(This article belongs to the Special Issue Computer Vision and Machine Learning in Medical Applications, 2nd Edition)
Browse Figures
Versions Notes

Abstract

Background: Chemokines play a pivotal role in the progression of osteoarthritis (OA), but their exact mechanisms remain unclear. This study aimed to identify potential chemokine-associated biomarkers and investigate their causal relationships with OA. Methods: Transcriptome and genome-wide association study (GWAS) data were obtained from public databases, while chemokine-related genes (CRGs) were sourced from the literature. Initially, CRGs were expanded, followed by Mendelian randomization (MR) analysis, differential expression analysis, machine learning, and receiver operating characteristic (ROC) curve plotting to identify potential biomarkers. The causal relationships between these biomarkers and OA, as well as their biological functions, were further explored. Results: Fourteen candidate genes were identified for machine learning analysis, with DDIT3, CEBPB, CX3CR1, and ARHGAP25 emerging as feature genes. CEBPB and CX3CR1, which exhibited AUCs > 0.7 in the GSE55235 and GSE55457 datasets, were selected as potential biomarkers. Notably, CEBPB expression was lower, while CX3CR1 expression was elevated in the case group. Furthermore, both genes were co-enriched in spliceosome, lysosome, and cell adhesion molecule pathways. MR analysis confirmed that CEBPB and CX3CR1 were causally linked to OA and acted as protective factors (IVW model for CEBPB: OR = 0.9051, p = 0.0001; IVW model for CX3CR1: OR = 0.8141, p = 0.0282). Conclusions: CEBPB and CX3CR1 were identified as potential chemokine-related biomarkers, offering insights into OA and suggesting new avenues for further investigation.

Graphical Abstract

1. Introduction

Osteoarthritis (OA), a prevalent chronic joint disorder, is primarily characterized by the degeneration of joint cartilage and inflammation of surrounding tissues [1]. This condition results in joint pain, stiffness, swelling, and functional impairment, significantly reducing patients’ quality of life [2]. OA ranks among the most common joint diseases globally, with a higher prevalence in the elderly population [3]. However, it also affects middle-aged and younger individuals, particularly those with a history of joint injuries or deformities. The incidence of OA is influenced by several factors, including aging, obesity, genetic predisposition, and joint trauma. OA progression is gradual, with symptoms worsening over time, and severe cases may lead to disability [4]. Current treatment approaches primarily focus on symptom management, including pain relief, anti-inflammatory drugs, physical therapy, and joint replacement surgery [5,6,7]. These therapies, however, provide only symptomatic relief and do not prevent disease progression or repair damaged joint cartilage. Consequently, identifying new therapeutic strategies has become a critical focus in OA research. Recently, an increasing number of studies have concentrated on the discovery and application of biomarkers to aid in diagnosing OA, assessing disease progression, and guiding personalized treatment [8,9].
Chemokines are small molecular proteins that play a pivotal role in cell chemotaxis, activation, and the regulation of immune cell migration [10]. In the context of inflammation and immune responses, chemokines are essential for cell regulation and tissue localization [11]. The chemokine family includes several subtypes, such as CC, CXC, C, and CX3C, with CC motif chemokine ligands (CCL) and their corresponding receptors (CCR) being particularly important [12] in disease onset and progression [13,14,15,16]. Abnormal expression of CCL and CCR is frequently associated with exacerbated inflammation and disease progression in various inflammatory conditions [17]. Despite extensive research into the roles of chemokines and related molecules across numerous diseases, their precise mechanisms in joint diseases like OA remain unclear and warrant further investigation [18].
Mendelian randomization (MR) has emerged as a key method for investigating the causal mechanisms of disease [19]. By simulating randomized controlled trials using naturally occurring genetic variations, MR evaluates the causal effects of specific genes on phenotypic outcomes [20]. However, MR studies focused on OA remain limited, and the influence of chemokine-related genes (CRGs) on OA is not yet fully understood [21]. Therefore, further MR studies are essential for a comprehensive understanding of the role of chemokines in OA.
For the first time, this study integrates transcriptomic, GWAS, and eQTL data related to OA, combined with Mendelian randomization analysis and machine learning approaches, to identify potential chemokine-related biomarkers and establish a novel framework for OA research. On this basis, functional, regulatory mechanism, and causal relationship analyses were conducted to provide a theoretical foundation for the diagnostic value of chemokines in OA. This work aims to offer new perspectives and methodologies for OA etiology and treatment, thereby proposing more effective diagnostic and therapeutic strategies to improve the overall management of OA.

2. Materials and Methods

2.1. Data Sources

The training set (GSE55235) and validation set (GSE55457) were sourced from the GEO database (https://www.ncbi.nlm.nih.gov/gds, accessed on 8 September 2023). These datasets included ten synovial tissue samples from osteoarthritic joints and ten synovial tissue samples from healthy joints (control group), respectively. CRGs, including CCL and CCR, were obtained from prior studies and encompassed inflammatory chemokines, homeostatic chemokines, and bifunctional chemokines [18]. Summary-level data for OA (ebi-a-GCST005810) and potential biomarkers (eqtl-a-ENSG00000172216 and eqtl-a-ENSG00000168329) were downloaded from the IEU OpenGWAS database (https://gwas.mrcieu.ac.uk/, accessed on 8 September 2023). The ebi-a-GCST005810 dataset consisted of 15,543,628 single-nucleotide polymorphisms (SNPs) from 11,989 European samples.

2.2. Weighted Gene Coexpression Network Construction Analysis (WGCNA)

The ssGSEA algorithm in the “GSVA” package was used to calculate the CCL and CCR scores for each sample in the GSE55235 dataset to identify module genes associated with CC chemokine ligands and receptors [22]. The “GoodSamplesGenes” function in the “WGCNA” package was then applied to cluster all samples in the training set, and outlier samples were identified and excluded to ensure the accuracy of the analysis [23]. To assess whether genes exhibited similar expression patterns, the scale-free fit index (R2) was set to 0.80. Optimal soft thresholding, with values exceeding 0.80 and a mean connectivity near 0, was selected to construct a scale-free co-expression network. Based on this threshold, the minModuleSize was set to 200, and gene modules were obtained using the hybrid dynamic tree cutting algorithm. CCL and CCR scores were introduced as traits, and Pearson correlation analysis was performed between traits and module genes using the “cor” function in the “corrplot” package (|correlation| > 0.3, p < 0.05) [24]. A correlation heatmap was generated using the “ggplot2” package. The module with the highest correlation was selected for subsequent analysis.

2.3. Differential Expression Analysis

Differentially expressed genes (DEGs) between OA and control samples were identified through differential expression analysis in GSE55235 using the “limma” package (|log2FC| > 0.5, adj. p < 0.05) [25]. Volcano plots and heatmaps were created to visualize the expression of these DEGs.

2.4. Identification of Candidate Genes

To further narrow down candidate genes, DEGs and key module genes were overlapped to identify DE-CRGs. MR analysis was then performed using the “TwoSampleMR” package to select genes with p-values < 0.05 for IVW analysis and level effects > 0.05 for subsequent evaluation [26]. Gene Ontology (GO) term and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses were conducted using the “ClusterProfiler” package to explore the functional roles of these genes (adj. p < 0.05) [27]. Genes identified through MR analysis were imported into the STRING database to investigate gene interactions and construct a protein–protein interaction (PPI) network using “Cytoscape 3.8.0 software” [28]. The top 20 genes, ranked by normalized cross-correlation (NCC) and DNMC methods in the cytoHubba plug-in, were intersected to select candidate genes.

2.5. Identification of Potential Biomarkers, Establishment of Nomogram, and Expression Validation

To further identify genes closely related to OA, candidate genes in the training set were screened using two machine learning algorithms: LASSO analysis and SVM-RFE. LASSO, with 10-fold cross-validation, was performed using the “glmnet” package [29], and the optimal genes were selected when the lambda value was minimized. SVM-RFE analysis was carried out using the “caret” package [30], and the optimal gene combination was determined by selecting the point with the lowest error rate. The intersection of the LASSO and SVM-RFE genes was obtained using the “ggvenn” package [31] to identify feature genes. The diagnostic value of these feature genes was assessed with the “pROC” package by plotting ROC curves for each gene in the GSE55235 and GSE55457 datasets. Genes with areas under the curve (AUCs) > 0.7 were selected as potential biomarkers. A nomogram model was then constructed based on these biomarkers using the “rms” package [32], and calibration curves were generated to validate the model’s efficacy. The expression levels of the potential biomarkers were further validated in the GSE55235 and GSE55457 datasets, with the Wilcoxon test applied to compare differences between OA and control samples (p < 0.05).

2.6. Gene Set Enrichment Analysis (GSEA)

To investigate the biological pathways associated with the potential biomarkers, the “c2.cp.kegg.v7.4.symbols.gmt” reference gene set from the MSigDB database (https://www.gsea-msigdb.org/gsea/msigdb, accessed on 10 September 2023) was used. Pearson correlations between the potential biomarkers and other genes in each sample were calculated using the “psych” package [33], with genes sorted by correlation coefficients. GSEA was then performed using the “clusterProfiler” package, with a normalized enrichment score (NES) > 1 and adjusted p-value < 0.05. The top 5 pathways with the most significant p-values were displayed for analysis [27].

2.7. MR Analysis

Exposure factor reading and filtering were conducted using the “extract_instruments” function from the “TwoSampleMR” package with a p-value threshold of <5 × 10−8 for MR analysis [26]. SNPs for linkage disequilibrium analysis (LDA) were removed (clump = TRUE, r2 = 0.001, kb = 10000), and the F-statistic was calculated. SNPs were considered sufficiently robust when F > 10. Instrumental variables (IVs) strongly correlated with exposure factors were then selected for MR analysis. Three key assumptions underpin MR studies: (1) IVs must be strongly correlated with exposure factors, (2) IVs must be independent of other confounding factors, and (3) IVs must influence the outcome only through the exposure factors.
The “Harmonize_data” function from the “TwoSampleMR” package was used to harmonize effect equivalents and effect sizes. The primary MR methods employed were MR–Egger [34], weighted median [35], inverse-variance weighted (IVW) [36], simple mode [26], and weighted mode [37]. Among these, IVW was considered the most crucial method due to its superior ability to detect causal relationships. A risk factor was identified when the odds ratio (OR) exceeded 1, while an OR below 1 indicated a protective factor. Scatter plots, forest plots, and funnel plots were generated to visualize the results. To assess the reliability of the MR analysis, a sensitivity analysis was conducted. First, heterogeneity was tested using Cochran’s Q test (p > 0.05). Second, a horizontal pleiotropy test was performed (p > 0.05). Finally, the leave-one-out (LOO) method was applied, systematically removing each SNP. If the exclusion of any SNP did not significantly alter the outcome, this indicated the robustness of the MR analysis.

2.8. Network Construction

The ChEA3 database (https://maayanlab.cloud/chea3/, accessed on 10 September 2023) was used to predict transcription factor (TF)-targeting potential biomarkers. The starBase database (http://mirdb.org/, accessed on 10 September 2023) was used to identify the miRNAs targeting these potential biomarkers (pancancerNum ≥ 6). The “Cytoscape software” was then used to visualize the miRNA-biomarker-TF network. Furthermore, to explore the interactions between potential biomarkers and drugs, the CTD database (https://ctdbase.org/, accessed on 10 September 2023) was used to predict potential drugs associated with the biomarkers, and a biomarker-drug network was constructed.

2.9. Statistical Analysis

R software (v 4.2.3) was employed for data processing and analysis. Group comparisons were performed using the Wilcoxon test, with a p-value of < 0.05 considered statistically significant (p < 0.05).

3. Results

3.1. Recognition of DE-CRGs

To identify genes associated with the CCL and CCR scores, WGCNA was performed. Clustering analysis revealed no outlier samples, indicating that subsequent analyses could proceed (Figure 1A). The optimal soft threshold was determined to be seven, at which point the interactions among genes best conformed to a scale-free distribution (Figure 1B). Based on this threshold, nine modules were identified (Figure 1C). The MEbrown module, which showed a negative correlation with the CCL score (cor = −0.82), and the MEturquoise module, which exhibited a positive correlation with the CCR score (cor = 0.65), were selected for further analysis (Figure 1D). A total of 6339 genes in these modules were selected as key module genes for subsequent analysis. Differential expression analysis identified 1797 DEGs from the GSE55235 dataset (case vs. control), which included 1084 overexpressed genes and 713 underexpressed genes. A volcano plot and heatmap were generated to visualize the expression of DEGs (Figure 1E,F). By overlapping the DEGs with the key module genes, 1466 DE-CRGs were identified for further study.

3.2. CEBPB and CX3CR1 Were Identified as Potential Biomarkers

The DE-CRGs were incorporated into the MR analysis, resulting in the identification of 82 genes with a causal relationship to OA (Table 1), offering valuable insights into the disease’s pathogenesis. These genes were primarily involved in the apelin signaling pathway and the cellular response to biotic stimulus pathway (Figure 2A,B), suggesting that further exploration of these pathways could unveil novel therapeutic targets for OA. The PPI network analysis revealed intricate interactions between CEBPB, CX3CR1, and other genes, including ANK1 and CRLF3 (Figure 2C). After intersecting the genes identified through the NCC and DNMC algorithms, 14 candidate genes were ultimately selected for further analysis (Figure 2D).
The LASSO analysis identified five feature genes—DDIT3, TFAM, CEBPB, CX3CR1, and ARHGAP25—when the lambda value was set to 5 × 10−4. Similarly, the SVM-RFE analysis revealed six feature genes—DDIT3, CX3CR1, CEBPB, PTAFR, ARHGAP25, and MYO1F—at the point of lowest error rate. The intersection of the feature genes from both methods resulted in four common genes: DDIT3, CEBPB, CX3CR1, and ARHGAP25 (Figure 3A,B). ROC curve analysis in both the training and testing sets revealed that CEBPB and CX3CR1 had AUCs greater than 0.7, thus confirming them as potential biomarkers (Figure 3C,D). Gene expression analysis in the GSE55235 and GSE55457 datasets indicated that CEBPB was expressed at low levels, while CX3CR1 was highly expressed in the case group (Figure 3E).
A nomogram was constructed for CEBPB and CX3CR1, where each potential biomarker corresponded to a specific point, and the sum of these points represented the total score. This total score could predict the prevalence of OA and was positively correlated with its incidence (Figure 4A). Calibration curve analysis showed that the curve closely approximated 1, indicating the high diagnostic accuracy of the nomogram (Figure 4B). This tool offers a valuable resource for facilitating communication between doctors and patients, enabling physicians to use the nomogram to select tailored, personalized treatment plans. Additionally, GSEA was performed to explore the biological functions of the biomarkers in greater detail. The results demonstrated that CEBPB and CX3CR1 were co-enriched in spliceosome, lysosome, and cell-adhesion molecule (CAM) pathways (Figure 4C). Abnormal spliceosome function could disrupt gene expression accuracy, leading to dysfunctions in related cells such as chondrocytes. Changes in lysosomal function may impair the degradation and metabolism of cellular substances, further affecting cartilage tissue homeostasis. Alterations in CAMs could disrupt cell–cell interactions and adhesion between cells and the extracellular matrix, thereby influencing the structure and function of articular cartilage. The co-enrichment of CEBPB and CX3CR1 in these pathways suggests that they may contribute to OA pathogenesis through a synergistic effect.

3.3. CEBPB and CX3CR1 Were Causally Associated with OA

After IV screening, a total of four SNPs related to CEBPB and five SNPs related to CX3CR1 were identified (Table S1). The MR analysis demonstrated that CEBPB and CX3CR1 were causally associated with OA and identified as protective factors (IVW model for CEBPB: OR = 0.9051, p = 0.0001; IVW model for CX3CR1: OR = 0.8141, p = 0.0282) (Table 2 and Table 3). The negative slope observed in the IVW method’s scatter plot suggested that higher levels of CEBPB and CX3CR1 were associated with a reduced risk of developing OA (Figure 5A). This relationship was further validated by the forest plot (Figure 5B). Additionally, the SNPs displayed a roughly symmetrical distribution on both sides of the plot, supporting the alignment with the second law of Mendelian inheritance (Figure 5C). Sensitivity analysis indicated no heterogeneity, as confirmed by the Cochran’s Q test (p > 0.05) (Table 4). The horizontal pleiotropy test also revealed no pleiotropic effects in the MR analysis (p > 0.05) (Table 5). Furthermore, LOO analysis showed no significant bias, reinforcing the reliability of the overall findings (Figure 5D).

3.4. Complex Interactions Between Potential Biomarkers

To explore the regulatory mechanisms of the potential biomarkers, target TFs and miRNAs were predicted via a database, identifying 29 TFs and 11 miRNAs for CEBPB, and one TF and three miRNAs for CX3CR1 (Figure 6A,B). A regulatory network was constructed, illustrating interactions such as GATA2-CX3CR1-hsa-miR-1276 (Figure 6C). This highlights the complex roles these components play in gene expression regulation, providing insights into the molecular processes involved in the onset and progression of OA. Furthermore, a regulatory relationship was identified between valproic acid and benzo(a)pyrene, both of which are potential biomarkers with therapeutic implications for OA (Figure 6D).

4. Discussion

Recent research has firmly established that persistent, low-intensity inflammation, encompassing both innate and adaptive immune responses, significantly influences the onset and progression of OA [38,39]. The interplay between CCLs and CCRs leads to the recruitment of various immune cells into the injured joints, contributing to local inflammation [40,41]. Additionally, within the nerve endings of the knee joint, CCLs, CCRs, and cytokines initiate the release of spinal neurotransmitters, causing hyperalgesia [42,43]. Consequently, this study identified two potential biomarkers, CEBPB and CX3CR1, using bioinformatics methods. These biomarkers can serve as quantitative indicators to predict disease progression, joint function deterioration, and patient response to treatment more accurately, facilitating the development of more proactive intervention strategies. Moreover, although various treatment options exist for late-stage OA, including pharmacotherapy, physical therapy, and surgical interventions, patients’ responses can vary. Identifying potential biomarkers can help uncover individual biological characteristics, enabling the customization of the most effective treatment plans. These biomarkers can also aid in monitoring disease recurrence, assisting physicians in promptly assessing changes in the conditions of patients with late-stage OA, adjusting treatment plans accordingly, and preventing further disease progression.
CEBPB, or CCAAT/enhancer-binding protein beta, is a TF in the C/EBP family. It can be activated by various inflammatory stimuli such as IL-17 and LPS, subsequently modulating multiple genes involved in the inflammatory process [44]. The upregulation of CEBPB in Alzheimer’s disease promotes the expression of proinflammatory genes in microglia and affects macrophage activation [44]. CEBPB also plays a role in dendritic cells and in autoimmune disorders of the central nervous system [45]. In patients with amyotrophic lateral sclerosis (ALS), CEBPB expression was elevated in lymphocytes and nerve tissue, making it a potential marker for ALS progression [46,47]. This suggests a strong connection between CEBPB and inflammatory processes in nerve tissues, indicating its involvement in various neurological inflammatory responses. Autoimmunity and inflammation are closely linked to OA’s development and progression. Furthermore, CEBPB is associated with macrophage-related pathways [48], suggesting a correlation with OA’s pathological process. Notably, CEBPB is a gene connected to both OA and metabolic syndrome, and it holds diagnostic value for OA individuals with metabolic syndrome [49]. Wang et al. found that 5,7,3’,4’-tetramethoxyflavone inhibits extracellular matrix degradation in OA by modulating the C/EBPβ/ADAMTS5 signaling pathway [50]. Nevertheless, its precise function in inflammation-associated disorders is still a matter of debate and warrants more in-depth studies [51]. In the present study, MR analysis revealed that CEBPB serves as a therapeutic target for OA, showing a causal relationship with the disease. MR effectively minimizes confounding factors and reverse causation, identifying CEBPB as a protective factor for OA. This provides stronger evidence for the causal link, offering a deeper and more comprehensive understanding of the relationship between CEBPB and OA.
CX3CR1, or CX3C chemokine receptor 1, is a G protein-coupled receptor that primarily interacts with the chemokine CX3CL1 (also known as fractalkine or neurotactin) [52]. While an expression correlation analysis of clinical samples has identified CX3CL1 as a potential biomarker for knee osteoarthritis, its receptor CX3CR1 remains unreported in this regard, suggesting a gap that merits further investigation [53]. It is found on the inner lining of synovial fibroblasts in the knee joint, where CX3CR1-positive macrophages form a dense physical barrier with CX3CR1, isolating the joint space from the external environment and protecting the joint [54]. These findings suggest that CX3CR1 may influence the pathogenesis of OA. Notably, the MR analysis in this study revealed a causal relationship between CX3CR1 and OA, with CX3CR1 acting as a protective factor against the disease. This result is consistent with most previous studies, providing a solid theoretical basis for the clinical diagnosis, treatment, and prognosis of OA.
Through database searches, this study predicted the target TFs and miRNAs for the potential biomarkers. Previous research suggests that the target miRNA of CX3CR1, hsa-miR-1276, may be linked to the development of cardiovascular diseases [55,56]. Hsa-miR-33a-5p is potentially associated with chemoresistance in hepatocellular carcinoma [57], while hsa-miR-33b-5p may be related to type 2 diabetes, myocardial infarction, and other conditions [58,59]. Notably, the relationships between these target miRNAs of CX3CR1 and OA have not been explored. Our study is the first to propose that these three miRNAs could be involved in OA development via CX3CR1.
The database search identified 11 target miRNAs for CEBPB. Previous studies have suggested that hsa-miR-20b-5p is implicated in diseases such as atrial fibrillation and liver cirrhosis [60,61], while hsa-miR-106b-5p is linked to pulmonary hypertension and melanoma progression [62,63]. These 11 target miRNAs of CEBPB have not been previously associated with OA development. Our study is the first to propose a potential connection, offering new insights into the mechanisms underlying OA.
This study is the first to systematically construct a miRNA–mRNA regulatory network centered on CX3CR1 and CEBPB, identifying several miRNAs previously unreported in the context of osteoarthritis (OA). This provides a novel perspective for exploring post-transcriptional regulatory mechanisms in OA. For example, hsa-miR-1276, a predicted regulator of CX3CR1, has been associated with the pathogenesis of cardiovascular diseases [55,56]. hsa-miR-33a-5p may be involved in chemotherapy resistance in hepatocellular carcinoma [57], while hsa-miR-33b-5p plays significant roles in type 2 diabetes and myocardial infarction [58,59]. In addition, among the miRNAs targeting CEBPB, hsa-miR-20b-5p has been implicated in atrial fibrillation and liver cirrhosis [60,61], and hsa-miR-106b-5p is known to contribute to the progression of pulmonary arterial hypertension and melanoma [62,63]. Although these miRNAs have been demonstrated to exert important biological functions in various diseases, their relevance to OA has not yet been established. The findings of this study offer new directions and molecular clues for future mechanistic investigations in OA.
Additionally, 29 TFs that target CEBPB were predicted in this study. Several of these TFs have been confirmed to play roles in the development and progression of OA. For example, JUND promotes OA progression via the miR-423-5p/KDM5C axis and induces immune inflammation [64,65]. ATF3 has been identified as a potential diagnostic marker for OA and is involved in synovial immunity and chondrocyte death [66,67,68,69]. Upregulation of TCF12 is known to lead to OA progression [70], and GATA3 is associated with cartilage damage during OA development [71,72]. Both TBP and MXI1 are linked to the occurrence and progression of OA [73,74]. Previous studies have connected GATA2 to rheumatism [75], and PRDM1 may be associated with Alzheimer’s disease development [76]. However, other target TFs of CEBPB have not been previously discussed in relation to OA progression, suggesting new avenues for research into their involvement in OA.
Both valproic acid and benzo[a]pyrene were found to have regulatory relationships with two potential biomarkers, suggesting their therapeutic potential for OA. Valproic acid, an anticonvulsant and mood stabilizer, operates through multiple mechanisms and is mainly used to treat epilepsy and bipolar disorder. Previous studies have demonstrated that VPA can influence neurotransmitter levels and regulate gene expression, though the precise mechanisms, particularly regarding specific pathways and targets in various disease states, remain unclear [77]. Further research could shed light on these mechanisms. Benzo[a]pyrene, primarily studied for its carcinogenic effects, has been shown to induce DNA double-strand breaks [78]. Future research could explore whether its therapeutic effects on OA involve gene regulation in synovial cells.
This study identified two potential biomarkers, CEBPB and CX3CR1, through bioinformatics analysis. MR analysis revealed a significant causal relationship between these biomarkers and OA, establishing that both genes serve as protective factors against OA. Based on these findings, a series of in-depth analyses were conducted to explore the functions and potential regulatory mechanisms of these genes. Although bioinformatics analysis has provided significant insights and direction for our research, the certainty and broad applicability of these results are somewhat limited due to the lack of validation through biological experiments. We are fully aware of the indispensable nature of experimental validation in biology. The reason for not conducting related experiments in this study mainly stems from resource limitations, time constraints, and difficulties in sample collection. Despite these challenges, we are firmly committed to conducting experimental validations and have developed a detailed future work plan. We will verify our findings through animal and cell experiments in the future, providing more effective strategies and methods for the diagnosis and treatment of OA. Additionally, the current data and types may not fully support all conclusions related to the early detection of OA biomarkers. To address this limitation, future research will focus on collaboration with other research teams or medical institutions, enabling access to broader and more representative patient data through data sharing or joint research initiatives. This collaboration will enhance the database and deepen the investigation into early biomarkers of OA.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/bioengineering12090930/s1, Table S1: IV screening of CEBPB and CX3CR1.

Author Contributions

All the authors were involved in conceiving the work and its main ideas. H.G. initially wrote the main body of the text. X.G. made revisions to the manuscript and created the figures and tables. C.H. and J.H. oversaw the work, offering comments and supplementary scientific details. H.G. and X.G. had equal contributions to this study. All authors have read and agreed to the published version of the manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (grant number: 82402993), the 1.3.5 project for disciplines of excellence, West China Hospital, Sichuan University (grant number: ZYGD23014), and the Natural Science Foundation of Sichuan Province (grant number: 2024NSFSC1580).

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

The original contributions presented in this study are included in the article/supplementary material. Further inquiries can be directed to the corresponding authors.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Martel-Pelletier, J.; Barr, A.J.; Cicuttini, F.M.; Conaghan, P.G.; Cooper, C.; Goldring, M.B.; Goldring, S.R.; Jones, G.; Teichtahl, A.J.; Pelletier, J.P. Osteoarthritis. Nat. Rev. Dis. Primers 2016, 2, 16072. [Google Scholar] [CrossRef]
  2. Jiang, Y. Osteoarthritis year in review 2021: Biology. Osteoarthr. Cartil. 2022, 30, 207–215. [Google Scholar] [CrossRef]
  3. Minnig, M.C.C.; Golightly, Y.M.; Nelson, A.E. Epidemiology of osteoarthritis: Literature update 2022–2023. Curr. Opin. Rheumatol. 2024, 36, 108–112. [Google Scholar] [CrossRef] [PubMed]
  4. Perruccio, A.V.; Young, J.J.; Wilfong, J.M.; Denise Power, J.; Canizares, M.; Badley, E.M. Osteoarthritis year in review 2023: Epidemiology & therapy. Osteoarthr. Cartil. 2024, 32, 159–165. [Google Scholar] [CrossRef]
  5. Kolasinski, S.L.; Neogi, T.; Hochberg, M.C.; Oatis, C.; Guyatt, G.; Block, J.; Callahan, L.; Copenhaver, C.; Dodge, C.; Felson, D.; et al. 2019 American College of Rheumatology/Arthritis Foundation Guideline for the Management of Osteoarthritis of the Hand, Hip, and Knee. Arthritis Care Res. 2020, 72, 149–162. [Google Scholar] [CrossRef] [PubMed]
  6. Sharma, L. Osteoarthritis of the Knee. N. Engl. J. Med. 2021, 384, 51–59. [Google Scholar] [CrossRef]
  7. Whittaker, J.L.; Truong, L.K.; Dhiman, K.; Beck, C. Osteoarthritis year in review 2020: Rehabilitation and outcomes. Osteoarthr. Cartil. 2021, 29, 190–207. [Google Scholar] [CrossRef]
  8. Mobasheri, A.; Thudium, C.S.; Bay-Jensen, A.C.; Maleitzke, T.; Geissler, S.; Duda, G.N.; Winkler, T. Biomarkers for osteoarthritis: Current status and future prospects. Best Pract. Res. Clin. Rheumatol. 2023, 37, 101852. [Google Scholar] [CrossRef]
  9. Watt, F.E. Osteoarthritis biomarkers: Year in review. Osteoarthr. Cartil. 2018, 26, 312–318. [Google Scholar] [CrossRef] [PubMed]
  10. Luster, A.D. Chemokines--chemotactic cytokines that mediate inflammation. N. Engl. J. Med. 1998, 338, 436–445. [Google Scholar] [CrossRef]
  11. Hughes, C.E.; Nibbs, R.J.B. A guide to chemokines and their receptors. FEBS J. 2018, 285, 2944–2971. [Google Scholar] [CrossRef]
  12. Lei, W.; Jia, L.; Wang, Z.; Liang, Z.; Zhao, A.; Liu, Y.; Tian, Y.; Zhao, L.; Chen, Y.; Shi, G.; et al. CC chemokines family in fibrosis and aging: From mechanisms to therapy. Ageing Res. Rev. 2023, 87, 101900. [Google Scholar] [CrossRef]
  13. Bakogiannis, C.; Sachse, M.; Stamatelopoulos, K.; Stellos, K. Platelet-derived chemokines in inflammation and atherosclerosis. Cytokine 2019, 122, 154157. [Google Scholar] [CrossRef]
  14. Chen, C.; Chu, S.F.; Liu, D.D.; Zhang, Z.; Kong, L.L.; Zhou, X.; Chen, N.H. Chemokines play complex roles in cerebral ischemia. Neurochem. Int. 2018, 112, 146–158. [Google Scholar] [CrossRef]
  15. Markl, F.; Huynh, D.; Endres, S.; Kobold, S. Utilizing chemokines in cancer immunotherapy. Trends Cancer 2022, 8, 670–682. [Google Scholar] [CrossRef] [PubMed]
  16. Rakotoarivelo, V.; Variya, B.; Langlois, M.F.; Ramanathan, S. Chemokines in human obesity. Cytokine 2020, 127, 154953. [Google Scholar] [CrossRef] [PubMed]
  17. Griffith, J.W.; Sokol, C.L.; Luster, A.D. Chemokines and chemokine receptors: Positioning cells for host defense and immunity. Annu. Rev. Immunol. 2014, 32, 659–702. [Google Scholar] [CrossRef]
  18. Zhang, Y.; Liu, D.; Vithran, D.T.A.; Kwabena, B.R.; Xiao, W.; Li, Y. CC chemokines and receptors in osteoarthritis: New insights and potential targets. Arthritis Res. Ther. 2023, 25, 113. [Google Scholar] [CrossRef]
  19. Birney, E. Mendelian Randomization. Cold Spring Harb. Perspect. Med. 2022, 12, a041302. [Google Scholar] [CrossRef]
  20. Sekula, P.; Del Greco, M.F.; Pattaro, C.; Kottgen, A. Mendelian Randomization as an Approach to Assess Causality Using Observational Data. J. Am. Soc. Nephrol. 2016, 27, 3253–3265. [Google Scholar] [CrossRef] [PubMed]
  21. Pang, W.W.; Cai, Y.S.; Cao, C.; Zhang, F.R.; Zeng, Q.; Liu, D.Y.; Wang, N.; Qu, X.C.; Chen, X.D.; Deng, H.W.; et al. Mendelian randomization and transcriptome analysis identified immune-related biomarkers for osteoarthritis. Front. Immunol. 2024, 15, 1334479. [Google Scholar] [CrossRef]
  22. Hanzelmann, S.; Castelo, R.; Guinney, J. GSVA: Gene set variation analysis for microarray and RNA-seq data. BMC Bioinform. 2013, 14, 7. [Google Scholar] [CrossRef]
  23. Langfelder, P.; Horvath, S. WGCNA: An R package for weighted correlation network analysis. BMC Bioinform. 2008, 9, 559. [Google Scholar] [CrossRef]
  24. Zhang, X.Q.; Chao, P.; Zhang, L.; Xu, L.; Cui, X.Y.; Wang, S.S.; Wusiman, M.; Jiang, H.; Lu, C. Single-cell RNA and transcriptome sequencing profiles identify immune-associated key genes in the development of diabetic kidney disease. Front. Immunol. 2023, 14, 1030198. [Google Scholar] [CrossRef]
  25. Ritchie, M.E.; Phipson, B.; Wu, D.; Hu, Y.; Law, C.W.; Shi, W.; Smyth, G.K. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 2015, 43, e47. [Google Scholar] [CrossRef]
  26. Hemani, G.; Zhengn, J.; Elsworth, B.; Wade, K.H.; Haberland, V.; Baird, D.; Laurin, C.; Burgess, S.; Bowden, J.; Langdon, R.; et al. The MR-Base platform supports systematic causal inference across the human phenome. Elife 2018, 7, e34408. [Google Scholar] [CrossRef] [PubMed]
  27. Yu, G.; Wang, L.G.; Han, Y.; He, Q.Y. clusterProfiler: An R package for comparing biological themes among gene clusters. OMICS 2012, 16, 284–287. [Google Scholar] [CrossRef] [PubMed]
  28. Liu, P.; Xu, H.C.; Shi, Y.C.; Deng, L.; Chen, X.Y. Potential Molecular Mechanisms of Plantain in the Treatment of Gout and Hyperuricemia Based on Network Pharmacology. Evid.-Based Complement. Altern. Med. 2020, 2020, 3023127. [Google Scholar] [CrossRef] [PubMed]
  29. Li, Y.; Lu, F.G.; Yin, Y.N. Applying logistic LASSO regression for the diagnosis of atypical Crohn’s disease. Sci. Rep. 2022, 12, 113400. [Google Scholar] [CrossRef]
  30. Zhang, Z.H.; Zhao, Y.M.; Canes, A.; Steinberg, D.; Lyashevska, O.; Collab, A.B.-D.C.T. Predictive analytics with gradient boosting in clinical medicine. Ann. Transl. Med. 2019, 7, 152. [Google Scholar] [CrossRef]
  31. Zheng, Y.; Gao, W.Q.; Zhang, Q.; Cheng, X.; Liu, Y.W.; Qi, Z.C.; Li, T. Ferroptosis and Autophagy-Related Genes in the Pathogenesis of Ischemic Cardiomyopathy. Front. Cardiovasc. Med. 2022, 9, 906753. [Google Scholar] [CrossRef]
  32. Sachs, M.C. plotROC: A Tool for Plotting ROC Curves. J. Stat. Softw. 2017, 79, 2231–2264. [Google Scholar] [CrossRef]
  33. Orifjon, S.; Jammatov, J.; Sousa, C.; Barros, R.; Vasconcelos, O.; Rodrigues, P. Translation and Adaptation of the Adult Developmental Coordination Disorder/Dyspraxia Checklist (ADC) into Asian Uzbekistan. Sports 2023, 11, 135. [Google Scholar] [CrossRef] [PubMed]
  34. Bowden, J.; Davey Smith, G.; Burgess, S. Mendelian randomization with invalid instruments: Effect estimation and bias detection through Egger regression. Int. J. Epidemiol. 2015, 44, 512–525. [Google Scholar] [CrossRef]
  35. Bowden, J.; Davey Smith, G.; Haycock, P.C.; Burgess, S. Consistent Estimation in Mendelian Randomization with Some Invalid Instruments Using a Weighted Median Estimator. Genet. Epidemiol. 2016, 40, 304–314. [Google Scholar] [CrossRef] [PubMed]
  36. Burgess, S.; Scott, R.A.; Timpson, N.J.; Davey Smith, G.; Thompson, S.G.; Consortium, E.-I. Using published data in Mendelian randomization: A blueprint for efficient identification of causal risk factors. Eur. J. Epidemiol. 2015, 30, 543–552. [Google Scholar] [CrossRef]
  37. Hartwig, F.P.; Davey Smith, G.; Bowden, J. Robust inference in summary data Mendelian randomization via the zero modal pleiotropy assumption. Int. J. Epidemiol. 2017, 46, 1985–1998. [Google Scholar] [CrossRef] [PubMed]
  38. Goldring, M.B.; Otero, M. Inflammation in osteoarthritis. Curr. Opin. Rheumatol. 2011, 23, 471–478. [Google Scholar] [CrossRef]
  39. Liu-Bryan, R.; Terkeltaub, R. Emerging regulators of the inflammatory process in osteoarthritis. Nat. Rev. Rheumatol. 2015, 11, 35–44. [Google Scholar] [CrossRef]
  40. Raghu, H.; Lepus, C.M.; Wang, Q.; Wong, H.H.; Lingampalli, N.; Oliviero, F.; Punzi, L.; Giori, N.J.; Goodman, S.B.; Chu, C.R.; et al. CCL2/CCR2, but not CCL5/CCR5, mediates monocyte recruitment, inflammation and cartilage destruction in osteoarthritis. Ann. Rheum. Dis. 2017, 76, 914–922. [Google Scholar] [CrossRef]
  41. Zhao, X.; Gu, M.; Xu, X.; Wen, X.; Yang, G.; Li, L.; Sheng, P.; Meng, F. CCL3/CCR1 mediates CD14(+)CD16(-) circulating monocyte recruitment in knee osteoarthritis progression. Osteoarthr. Cartil. 2020, 28, 613–625. [Google Scholar] [CrossRef]
  42. Ishihara, S.; Obeidat, A.M.; Wokosin, D.L.; Ren, D.; Miller, R.J.; Malfait, A.M.; Miller, R.E. The role of intra-articular neuronal CCR2 receptors in knee joint pain associated with experimental osteoarthritis in mice. Arthritis Res. Ther. 2021, 23, 103. [Google Scholar] [CrossRef]
  43. Valdrighi, N.; Vago, J.P.; Blom, A.B.; van de Loo, F.A.J.; Blaney Davidson, E.N. Innate Immunity at the Core of Sex Differences in Osteoarthritic Pain? Front. Pharmacol. 2022, 13, 881500. [Google Scholar] [CrossRef]
  44. Sterken, B.A.; Ackermann, T.; Muller, C.; Zuidhof, H.R.; Kortman, G.; Hernandez-Segura, A.; Broekhuis, M.; Spierings, D.; Guryev, V.; Calkhoven, C.F. C/EBPbeta isoform-specific regulation of migration and invasion in triple-negative breast cancer cells. NPJ Breast Cancer 2022, 8, 11. [Google Scholar] [CrossRef]
  45. Xia, Y.; Wang, Z.H.; Zhang, J.; Liu, X.; Yu, S.P.; Ye, K.X.; Wang, J.Z.; Ye, K.; Wang, X.C. C/EBPbeta is a key transcription factor for APOE and preferentially mediates ApoE4 expression in Alzheimer’s disease. Mol. Psychiatry 2021, 26, 6002–6022. [Google Scholar] [CrossRef] [PubMed]
  46. Cooper-Knock, J.; Green, C.; Altschuler, G.; Wei, W.; Bury, J.J.; Heath, P.R.; Wyles, M.; Gelsthorpe, C.; Highley, J.R.; Lorente-Pons, A.; et al. A data-driven approach links microglia to pathology and prognosis in amyotrophic lateral sclerosis. Acta Neuropathol. Commun. 2017, 5, 23. [Google Scholar] [CrossRef] [PubMed]
  47. Sun, H.; Li, M.; Ji, Y.; Zhu, J.; Chen, Z.; Zhang, L.; Deng, C.; Cheng, Q.; Wang, W.; Shen, Y.; et al. Identification of Regulatory Factors and Prognostic Markers in Amyotrophic Lateral Sclerosis. Antioxidants 2022, 11, 303. [Google Scholar] [CrossRef]
  48. Hu, P.; Li, B.; Yin, Z.; Peng, P.; Cao, J.; Xie, W.; Liu, L.; Cao, F.; Zhang, B. Multi-omics characterization of macrophage polarization-related features in osteoarthritis based on a machine learning computational framework. Heliyon 2024, 10, e30335. [Google Scholar] [CrossRef]
  49. Huang, J.; Wang, L.; Zhou, J.; Dai, T.; Zhu, W.; Wang, T.; Wang, H.; Zhang, Y. Unveiling the ageing-related genes in diagnosing osteoarthritis with metabolic syndrome by integrated bioinformatics analysis and machine learning. Artif. Cells Nanomed. Biotechnol. 2025, 53, 57–68. [Google Scholar] [CrossRef]
  50. Wang, Z.; Shi, W.; Wu, L.; Xiao, Y.; Wang, M.; Zhang, S.; Chen, Z.; Yin, G.; Xie, X.; Bi, S.; et al. TMF inhibits extracellular matrix degradation by regulating the C/EBPbeta/ADAMTS5 signaling pathway in osteoarthritis. Biomed. Pharmacother. 2024, 174, 116501. [Google Scholar] [CrossRef] [PubMed]
  51. Ren, Q.; Liu, Z.; Wu, L.; Yin, G.; Xie, X.; Kong, W.; Zhou, J.; Liu, S. C/EBPβ: The structure, regulation, and its roles in inflammation-related diseases. Biomed. Pharmacother. 2023, 169, 115938. [Google Scholar] [CrossRef]
  52. Ferretti, E.; Pistoia, V.; Corcione, A. Role of fractalkine/CX3CL1 and its receptor in the pathogenesis of inflammatory and malignant diseases with emphasis on B cell malignancies. Mediat. Inflamm. 2014, 2014, 480941. [Google Scholar] [CrossRef]
  53. Huo, L.W.; Ye, Y.L.; Wang, G.W.; Ye, Y.G. Fractalkine (CX3CL1): A biomarker reflecting symptomatic severity in patients with knee osteoarthritis. J. Investig. Med. 2015, 63, 626–631. [Google Scholar] [CrossRef]
  54. Culemann, S.; Gruneboom, A.; Nicolas-Avila, J.A.; Weidner, D.; Lammle, K.F.; Rothe, T.; Quintana, J.A.; Kirchner, P.; Krljanac, B.; Eberhardt, M.; et al. Locally renewing resident synovial macrophages provide a protective barrier for the joint. Nature 2019, 572, 670–675. [Google Scholar] [CrossRef]
  55. Aharon-Yariv, A.; Wang, Y.; Ahmed, A.; Delgado-Olguin, P. Integrated small RNA, mRNA and protein omics reveal a miRNA network orchestrating metabolic maturation of the developing human heart. BMC Genom. 2023, 24, 709. [Google Scholar] [CrossRef]
  56. Song, C.; Wei, S.; Fan, Y.; Jiang, S. Bioinformatic-based Identification of Genes Associated with Aortic Valve Stenosis. Heart Surg. Forum 2022, 25, E069–E078. [Google Scholar] [CrossRef]
  57. Meng, W.; Tai, Y.; Zhao, H.; Fu, B.; Zhang, T.; Liu, W.; Li, H.; Yang, Y.; Zhang, Q.; Feng, Y.; et al. Downregulation of miR-33a-5p in Hepatocellular Carcinoma: A Possible Mechanism for Chemotherapy Resistance. Med. Sci. Monit. 2017, 23, 1295–1304. [Google Scholar] [CrossRef] [PubMed]
  58. De Silva, K.; Demmer, R.T.; Jonsson, D.; Mousa, A.; Forbes, A.; Enticott, J. A data-driven biocomputing pipeline with meta-analysis on high throughput transcriptomics to identify genome-wide miRNA markers associated with type 2 diabetes. Heliyon 2022, 8, e08886. [Google Scholar] [CrossRef] [PubMed]
  59. Wang, S.; Cao, N. Uncovering potential differentially expressed miRNAs and targeted mRNAs in myocardial infarction based on integrating analysis. Mol. Med. Rep. 2020, 22, 4383–4395. [Google Scholar] [CrossRef] [PubMed]
  60. Harada, M.; Okuzaki, D.; Yamauchi, A.; Ishikawa, S.; Nomura, Y.; Nishimura, A.; Motoike, Y.; Koshikawa, M.; Hitachi, K.; Tsuchida, K.; et al. Circulating miR-20b-5p and miR-330-3p are novel biomarkers for progression of atrial fibrillation: Intracardiac/extracardiac plasma sample analysis by small RNA sequencing. PLoS ONE 2023, 18, e0283942. [Google Scholar] [CrossRef]
  61. Zhang, L.; Dong, X.; Zhan, Y.; Ma, S.; Liu, C.; Gao, Y. Expression profile of microRNAs in patients with decompensated cirrhosis by small RNA deep sequencing. Clin. Biochem. 2024, 123, 110705. [Google Scholar] [CrossRef]
  62. Miao, R.; Dong, X.; Gong, J.; Wang, Y.; Guo, X.; Li, Y.; Liu, M.; Wan, J.; Li, J.; Yang, S.; et al. hsa-miR-106b-5p participates in the development of chronic thromboembolic pulmonary hypertension via targeting matrix metalloproteinase 2. Pulm. Circ. 2020, 10, 2045894020928300. [Google Scholar] [CrossRef]
  63. Wu, X.; Wang, Y.; Chen, C.; Xue, Y.; Zheng, S.; Cai, L. Hsa-let-7c-5p, hsa-miR-130b-3p, and hsa-miR-142-3p as Novel miRNA Biomarkers for Melanoma Progression. Genet. Res. 2022, 2022, 5671562. [Google Scholar] [CrossRef] [PubMed]
  64. Chen, K.; Fang, H.; Xu, N. LncRNA LOXL1-AS1 is transcriptionally activated by JUND and contributes to osteoarthritis progression via targeting the miR-423-5p/KDM5C axis. Life Sci. 2020, 258, 118095. [Google Scholar] [CrossRef]
  65. Zhou, J.; Huang, J.; Li, Z.; Song, Q.; Yang, Z.; Wang, L.; Meng, Q. Identification of aging-related biomarkers and immune infiltration characteristics in osteoarthritis based on bioinformatics analysis and machine learning. Front. Immunol. 2023, 14, 1168780. [Google Scholar] [CrossRef] [PubMed]
  66. He, X.; Deng, L.; Zou, K.; Tian, Y.; Tang, X. ATF3 Modulates the Proliferation, Migration, and Apoptosis of Synovial Fibroblasts after Arthroscopy by Promoting RGS1 Transcription. Curr. Mol. Med. 2023, 23, 981–990. [Google Scholar] [CrossRef] [PubMed]
  67. Li, S.; Wang, X.; Li, J.; Wei, W.; Liu, J. Identification and Molecular Mechanisms Study of Genes Associated with Osteoarthritis: A Comprehensive Bioinformatic Study of Cartilage and Synovium. Crit. Rev. Eukaryot. Gene Expr. 2022, 32, 25–38. [Google Scholar] [CrossRef]
  68. Xia, L.; Gong, N. Identification and verification of ferroptosis-related genes in the synovial tissue of osteoarthritis using bioinformatics analysis. Front. Mol. Biosci. 2022, 9, 992044. [Google Scholar] [CrossRef]
  69. Yang, J.; Fan, Y.; Liu, S. ATF3 as a potential diagnostic marker of early-stage osteoarthritis and its correlation with immune infiltration through bioinformatics analysis. Bone Jt. Res. 2022, 11, 679–689. [Google Scholar] [CrossRef]
  70. Zheng, H.; Fang, J.; Lu, W.; Liu, Y.; Chen, S.; Huang, G.; Zou, Y.; Hu, S.; Zheng, Y.; Fang, H.; et al. TCF12 regulates the TGF-beta/Smad2/3 signaling pathway to accelerate the progression of osteoarthritis by targeting CXCR4. J. Orthop. Transl. 2024, 44, 35–46. [Google Scholar] [CrossRef]
  71. Poursamimi, J.; Shariati-Sarabi, Z.; Tavakkol-Afshari, J.; Mohammadi, M. A Significant Increase in the Gene Expression of GATA-3 Following the Treatment of Osteoarthritis Patients with Crocin. Iran. J. Allergy Asthma Immunol. 2022, 21, 35–43. [Google Scholar] [CrossRef]
  72. Singh, P.N.P.; Yadav, U.S.; Azad, K.; Goswami, P.; Kinare, V.; Bandyopadhyay, A. NFIA and GATA3 are crucial regulators of embryonic articular cartilage differentiation. Development 2018, 145, dev156554. [Google Scholar] [CrossRef]
  73. Li, X.; Xiao, S.; Li, F.; Fang, K.; Wen, J.; Gong, H. Max interacting protein 1 induces IL-17-producing T helper/regulatory T imbalance in osteoarthritis by upregulating tectonic family member 2. Tissue Cell 2022, 78, 101906. [Google Scholar] [CrossRef]
  74. Ragni, E.; Piccolo, S.; Papait, A.; De Luca, P.; Taiana, M.; Grieco, G.; Silini, A.R.; Parolini, O.; de Girolamo, L. Stable Housekeeping Genes in Bone Marrow, Adipose Tissue, and Amniotic Membrane-Derived Mesenchymal Stromal Cells for Orthopedic Regenerative Medicine Approaches. Int. J. Mol. Sci. 2024, 25, 1461. [Google Scholar] [CrossRef]
  75. Amarnani, A.A.; Poladian, K.R.; Marciano, B.E.; Daub, J.R.; Williams, S.G.; Livinski, A.A.; Hsu, A.P.; Palmer, C.L.; Kenney, C.M.; Avila, D.N.; et al. A Panoply of Rheumatological Manifestations in Patients with GATA2 Deficiency. Sci. Rep. 2020, 10, 8305. [Google Scholar] [CrossRef] [PubMed]
  76. Liu, J.; Yuan, S.; Niu, X.; Kelleher, R.; Sheridan, H. ESR1 dysfunction triggers neuroinflammation as a critical upstream causative factor of the Alzheimer’s disease process. Aging 2022, 14, 8595–8614. [Google Scholar] [CrossRef] [PubMed]
  77. Zhu, M.M.; Li, H.L.; Shi, L.H.; Chen, X.P.; Luo, J.; Zhang, Z.L. The pharmacogenomics of valproic acid. J. Hum. Genet. 2017, 62, 1009–1014. [Google Scholar] [CrossRef] [PubMed]
  78. Han, Y.; Wang, X.; He, X.; Jia, M.; Pan, H.; Chen, J. Excited State Kinetics of Benzo[a]pyrene Is Affected by Oxygen and DNA. Molecules 2023, 28, 5269. [Google Scholar] [CrossRef]
Bioengineering 12 00930 g001
Figure 1. WGCNA and differential expression analysis in the training set. (A) Sample clustering tree. (B) Identification of the soft threshold (the horizontal axis in both cases represented the weighting parameter, the power value. In the left—hand figure, the vertical axis represented the square of the fitting coefficient between log(k) and log(p(k)) in the corresponding network, namely signedR2. The higher the square of the correlation coefficient, the closer the network approximates a scale-free distribution. In the right-hand figure, the vertical axis represented the mean value of the adjacency functions of all genes in the corresponding gene module). (C) Cluster dendrogram (different colors represented different modules. By default, genes that cannot be classified into any module were colored in gray). (D) Relationships between modules and traits (CCL and CCR) (the vertical axis represented different modules, and the horizontal axis represented clinical traits. Each square denoted the correlation coefficient between a certain module and a certain trait. Blue indicated negative correlation, while red indicated positive correlation). (E) Volcano plot of DEGs (the horizontal axis represented the fold change in gene expression, and the vertical axis represented the adj.p.value). (F) Heatmap of DEGs (the top graph showed the density distribution of the expression levels of differentially expressed genes. The bottom graph showed the heatmap of the expression levels of differentially expressed genes).
Figure 1. WGCNA and differential expression analysis in the training set. (A) Sample clustering tree. (B) Identification of the soft threshold (the horizontal axis in both cases represented the weighting parameter, the power value. In the left—hand figure, the vertical axis represented the square of the fitting coefficient between log(k) and log(p(k)) in the corresponding network, namely signedR2. The higher the square of the correlation coefficient, the closer the network approximates a scale-free distribution. In the right-hand figure, the vertical axis represented the mean value of the adjacency functions of all genes in the corresponding gene module). (C) Cluster dendrogram (different colors represented different modules. By default, genes that cannot be classified into any module were colored in gray). (D) Relationships between modules and traits (CCL and CCR) (the vertical axis represented different modules, and the horizontal axis represented clinical traits. Each square denoted the correlation coefficient between a certain module and a certain trait. Blue indicated negative correlation, while red indicated positive correlation). (E) Volcano plot of DEGs (the horizontal axis represented the fold change in gene expression, and the vertical axis represented the adj.p.value). (F) Heatmap of DEGs (the top graph showed the density distribution of the expression levels of differentially expressed genes. The bottom graph showed the heatmap of the expression levels of differentially expressed genes).
Bioengineering 12 00930 g001
Bioengineering 12 00930 g002
Figure 2. Functional analysis of 82 genes and identification of candidate genes. (A) GO enrichment results for 82 genes. (B) KEGG enrichment results for 82 genes. (C) PPI results of 82 genes. (D) Venn diagram of the NCC and DNMC algorithms.
Figure 2. Functional analysis of 82 genes and identification of candidate genes. (A) GO enrichment results for 82 genes. (B) KEGG enrichment results for 82 genes. (C) PPI results of 82 genes. (D) Venn diagram of the NCC and DNMC algorithms.
Bioengineering 12 00930 g002
Bioengineering 12 00930 g003
Figure 3. Screening for potential biomarkers. (A) Results of LASSO regression analysis for 14 candidate genes (in the left graph, the horizontal axis represents the log(lambda) value, and the vertical axis represents the degree of freedom. In the right graph, the horizontal axis represents the log(lambda), and the vertical axis represents the coefficient of the gene). (B) Results of SVM-RFE analysis for 14 candidate genes (the vertical axis is labeled as “10xCV Error”, which represents the ten-fold cross-validation error that was used to evaluate the generalization ability of the model. The curve shows the fluctuation of the ten-fold cross-validation error as the variable on the horizontal axis changed) and Venn diagram of two machine learning algorithms. (C,D) ROC curves analysis (the vertical axis represents sensitivity, and the horizontal axis represents specificity). (E) Results of gene expression analyses in the training and testing sets (the horizontal axis represents potential biomarkers, and the vertical axis represents the expression level).
Figure 3. Screening for potential biomarkers. (A) Results of LASSO regression analysis for 14 candidate genes (in the left graph, the horizontal axis represents the log(lambda) value, and the vertical axis represents the degree of freedom. In the right graph, the horizontal axis represents the log(lambda), and the vertical axis represents the coefficient of the gene). (B) Results of SVM-RFE analysis for 14 candidate genes (the vertical axis is labeled as “10xCV Error”, which represents the ten-fold cross-validation error that was used to evaluate the generalization ability of the model. The curve shows the fluctuation of the ten-fold cross-validation error as the variable on the horizontal axis changed) and Venn diagram of two machine learning algorithms. (C,D) ROC curves analysis (the vertical axis represents sensitivity, and the horizontal axis represents specificity). (E) Results of gene expression analyses in the training and testing sets (the horizontal axis represents potential biomarkers, and the vertical axis represents the expression level).
Bioengineering 12 00930 g003
Bioengineering 12 00930 g004
Figure 4. Construction and validation of the nomogram and functional analysis of potential biomarkers. (A) Nomogram of potential biomarkers. (B) Calibration curve of the nomogram. (C) GSEA results for potential biomarkers (CEBPB and CX3CR1).
Figure 4. Construction and validation of the nomogram and functional analysis of potential biomarkers. (A) Nomogram of potential biomarkers. (B) Calibration curve of the nomogram. (C) GSEA results for potential biomarkers (CEBPB and CX3CR1).
Bioengineering 12 00930 g004
Bioengineering 12 00930 g005
Figure 5. MR results. (A) Scatter plots of SNPs associated with two potential biomarkers and OA (the lines in the figure represented five algorithms). (B) Forest plot results for the two potential biomarkers. (C) Funnel plot results for the two potential biomarkers (the horizontal axis is βiv, which represents the effect estimate calculated by the inverse-variance weighted method. The vertical axis was 1/SEiv, where SEiv represented the standard error. The 1/SEiv could reflect the precision of the effect estimate, and a larger value indicates higher precision). (D) Leave-one-out results for two potential biomarkers (the horizontal axis represents the estimated effect, and the vertical axis represents different single-nucleotide polymorphism (SNP) loci).
Figure 5. MR results. (A) Scatter plots of SNPs associated with two potential biomarkers and OA (the lines in the figure represented five algorithms). (B) Forest plot results for the two potential biomarkers. (C) Funnel plot results for the two potential biomarkers (the horizontal axis is βiv, which represents the effect estimate calculated by the inverse-variance weighted method. The vertical axis was 1/SEiv, where SEiv represented the standard error. The 1/SEiv could reflect the precision of the effect estimate, and a larger value indicates higher precision). (D) Leave-one-out results for two potential biomarkers (the horizontal axis represents the estimated effect, and the vertical axis represents different single-nucleotide polymorphism (SNP) loci).
Bioengineering 12 00930 g005
Bioengineering 12 00930 g006
Figure 6. TF, miRNA, and drug prediction. (A): CEBPB and CX3CR1 target transcription factors (red color represents potential biomarkers, and blue color represents TFs), (B): CEBPB and CX3CR1 target miRNAs (red color represents potential biomarkers, and blue color represents miRNAs), (C): visualization of miRNA-biomarker-TF networks (red color represents potential biomarkers, blue color represents miRNAs, and green color represents TFs), (D): CTD constructing biomarker–drug networks (red color represents potential biomarkers, and blue color represents drugs).
Figure 6. TF, miRNA, and drug prediction. (A): CEBPB and CX3CR1 target transcription factors (red color represents potential biomarkers, and blue color represents TFs), (B): CEBPB and CX3CR1 target miRNAs (red color represents potential biomarkers, and blue color represents miRNAs), (C): visualization of miRNA-biomarker-TF networks (red color represents potential biomarkers, blue color represents miRNAs, and green color represents TFs), (D): CTD constructing biomarker–drug networks (red color represents potential biomarkers, and blue color represents drugs).
Bioengineering 12 00930 g006
Table 1. Mendelian randomization analysis of DE-CRGs.
Table 1. Mendelian randomization analysis of DE-CRGs.
Geneid.Exposureid.OutcomeOutcomeExposureMethodnsnpbsepvalp_noLevel Test
SKAP2eqtl-a-ENSG00000005020ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000005020 || id:eqtl-a-ENSG00000005020Inverse-variance weighted (multiplicative random effects)12−0.063550.0290.028MR Egger weighted median0.964302
ANK1eqtl-a-ENSG00000029534ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000029534 || id:eqtl-a-ENSG00000029534Inverse-variance weighted (multiplicative random effects)4−0.288750.064930MR Egger0.604415
TFB1Meqtl-a-ENSG00000029639ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000029639 || id:eqtl-a-ENSG00000029639Inverse-variance weighted (multiplicative random effects)3−0.221940.0519880MR Egger0.787497
HSPA5eqtl-a-ENSG00000044574ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000044574 || id:eqtl-a-ENSG00000044574Inverse-variance weighted (multiplicative random effects)50.1569070.045020MR Egger weighted median0.953288
ASB1eqtl-a-ENSG00000065802ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000065802 || id:eqtl-a-ENSG00000065802Inverse-variance weighted (multiplicative random effects)30.0955620.0481240.047MR Egger weighted median0.476947
SEMA3Aeqtl-a-ENSG00000075213ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000075213 || id:eqtl-a-ENSG00000075213Inverse-variance weighted (multiplicative random effects)3−0.277090.0171050MR Egger0.986775
SP140eqtl-a-ENSG00000079263ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000079263 || id:eqtl-a-ENSG00000079263Inverse-variance weighted (multiplicative random effects)3−0.171960.0865010.047MR Egger0.840832
EPB41L2eqtl-a-ENSG00000079819ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000079819 || id:eqtl-a-ENSG00000079819Inverse-variance weighted (multiplicative random effects)30.1219460.0464390.009MR Egger weighted median0.62119
MEF2Ceqtl-a-ENSG00000081189ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000081189 || id:eqtl-a-ENSG00000081189Inverse-variance weighted (multiplicative random effects)5−0.136350.0691890.049MR Egger weighted median0.38673
OVGP1eqtl-a-ENSG00000085465ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000085465 || id:eqtl-a-ENSG00000085465Inverse-variance weighted (multiplicative random effects)3−0.098360.038390.01MR Egger weighted median0.725835
NRP1eqtl-a-ENSG00000099250ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000099250 || id:eqtl-a-ENSG00000099250Inverse-variance weighted (multiplicative random effects)6−0.129290.0378490.001MR Egger weighted median0.866044
CYTH4eqtl-a-ENSG00000100055ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000100055 || id:eqtl-a-ENSG00000100055Inverse-variance weighted (multiplicative random effects)5−0.170780.0796440.032MR Egger weighted median0.716136
SYNGR1eqtl-a-ENSG00000100321ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000100321 || id:eqtl-a-ENSG00000100321Inverse-variance weighted (multiplicative random effects)5−0.062840.0291770.031MR Egger weighted median0.531044
KIAA0930eqtl-a-ENSG00000100364ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000100364 || id:eqtl-a-ENSG00000100364Inverse-variance weighted (multiplicative random effects)4−0.145080.0270170MR Egger weighted median0.75393
HCKeqtl-a-ENSG00000101336ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000101336 || id:eqtl-a-ENSG00000101336Inverse-variance weighted (multiplicative random effects)3−0.072360.0139910MR Egger weighted median0.902966
FNDC3Aeqtl-a-ENSG00000102531ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000102531 || id:eqtl-a-ENSG00000102531Inverse-variance weighted (multiplicative random effects)4−0.096550.0295650.001MR Egger weighted median0.745019
NOMO3eqtl-a-ENSG00000103226ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000103226 || id:eqtl-a-ENSG00000103226Inverse-variance weighted (multiplicative random effects)3−0.112210.0290730MR Egger weighted median0.709799
RIPK2eqtl-a-ENSG00000104312ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000104312 || id:eqtl-a-ENSG00000104312Inverse-variance weighted (multiplicative random effects)3−0.143120.0349840MR Egger weighted median0.825168
CD37eqtl-a-ENSG00000104894ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000104894 || id:eqtl-a-ENSG00000104894Inverse-variance weighted (multiplicative random effects)4−0.125210.0487950.01MR Egger weighted median0.440039
NKG7eqtl-a-ENSG00000105374ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000105374 || id:eqtl-a-ENSG00000105374Inverse-variance weighted (multiplicative random effects)30.1791320.0579990.002MR Egger weighted median0.874258
ZKSCAN1eqtl-a-ENSG00000106261ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000106261 || id:eqtl-a-ENSG00000106261Inverse-variance weighted (multiplicative random effects)3−0.045860.0096490MR Egger weighted median0.916197
TNFSF8eqtl-a-ENSG00000106952ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000106952 || id:eqtl-a-ENSG00000106952Inverse-variance weighted (multiplicative random effects)3−0.14120.0375910MR Egger weighted median0.70572
PTGDSeqtl-a-ENSG00000107317ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000107317 || id:eqtl-a-ENSG00000107317Inverse-variance weighted (multiplicative random effects)30.2340730.0598620MR Egger0.564796
TFAMeqtl-a-ENSG00000108064ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000108064 || id:eqtl-a-ENSG00000108064Inverse-variance weighted (multiplicative random effects)70.0667430.0299340.026MR Egger weighted median0.473236
TMEM97eqtl-a-ENSG00000109084ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000109084 || id:eqtl-a-ENSG00000109084Inverse-variance weighted (multiplicative random effects)6−0.068770.0268080.01MR Egger weighted median0.829951
PPARGC1Aeqtl-a-ENSG00000109819ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000109819 || id:eqtl-a-ENSG00000109819Inverse-variance weighted (multiplicative random effects)30.0422770.0064390MR Egger weighted median0.929293
PANX1eqtl-a-ENSG00000110218ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000110218 || id:eqtl-a-ENSG00000110218Inverse-variance weighted (multiplicative random effects)5−0.080890.0358940.024MR Egger weighted median0.501225
SLC22A18eqtl-a-ENSG00000110628ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000110628 || id:eqtl-a-ENSG00000110628Inverse-variance weighted (multiplicative random effects)30.0405910.0202050.045MR Egger weighted median0.787391
CBLBeqtl-a-ENSG00000114423ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000114423 || id:eqtl-a-ENSG00000114423Inverse-variance weighted (multiplicative random effects)40.0629250.0287860.029MR Egger weighted median0.754687
TP53I3eqtl-a-ENSG00000115129ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000115129 || id:eqtl-a-ENSG00000115129Inverse-variance weighted (multiplicative random effects)40.0527810.016870.002MR Egger weighted median0.79608
IL1R1eqtl-a-ENSG00000115594ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000115594 || id:eqtl-a-ENSG00000115594Inverse-variance weighted (multiplicative random effects)3−0.135320.0265440MR Egger weighted median0.906106
SCP2eqtl-a-ENSG00000116171ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000116171 || id:eqtl-a-ENSG00000116171Inverse-variance weighted (multiplicative random effects)50.0840150.0362150.02MR Egger weighted median0.53642
ITGB1BP1eqtl-a-ENSG00000119185ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000119185 || id:eqtl-a-ENSG00000119185Inverse-variance weighted (multiplicative random effects)4−0.124990.0618580.043MR Egger weighted median0.697252
ADCY7eqtl-a-ENSG00000121281ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000121281 || id:eqtl-a-ENSG00000121281Inverse-variance weighted (multiplicative random effects)30.0804490.009580MR Egger weighted median0.90424
NQO2eqtl-a-ENSG00000124588ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000124588 || id:eqtl-a-ENSG00000124588Inverse-variance weighted (multiplicative random effects)60.1031650.0362880.004MR Egger0.421145
ATXN1eqtl-a-ENSG00000124788ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000124788 || id:eqtl-a-ENSG00000124788Inverse-variance weighted (multiplicative random effects)30.0837810.0380750.028MR Egger weighted median0.697622
DOCK4eqtl-a-ENSG00000128512ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000128512 || id:eqtl-a-ENSG00000128512Inverse-variance weighted (multiplicative random effects)5−0.378260.1478650.011MR Egger weighted median0.638297
KLHDC10eqtl-a-ENSG00000128607ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000128607 || id:eqtl-a-ENSG00000128607Inverse-variance weighted (multiplicative random effects)3−0.072080.0354920.042MR Egger weighted median0.607087
CALML4eqtl-a-ENSG00000129007ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000129007 || id:eqtl-a-ENSG00000129007Inverse-variance weighted (multiplicative random effects)4−0.027550.0106730.01MR Egger weighted median0.837662
CD68eqtl-a-ENSG00000129226ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000129226 || id:eqtl-a-ENSG00000129226Inverse-variance weighted (multiplicative random effects)3−0.261710.0152440MR Egger0.893517
CDO1eqtl-a-ENSG00000129596ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000129596 || id:eqtl-a-ENSG00000129596Inverse-variance weighted (multiplicative random effects)50.0726090.0065020MR Egger weighted median0.927536
LDLReqtl-a-ENSG00000130164ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000130164 || id:eqtl-a-ENSG00000130164Inverse-variance weighted (multiplicative random effects)80.200670.045220MR Egger weighted median0.682704
GCH1eqtl-a-ENSG00000131979ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000131979 || id:eqtl-a-ENSG00000131979Inverse-variance weighted (multiplicative random effects)3−0.102840.0324840.002MR Egger weighted median0.68823
EGLN1eqtl-a-ENSG00000135766ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000135766 || id:eqtl-a-ENSG00000135766Inverse-variance weighted (multiplicative random effects)3−0.16140.0656310.014MR Egger weighted median0.562518
PLXNC1eqtl-a-ENSG00000136040ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000136040 || id:eqtl-a-ENSG00000136040Inverse-variance weighted (multiplicative random effects)3−0.178680.0726460.014MR Egger weighted median0.727798
C7orf25eqtl-a-ENSG00000136197ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000136197 || id:eqtl-a-ENSG00000136197Inverse-variance weighted (multiplicative random effects)8−0.11080.0447460.013MR Egger weighted median0.44153
IFI44eqtl-a-ENSG00000137965ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000137965 || id:eqtl-a-ENSG00000137965Inverse-variance weighted (multiplicative random effects)70.1794050.0502540MR Egger weighted median0.827067
ADCY3eqtl-a-ENSG00000138031ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000138031 || id:eqtl-a-ENSG00000138031Inverse-variance weighted (multiplicative random effects)5−0.212490.1016040.036MR Egger weighted median0.586261
RAB15eqtl-a-ENSG00000139998ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000139998 || id:eqtl-a-ENSG00000139998Inverse-variance weighted (multiplicative random effects)4−0.254420.0806830.002MR Egger0.948296
MFGE8eqtl-a-ENSG00000140545ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000140545 || id:eqtl-a-ENSG00000140545Inverse-variance weighted (multiplicative random effects)40.0931750.0442740.035MR Egger weighted median0.800747
MYO1Feqtl-a-ENSG00000142347ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000142347 || id:eqtl-a-ENSG00000142347Inverse-variance weighted (multiplicative random effects)30.3503130.0892570MR Egger weighted median0.64159
HNMTeqtl-a-ENSG00000150540ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000150540 || id:eqtl-a-ENSG00000150540Inverse-variance weighted (multiplicative random effects)4−0.070730.006560MR Egger weighted median0.979268
ING1eqtl-a-ENSG00000153487ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000153487 || id:eqtl-a-ENSG00000153487Inverse-variance weighted (multiplicative random effects)3−0.029850.0088050.001MR Egger weighted median0.96251
ARHGAP25eqtl-a-ENSG00000163219ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000163219 || id:eqtl-a-ENSG00000163219Inverse-variance weighted (multiplicative random effects)30.1623160.0261590MR Egger weighted median0.823096
TGFBR2eqtl-a-ENSG00000163513ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000163513 || id:eqtl-a-ENSG00000163513Inverse-variance weighted (multiplicative random effects)4−0.157540.0707850.026MR Egger0.477357
CITED2eqtl-a-ENSG00000164442ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000164442 || id:eqtl-a-ENSG00000164442Inverse-variance weighted (multiplicative random effects)30.1529190.0694010.028MR Egger weighted median0.57333
GALNT10eqtl-a-ENSG00000164574ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000164574 || id:eqtl-a-ENSG00000164574Inverse-variance weighted (multiplicative random effects)40.3071270.0865860MR Egger0.612881
SYKeqtl-a-ENSG00000165025ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000165025 || id:eqtl-a-ENSG00000165025Inverse-variance weighted (multiplicative random effects)40.0412770.0177230.02MR Egger weighted median0.826421
NCF1Ceqtl-a-ENSG00000165178ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000165178 || id:eqtl-a-ENSG00000165178Inverse-variance weighted (multiplicative random effects)4−0.111620.0254920MR Egger weighted median0.695435
TRANK1eqtl-a-ENSG00000168016ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000168016 || id:eqtl-a-ENSG00000168016Inverse-variance weighted (multiplicative random effects)3−0.043240.0193820.026MR Egger weighted median0.973751
CX3CR1eqtl-a-ENSG00000168329ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000168329 || id:eqtl-a-ENSG00000168329Inverse-variance weighted (multiplicative random effects)5−0.205610.0936910.028MR Egger weighted median0.22481
RAB31eqtl-a-ENSG00000168461ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000168461 || id:eqtl-a-ENSG00000168461Inverse-variance weighted (multiplicative random effects)8−0.125150.062570.045MR Egger weighted median0.520829
PTAFReqtl-a-ENSG00000169403ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000169403 || id:eqtl-a-ENSG00000169403Inverse-variance weighted (multiplicative random effects)40.2692850.1091840.014MR Egger weighted median0.535136
NPAS2eqtl-a-ENSG00000170485ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000170485 || id:eqtl-a-ENSG00000170485Inverse-variance weighted (multiplicative random effects)40.0862530.0201960MR Egger weighted median0.827519
PKIAeqtl-a-ENSG00000171033ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000171033 || id:eqtl-a-ENSG00000171033Inverse-variance weighted (multiplicative random effects)10−0.083690.041370.043MR Egger weighted median0.569548
INSReqtl-a-ENSG00000171105ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000171105 || id:eqtl-a-ENSG00000171105Inverse-variance weighted (multiplicative random effects)30.4294770.1173370MR Egger0.686507
CEBPBeqtl-a-ENSG00000172216ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000172216 || id:eqtl-a-ENSG00000172216Inverse-variance weighted (multiplicative random effects)4−0.099690.0260820MR Egger weighted median0.851249
DDIT3eqtl-a-ENSG00000175197ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000175197 || id:eqtl-a-ENSG00000175197Inverse-variance weighted (multiplicative random effects)30.2284110.0408890MR Egger weighted median0.875236
MRPL48eqtl-a-ENSG00000175581ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000175581 || id:eqtl-a-ENSG00000175581Inverse-variance weighted (multiplicative random effects)30.1078870.0099950MR Egger weighted median0.906152
CRLF3eqtl-a-ENSG00000176390ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000176390 || id:eqtl-a-ENSG00000176390Inverse-variance weighted (multiplicative random effects)40.0757770.0233690.001MR Egger weighted median0.743505
ADAP2eqtl-a-ENSG00000184060ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000184060 || id:eqtl-a-ENSG00000184060Inverse-variance weighted (multiplicative random effects)30.4507660.2170430.038MR Egger weighted median0.474219
FOXO4eqtl-a-ENSG00000184481ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000184481 || id:eqtl-a-ENSG00000184481Inverse-variance weighted (multiplicative random effects)3−0.386960.0349770MR Egger weighted median0.935967
PDE4Beqtl-a-ENSG00000184588ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000184588 || id:eqtl-a-ENSG00000184588Inverse-variance weighted (multiplicative random effects)30.2369310.0414290MR Egger weighted median0.735599
INSIG1eqtl-a-ENSG00000186480ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000186480 || id:eqtl-a-ENSG00000186480Inverse-variance weighted (multiplicative random effects)70.1112390.0431960.01MR Egger weighted median0.611197
FPR3eqtl-a-ENSG00000187474ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000187474 || id:eqtl-a-ENSG00000187474Inverse-variance weighted (multiplicative random effects)6−0.210360.0671080.002MR Egger weighted median0.703622
CARD9eqtl-a-ENSG00000187796ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000187796 || id:eqtl-a-ENSG00000187796Inverse-variance weighted (multiplicative random effects)6−0.077160.0301940.011MR Egger weighted median0.481819
ATG7eqtl-a-ENSG00000197548ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000197548 || id:eqtl-a-ENSG00000197548Inverse-variance weighted (multiplicative random effects)4−0.171770.0708710.015MR Egger weighted median0.522111
CCDC69eqtl-a-ENSG00000198624ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000198624 || id:eqtl-a-ENSG00000198624Inverse-variance weighted (multiplicative random effects)30.205860.0603690.001MR Egger weighted median0.863129
HSPA1Beqtl-a-ENSG00000204388ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000204388 || id:eqtl-a-ENSG00000204388Inverse-variance weighted (multiplicative random effects)4−0.185740.0481950MR Egger weighted median0.737545
PPP1CBeqtl-a-ENSG00000213639ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000213639 || id:eqtl-a-ENSG00000213639Inverse-variance weighted (multiplicative random effects)40.1402120.0646860.03MR Egger0.264393
ANGeqtl-a-ENSG00000214274ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000214274 || id:eqtl-a-ENSG00000214274Inverse-variance weighted (multiplicative random effects)9−0.099340.0503570.049MR Egger weighted median0.457848
APOBEC3Geqtl-a-ENSG00000239713ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000239713 || id:eqtl-a-ENSG00000239713Inverse-variance weighted (multiplicative random effects)6−0.113150.0267960MR Egger weighted median0.767311
Table 2. IVW model of the CEBPB.
Table 2. IVW model of the CEBPB.
id.Exposureid.OutcomeOutcomeExposureMethodnsnpbsepvallo_ciup_cioror_lci95or_uci95
eqtl-a-ENSG00000172216ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000172216 || id:eqtl-a-ENSG00000172216MR Egger4−0.02390.37510.9551−0.758970.7112650.97640.4681482.036566
eqtl-a-ENSG00000172216ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000172216 || id:eqtl-a-ENSG00000172216Inverse-variance weighted (multiplicative random effects)4−0.09970.02610.0001−0.15081−0.048570.90510.8600090.952588
eqtl-a-ENSG00000172216ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000172216 || id:eqtl-a-ENSG00000172216Weighted median4−0.10270.12440.4089−0.346420.1410370.90240.7072181.151467
eqtl-a-ENSG00000172216ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000172216 || id:eqtl-a-ENSG00000172216Simple mode4−0.13230.1670.486−0.459610.1949570.87610.6315281.215259
eqtl-a-ENSG00000172216ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000172216 || id:eqtl-a-ENSG00000172216Weighted mode4−0.10750.14560.5138−0.392850.1778320.89810.6751331.194625
Table 3. IVW model of CX3CR1.
Table 3. IVW model of CX3CR1.
id.Exposureid.OutcomeOutcomeExposureMethodnsnpbsepvallo_ciup_cioror_lci95or_uci95
eqtl-a-ENSG00000168329ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000168329 || id:eqtl-a-ENSG00000168329MR Egger5−0.3960.1530.0812−0.6959−0.096120.6730.4986240.908354
eqtl-a-ENSG00000168329ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000168329 || id:eqtl-a-ENSG00000168329Inverse-variance weighted (multiplicative random effects)5−0.20560.09370.0282−0.38925−0.021980.81410.6775670.978261
eqtl-a-ENSG00000168329ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000168329 || id:eqtl-a-ENSG00000168329Weighted median5−0.20150.10420.0531−0.40580.0027260.81750.6664461.002729
eqtl-a-ENSG00000168329ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000168329 || id:eqtl-a-ENSG00000168329Simple mode5−0.18560.13470.2405−0.449640.0785360.83060.6378571.081703
eqtl-a-ENSG00000168329ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000168329 || id:eqtl-a-ENSG00000168329Weighted mode5−0.2210.12480.1512−0.465560.0235090.80170.6277861.023788
Table 4. Sensitivity analysis.
Table 4. Sensitivity analysis.
Geneid.Exposureid.OutcomeOutcomeExposureMethodQQ_dfQ_pval
CEBPBeqtl-a-ENSG00000172216ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000172216 || id:eqtl-a-ENSG00000172216MR Egger0.105120.9488
CEBPBeqtl-a-ENSG00000172216ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000172216 || id:eqtl-a-ENSG00000172216Inverse-variance weighted0.150330.9852
CX3CR1eqtl-a-ENSG00000168329ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000168329 || id:eqtl-a-ENSG00000168329MR Egger2.171830.5375
CX3CR1eqtl-a-ENSG00000168329ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000168329 || id:eqtl-a-ENSG00000168329Inverse-variance weighted4.495640.3431
Table 5. Horizontal pleiotropy test.
Table 5. Horizontal pleiotropy test.
Geneid.Exposureid.OutcomeOutcomeExposureEgger_Interceptsepval
CEBPBeqtl-a-ENSG00000172216ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000172216 || id:eqtl-a-ENSG00000172216−0.01350.06360.8512
CX3CR1eqtl-a-ENSG00000168329ebi-a-GCST005810Osteoarthritis of the hip (hospital-diagnosed) || id:ebi-a-GCST005810ENSG00000168329 || id:eqtl-a-ENSG000001683290.0510.03350.2248
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Gao, H.; Gan, X.; He, J.; He, C. Mendelian Randomization and Transcriptome Analyses Reveal Important Roles for CEBPB and CX3CR1 in Osteoarthritis. Bioengineering 2025, 12, 930. https://doi.org/10.3390/bioengineering12090930

AMA Style

Gao H, Gan X, He J, He C. Mendelian Randomization and Transcriptome Analyses Reveal Important Roles for CEBPB and CX3CR1 in Osteoarthritis. Bioengineering. 2025; 12(9):930. https://doi.org/10.3390/bioengineering12090930

Chicago/Turabian Style

Gao, Hui, Xinling Gan, Jing He, and Chengqi He. 2025. "Mendelian Randomization and Transcriptome Analyses Reveal Important Roles for CEBPB and CX3CR1 in Osteoarthritis" Bioengineering 12, no. 9: 930. https://doi.org/10.3390/bioengineering12090930

APA Style

Gao, H., Gan, X., He, J., & He, C. (2025). Mendelian Randomization and Transcriptome Analyses Reveal Important Roles for CEBPB and CX3CR1 in Osteoarthritis. Bioengineering, 12(9), 930. https://doi.org/10.3390/bioengineering12090930

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop