Identification of hub genes in osteoarthritis via integrating bioinformatics analysis and machine learning

Identification of hub genes in osteoarthritis via integrating bioinformatics analysis and machine learning | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Identification of hub genes in osteoarthritis via integrating bioinformatics analysis and machine learning Yanqing Wang, Yizhou Xu, Gang Deng, Shuyi Xu, Xican Li, Xianxi Zhou, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-4877796/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 3 You are reading this latest preprint version Abstract Osteoarthritis (OA) represents a significant burden on global healthcare systems that causes pain and functional impairment by affecting joint tissue. Ubiquitin-specific peptidase 53 (USP53), a member of the ubiquitin-specific protease (USP) family, is involved in the progression of various disease states, but its role in OA has not been investigated. This study employed bioinformatics analysis to identify 92 common genes were identified in the positivity related modules of OA. Further studies demonstrated that the DEGs enriched in 20 pathways, mainly including the Focal adhesion, PI3K-Akt signaling pathway, ECM-receptor interaction, Human papillomavirus infection and Pyrimidine metabolism in KEGG enrichment analysis. Furthermore, machine learning results show that USP53, CA12 and PON could be effective biomarkers for OA diagnosis. Preliminary laboratory investigations showed that compared with the control group, the expression of USP53 in the OA group showed significant changes in vivo. The clinical specimen test results showed the protein expression of USP53 in cartilage tissue of OA patients. All in all, this study indicated the hub genes USP53 could be a potential biomarker for OA, thus providing a novel insight into the modulation of ubiquitin in OA clinical diagnosis and treatment. OA USP53 biomarker bioinformatics analysis machine learning Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 1. Introduction Osteoarthritis (OA) is a chronic disease that causes pain and functional impairment by affecting joint tissue. The primary symptoms of OA are joint pain and loss of function which may lead to disability and the need for joint replacement [1, 2] . Epidemiological studies have shown that OA results from a combination of individual and joint factors, including age, gender, obesity, genetics, diet, injury, joint malalignment, and abnormal joint loading [2–4] . As the population ages and obesity rates rise, the prevalence of OA is expected to increase and will cause significant economic losses and property damage [5] . However, there is no effective therapeutic to manage symptoms, improve joint function, and enhance overall quality of life for OA yet [4] . Despite extensive research, the underlying pathogenesis and clinical treatment of OA remain an area of ongoing investigation. In the vast landscape of biological sciences, the emergence of bioinformatics stands as a beacon of innovation and progress. With the exponential growth of biological data stemming from genomics, proteomics, transcriptomics, and beyond, the need for computational tools to analyze, interpret, and derive insights from this wealth of information has never been greater [6–9] . Bioinformatics, at its core, represents the synergy between biology and computational science, offering a powerful lens through which we can unravel the mysteries of life. At its essence, bioinformatics empowers researchers to navigate the vast ocean of biological data, providing tools and methodologies to analyze genomes, decipher genetic codes, elucidate molecular pathways, and understand the complexities of living systems [10, 11] . By harnessing the power of computational algorithms, machine learning, and data visualization techniques, bioinformatics enables us to extract meaningful patterns, identify biological markers, and unveil the underlying principles governing life processes [12] . With the development and widespread use of microarray and high-throughput sequencing technology, bioinformatics analysis can be used to identify novel genes and biomarkers for many diseases [6, 13, 14] . Machine learning allows us to get a better grasp of disease pathobiology quickly and non-invasively [15–17] . Machine learning finds applications across diverse healthcare domains, including but not limited to aids in disease diagnosis, personalized treatment planning, and drug discovery [18–21] . Screening OA through machine learning and bioinformatics will help to quickly identify key molecules in the pathological mechanism of osteoarthritis and promote the mechanism research of OA. Ubiquitin, a small, highly conserved protein found in eukaryotic cells, has emerged as a versatile signaling molecule, participating in a multitude of cellular processes [22–24] . Ubiquitination plays a pivotal role in shaping cellular behavior and response to internal and external cues such as protein degradation, DNA repair mechanisms, transcriptional regulation, and vesicular trafficking [25–27] . The ubiquitin system (UPS) comprises an intricate network of enzymes, substrates, and regulatory factors that meticulously govern the attachment, modification, and removal of ubiquitin moieties [23] . Action of the UPS is initiated by a three-step enzyme cascade, consisted of ubiquitin-activating enzymes (E1), ubiquitin-conjugating enzymes (E2), and ubiquitin ligases (E3). The UPS modifications can adopt various topologies, including mono-ubiquitination or polyubiquitination via different lysine linkages, each imparting distinct regulatory consequence [28] . Recent studies have further demonstrated the crucial contribution of ubiquitin system in the development of OA. However, the mechanism of how the ubiquitination system is involved in osteoarthritis remains unclear [29, 30] . This study was integrated and analyzed hub gene of OA from the public databases and verified through various experimental methods in vitro, which could provide new insights into the biological mechanisms of OA. In the present study, the hub genes, which could serve as the biomarkers for disease diagnosis and therapeutic monitoring, were screened by bioinformatics and machine learning in OA, and to investigated whether ubiquitination system was involved in the occurrence and development of osteoarthritis. 2. Materials and Methods 2.1. Datum Origins We obtained data from the GEO database, which included a gene expression profile dataset (GSE129147 and GSE169077). The GSE129147 and GSE169077 dataset contained 30 samples, 16 of which were OA samples and 14 were normal samples. 2.2. Filtration and Annotation of Data Annotation of the GEO datasets is necessary before conducting variation analysis and WGCNA when the data are enormous, obscure, and unmatched. The platform files and the probe matrix file datasets which downloaded from the GEO database were prepared as the input documents, and then, the PERL (version5.32.1) arithmetic software processed the data by detecting the gene name and matching it with its specific probe according to the relationship between the gene name and the probe matrix, ultimately transforming the probe matrix into the gene matrix. 2.3. Detection of Differentially Expressed Genes(DEGs) EdgR (version 3.32.1) and limma[16] (version: 3.46.0) packages in Bioconductor were utilized to analyze DEG expression. We divided the genes into two groups with the different expression trends based on log fold change (FC) and calculated the mean value of the expressed genes in glioma to obtain the t value. We calculated the p value based on the t value and adjusted it using the false discovery rate (FDR) method. The DEGs were filtered under the condition of∣logFC∣>1 and adjusted p < 0.05. Moreover, we visualized DEGs by plotting heatmaps and volcano plots using the heatmap and ggplot packages in R, respectively. To identify the glioma-related gene modules, we conducted WGCNA based on R programming language (3.6.3). The GO.db (version 3.12.1), preprocessCore (version1.52.1), impute (version 1.64.0), and limma packages in Bioconductor were used to save and process the obtained datasets. The WGCNA package [17] was applied to identify the highly cooperative genes. Using the coefficient of association and the corresponding p value, we obtained several modules that reflected the relationships between tumor tissues and normal ones. According to all modules we obtained from variation analysis and WGCNA, we selected the best glioma-related modules based on the most conspicuous coefficient of association and fetched information regarding the genes from the modules. The VennDiagram [18] package in R should be installed to identify the intersection among modules and plot the Venn diagram to visualize the consequences of the intersection. 2.4. Function Cognition and Pathway Enrichment Analysis To detect how the OA-related genes functioned in glioma and which sites and pathways they may act on, KEGG enrichment analyses were conducted. Dose [19](3.16.0), clusterProfiler [20] (3.18.1), and enrichplot (1.10.2) packages in Bioconductor and colorspace, stringi, and ggplot2 packages in R were applied for analyses. Results of KEGG enrichment analyses were output as two diagrams: a bar plot and a bubble diagram. Diagrams of KEGG results revealed several pathways and target sites where the DEGs may be enriched. The results of KEGG enrichment analyses were calculated based on p < 0:05. 2.5. Configuration of the PPI Network and Developing a Network of Hub Genes After filtering glioma-related genes, we attempted to identify the potential interaction among these genes and subsequently developed a protein–protein interaction (PPI) network. The hub genes were identified using STRING [21] (version 11.0) and the CytoHubba [22] plug-in in Cytoscape [23] software. The application of Cytoscape software is aimed at abstracting DEGs encoding proteins and establishing a network scaffold. The CytoHubba plug-in can detect and locate 10 of the most relevant DEGs using the maximal clique centrality (MCC) and mark them in red (high correlation), orange (medium correlation), and yellow (low correlation) colors based on their correlation with glioma. 2.6. Correlation Analysis between the Target Gene and Glioma USP53, CA12 and PON were selected from the 9 hub genes and selected as the optimum gene. Confirmation regarding whether it is associated with the occurrence of gliomas is required. Module USP53 and PON in the Gene Expression Profiling Interactive Analysis database provides a macro perspective of the difference in the gene expression between OA and normal tissues. 2.7. Verification of the Protein Expression of Hub Genes The expression of core genes was verified to identify the differential expression of USP53 between the normal tissue and OA tissue. We input the hub gene into the software and chose the tissue module and pathology module to run the analysis. The cerebral cortex was selected, and the result with immunohistochemical images was generated automatically. The survival and survminer packages in R were applied to analyze the overall survival (OS) using the best cut-off criteria based on the gene samples. We divided the samples based on the expression level of the target genes and conducted the KM analysis to check the difference in the survival rate between the groups. A p value of < 0.05 was considered statistically significant. Subsequently, a survival curve diagram was plotted to visualize the survival analysis results. The GEPIA [25] database was used to analyze the expression difference among GBM, LGG, and normal brain tissues. In order to understand the correlation between GRIN1 and glioma more comprehensively, we consulted the LOGpc [26] database to verify the expression difference of GRIN1 in a high-grade glioma and a low-grade glioma. 2.9. collagen2/DAPI stained Chondrocytes were cultured in primary culture and then experiments were carried out. Chondrocytes cells were grown in 6 well-plates and divided into primary cultured cells (P0) and the sixth passage (P6). Each group was stained with collagen2/DAPI and examined under a microscope. Subsequent protein extraction for western blot detection. 2.10 Western blotting analysis Chondrocytes cells were digested with 0.25% Trypsin and resuspended in 500 µL of binding buffer. Cell debris was removed by centrifuging at 12,000g for 10 min at 4℃, and the supernatants were harvested for further evaluation and stored at ‑80℃. The bicinchoninic acid method was used to detect protein concentration. Total protein (30 µg) was separated by SDS-PAGE, using a 10% gel, and transferred to 0.45 µm polyvinylidene difluoride (PVDF) membranes (EMD Millipore, Billerica, MA, USA). After blocking with 5% BSA for 2 h at room temperature, the membranes were incubated with primary antibodies (USP53-ABclonal, China, Cat#A24873, Rabbit,1:1000; GAPDH- Abcam, UK, Cat#ab8245, Rabbit,1:1000; Abcam, Inc., Cambridge, UK) at 4℃ overnight. After rinsing with Tris-buffered saline with 0.01% Tween-20, the PVDF membranes were incubated with the secondary antibodies for 2 h at room temperature. Enhanced chemiluminescence (ECL) western blotting substrate (Pierce; Thermo Fisher Scientific, Inc., Wilmington, USA) was applied to visualize the protein bands. Image acquisition was performed using a Tanon-6200 gel imaging system (Tanon Science and Technology Co., Ltd., Shanghai, China). Image processing software (ImageJ; Version 14.8; National Institutes of Health) was used to analyze the images. Each experiment was repeated three times, and the data are presented as the mean ± SEM. 2.11 Experimental animals and human OA samples Cartilage samples were collected from four OA patients (average age = 65 years, 2 males and 2 females). These patients had received total knee replacements at Zhujiang Hospital, and were used to establish primary cartilage explant and chondrocyte cultures. 2.12 Primary culture of chondrocytes from OA patients Chondrocytes were collected asdescribed [31] and propagated in DMEM/F12 medium (Cat#11330057, Corning, USA) containing 10% FBS (Cat#F8318, Sigma, USA). Chondrocytes were used as senescence models after six passages [32] . Cartilage fragments from OA patients were washed in PBS and incubated with 0.1% collagenase II in DMEM overnight. The cells were then harvested and cultured with IL-1β (10 ng/mL) for 72 h to produce senescence, as previously described [33] . 3. Results 2.1 Screening of DEGs The datasets were downloaded from the GEO database and the expression matrices of two datasets were normalized (Fig. 1 A). The result showed that 155 DEGs were significant difference (| log2 (FC) | > 1 and P < 0.05), which including 41 down-regulated genes and 114 up-regulated genes (Fig. 1 B). The expression of the top 20 expression rankings DEGs was shown in Heatmaps (Fig. 1 C). Meanwhile, we performed GSEA to investigate the biological signaling pathway related to up-regulated genes and down-regulated genes, respectively. The up-regulated genes were enriched in pathways, such as Asthma, Intestinal immune network for IgA production, mucin type O-glycan biosynthesis, protein digestion and absorption, rheumatoid arthritis (Fig. 1 D). The down-regulated genes were enriched in pathways, such as endocytosis, Epstein-barr virus infection, human T-cell leukemia virus 1 infection, insulin signaling pathway and pathways in cancer. 2.2 Variation Analysis and WGCNA of DEGs To identify DEGs in the TCGA datasets and GEO datasets, variation analysis and WGCNA must be conducted, and then, a network based on the gene’s cooperation relationship must be constructed. To visualize the results, the gene clusters were placed into different modules by plotting dendrograms (Figs. 2 A). Clustering analysis was poorly clustered (Fig. 2 B). Therefore, this sample was excluded as an outlier in the WGCNA analysis. The analysis of soft threshold selection revealed that gene associations were maximally consistent with the scale-free distribution and when b = 7. Then, the total modules were identified in the weighted gene co-expression network by merging modules with feature factors greater than 0.5 and setting the minimum number of genes in a module. Module-trait diagrams were also obtained based on R. These dendrograms illustrated the association of the gene cluster between the normal and OA tissues. After WGCNA of GEO datasets, a diagram containing 17 colors representing different modules were constructed (Fig. 1 C). The turquoise module was declared to have the highest association with the OA tissue after GEO analysis (Fig. 2 D). These diagrams exhibited solid evidence illustrating the relationships between DEGs and tissues. 2.3 Enrichment analysis of common genes from WGCNA The common genes were screened between OA positively related modules (MEturquoise and MEpink modules) and normal related modules (MEblue module). Then, 92 common genes were identified in the positivity related modules of OA (Fig. 3 A). DEGs between the high-risk group and the low-risk group were further subjected to KEGG analysis and GO-BP analysis. Through functional annotation of 92 key osteoarthritis genes screened, DEGs enriched in 20 pathways, mainly including the Focal adhesion, PI3K-Akt signaling pathway, ECM-receptor interaction, Human papillomavirus infection and Pyrimidine metabolism in KEGG enrichment analysis (Fig. 3 B). The GO-BP analysis showed that DEGs were significantly enriched in extracellular matrix organization, extracellular structure organization, external encapsulating structure organization. By the CC analysis of DEGs identify that DEGs were mainly enriched in collagen-containing extracellular matrix, endoplasmic reticulum lumen, basement membrane, collagen trimer. The MF analysis revealed that the most significantly enriched terms are extracellular matrix structural constituent, sulfur compound binding, integrin binding (Fig. 3 C). 2.4 The feature genes and construction of a diagnostic model for OA Using a Venn diagram to compare the overlapping regions of DEGs and key module genes, we were able to identify 92 overlapping gene regions (Fig. 2 A). We use LASSO regression analysis to select 9 predicted genes from the 92 osteoarthritis genes: ANGPTL7、CA12、IGFBP3、LOXL2、MARCO、PDE3B、PON3、SERPINE2、USP53(Figs. 4 A). Then use boxplots to compare the 9 hub genes expression between control and osteoarthritis groups (Fig. 4 B). Further correlation analysis of these nine genes using the heatmap of correlation, the result showed that USP53 had the largest positive correlation with PON 3 and the largest negative correlation with CA12 (Fig. 4 C). To further validation of its diagnostic value of three hub genes, We use ROC curve analysis and calculate the area under the curve (AUC). The results showed that the AUC values of the three hub genes were greater than 0.5 (USP53: AUC = 0.781; CA12: AUC = 0.924; PON: AUC = 0.906;), indicating that USP53, CA12 and PON might be effective biomarkers for OA diagnosis (Fig. 4 D). 2.5 The expression levels of USP53 was decreased in P6 chondrocytes and OA patient In order to verify the results of bioinformatics and machine learning, the expression of USP53 was detected in cartilage senescent cells. β-galactosidase staining was used to detect the aging level of P6 chondrocytes. The results showed that P6 chondrocytes were seriously aged (Fig. 5 C). The results of qPCR showed that, compared with the control group, the mRNA expression of USP53 decreased in P6 chondrocytes. Meanwhile, the mRNA expression of PON3 and CA12 had no significant change (Fig. 5 D). Consistently, the USP53 protein expression level decreased (Fig. 5 E). In order to further study the relationship between USP53 and osteoarthritis, MRI images and cartilage tissue of OA patients collected according to clinical classification 0–4 and the expression level of USP53 in patients' isolated cartilage tissue was tested. The results showed that the protein expression level of USP53 decreased (Fig. 5 F, 5 G). These results are consistent with bioinformatics and machine learning, indicating that USP53 is involved in the development of osteoarthritis. 4. Discussion OA represents a significant burden on global healthcare systems, necessitating the urgent need for reliable biomarkers to aid in its diagnosis. A series of studies have further demonstrated the crucial contribution of ubiquitin system in the development of OA in decades. However, it is unclear that which protein of UPS is the hub genes in OA progression. In this study, we use bioinformatics and machine learning methods to investigate the role of UPS-related genes in OA, as this may provide new ideas for diagnosis and treatment for OA. The result showed the novel evidence that UPS53 is the hub gene in the regulation of OA. The major findings of the present study were: there are 155 DEGs in the training set by Screening of DEGs. The top 20 expression rankings DEGs were obtained through bioinformatics analysis. The variation and WGCNA analysis were used to detecte the association of the gene cluster between the normal and OA. The result revealed that the most significantly enriched terms are extracellular matrix structural constituent, sulfur compound binding, integrin binding. There are 92 common genes were identified in the positivity related modules of OA. For further research, LASSO regression analysis was used to select 9 predicted genes from the 92 osteoarthritis genes: ANGPTL7、CA12、IGFBP3、LOXL2、MARCO、PDE3B、PON3、SERPINE2、USP53. The result showed that USP53, CA12 and PON might be effective biomarkers for OA diagnosis. Finally, the results showed that compared with the control group, the expression of USP53 in the OA group showed significant changes in vitro. This article jointly verifies that USP53 can be used as a diagnostic indicator for OA disease through bioinformatics, machine learning and in vitro experiments. This will provide a reference for the clinical diagnosis and application of this disease. Ubiquitin-specific peptidase 53 (USP53) is a member of the ubiquitin-specific protease (USP) family, which it plays critical roles in fundamental cellular processes and has implications in various disease states, such as neurodegenerative diseases and cancer. The deubiquitination process mediated by USP53 involves the removal of ubiquitin moieties from protein substrates, thereby influencing their stability, localization, and function within the cell. Existing research shows that USP53 may be involved in the regulation of DNA damage response pathways, potentially impacting genome stability and cell survival mechanisms. Additionally, USP53 has been implicated in the control of mitotic progression and centrosome regulation, suggesting its importance in cell cycle control and maintenance of chromosomal integrity. However, there are not studies explored the relationship between USP53 and OA. This article discovered that, compared with normal tissues, the expression of USP53 in OA tissues is significantly increased and USP53 may become a potential marker for the diagnosis and treatment of OA. Further research is warranted to elucidate the molecular mechanisms underlying USP53 function and its potential as a therapeutic target for OA disease intervention. Bone homeostasis is maintained by the balanced activities of osteoblasts and osteoclasts during bone remodeling. The disorder of Bone homeostasis led to OA. Research shows that USP53 regulates osteoblast-dependent osteoclasts by controlling RANKL expression, thereby affecting bone balance. However, it is not clear whether USP53 affects the occurrence and development of OA disease in the same pathway. USP53 belongs to the ubiquitin-specific protease (USP) superfamily of deubiquitinating enzymes. This study determined that USP53 is involved in the occurrence and development of OA disease firstly and provides new target molecules for the clinical diagnosis and treatment of OA and worth the further exploration. Subsequently, clinical specimens can be combined to detect the stability and difference of USP53 in pathological specimens to explore the possibility of this protein as a diagnostic molecule. Further study the signaling pathways involved in the occurrence and development of OA disease by USP53, exploring its pathological mechanism, and providing reference for the study of the mechanism of OA. In summary, this article discovered the potential value of USP53 as a diagnostic tool for OA disease, provided a reference for the study of the mechanism of OA, and will be helpful for the clinical treatment of OA. Declarations All data and materials are relevant to manuscript. The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Conflict of interest All authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Ethics approval Animal ethical approval was obtained from the Animal Care Committee of the University of Guangzhou University of Chinese Medicine (Guangzhou, China). The tissue collection was authorized by the Ethical Board of Zhujiang Hospital, Southern Medical University (2022-KY-190-01) Funding This work was supported by The National Natural Science Foundation of China (grant no. 82374485 ) Author Contribution Yanqing Wang. Gang Deng. and Xianxi Zhou. wrote the main manuscript text；Gang Deng.Shuyi Xu. and Xican Li. prepared figures 1-5. All authors reviewed the manuscript Data availability The datasets used or analyzed during the current study are available from the corresponding author upon reasonable request. References Chen Y, Hu Y, Yu Y E, et al. Subchondral Trabecular Rod Loss and Plate Thickening in the Development of Osteoarthritis[J]. J Bone Miner Res. 2018, 33(2): 316-327. Ondrésik M, Azevedo M F, Da S M A, et al. Management of knee osteoarthritis. Current status and future trends[J]. Biotechnol Bioeng. 2017, 114(4): 717-739. Palazzo C, Nguyen C, Lefevre-Colau M M, et al. Risk factors and burden of osteoarthritis[J]. Ann Phys Rehabil Med. 2016, 59(3): 134-138. Abramoff B, Caldera F E. Osteoarthritis: Pathology, Diagnosis, and Treatment Options[J]. Med Clin North Am. 2020, 104(2): 293-311. Wallace I J, Worthington S, Felson D T, et al. Knee osteoarthritis has doubled in prevalence since the mid-20th century[J]. Proc Natl Acad Sci U S A. 2017, 114(35): 9332-9336. Hu X, Ni S, Zhao K, et al. Bioinformatics-Led Discovery of Osteoarthritis Biomarkers and Inflammatory Infiltrates[J]. Front Immunol. 2022, 13: 871008. Fagan A, Culhane A C, Higgins D G. A multivariate analysis approach to the integration of proteomic and gene expression data[J]. Proteomics. 2007, 7(13): 2162-2171. Gwynne P, Heebner G. Proteomics[J]. Science. 2005, 308(5722): 707-708, 712, 715. Posada D. Modeltest : testing the model DNA substitution[J]. Bioinformatics. 1998, 14. Jr G D, Sherman B T, Hosack D A, et al. DAVID: Database for Annotation, Visualization, and Integrated Discovery[J]. Genome Biology. 2003, 4(9). Haw R, Hermjakob H, D'Eustachio P, et al. Reactome pathway analysis to enrich biological discovery in proteomics data sets[J]. Proteomics. 2011, 11(18). Awada W, Khoshgoftaar T M, Dittman D, et al. A review of the stability of feature selection techniques for bioinformatics data[C]. 2012. Qiu Y, Chen X, Yan X, et al. Gut microbiota perturbations and neurodevelopmental impacts in offspring rats concurrently exposure to inorganic arsenic and fluoride[J]. Environ Int. 2020, 140: 105763. He Y, Zhang H, Ma L, et al. Identification of TIMP1 as an inflammatory biomarker associated with temporal lobe epilepsy based on integrated bioinformatics and experimental analyses[J]. J Neuroinflammation. 2023, 20(1): 151. Camacho D M, Collins K M, Powers R K, et al. Next-Generation Machine Learning for Biological Networks[J]. Cell. 2018, 173(7): 1581-1592. Ahuja A S. The impact of artificial intelligence in medicine on the future role of the physician[J]. PeerJ. 2019, 7: e7702. Orlov Y L, Anashkina A A, Klimontov V V, et al. Medical Genetics, Genomics and Bioinformatics Aid in Understanding Molecular Mechanisms of Human Diseases[J]. Int J Mol Sci. 2021, 22(18). Goecks J, Jalili V, Heiser L M, et al. How Machine Learning Will Transform Biomedicine[J]. Cell. 2020, 181(1): 92-101. Arora S, Venkataraman V, Zhan A, et al. Detecting and monitoring the symptoms of Parkinson's disease using smartphones: A pilot study[J]. Parkinsonism Relat Disord. 2015, 21(6): 650-653. van Driel M A, Brunner H G. Bioinformatics methods for identifying candidate disease genes[J]. Hum Genomics. 2006, 2(6): 429-432. Sammut S J, Crispin-Ortuzar M, Chin S F, et al. Multi-omic machine learning predictor of breast cancer therapy response[J]. Nature. 2022, 601(7894): 623-629. Sadowski M, Suryadinata R, Tan A R, et al. Protein monoubiquitination and polyubiquitination generate structural diversity to control distinct biological processes[J]. IUBMB Life. 2012, 64(2): 136-142. Glickman M H, Ciechanover A. The Ubiquitin-Proteasome Proteolytic Pathway: Destruction for the Sake of Construction[J]. Physiological Reviews. 2002, 82(2): 373-428. Xiong Y, Yu C, Zhang Q. Ubiquitin-Proteasome System–Regulated Protein Degradation in Spermatogenesis[J]. Cells. 2022, 11(6): 1058. Nijman S M, Luna-Vargas M P, Velds A, et al. A genomic and functional inventory of deubiquitinating enzymes[J]. Cell. 2005, 123(5): 773-786. Cockram P E, Kist M, Prakash S, et al. Ubiquitination in the regulation of inflammatory cell death and cancer[J]. Cell Death Differ. 2021, 28(2): 591-605. Nakazawa Y, Hara Y, Oka Y, et al. Ubiquitination of DNA Damage-Stalled RNAPII Promotes Transcription-Coupled Repair[J]. Cell. 2020, 180(6): 1228-1244. Komander D. The emerging complexity of protein ubiquitination[J]. Biochem Soc Trans. 2009, 37(Pt 5): 937-953. Liu Y, Molchanov V, Yang T. Enzymatic Machinery of Ubiquitin and Ubiquitin-Like Modification Systems in Chondrocyte Homeostasis and Osteoarthritis[J]. Curr Rheumatol Rep. 2021, 23(8): 62. Wu Q, Zhu M, Rosier R N, et al. Beta-catenin, cartilage, and osteoarthritis[J]. Ann N Y Acad Sci. 2010, 1192(1): 344-350. Arra M, Swarnkar G, Alippe Y, et al. IκB-ζ signaling promotes chondrocyte inflammatory phenotype, senescence, and erosive joint pathology[J]. Bone Research. 2022, 10(1): 12. Wu Y, Chen L, Wang Y, et al. Overexpression of Sirtuin 6 suppresses cellular senescence and NF-κB mediated inflammatory responses in osteoarthritis development[J]. Scientific Reports. 2015, 5(1): 17602. Lu Y, Liu L, Pan J, et al. MFG-E8 regulated by miR-99b-5p protects against osteoarthritis by targeting chondrocyte senescence and macrophage reprogramming via the NF-κB pathway[J]. Cell Death & Disease. 2021, 12(6): 533. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Editor assigned by journal 12 Aug, 2024 Submission checks completed at journal 11 Aug, 2024 First submitted to journal 07 Aug, 2024 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-4877796","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":339014366,"identity":"7a8564bc-039a-4667-b687-b496e7c474ec","order_by":0,"name":"Yanqing Wang","email":"","orcid":"","institution":"Guangzhou University of Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Yanqing","middleName":"","lastName":"Wang","suffix":""},{"id":339014367,"identity":"b8a6f660-c404-4631-92ee-0695cd0fecf3","order_by":1,"name":"Yizhou Xu","email":"","orcid":"","institution":"Orthopedic Medical Center, Southern Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yizhou","middleName":"","lastName":"Xu","suffix":""},{"id":339014368,"identity":"996153a1-ab72-4a54-a451-01ecd9b011c9","order_by":2,"name":"Gang Deng","email":"","orcid":"","institution":"Southern Medical University","correspondingAuthor":false,"prefix":"","firstName":"Gang","middleName":"","lastName":"Deng","suffix":""},{"id":339014369,"identity":"05e6380b-adf0-4840-918d-ce0da81483b5","order_by":3,"name":"Shuyi Xu","email":"","orcid":"","institution":"Southern Medical University","correspondingAuthor":false,"prefix":"","firstName":"Shuyi","middleName":"","lastName":"Xu","suffix":""},{"id":339014370,"identity":"83b21cee-2e61-4544-a137-ee51eaef6e43","order_by":4,"name":"Xican Li","email":"","orcid":"","institution":"Guangzhou University of Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Xican","middleName":"","lastName":"Li","suffix":""},{"id":339014371,"identity":"49c2f4f2-b271-46e2-a519-3fd04e8b2190","order_by":5,"name":"Xianxi Zhou","email":"","orcid":"","institution":"Guangzhou University of Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Xianxi","middleName":"","lastName":"Zhou","suffix":""},{"id":339014372,"identity":"2a4a3624-99bd-4be2-9e95-0aaa56a84c80","order_by":6,"name":"Jiasong Guo","email":"","orcid":"","institution":"Southern Medical University","correspondingAuthor":false,"prefix":"","firstName":"Jiasong","middleName":"","lastName":"Guo","suffix":""},{"id":339014373,"identity":"8c20aa48-10fb-4878-9157-0908ae5ac367","order_by":7,"name":"Dongfeng Chen","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAuUlEQVRIie3NMQrCMBSA4RcC7RLp+lw8Q6BQHD2GY0sGR9dOmiLUxeIRvILg4viKaw7gaMngGYoidnUQnptD/v3jBwiF/rDEFp1/ljhJ4sryCJJJtXLTdLy7MImemQxHdVkcrwsuiQ0gRihOVFjozxyiPGitUGbUWtE4DsEl5fnwydrKSlGziAEijSrdCC6ZG1HZYaMjLkHnpQRCjU7YtuGQZGvih3it1vvDvbv1HPIR/QpCoVAo9KU36gY2Z6TgqhQAAAAASUVORK5CYII=","orcid":"","institution":"Guangzhou University of Chinese Medicine","correspondingAuthor":true,"prefix":"","firstName":"Dongfeng","middleName":"","lastName":"Chen","suffix":""}],"badges":[],"createdAt":"2024-08-08 03:24:28","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-4877796/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-4877796/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":66329473,"identity":"3c6d3e99-5a5f-4206-aa64-f18df0032a77","added_by":"auto","created_at":"2024-10-10 13:15:33","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":1382140,"visible":true,"origin":"","legend":"\u003cp\u003eDEGs were Screened by bioinformatics analysis. A. two datasets were normalized; B. there are 155 DEGs were significant difference; C. the top 20 expression rankings DEGs. D. the biological signaling pathway related to DEGs genes.\u003c/p\u003e","description":"","filename":"fig1.png","url":"https://assets-eu.researchsquare.com/files/rs-4877796/v1/02247646fa1b89e50d7427d9.png"},{"id":66327428,"identity":"e6df2083-6aa6-4483-ad4e-28195c05bd56","added_by":"auto","created_at":"2024-10-10 13:07:33","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":1120932,"visible":true,"origin":"","legend":"\u003cp\u003eVariation Analysis and WGCNA of DEGs. A. WGCNA and DEG analysis screened out 92 differential genes. B. Clustering analysis. C. GO-BP analysis these differential genes.\u003c/p\u003e","description":"","filename":"fig2.png","url":"https://assets-eu.researchsquare.com/files/rs-4877796/v1/5d416aa997a103ccf7db68d1.png"},{"id":66329474,"identity":"3c98fef4-3c61-4f39-87cd-c38e1245da00","added_by":"auto","created_at":"2024-10-10 13:15:33","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":860059,"visible":true,"origin":"","legend":"\u003cp\u003eThe 92 differential genes were identified in OA. \u0026nbsp;A. WGCNA and DEG analysis screened out 92 differential genes. B. KEGG analysis these differential genes. C. GO-BP analysis these differential genes.\u003c/p\u003e","description":"","filename":"fig3.png","url":"https://assets-eu.researchsquare.com/files/rs-4877796/v1/ae88a9a5071e6e253b9a5e19.png"},{"id":66327433,"identity":"cbdcb019-a855-4f97-a973-768c19d7fc48","added_by":"auto","created_at":"2024-10-10 13:07:36","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":606784,"visible":true,"origin":"","legend":"\u003cp\u003eThe hub gene was selected as feature genes for OA. A. LASSO regression analysis to select 9 predicted genes. B. The gene expression between control and osteoarthritis groups. C. USP53 had the largest correlation for OA. D. ROC curve analysis of USP53, CA12, PON3.\u003c/p\u003e","description":"","filename":"fig4.png","url":"https://assets-eu.researchsquare.com/files/rs-4877796/v1/226bde16e01a0dfcbb0006d7.png"},{"id":66327432,"identity":"ecf6c201-1640-4ed8-bacb-314e6858a62f","added_by":"auto","created_at":"2024-10-10 13:07:33","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":1655136,"visible":true,"origin":"","legend":"\u003cp\u003eUSP53 was downregulate in P6 chondrocytes. A, B. Identification of mouse chondrocytes (P0: primary chondrocytes, P6: sixth generation chondrocytes) by Collagen2 immunofluorescence staining. C. β-galactosidase staining detects the senescence level of P6 chondrocytes. D. qPCR detection compares the RNA expression levels of PON3, CA12, USP53 in P0 chondrocytes and P6 chondrocytes. E. Western blot detects changes in USP53 protein expression levels in vitro. F. MRI images of OA patients collected according to clinical classification 0-4. G. Western blot detects changes in the protein expression level of USP53 in of OA patients. Statistical analysis used one-way analysis of variance, and differences between groups used Tukey's post hoc test. *\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.05, N.S: no statistical difference。\u003c/p\u003e","description":"","filename":"fig5.png","url":"https://assets-eu.researchsquare.com/files/rs-4877796/v1/249730153151c0b41644cf4f.png"},{"id":66330105,"identity":"b3ae9872-270a-4e62-9ef8-b641260c13cd","added_by":"auto","created_at":"2024-10-10 13:23:37","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":6045474,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4877796/v1/0b8e7d8f-5052-43f5-abc5-7d2f91046a34.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Identification of hub genes in osteoarthritis via integrating bioinformatics analysis and machine learning","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eOsteoarthritis (OA) is a chronic disease that causes pain and functional impairment by affecting joint tissue. The primary symptoms of OA are joint pain and loss of function which may lead to disability and the need for joint replacement\u003csup\u003e[1, 2]\u003c/sup\u003e. Epidemiological studies have shown that OA results from a combination of individual and joint factors, including age, gender, obesity, genetics, diet, injury, joint malalignment, and abnormal joint loading\u003csup\u003e[2–4]\u003c/sup\u003e. As the population ages and obesity rates rise, the prevalence of OA is expected to increase and will cause significant economic losses and property damage\u003csup\u003e[5]\u003c/sup\u003e. However, there is no effective therapeutic to manage symptoms, improve joint function, and enhance overall quality of life for OA yet\u003csup\u003e[4]\u003c/sup\u003e. Despite extensive research, the underlying pathogenesis and clinical treatment of OA remain an area of ongoing investigation.\u003c/p\u003e \u003cp\u003eIn the vast landscape of biological sciences, the emergence of bioinformatics stands as a beacon of innovation and progress. With the exponential growth of biological data stemming from genomics, proteomics, transcriptomics, and beyond, the need for computational tools to analyze, interpret, and derive insights from this wealth of information has never been greater\u003csup\u003e[6–9]\u003c/sup\u003e. Bioinformatics, at its core, represents the synergy between biology and computational science, offering a powerful lens through which we can unravel the mysteries of life. At its essence, bioinformatics empowers researchers to navigate the vast ocean of biological data, providing tools and methodologies to analyze genomes, decipher genetic codes, elucidate molecular pathways, and understand the complexities of living systems\u003csup\u003e[10, 11]\u003c/sup\u003e. By harnessing the power of computational algorithms, machine learning, and data visualization techniques, bioinformatics enables us to extract meaningful patterns, identify biological markers, and unveil the underlying principles governing life processes\u003csup\u003e[12]\u003c/sup\u003e. With the development and widespread use of microarray and high-throughput sequencing technology, bioinformatics analysis can be used to identify novel genes and biomarkers for many diseases\u003csup\u003e[6, 13, 14]\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eMachine learning allows us to get a better grasp of disease pathobiology quickly and non-invasively \u003csup\u003e[15–17]\u003c/sup\u003e. Machine learning finds applications across diverse healthcare domains, including but not limited to aids in disease diagnosis, personalized treatment planning, and drug discovery\u003csup\u003e[18–21]\u003c/sup\u003e. Screening OA through machine learning and bioinformatics will help to quickly identify key molecules in the pathological mechanism of osteoarthritis and promote the mechanism research of OA.\u003c/p\u003e \u003cp\u003eUbiquitin, a small, highly conserved protein found in eukaryotic cells, has emerged as a versatile signaling molecule, participating in a multitude of cellular processes\u003csup\u003e[22–24]\u003c/sup\u003e. Ubiquitination plays a pivotal role in shaping cellular behavior and response to internal and external cues such as protein degradation, DNA repair mechanisms, transcriptional regulation, and vesicular trafficking\u003csup\u003e[25–27]\u003c/sup\u003e. The ubiquitin system (UPS) comprises an intricate network of enzymes, substrates, and regulatory factors that meticulously govern the attachment, modification, and removal of ubiquitin moieties\u003csup\u003e[23]\u003c/sup\u003e. Action of the UPS is initiated by a three-step enzyme cascade, consisted of ubiquitin-activating enzymes (E1), ubiquitin-conjugating enzymes (E2), and ubiquitin ligases (E3). The UPS modifications can adopt various topologies, including mono-ubiquitination or polyubiquitination via different lysine linkages, each imparting distinct regulatory consequence\u003csup\u003e[28]\u003c/sup\u003e. Recent studies have further demonstrated the crucial contribution of ubiquitin system in the development of OA. However, the mechanism of how the ubiquitination system is involved in osteoarthritis remains unclear\u003csup\u003e[29, 30]\u003c/sup\u003e. This study was integrated and analyzed hub gene of OA from the public databases and verified through various experimental methods in vitro, which could provide new insights into the biological mechanisms of OA. In the present study, the hub genes, which could serve as the biomarkers for disease diagnosis and therapeutic monitoring, were screened by bioinformatics and machine learning in OA, and to investigated whether ubiquitination system was involved in the occurrence and development of osteoarthritis.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e\u003c/p\u003e "},{"header":"2. Materials and Methods","content":"\u003cp\u003e2.1. Datum Origins\u003c/p\u003e\u003cp\u003eWe obtained data from the GEO database, which included a gene expression profile dataset (GSE129147 and GSE169077). The GSE129147 and GSE169077 dataset contained 30 samples, 16 of which were OA samples and 14 were normal samples.\u003c/p\u003e\u003cp\u003e2.2. Filtration and Annotation of Data\u003c/p\u003e\u003cp\u003eAnnotation of the GEO datasets is necessary before conducting variation analysis and WGCNA when the data are enormous, obscure, and unmatched. The platform files and the probe matrix file datasets which downloaded from the GEO database were prepared as the input documents, and then, the PERL (version5.32.1) arithmetic software processed the data by detecting the gene name and matching it with its specific probe according to the relationship between the gene name and the probe matrix, ultimately transforming the probe matrix into the gene matrix.\u003c/p\u003e\u003cp\u003e2.3. Detection of Differentially Expressed Genes(DEGs)\u003c/p\u003e\u003cp\u003eEdgR (version 3.32.1) and limma[16] (version: 3.46.0) packages in Bioconductor were utilized to analyze DEG expression. We divided the genes into two groups with the different expression trends based on log fold change (FC) and calculated the mean value of the expressed genes in glioma to obtain the t value. We calculated the p value based on the t value and adjusted it using the false discovery rate (FDR) method. The DEGs were filtered under the condition of∣logFC∣\u0026gt;1 and adjusted p \u0026lt; 0.05. Moreover, we visualized DEGs by plotting heatmaps and volcano plots using the heatmap and ggplot packages in R, respectively. To identify the glioma-related gene modules, we conducted WGCNA based on R programming language (3.6.3). The GO.db (version 3.12.1), preprocessCore (version1.52.1), impute (version 1.64.0), and limma packages in Bioconductor were used to save and process the obtained datasets. The WGCNA package [17] was applied to identify the highly cooperative genes. Using the coefficient of association and the corresponding p value, we obtained several modules that reflected the relationships between tumor tissues and normal ones. According to all modules we obtained from variation analysis and WGCNA, we selected the best glioma-related modules based on the most conspicuous coefficient of association and fetched information regarding the genes from the modules. The VennDiagram [18] package in R should be installed to identify the intersection among modules and plot the Venn diagram to visualize the consequences of the intersection.\u003c/p\u003e\u003cp\u003e2.4. Function Cognition and Pathway Enrichment Analysis\u003c/p\u003e\u003cp\u003eTo detect how the OA-related genes functioned in glioma and which sites and pathways they may act on, KEGG enrichment analyses were conducted. Dose [19](3.16.0), clusterProfiler [20] (3.18.1), and enrichplot (1.10.2) packages in Bioconductor and colorspace, stringi, and ggplot2 packages in R were applied for analyses. Results of KEGG enrichment analyses were output as two diagrams: a bar plot and a bubble diagram. Diagrams of KEGG results revealed several pathways and target sites where the DEGs may be enriched. The results of KEGG enrichment analyses were calculated based on p \u0026lt; 0:05.\u003c/p\u003e\u003cp\u003e2.5. Configuration of the PPI Network and Developing a Network of Hub Genes\u003c/p\u003e\u003cp\u003eAfter filtering glioma-related genes, we attempted to identify the potential interaction among these genes and subsequently developed a protein–protein interaction (PPI) network. The hub genes were identified using STRING [21] (version 11.0) and the CytoHubba [22] plug-in in Cytoscape [23] software. The application of Cytoscape software is aimed at abstracting DEGs encoding proteins and establishing a network scaffold. The CytoHubba plug-in can detect and locate 10 of the most relevant DEGs using the maximal clique centrality (MCC) and mark them in red (high correlation), orange (medium correlation), and yellow (low correlation) colors based on their correlation with glioma.\u003c/p\u003e\u003cp\u003e2.6. Correlation Analysis between the Target Gene and Glioma\u003c/p\u003e\u003cp\u003eUSP53, CA12 and PON were selected from the 9 hub genes and selected as the optimum gene. Confirmation regarding whether it is associated with the occurrence of gliomas is required. Module USP53 and PON in the Gene Expression Profiling Interactive Analysis database provides a macro perspective of the difference in the gene expression between OA and normal tissues.\u003c/p\u003e\u003cp\u003e2.7. Verification of the Protein Expression of Hub Genes\u003c/p\u003e\u003cp\u003eThe expression of core genes was verified to identify the differential expression of USP53 between the normal tissue and OA tissue. We input the hub gene into the software and chose the tissue module and pathology module to run the analysis. The cerebral cortex was selected, and the result with immunohistochemical images was generated automatically.\u003c/p\u003e\u003cp\u003eThe survival and survminer packages in R were applied to analyze the overall survival (OS) using the best cut-off criteria based on the gene samples. We divided the samples based on the expression level of the target genes and conducted the KM analysis to check the difference in the survival rate between the groups. A p value of \u0026lt; 0.05 was considered statistically significant. Subsequently, a survival curve diagram was plotted to visualize the survival analysis results. The GEPIA [25] database was used to analyze the expression difference among GBM, LGG, and normal brain tissues. In order to understand the correlation between GRIN1 and glioma more comprehensively, we consulted the LOGpc [26] database to verify the expression difference of GRIN1 in a high-grade glioma and a low-grade glioma.\u003c/p\u003e\u003cp\u003e2.9. collagen2/DAPI stained\u003c/p\u003e\u003cp\u003eChondrocytes were cultured in primary culture and then experiments were carried out. Chondrocytes cells were grown in 6 well-plates and divided into primary cultured cells (P0) and the sixth passage (P6). Each group was stained with collagen2/DAPI and examined under a microscope. Subsequent protein extraction for western blot detection.\u003c/p\u003e\u003cp\u003e2.10 Western blotting analysis\u003c/p\u003e\u003cp\u003eChondrocytes cells were digested with 0.25% Trypsin and resuspended in 500 µL of binding buffer. Cell debris was removed by centrifuging at 12,000g for 10 min at 4℃, and the supernatants were harvested for further evaluation and stored at ‑80℃. The bicinchoninic acid method was used to detect protein concentration. Total protein (30 µg) was separated by SDS-PAGE, using a 10% gel, and transferred to 0.45 µm polyvinylidene difluoride (PVDF) membranes (EMD Millipore, Billerica, MA, USA). After blocking with 5% BSA for 2 h at room temperature, the membranes were incubated with primary antibodies (USP53-ABclonal, China, Cat#A24873, Rabbit,1:1000; GAPDH- Abcam, UK, Cat#ab8245, Rabbit,1:1000; Abcam, Inc., Cambridge, UK) at 4℃ overnight. After rinsing with Tris-buffered saline with 0.01% Tween-20, the PVDF membranes were incubated with the secondary antibodies for 2 h at room temperature. Enhanced chemiluminescence (ECL) western blotting substrate (Pierce; Thermo Fisher Scientific, Inc., Wilmington, USA) was applied to visualize the protein bands. Image acquisition was performed using a Tanon-6200 gel imaging system (Tanon Science and Technology Co., Ltd., Shanghai, China). Image processing software (ImageJ; Version 14.8; National Institutes of Health) was used to analyze the images. Each experiment was repeated three times, and the data are presented as the mean ± SEM.\u003c/p\u003e\u003cp\u003e2.11 Experimental animals and human OA samples\u003c/p\u003e\u003cp\u003eCartilage samples were collected from four OA patients (average age = 65 years, 2 males and 2 females). These patients had received total knee replacements at Zhujiang Hospital, and were used to establish primary cartilage explant and chondrocyte cultures.\u003c/p\u003e\u003cp\u003e2.12 Primary culture of chondrocytes from OA patients\u003c/p\u003e\u003cp\u003eChondrocytes were collected asdescribed\u003csup\u003e[31]\u003c/sup\u003eand propagated in DMEM/F12 medium (Cat#11330057, Corning, USA) containing 10% FBS (Cat#F8318, Sigma, USA). Chondrocytes were used as senescence models after six passages\u003csup\u003e[32]\u003c/sup\u003e. Cartilage fragments from OA patients were washed in PBS and incubated with 0.1% collagenase II in DMEM overnight. The cells were then harvested and cultured with IL-1β (10 ng/mL) for 72 h to produce senescence, as previously described\u003csup\u003e[33]\u003c/sup\u003e.\u003c/p\u003e"},{"header":"3. Results","content":"\u003cp\u003e2.1 Screening of DEGs\u003c/p\u003e \u003cp\u003eThe datasets were downloaded from the GEO database and the expression matrices of two datasets were normalized (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eA). The result showed that 155 DEGs were significant difference (| log2 (FC) | \u0026gt; 1 and P\u0026thinsp;\u0026lt;\u0026thinsp;0.05), which including 41 down-regulated genes and 114 up-regulated genes (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eB). The expression of the top 20 expression rankings DEGs was shown in Heatmaps (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eC). Meanwhile, we performed GSEA to investigate the biological signaling pathway related to up-regulated genes and down-regulated genes, respectively. The up-regulated genes were enriched in pathways, such as Asthma, Intestinal immune network for IgA production, mucin type O-glycan biosynthesis, protein digestion and absorption, rheumatoid arthritis (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eD). The down-regulated genes were enriched in pathways, such as endocytosis, Epstein-barr virus infection, human T-cell leukemia virus 1 infection, insulin signaling pathway and pathways in cancer.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e2.2 Variation Analysis and WGCNA of DEGs\u003c/p\u003e \u003cp\u003eTo identify DEGs in the TCGA datasets and GEO datasets, variation analysis and WGCNA must be conducted, and then, a network based on the gene\u0026rsquo;s cooperation relationship must be constructed. To visualize the results, the gene clusters were placed into different modules by plotting dendrograms (Figs.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA). Clustering analysis was poorly clustered (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB). Therefore, this sample was excluded as an outlier in the WGCNA analysis. The analysis of soft threshold selection revealed that gene associations were maximally consistent with the scale-free distribution and when b\u0026thinsp;=\u0026thinsp;7. Then, the total modules were identified in the weighted gene co-expression network by merging modules with feature factors greater than 0.5 and setting the minimum number of genes in a module. Module-trait diagrams were also obtained based on R. These dendrograms illustrated the association of the gene cluster between the normal and OA tissues. After WGCNA of GEO datasets, a diagram containing 17 colors representing different modules were constructed (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eC). The turquoise module was declared to have the highest association with the OA tissue after GEO analysis (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eD). These diagrams exhibited solid evidence illustrating the relationships between DEGs and tissues.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e2.3 Enrichment analysis of common genes from WGCNA\u003c/p\u003e \u003cp\u003eThe common genes were screened between OA positively related modules (MEturquoise and MEpink modules) and normal related modules (MEblue module). Then, 92 common genes were identified in the positivity related modules of OA (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eA). DEGs between the high-risk group and the low-risk group were further subjected to KEGG analysis and GO-BP analysis. Through functional annotation of 92 key osteoarthritis genes screened, DEGs enriched in 20 pathways, mainly including the Focal adhesion, PI3K-Akt signaling pathway, ECM-receptor interaction, Human papillomavirus infection and Pyrimidine metabolism in KEGG enrichment analysis (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eB). The GO-BP analysis showed that DEGs were significantly enriched in extracellular matrix organization, extracellular structure organization, external encapsulating structure organization. By the CC analysis of DEGs identify that DEGs were mainly enriched in collagen-containing extracellular matrix, endoplasmic reticulum lumen, basement membrane, collagen trimer. The MF analysis revealed that the most significantly enriched terms are extracellular matrix structural constituent, sulfur compound binding, integrin binding (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eC).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e2.4 The feature genes and construction of a diagnostic model for OA\u003c/p\u003e \u003cp\u003eUsing a Venn diagram to compare the overlapping regions of DEGs and key module genes, we were able to identify 92 overlapping gene regions (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA). We use LASSO regression analysis to select 9 predicted genes from the 92 osteoarthritis genes: ANGPTL7、CA12、IGFBP3、LOXL2、MARCO、PDE3B、PON3、SERPINE2、USP53(Figs.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eA). Then use boxplots to compare the 9 hub genes expression between control and osteoarthritis groups (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eB). Further correlation analysis of these nine genes using the heatmap of correlation, the result showed that USP53 had the largest positive correlation with PON 3 and the largest negative correlation with CA12 (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eC). To further validation of its diagnostic value of three hub genes, We use ROC curve analysis and calculate the area under the curve (AUC). The results showed that the AUC values of the three hub genes were greater than 0.5 (USP53: AUC\u0026thinsp;=\u0026thinsp;0.781; CA12: AUC\u0026thinsp;=\u0026thinsp;0.924; PON: AUC\u0026thinsp;=\u0026thinsp;0.906;), indicating that USP53, CA12 and PON might be effective biomarkers for OA diagnosis (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eD).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e2.5 The expression levels of USP53 was decreased in P6 chondrocytes and OA patient\u003c/p\u003e \u003cp\u003eIn order to verify the results of bioinformatics and machine learning, the expression of USP53 was detected in cartilage senescent cells. β-galactosidase staining was used to detect the aging level of P6 chondrocytes. The results showed that P6 chondrocytes were seriously aged (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eC). The results of qPCR showed that, compared with the control group, the mRNA expression of USP53 decreased in P6 chondrocytes. Meanwhile, the mRNA expression of PON3 and CA12 had no significant change (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eD). Consistently, the USP53 protein expression level decreased (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eE). In order to further study the relationship between USP53 and osteoarthritis, MRI images and cartilage tissue of OA patients collected according to clinical classification 0\u0026ndash;4 and the expression level of USP53 in patients' isolated cartilage tissue was tested. The results showed that the protein expression level of USP53 decreased (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eF,\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eG). These results are consistent with bioinformatics and machine learning, indicating that USP53 is involved in the development of osteoarthritis.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e"},{"header":"4. Discussion","content":"\u003cp\u003eOA represents a significant burden on global healthcare systems, necessitating the urgent need for reliable biomarkers to aid in its diagnosis. A series of studies have further demonstrated the crucial contribution of ubiquitin system in the development of OA in decades. However, it is unclear that which protein of UPS is the hub genes in OA progression. In this study, we use bioinformatics and machine learning methods to investigate the role of UPS-related genes in OA, as this may provide new ideas for diagnosis and treatment for OA. The result showed the novel evidence that UPS53 is the hub gene in the regulation of OA. The major findings of the present study were: there are 155 DEGs in the training set by Screening of DEGs. The top 20 expression rankings DEGs were obtained through bioinformatics analysis. The variation and WGCNA analysis were used to detecte the association of the gene cluster between the normal and OA. The result revealed that the most significantly enriched terms are extracellular matrix structural constituent, sulfur compound binding, integrin binding. There are 92 common genes were identified in the positivity related modules of OA. For further research, LASSO regression analysis was used to select 9 predicted genes from the 92 osteoarthritis genes: ANGPTL7、CA12、IGFBP3、LOXL2、MARCO、PDE3B、PON3、SERPINE2、USP53. The result showed that USP53, CA12 and PON might be effective biomarkers for OA diagnosis. Finally, the results showed that compared with the control group, the expression of USP53 in the OA group showed significant changes in vitro. This article jointly verifies that USP53 can be used as a diagnostic indicator for OA disease through bioinformatics, machine learning and in vitro experiments. This will provide a reference for the clinical diagnosis and application of this disease.\u003c/p\u003e \u003cp\u003eUbiquitin-specific peptidase 53 (USP53) is a member of the ubiquitin-specific protease (USP) family, which it plays critical roles in fundamental cellular processes and has implications in various disease states, such as neurodegenerative diseases and cancer. The deubiquitination process mediated by USP53 involves the removal of ubiquitin moieties from protein substrates, thereby influencing their stability, localization, and function within the cell. Existing research shows that USP53 may be involved in the regulation of DNA damage response pathways, potentially impacting genome stability and cell survival mechanisms. Additionally, USP53 has been implicated in the control of mitotic progression and centrosome regulation, suggesting its importance in cell cycle control and maintenance of chromosomal integrity. However, there are not studies explored the relationship between USP53 and OA. This article discovered that, compared with normal tissues, the expression of USP53 in OA tissues is significantly increased and USP53 may become a potential marker for the diagnosis and treatment of OA. Further research is warranted to elucidate the molecular mechanisms underlying USP53 function and its potential as a therapeutic target for OA disease intervention.\u003c/p\u003e \u003cp\u003eBone homeostasis is maintained by the balanced activities of osteoblasts and osteoclasts during bone remodeling. The disorder of Bone homeostasis led to OA. Research shows that USP53 regulates osteoblast-dependent osteoclasts by controlling RANKL expression, thereby affecting bone balance. However, it is not clear whether USP53 affects the occurrence and development of OA disease in the same pathway. USP53 belongs to the ubiquitin-specific protease (USP) superfamily of deubiquitinating enzymes. This study determined that USP53 is involved in the occurrence and development of OA disease firstly and provides new target molecules for the clinical diagnosis and treatment of OA and worth the further exploration. Subsequently, clinical specimens can be combined to detect the stability and difference of USP53 in pathological specimens to explore the possibility of this protein as a diagnostic molecule. Further study the signaling pathways involved in the occurrence and development of OA disease by USP53, exploring its pathological mechanism, and providing reference for the study of the mechanism of OA. In summary, this article discovered the potential value of USP53 as a diagnostic tool for OA disease, provided a reference for the study of the mechanism of OA, and will be helpful for the clinical treatment of OA.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003eAll data and materials are relevant to manuscript. The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConflict of interest\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAnimal ethical approval was obtained from the Animal Care Committee of the University of Guangzhou University of Chinese Medicine (Guangzhou, China). The tissue collection was authorized by the Ethical Board of Zhujiang Hospital, Southern Medical University (2022-KY-190-01)\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was supported by The National Natural Science Foundation of China (grant no. 82374485 )\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor Contribution\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eYanqing Wang. Gang Deng. and Xianxi Zhou. wrote the main manuscript text；Gang Deng.Shuyi Xu. and Xican Li. prepared figures 1-5. All authors reviewed the manuscript\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets used or analyzed during the current study are available from the corresponding author upon reasonable request.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eChen Y, Hu Y, Yu Y E, et al. Subchondral Trabecular Rod Loss and Plate Thickening in the Development of Osteoarthritis[J]. J Bone Miner Res. 2018, 33(2): 316-327.\u003c/li\u003e\n\u003cli\u003eOndr\u0026eacute;sik M, Azevedo M F, Da S M A, et al. Management of knee osteoarthritis. Current status and future trends[J]. Biotechnol Bioeng. 2017, 114(4): 717-739.\u003c/li\u003e\n\u003cli\u003ePalazzo C, Nguyen C, Lefevre-Colau M M, et al. Risk factors and burden of osteoarthritis[J]. Ann Phys Rehabil Med. 2016, 59(3): 134-138.\u003c/li\u003e\n\u003cli\u003eAbramoff B, Caldera F E. Osteoarthritis: Pathology, Diagnosis, and Treatment Options[J]. Med Clin North Am. 2020, 104(2): 293-311.\u003c/li\u003e\n\u003cli\u003eWallace I J, Worthington S, Felson D T, et al. Knee osteoarthritis has doubled in prevalence since the mid-20th century[J]. Proc Natl Acad Sci U S A. 2017, 114(35): 9332-9336.\u003c/li\u003e\n\u003cli\u003eHu X, Ni S, Zhao K, et al. Bioinformatics-Led Discovery of Osteoarthritis Biomarkers and Inflammatory Infiltrates[J]. Front Immunol. 2022, 13: 871008.\u003c/li\u003e\n\u003cli\u003eFagan A, Culhane A C, Higgins D G. A multivariate analysis approach to the integration of proteomic and gene expression data[J]. Proteomics. 2007, 7(13): 2162-2171.\u003c/li\u003e\n\u003cli\u003eGwynne P, Heebner G. Proteomics[J]. Science. 2005, 308(5722): 707-708, 712, 715.\u003c/li\u003e\n\u003cli\u003ePosada D. Modeltest : testing the model DNA substitution[J]. Bioinformatics. 1998, 14.\u003c/li\u003e\n\u003cli\u003eJr G D, Sherman B T, Hosack D A, et al. DAVID: Database for Annotation, Visualization, and Integrated Discovery[J]. Genome Biology. 2003, 4(9).\u003c/li\u003e\n\u003cli\u003eHaw R, Hermjakob H, D\u0026apos;Eustachio P, et al. Reactome pathway analysis to enrich biological discovery in proteomics data sets[J]. Proteomics. 2011, 11(18).\u003c/li\u003e\n\u003cli\u003eAwada W, Khoshgoftaar T M, Dittman D, et al. A review of the stability of feature selection techniques for bioinformatics data[C]. 2012.\u003c/li\u003e\n\u003cli\u003eQiu Y, Chen X, Yan X, et al. Gut microbiota perturbations and neurodevelopmental impacts in offspring rats concurrently exposure to inorganic arsenic and fluoride[J]. Environ Int. 2020, 140: 105763.\u003c/li\u003e\n\u003cli\u003eHe Y, Zhang H, Ma L, et al. Identification of TIMP1 as an inflammatory biomarker associated with temporal lobe epilepsy based on integrated bioinformatics and experimental analyses[J]. J Neuroinflammation. 2023, 20(1): 151.\u003c/li\u003e\n\u003cli\u003eCamacho D M, Collins K M, Powers R K, et al. Next-Generation Machine Learning for Biological Networks[J]. Cell. 2018, 173(7): 1581-1592.\u003c/li\u003e\n\u003cli\u003eAhuja A S. The impact of artificial intelligence in medicine on the future role of the physician[J]. PeerJ. 2019, 7: e7702.\u003c/li\u003e\n\u003cli\u003eOrlov Y L, Anashkina A A, Klimontov V V, et al. Medical Genetics, Genomics and Bioinformatics Aid in Understanding Molecular Mechanisms of Human Diseases[J]. Int J Mol Sci. 2021, 22(18).\u003c/li\u003e\n\u003cli\u003eGoecks J, Jalili V, Heiser L M, et al. How Machine Learning Will Transform Biomedicine[J]. Cell. 2020, 181(1): 92-101.\u003c/li\u003e\n\u003cli\u003eArora S, Venkataraman V, Zhan A, et al. Detecting and monitoring the symptoms of Parkinson\u0026apos;s disease using smartphones: A pilot study[J]. Parkinsonism Relat Disord. 2015, 21(6): 650-653.\u003c/li\u003e\n\u003cli\u003evan Driel M A, Brunner H G. Bioinformatics methods for identifying candidate disease genes[J]. Hum Genomics. 2006, 2(6): 429-432.\u003c/li\u003e\n\u003cli\u003eSammut S J, Crispin-Ortuzar M, Chin S F, et al. Multi-omic machine learning predictor of breast cancer therapy response[J]. Nature. 2022, 601(7894): 623-629.\u003c/li\u003e\n\u003cli\u003eSadowski M, Suryadinata R, Tan A R, et al. Protein monoubiquitination and polyubiquitination generate structural diversity to control distinct biological processes[J]. IUBMB Life. 2012, 64(2): 136-142.\u003c/li\u003e\n\u003cli\u003eGlickman M H, Ciechanover A. The Ubiquitin-Proteasome Proteolytic Pathway: Destruction for the Sake of Construction[J]. Physiological Reviews. 2002, 82(2): 373-428.\u003c/li\u003e\n\u003cli\u003eXiong Y, Yu C, Zhang Q. Ubiquitin-Proteasome System\u0026amp;ndash;Regulated Protein Degradation in Spermatogenesis[J]. Cells. 2022, 11(6): 1058.\u003c/li\u003e\n\u003cli\u003eNijman S M, Luna-Vargas M P, Velds A, et al. A genomic and functional inventory of deubiquitinating enzymes[J]. Cell. 2005, 123(5): 773-786.\u003c/li\u003e\n\u003cli\u003eCockram P E, Kist M, Prakash S, et al. Ubiquitination in the regulation of inflammatory cell death and cancer[J]. Cell Death Differ. 2021, 28(2): 591-605.\u003c/li\u003e\n\u003cli\u003eNakazawa Y, Hara Y, Oka Y, et al. Ubiquitination of DNA Damage-Stalled RNAPII Promotes Transcription-Coupled Repair[J]. Cell. 2020, 180(6): 1228-1244.\u003c/li\u003e\n\u003cli\u003eKomander D. The emerging complexity of protein ubiquitination[J]. Biochem Soc Trans. 2009, 37(Pt 5): 937-953.\u003c/li\u003e\n\u003cli\u003eLiu Y, Molchanov V, Yang T. Enzymatic Machinery of Ubiquitin and Ubiquitin-Like Modification Systems in Chondrocyte Homeostasis and Osteoarthritis[J]. Curr Rheumatol Rep. 2021, 23(8): 62.\u003c/li\u003e\n\u003cli\u003eWu Q, Zhu M, Rosier R N, et al. Beta-catenin, cartilage, and osteoarthritis[J]. Ann N Y Acad Sci. 2010, 1192(1): 344-350.\u003c/li\u003e\n\u003cli\u003eArra M, Swarnkar G, Alippe Y, et al. I\u0026kappa;B-\u0026zeta; signaling promotes chondrocyte inflammatory phenotype, senescence, and erosive joint pathology[J]. Bone Research. 2022, 10(1): 12.\u003c/li\u003e\n\u003cli\u003eWu Y, Chen L, Wang Y, et al. Overexpression of Sirtuin 6 suppresses cellular senescence and NF-\u0026kappa;B mediated inflammatory responses in osteoarthritis development[J]. Scientific Reports. 2015, 5(1): 17602.\u003c/li\u003e\n\u003cli\u003eLu Y, Liu L, Pan J, et al. MFG-E8 regulated by miR-99b-5p protects against osteoarthritis by targeting chondrocyte senescence and macrophage reprogramming via the NF-\u0026kappa;B pathway[J]. Cell Death \u0026amp; Disease. 2021, 12(6): 533.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"journal-of-orthopaedic-surgery-and-research","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"josr","sideBox":"Learn more about [Journal of Orthopaedic Surgery and Research](http://josr-online.biomedcentral.com)","snPcode":"13018","submissionUrl":"https://submission.nature.com/new-submission/13018/3","title":"Journal of Orthopaedic Surgery and Research","twitterHandle":"@MSKmedBMC","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"OA, USP53, biomarker, bioinformatics analysis, machine learning","lastPublishedDoi":"10.21203/rs.3.rs-4877796/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-4877796/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eOsteoarthritis (OA) represents a significant burden on global healthcare systems that causes pain and functional impairment by affecting joint tissue. Ubiquitin-specific peptidase 53 (USP53), a member of the ubiquitin-specific protease (USP) family, is involved in the progression of various disease states, but its role in OA has not been investigated. This study employed bioinformatics analysis to identify 92 common genes were identified in the positivity related modules of OA. Further studies demonstrated that the DEGs enriched in 20 pathways, mainly including the Focal adhesion, PI3K-Akt signaling pathway, ECM-receptor interaction, Human papillomavirus infection and Pyrimidine metabolism in KEGG enrichment analysis. Furthermore, machine learning results show that USP53, CA12 and PON could be effective biomarkers for OA diagnosis. Preliminary laboratory investigations showed that compared with the control group, the expression of USP53 in the OA group showed significant changes in vivo. The clinical specimen test results showed the protein expression of USP53 in cartilage tissue of OA patients. All in all, this study indicated the hub genes USP53 could be a potential biomarker for OA, thus providing a novel insight into the modulation of ubiquitin in OA clinical diagnosis and treatment.\u003c/p\u003e","manuscriptTitle":"Identification of hub genes in osteoarthritis via integrating bioinformatics analysis and machine learning","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-10-10 13:07:28","doi":"10.21203/rs.3.rs-4877796/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"editorAssigned","content":"","date":"2024-08-12T08:08:35+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2024-08-12T01:09:28+00:00","index":"","fulltext":""},{"type":"submitted","content":"Journal of Orthopaedic Surgery and Research","date":"2024-08-08T03:23:13+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"journal-of-orthopaedic-surgery-and-research","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"josr","sideBox":"Learn more about [Journal of Orthopaedic Surgery and Research](http://josr-online.biomedcentral.com)","snPcode":"13018","submissionUrl":"https://submission.nature.com/new-submission/13018/3","title":"Journal of Orthopaedic Surgery and Research","twitterHandle":"@MSKmedBMC","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"279432b6-4fbb-486a-a0a9-3777b64ee645","owner":[],"postedDate":"October 10th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2024-11-27T11:38:19+00:00","versionOfRecord":[],"versionCreatedAt":"2024-10-10 13:07:28","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-4877796","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-4877796","identity":"rs-4877796","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}