ATP2A3 promotes inflammatory cytokine secretion in rheumatoid arthritis fibroblast-like synoviocytes via activation of the STING signaling pathway

ATP2A3 promotes inflammatory cytokine secretion in rheumatoid arthritis fibroblast-like synoviocytes via activation of the STING signaling pathway | 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 ATP2A3 promotes inflammatory cytokine secretion in rheumatoid arthritis fibroblast-like synoviocytes via activation of the STING signaling pathway Yujie Cai, Donghong Guo, Songyuan Zheng, Dingji Zhu, Jinjun Zhao, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8165027/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background Rheumatoid arthritis (RA) is an autoimmune disorder characterized by synovial inflammation and progressive bone destruction. Endoplasmic reticulum (ER) stress plays a key role in the pathogenesis of RA. However, the biomarkers of ER stress in RA remain elusive. We aimed to identify key ER stress-related genes (ERSRGs) in RA, evaluate their diagnostic potential, and investigate their functional roles. Methods. Transcriptomic data from six RA datasets were analyzed using machine learning algorithms. ATPase Sarcoplasmic/Endoplasmic Reticulum Ca2 + Transporting 3 (ATP2A3) expression was validated via immunohistochemistry (IHC). Cytokine secretion and signaling pathways in RA fibroblast-like synoviocytes (FLS) were examined using enzyme-linked immunosorbent assay (ELISA) and Western blot analysis, respectively. Results Seventeen ERSRGs were identified by intersecting differentially expressed genes (DEGs) between RA and control samples with ERSRGs from the Gene Ontology Consortium. RA patients were classified into two molecular subtypes based on the expression pattern of these 17 genes. Using machine learning, ATP2A3 emerged as key diagnostic biomarker in RA and was validated in external datasets. IHC further confirmed ATP2A3 was highly expressed in RA synovial tissues. Importantly, we found that ATP2A3 promotes the secretion of inflammatory cytokines in RA FLS. Mechanistically, ATP2A3 enhances the phosphorylation of stimulator of interferon genes (STING), thereby activating the STING signaling pathway and leading to increased inflammatory cytokine secretion. Conclusions We establish ATP2A3 as a promising diagnostic biomarker for ER stress in RA and demonstrate that it promotes the secretion of inflammatory cytokines in RA FLS through STING pathway activation. Cell Communication and Signaling Rheumatoid arthritis Endoplasmic reticulum stress Fibroblast-like synoviocytes ATPase Sarcoplasmic/Endoplasmic Reticulum Ca2 + Transporting 3 Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Introduction Rheumatoid Arthritis (RA) is a chronic autoimmune disease, characterized by persistent inflammation of the synovial membrane, leading to joint destruction, cartilage loss, and bone erosion [ 1 ]. This progressive joint damage often results in significant disability and a reduced quality of life [ 2 ]. Moreover, beyond joint involvement, RA can also affect extra-articular systems, highlighting its systemic nature [ 3 ]. Despite advances in therapeutic options, including disease-modifying anti-rheumatic drugs (DMARDs), RA continues to place a substantial physical and economic burden on patients and healthcare systems worldwide [ 4 , 5 ]. Therefore, early diagnosis and personalized treatment plans remain critical for improving outcomes and reducing disease-related complications. Endoplasmic reticulum (ER) stress plays an important role in RA pathology that links cellular stress to immune dysfunction [ 6 ]. ER stress occurs when the ER's protein-folding capacity is overwhelmed, activating the unfolded protein response (UPR) to restore homeostasis [ 7 , 8 ]. In RA, reactive oxygen species (ROS), hypoxia, nutrient deprivation, and inflammatory cytokines contributes to chronic ER stress [ 9 ]. This persistent ER stress enhances the survival and proliferation of fibroblast-like synoviocytes (FLS), driving pannus formation and joint damage [ 10 ]. Furthermore, the UPR exacerbates inflammation by increasing cytokine production and altering immune cell activity, amplifying systemic and local immune dysregulation in RA [ 11 ]. This highlights ER stress and the consequent cellular response as a key contributor to pathology and a potential therapeutic target. Despite this potential, the role of ER stress in RA remains underexplored, particularly through the lens of bioinformatics, which is increasingly leveraged to uncover prognostic and predictive biomarkers at the gene level [ 12 ]. FLS, which have an aggressive phenotype, play an important role in the pathological processes of RA [ 13 ]. FLS not only produce the extracellular matrix but also secrete inflammatory cytokines and proteases that contribute to disease pathogenesis and perpetuation [ 14 ]. For example, FLS secrete inflammatory cytokines such as IL-6 to recruit immune cells, thereby promoting joint inflammation [ 15 ]. Moreover, ITGA5 + synovial fibroblasts secrete TGF-β1 to induce the differentiation of naive CD4 + T cells into CXCL13 hi PD-1 hi peripheral helper T cells, thereby remodeling the pro-inflammatory microenvironment and facilitating the progression of RA [ 16 ]. Although several studies have indicated that ER stress is involved in the activation of FLS, the molecular mechanisms by which ER stress induces this activation are largely unknown. In this study, bioinformatics approaches were used to identify ER stress response-related genes (ERSRGs) in synovial tissues of RA patients using Gene Expression Omnibus (GEO) database datasets. Machine learning algorithms were applied to refine and prioritize key biomarkers. Pathway enrichment analyses further explored their functional roles. We identified ATPase Sarcoplasmic/Endoplasmic Reticulum Ca2 + Transporting 3 (ATP2A3) as a potential diagnostic biomarker of ER stress in RA. ATP2A3 can promote the secretion of inflammatory cytokines in RA FLS through activation of the stimulator of interferon genes (STING) signaling pathway. Methods & Materials Datasets Collection and Preprocessing Six publicly available gene expression datasets (GSE55235, GSE55457, GSE77298, GSE1919, GSE236924, and GSE89408) were retrieved from the GEO database ( https://www.ncbi.nlm.nih.gov/geo/ ). Among these, GSE55235, GSE55457, and GSE77298, which contained balanced numbers of RA and healthy control samples, were designated as the training set. The remaining datasets (GSE1919, GSE236924, and GSE89408) were reserved for external validation. Details of the sample sizes and platforms used in each dataset are summarized in Supplementary Table 1. The datasets originated from various microarray platforms: GSE55235 and GSE55457 utilized the GPL96 ([HG-U133A] Affymetrix Human Genome U133A Array), GSE77298 and GSE236924 were based on GPL570 ([HG-U133_Plus_2] Affymetrix Human Genome U133 Plus 2.0 Array), and GSE1919 used GPL91 ([HG_U95A] Affymetrix Human Genome U95A Array). GSE89408 was generated using high-throughput sequencing on the GPL11154 (Illumina HiSeq 2000) platform. Raw gene expression data were preprocessed using custom Perl scripts to extract and annotate expression matrices. To ensure comparability, only the intersection of common genes across all datasets was retained. For each dataset, probes were averaged to gene-level expression using the avereps function, and sample names were standardized to include dataset identifiers. To correct for platform and batch-related variability, expression values were normalized using the limma package in R, and batch effects were adjusted using the ComBat algorithm implemented in the sva package. Each sample was assigned a batch label based on its dataset of origin. The ComBat function applied empirical Bayes methods to model batch effects, assuming parametric prior distributions and preserving biological variation. Principal component analysis (PCA) was conducted before and after ComBat adjustment to assess data structure and batch correction efficacy. The PCA was performed using the prcomp function on the transposed expression matrix. Sample clustering was visualized with ggpubr::ggscatter, displaying 95% confidence ellipses for each batch. Prior to correction, clear batch-specific clustering was observed. After ComBat correction, samples distributed more homogenously, indicating successful mitigation of batch effects. Differential expression analysis Differential expression analysis was conducted using the limma R package to identify differentially expressed genes (DEGs) between risk subgroups. The criteria for defining the DEGs was log fold change (logFC) = 0.585 and false discovery rate (FDR) < 0.05. Endoplasmic reticulum stress response-related genes The ERSRGs were sourced from the Gene Ontology Consortium pathway “GOBP_RESPONSE_TO_ENDOPLASMIC_RETICULUM_STRESS” available through the Molecular Signatures Database (MSigDB) ( https://www.gsea-msigdb.org/gsea/msigdb ). This gene set contains a total of 267 ERSRGs, which were extracted for further analysis. Pathway enrichment analysis Functional annotation of the identified DEGs was performed through Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses to investigate their biological roles and associated pathways. These analyses were conducted using the "clusterProfiler" R package [ 17 ]. Additionally, GO enrichment information for individual ER stress biomarkers was retrieved from the STRING database using its "Analysis" module ( https://string-db.org/ ). Protein-protein interaction analysis The interaction network of ER stress biomarkers associated with RA was analyzed using the STRING database ( https://string-db.org/ ) [ 18 ]. A medium confidence threshold (score ≥ 4.00) was applied to ensure reliable interactions. Active interaction sources included evidence from text mining, experimental data, curated databases, co-expression patterns, genomic neighborhood, gene fusion events, and co-occurrence. This comprehensive analysis facilitated the identification of functional connections and potential regulatory networks involving ER stress biomarkers. Consensus clustering To stratify patients based on the expression patterns of ERSRGs, an unsupervised clustering approach using consensus clustering was applied. This method systematically evaluated patient groupings across a range of cluster numbers (k), varying from 2 to 11. The optimal cluster number was determined by assessing the consensus stability, utilizing cumulative distribution function (CDF) plots to identify the value of k that offered the most robust clustering solution. The analysis was conducted using the Consensus ClusterPlus R package, set to 1000 iterations to ensure reliability and reproducibility of the clustering results [ 19 ]. Discovery of ER stress biomarkers using machine learning algorithms To ensure robust identification of ER stress biomarkers in RA, we employed a multi-model machine learning framework using Least Absolute Shrinkage and Selection Operator (LASSO) regression, Support Vector Machine-Recursive Feature Elimination (SVM-RFE), and Random Forest (RF). Each algorithm incorporated appropriate cross-validation, model diagnostics, and feature selection techniques to prevent overfitting and maximize generalizability. Model Selection, Assumptions, and Diagnostics LASSO regression was conducted using the glmnet R package [ 20 ]. The gene expression matrix was transposed such that rows represented samples and columns represented genes, with binary labels assigned as control = 0 and RA = 1. The optimal penalty parameter (lambda) was selected by 10-fold cross-validation using cv.glmnet, minimizing binomial deviance. Genes with non-zero coefficients at lambda.min were retained. To quantify diagnostic performance and uncertainty, we computed sample-wise risk scores as the weighted sum of selected gene expressions and coefficients, then evaluated model performance using Recursive operator curve (ROC) and bootstrapped AUC (n = 1000) via the pROC R package. SVM-RFE was performed using the caret and e1071 packages with an RBF kernel (svmRadial) [ 21 ]. A 10-fold cross-validation strategy was applied using rfeControl, evaluating model accuracy across feature subsets of increasing size. The optimal feature set was selected based on maximal cross-validated accuracy. Genes selected through this procedure were exported, and diagnostic validity was assessed via ROC-AUC analysis using predicted labels from the final model. This approach captures non-linear relationships and is particularly suitable for high-dimensional biological data. RF analysis was implemented using the randomForest package with 500 decision trees. The expression matrix was again transposed, and group labels were extracted from sample names. The optimal number of trees was identified by selecting the point of minimum error from the out-of-bag (OOB) error rate. Feature importance was measured using the Mean Decrease Gini index. Genes were ranked accordingly, and the top 30 most informative features were retained. A variable importance plot was generated to visualize key contributors to classification. Expression profiles of these top genes were exported for downstream validation and visualization. Together, these three complementary approaches enabled robust and reproducible feature selection. LASSO ensured sparse linear modeling with embedded regularization; SVM-RFE captured non-linear margins with recursive feature filtering; and Random Forest leveraged ensemble learning and internal validation via OOB error estimation. Final candidate biomarkers were identified by intersecting outputs across all three methods using a Venn diagram approach, ensuring selection of consistently predictive features across distinct algorithms. Recursive operator curve analysis The diagnostic performance of the selected biomarkers was evaluated using ROC analysis, with AUC values calculated via the pROC R package. This multi-algorithm approach ensured reliable identification of biomarkers with strong diagnostic potential for ERSRGs-RA. Clinical specimens We collected synovial tissue samples from 9 RA patients and 3 healthy controls between July 2023 and August 2024 at the Department of Orthopedic Surgery, Nanfang Hospital, Southern Medical University. Informed consent was obtained from the patients and healthy controls, and approval was granted by the internal review and ethics boards of the Southern Medical University. Immunohistochemistry Immunohistochemistry (IHC) analysis was conducted on 4 µm thick sections of formalin-fixed, paraffin-embedded synovial tissue from RA patients and healthy controls. Tissue sections were deparaffinized with xylene and ethanol, followed by antigen retrieval using citrate buffer (pH 6.0) through microwave heating. Endogenous peroxidase activity was quenched with 0.3% hydrogen peroxide, and the sections were blocked with 5% bovine serum albumin in phosphate-buffered saline (PBS). Primary antibodies for ATP2A3 (Proteintech, #13619-1-AP, rabbit, 1:50), MARCKS (Proteintech, #10004-2-Ig, rabbit, 1:50), and UBE2J1 (Affinity, #MG805029, rabbit, 1:100) were applied, and samples were incubated overnight at 4°C. Biotinylated secondary antibodies were then applied at room temperature, and staining was visualized using a 3,5-diaminobenzidine (DAB) substrate with hematoxylin counterstaining. Staining intensity was scored on a 0–3 scale (0: negative, 1: weak, 2: moderate, 3: strong), and positive cell frequency on a 0–4 scale (0: 75%). The IHC score was calculated as the product of intensity and frequency. For heterogeneous tissues, individual regions were scored and combined for the final result. Culture of fibroblast-like synoviocytes The fresh human synovial tissues were digested to isolate FLS, which were subsequently cultured in Dulbecco's Modified Eagle Medium (DMEM; Thermo Fisher Scientific, Waltham, MA) containing 10% fetal bovine serum (FBS; Gibco, Grand Island, NY). Cells from passages four to six were employed in all experiments. Homogeneity and purity (> 98%) of the FLS population were confirmed via vimentin-specific immunofluorescence staining. Western blot analysis Primary antibodies anti-STING, anti-p-STING, anti-IRF3, anti-p-IRF3, anti-TBK1, anti-p-TBK1, anti-p65, anti-p-p65 were purchased from Cell Signaling Technology (Beverly, MA, USA). Primary antibody anti-β-actin were obtained from Protein Tech Group (Rosemont, IL). Briefly, RA-FLS were lysed in ice-cold radioimmunoprecipitation assay (RIPA) buffer (BestBio, Shanghai). Proteins were resolved by 10% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto polyvinylidene difluoride (PVDF) membranes (Millipore, USA). Immunoreactive bands were quantified using Quantity One software (Bio-Rad). siRNA and adenovirus transfection The siRNAs target for ATP2A3 were purchased from RiboBio (Guangzhou, China). Adenovirus vector containing the coding sequences for ATP2A3 were obtain form Genepharma (Shanghai, China). The transfection process was performed using Lipofectamine 3000 (Hanbio Biotechnology, Shanghai, China) according to the manufacturer’s protocol. The expression level of ATP2A3 were confirmed by Western Blot Analysis. Enzyme-linked immunosorbent assay (ELISA) The levels of TNFα, IFNα, CXCL10, IL-1β, IL-6, IL-8 in the supernatants of RA-FLS were detected by Enzyme-linked immunosorbent assay (ELISA). The ELISA kits were purchased from Boster Biological Technology (Pleasanton, CA, USA). Cells were seeded into 6-well plates and grown to 80% confluence. They were then cultured in fresh serum-free media, and supernatant was harvested at 24 hours. Measurements were according to the manufacturer's instructions. Statistical analysis Statistical analyses were conducted using R software (v4.0.3, http://www.r-project.org ). Data are presented as the means standard deviation and were analyzed using a Student t test. For comparisons of gene expression/enrichment/IHC scores between two groups, the non-parametric Wilcoxon Rank-Sum Test (equivalent to the Mann-Whitney U Test) was applied. Correlation analyses were performed using either Spearman's rank/Pearson's correlation tests. p < 0.05 was considered statistical significance. Results Differential expression and functional analysis in RA The research strategy and methodology employed in this study are illustrated in Fig. 1 A. Key steps included: performing differential expression analysis; identifying diagnostic and predictive genes associated with response to ER stress; constructing molecular subtypes; conducting pathway enrichment and validating findings through external datasets; and experimental approaches. Transcriptomic data from RA patients were obtained from three GEO microarray datasets (GSE55235, GSE55457, GSE77298). After merging the datasets and performing batch correction, a meta-cohort was created. Principal component analysis (PCA) before batch correction revealed distinct clustering of samples by dataset (Fig. 1 B). Post-ComBat PCA showed a significant reduction in inter-dataset variance, indicating successful harmonization (Fig. 1 C). Differential expression analysis was conducted between RA patients (n = 39) and healthy controls (n = 27), identifying 1433 DEGs, including 872 upregulated and 561 downregulated genes based on the criteria of logFC ≥ 0.585 and FDR ≤ 0.05 (Fig. 1 D). A heatmap displayed the top 50 DEGs, highlighting distinct expression patterns (Fig. 1 E). Pathway enrichment analysis revealed significant functional implications of the DEGs. The enriched pathways highlighted key mechanisms underlying RA pathology (Fig. 1 F-G). These include immune response activation (e.g., Th17 cell differentiation, leukocyte adhesion, chemokine signaling), osteoclast differentiation, and cell adhesion molecules, all of which contribute to chronic inflammation and tissue damage in RA. Additionally, pathways such as Epstein-Barr virus infection and hematopoietic cell lineage suggest a link between viral triggers, immune dysregulation, and disease progression in RA. Identify ERSRGs and ER stress-based molecular subtypes in RA The enrichment of the ERSRGs pathway obtained from Gene Ontology Consortium (GOBP_RESPONSE_TO_ENDOPLASMIC_RETICULUM_STRESS) corresponded to the upregulated DEGs in RA (Fig. 2 A). An intersection of DEGs (n = 1433) with ERSRGs (n = 267) revealed 17 overlapping ERSRGs (Fig. 2 B). The heatmap shows their expression patterns, highlighting 13 genes upregulated and 4 downregulated in RA (Fig. 2 C). This result underscores the involvement of ERSRGs in RA pathology. As these 17 ERSRGs were significantly differentially expressed in RA, we employed consensus clustering to obtain molecular subtypes of RA and identify the implications of ER stress. The RA patients were categorized into two subgroups based on the expression of these 17 genes as indicated by the observation of consensus matrix and CDF matrix at K = 2 (Fig. 2 D-E). The 39 RA samples were stratified into two clusters: Cluster A (n = 25) and Cluster B (n = 14). The heatmap illustrates the similarity in the expression of upregulated and downregulated genes between the clusters to the expression pattern in control versus RA patients (Fig. 2 F). Cluster B exhibited elevated expression of JUN, PPP1R15A, ATF3, and PIK3R1—genes that were also highly expressed in healthy controls. This suggests that the ER stress profile in Cluster B patients may share some features with that of healthy individuals. Analysis of the features driving cluster separation identified 12 key genes: 2 that were downregulated (JUN, PPP1R15A) and 10 that were upregulated (Fig. 2 G). Pathway enrichment analysis of the DEGs between the clusters also highlighted similar functional differences seen in control versus RA, with enhanced enrichment of immune response activation and inflammatory functions (Fig. 2 H-I). Identification of ER stress diagnostic biomarkers To explore the potential diagnostic role of ERSRGs in distinguishing RA patients from healthy individuals, we employed a robust multi-algorithm approach. This involved three machine learning techniques: LASSO regression, SVM-RFE, and RF. LASSO regression selected 10 genes from an initial pool of 17 candidates (Fig. 3 A-B). SVM-RFE identified 8 ERSRGs with high precision (0.881) and minimal root mean square error (RMSE = 0.119) (Fig. 3 C-D). Random Forest analysis ranked five genes based on an importance score threshold of > 2 (Fig. 3 E-F). Integration of these results revealed three key genes—ATP2A3, MARCKS (Myristoylated Alanine-Rich C-Kinase Substrate), and UBE2J1 (ubiquitin conjugating enzyme E2 J1)—as significant biomarkers (Fig. 3 G). ROC curve analysis validated their diagnostic accuracy, with each gene achieving an AUC greater than 0.80 (Fig. 3 H-J). These results underscore their potential in differentiating RA patients from healthy individuals. Functional and immunological relevance of ER stress biomarkers To enhance our understanding of the functional roles of these three ERSRGs biomarkers, we first identified their protein-protein interaction networks using the STRING database, followed by the pathway enrichment analysis. GO analysis of the ATP2A3 and MARCKS networks indicated their critical role in calcium homeostasis with slightly opposite functions. ATP2A3 was associated with positive regulation of calcium transmembrane transport (Fig. 4 A-B) and MARCKS was associated with negative regulation of calcium ion export across plasma membrane and depletion of calcium ion (Fig. 4 C-D). Calcium is essential for maintaining ER homeostasis, and a significant depletion of Ca2 + in the ER serves as a potent inducer of ER stress [ 22 ]. Lastly, UBE2J1 network was enriched in ER-associated degradation (ERAD) pathway, essential for managing misfolded proteins under ER stress (Fig. 4 E-F). All three biomarkers are critical for ER stress and cellular response to ER stress. The association of these genes with immune cell infiltration varied: ATP2A3 was linked to increased activated CD4 memory T cells and plasma cells (Supplementary Fig. 1A); MARCKS was associated with neutrophils, plasma cells, and CD8 T cells (Supplementary Fig. 1B); while UBE2J1 was associated with plasma cells and gamma delta T cells (Supplementary Fig. 1C). These findings suggest that each gene may influence different immune cell subsets, contributing to the inflammatory microenvironment characteristic of RA. External validation of ER stress diagnostic biomarkers To validate the diagnostic utility of ER stress biomarkers for RA, we analyzed their expression across three independent cohorts from the GEO database. The datasets included GSE236924 (N = 43; 7 healthy controls, 36 RA), GSE1919 (N = 8; 4 healthy controls, 4 RA), and GSE89408 (N = 179; 27 healthy controls, 152 RA). The datasets were combined following normalization and batch effect correction. PCA conducted prior to batch correction demonstrated clear clustering of samples according to their originating datasets (Fig. 4 G). After applying the ComBat method, PCA revealed a marked decrease in inter-dataset variability, reflecting effective data harmonization (Fig. 4 H). ATP2A3, MARCKS, and UBE2J1 were significantly upregulated in RA samples compared to healthy controls (Fig. 4 I). ROC curve analysis for the meta-cohort revealed that ATP2A3, MARCKS, and UBE2J1 achieved AUC values of 0.650, 0.874, and 0.958, respectively, demonstrating their potential diagnostic value (Fig. 4 J). In individual datasets, MARCKS consistently showed upregulation across all three cohorts, with AUC values ranging from 0.666 to 1.000 (Supplementary Fig. 2A-C). UBE2J1 was significantly upregulated in two cohorts, achieving AUC values between 0.800 and 0.983 (Supplementary Fig. 2A-C). ATP2A3 displayed upregulation in one cohort, with AUC values ranging from 0.560 to 0.909 (Supplementary Fig. 2A-C). These findings underscore the strong diagnostic potential of MARCKS and UBE2J1 as biomarkers for RA, while ATP2A3 demonstrated a more variable but noteworthy performance. This validation supports the clinical relevance of these biomarkers in distinguishing RA from healthy individuals. ATP2A3 promotes inflammatory cytokine secretion in RA FLS We performed IHC on synovial tissue samples from nine RA patients and three healthy controls to validate the diagnostic utility of ER stress biomarkers. We found that the protein expression of ATP2A3, MARCKS, and UBE2J1 was significantly increased in RA synovial tissues compared to healthy controls (Fig. 5 A-B). These results suggested that ATP2A3, MARCKS, and UBE2J1 are involvement in ER stress pathways and RA pathology. Among these ER stress diagnostic biomarkers, ATP2A3 has attracted our attention. ATP2A3 belongs to the sarco/endoplasmic reticulum Ca2-ATPase (SERCA) family, which can be induced by ER stress [ 23 ]. It plays an important role in maintaining intracellular Ca2 + homeostasis and ER stress-associated apoptosis [ 24 ]. Moreover, it has been reported that ATP2A3 involved in T lymphocyte activation and cytokine secretion in head and neck squamous cell carcinoma [ 25 ]. We suspected that ATP2A3 might be involved in inflammatory cytokine secretion in RA FLS. Therefore, we selected ATP2A3 for further investigation. To investigate the role of ATP2A3 in the inflammatory response of RA FLS, we inhibited its expression using siRNA and overexpressed it with an adenoviral vector (Fig. 5 C). We then evaluated the effect of ATP2A3 modulation on inflammatory cytokine secretion. The results showed that inhibition of ATP2A3 significantly decreased the secretion of inflammatory cytokines, including TNFα, IFNα, CXCL10, IL-1β, IL-6, and IL-8, in RA FLS (Fig. 5 D). Conversely, overexpression of ATP2A3 significantly upregulated the secretion of these cytokines (Fig. 5 E). These results indicated that ATP2A3 promotes inflammatory cytokine secretion in RA FLS. ATP2A3 promotes inflammatory cytokine secretion by activating the STING signaling The SERCA family, which includes ATP2A1, ATP2A2, and ATP2A3, is composed of members that possess conserved structures and exhibit analogous functions [ 26 ]. Previous studies have indicated that ATP2A2 directly interacts with STING and regulates its phosphorylation, thereby activating STING-induced inflammatory response [ 27 ]. Given this functional conservation within the SERCA family, we hypothesized that ATP2A3 regulates STING phosphorylation, thereby activating STING-induced inflammatory response. Upon activation, STING recruits and activates the kinases TANK-binding kinase 1 (TBK1) and inhibitor of kappa B kinase (IKK); this leads to the phosphorylation of interferon regulatory factor 3 (IRF3) and nuclear factor kappa-B (NF-κB), thereby driving the expression of downstream inflammatory cytokines [ 28 ]. Consistently, we found that inhibition of ATP2A3 significantly decreased the phosphorylation of STING, TBK1, IRF3, p65 (Fig. 6 A). In contrast, overexpression of ATP2A3 enhanced the phosphorylation of STING, TBK1, IRF3, and p65 (Fig. 6 A). These results indicate that ATP2A3 enhances STING signaling pathway activation. To investigate whether the effect of ATP2A3 on inflammatory cytokine secretion is dependent on STING signaling in RA FLS, we overexpressed ATP2A3 and then treated the cells with the STING inhibitor SN-011. SN-011 is a STING inhibitor that can effectively inhibit STING phosphorylation [ 29 ]. We found that SN-011 blocked the phosphorylation of STING, TBK1, IRF3, and p65 induced by ATP2A3 overexpression (Fig. 6 B). Importantly, the inhibitor also blocked the upregulation of inflammatory cytokines (TNFα, IFNα, CXCL10, IL-1β, IL-6, and IL-8) induced by ATP2A3 overexpression (Fig. 6 C-H). These results demonstrate that ATP2A3 promotes inflammatory cytokine secretion by activating the STING signaling pathway in RA FLS. Discussion ER stress contributes to RA by disrupting cellular homeostasis and amplifying immune dysfunction [ 6 ]. In this study, we explored the role of ERSRGs in RA. Seventeen ERSRGs were identified as significantly dysregulated in RA, enabling consensus clustering to classify patients into two distinct subtypes with unique functional and immunological characteristics. Through machine learning, three key biomarkers—ATP2A3, MARCKS, and UBE2J1—were identified. Importantly, we found that ATP2A3 promotes the secretion of inflammatory cytokines in RA FLS. Mechanistically, ATP2A3 enhances the phosphorylation of STING, thereby activating the STING signaling pathway and leading to increased inflammatory cytokines production. This work provides novel insights into the molecular underpinnings of RA and highlights ATP2A3 as a potential target for therapeutic intervention. ER stress is implicated in a broad spectrum of diseases, including cancer, cardiovascular conditions, neurodegenerative disorders, and autoimmune diseases [ 30 , 31 ]. In RA, the synovial microenvironment—characterized by hypoxia, nutrient deprivation, and an abundance of pro-inflammatory cytokines—provides conditions that could trigger ER stress [ 32 , 33 ]. A detailed examination of ER stress could provide critical insights into the molecular mechanisms driving RA and identify new diagnostic biomarkers as well as target for RA therapy. We identified 17 dysregulated ERSRGs in RA. Essentially, the upregulation of 13 ERSRGs, including the three ER stress biomarkers (ATP2A3, MARCKS, and UBE2J1) and downregulation of 4 ERSRGs (JUN, ATF3, PIK3R1 and PPP1R15A) differentiated the RA from healthy controls. Molecular heterogeneity was evident when two molecular subtypes based on the expression pattern of these 17 ERSRGs were obtained. Interestingly, 12 of the 17 ERSRGs (10 upregulated and 2 downregulated) contributed to the cluster differentiation with rather consistent expression pattern. Future studies are warranted to determine whether these subtypes correlate with distinct clinical manifestations or treatment responses. Altogether, our findings highlight that ER stress induces specific alterations in ERSRGs, which likely contribute to RA pathogenesis. Among the upregulated ERSRGs, ATP2A3, MARCKS, and UBE2J1 were identified as the ER stress biomarkers with enhanced diagnostic potential for RA. ATP2A3 was specifically selected for further investigation. ATP2A3 is an important member of the SERCA family and can be induced by ER stress [ 23 ]. Previous study indicated that ATP2A3 plays a critical role in calcium homeostasis by catalyzing ATP hydrolysis to transport calcium from the cytosol into the sarcoplasmic reticulum lumen, facilitating processes such as muscle excitation and contraction [ 34 ]. But, the role of ATP2A3 in RA is largely unknown. Here, we identified that ATP2A3 enhances inflammatory cytokine production by activating the STING signaling pathway in RA FLS. To the best of our knowledge, this is the first study to identify ATP2A3 as a key gene that regulates inflammatory cytokine secretion in RA FLS. Our study highlights that the ER stress biomarker ATP2A3 is a key regulator in the inflammatory activation of RA FLS. However, the roles of other ER stress biomarkers, such as MARCKS and UBE2J1, in RA require further investigation in future studies. RA is characterized by inflammation of the joint synovium and the infiltration of multiple immune cells [ 35 ]. The STING signaling pathway, which is widely expressed in both immune and non-immune cells, plays a pivotal role in regulating inflammatory responses [ 36 ]. Numerous studies have demonstrated the implication of the STING signaling pathway in RA pathogenesis, highlighting its potential as a therapeutic target [ 37 ]. In present study, we observed a significant upregulation of ATP2A3 in RA synovial tissues compared to healthy controls. Importantly, we discovered that ATP2A3 enhances STING phosphorylation, thereby activating the STING signaling pathway and its downstream effectors, IRF3 and NF-κB. This cascade ultimately leads to increased secretion of inflammatory cytokines from RA-FLS. This study improves our understanding of the mechanisms underlying the activation of the inflammatory response in RA and offers a promising target for RA treatment. ER stress biomarker ATP2A3 serves as a link between ER stress, the STING signaling pathway, and inflammation of the joint synovium of RA. However, this study also has several limitations that should be acknowledged. First, the integration of multiple GEO datasets introduces heterogeneity due to differences in platforms, sample processing, and patient populations, which may affect the consistency of the findings. Although we applied the ComBat algorithm to correct for batch effects and assessed correction using PCA, residual inter-dataset variability may still persist. Second, the lack of prospective validation in independent clinical cohorts limits the generalizability and clinical translation of the identified biomarkers. Third, the exploratory nature of our analysis, primarily driven by machine learning algorithms, raises the possibility of overfitting despite cross-validation and robust feature selection methods. Additionally, the limited availability of clinical metadata across datasets restricted stratified analyses and precluded the assessment of biomarker associations with disease subtypes or treatment response. Conclusions This study provides valuable insights into the role of ERSRGs in RA. We establish ATP2A3 as a promising diagnostic biomarker for ER stress in RA and demonstrate that it promotes the secretion of inflammatory cytokines in RA FLS through STING pathway activation. ATP2A3 represents a potential therapeutic target for RA treatment. Abbreviations RA Rheumatoid arthritis ER Endoplasmic reticulum ERSRGs ER stress-related genes ATP2A3 ATPase Sarcoplasmic/Endoplasmic Reticulum Ca2 + Transporting 3 MARCKS Myristoylated Alanine-Rich C-Kinase Substrate UBE2J1 Ubiquitin conjugating enzyme E2J1 LASSO Least Absolute Shrinkage and Selection Operator SVM-RFE Support Vector Machine-Recursive Feature Elimination RF Random Forest IHC Immunohistochemistry DMARDs Disease-modifying anti-rheumatic drugs FLS Fibroblast-like synoviocytes ACPAs Anti-citrullinated protein antibodies UPR Unfolded protein response ROS Reactive oxygen species PCA Principal component analysis GO Gene Ontology ROC Recursive operator curve DEGs Differentially expressed genes MSigDB Molecular Signatures Database ERAD ER-associated degradation JUN Jun proto-oncogene PPP1R15A Protein phosphatase 1 regulatory subunit 15A ATF3 Activating transcription factor 3 PIK3R1 Phosphoinositide-3-kinase regulatory subunit 1 ELISA Enzyme-linked immunosorbent assay STING Stimulator of interferon genes TBK1 TANK binding kinase 1 IRF3 interferon regulatory factor 3 SERCA Sarco/endoplasmic reticulum Ca2-ATPase KEGG Kyoto Encyclopedia of Genes and Genomes. Declarations Ethics approval and consent to participate The studies involving human participants were reviewed and approved by the Internal Review and Ethics Boards of the Nanfang Hospital, Southern Medical University. The patients provided written informed consent to participate in the study. The study adhered to the ethical principles outlined in the Helsinki Declaration. Consent for publication Not applicable Clinical trial number Not applicable Competing interests Authors declare no conflict of interest Funding This work was supported by the Natural Science Foundation of Guangdong Province of China [2023A1515111036] and the President Foundation of Nanfang Hospital, Southern Medical University [grant number 2023B042]. Authors' contributions J.Z. and W.L. conceived and designed the study. Y.C. and D.G. drafted the manuscript and created the figure. S.Z. and D.Z. performed IHC staining and Western blot analysis. All authors reviewed and approved the final version of the manuscript. Acknowledgements We thank the patients with RA and healthy donors who contributed to this study. Availability of Data and Materials The datasets used in this study are available from the GEO database (https://www.ncbi.nlm.nih.gov/geo/). Raw data from our experiments can be requested from the corresponding author, Weinan Lai, upon reasonable request. References Di Matteo A, Bathon JM, Emery P (2023) Rheumatoid arthritis. Lancet 402(10416):2019–2033 Gravallese EM, Firestein GS (2023) Rheumatoid Arthritis - Common Origins, Divergent Mechanisms. NEW ENGL J MED 388(6):529–542 Conforti A, Di Cola I, Pavlych V, Ruscitti P, Berardicurti O, Ursini F, Giacomelli R, Cipriani P (2021) Beyond the joints, the extra-articular manifestations in rheumatoid arthritis. AUTOIMMUN REV 20(2):102735 Prasad P, Verma S, Surbhi, Ganguly NK, Chaturvedi V, Mittal SA (2023) Rheumatoid arthritis: advances in treatment strategies. MOL CELL BIOCHEM 478(1):69–88 Konzett V, Aletaha D (2024) Management strategies in rheumatoid arthritis. NAT REV RHEUMATOL 20(12):760–769 Rahmati M, Moosavi MA, McDermott MF (2018) ER Stress: A Therapeutic Target in Rheumatoid Arthritis? TRENDS PHARMACOL SCI 39(7):610–623 Acosta-Alvear D, Harnoss JM, Walter P, Ashkenazi A (2024) Homeostasis control in health and disease by the unfolded protein response. NAT REV MOL CELL BIO Wiseman RL, Mesgarzadeh JS, Hendershot LM (2022) Reshaping endoplasmic reticulum quality control through the unfolded protein response. MOL CELL 82(8):1477–1491 Miglioranza Scavuzzi B, Holoshitz J: Endoplasmic Reticulum Stress, Oxidative Stress, and, Diseases R (2022) ANTIOXIDANTS-BASEL 11(7) Ao Q, Hu H, Huang Y (2024) Ferroptosis and endoplasmic reticulum stress in rheumatoid arthritis. FRONT IMMUNOL 15:1438803 Navid F, Colbert RA (2017) Causes and consequences of endoplasmic reticulum stress in rheumatic disease. NAT REV RHEUMATOL 13(1):25–40 Li J, Cui J, Wu L, Liu Y, Wang Q (2023) Machine learning and molecular subtype analyses provide insights into PANoptosis-associated genes in rheumatoid arthritis. ARTHRITIS RES THER 25(1):233 Qian H, Deng C, Chen S, Zhang X, He Y, Lan J, Wang A, Shi G, Liu Y (2024) Targeting pathogenic fibroblast-like synoviocyte subsets in rheumatoid arthritis. ARTHRITIS RES THER 26(1):103 Nygaard G, Firestein GS (2020) Restoring synovial homeostasis in rheumatoid arthritis by targeting fibroblast-like synoviocytes. NAT REV RHEUMATOL 16(6):316–333 Symons RA, Colella F, Collins FL, Rafipay AJ, Kania K, McClure JJ, White N, Cunningham I, Ashraf S, Hay E et al (2022) Targeting the IL-6-Yap-Snail signalling axis in synovial fibroblasts ameliorates inflammatory arthritis. ANN RHEUM DIS 81(2):214–224 Zheng L, Gu M, Li X, Hu X, Chen C, Kang Y, Pan B, Chen W, Xian G, Wu X et al (2025) ITGA5(+) synovial fibroblasts orchestrate proinflammatory niche formation by remodelling the local immune microenvironment in rheumatoid arthritis. ANN RHEUM DIS 84(2):232–252 Wu T, Hu E, Xu S, Chen M, Guo P, Dai Z, Feng T, Zhou L, Tang W, Zhan L et al (2021) clusterProfiler 4.0: A universal enrichment tool for interpreting omics data. Innov (Camb) 2(3):100141 Szklarczyk D, Gable AL, Nastou KC, Lyon D, Kirsch R, Pyysalo S, Doncheva NT, Legeay M, Fang T, Bork P et al (2021) The STRING database in 2021: customizable protein-protein networks, and functional characterization of user-uploaded gene/measurement sets. NUCLEIC ACIDS RES 49(D1):D605–D612 Wilkerson MD, Hayes DN (2010) ConsensusClusterPlus: a class discovery tool with confidence assessments and item tracking. Bioinformatics 26(12):1572–1573 Friedman J, Hastie T, Tibshirani R (2010) Regularization Paths for Generalized Linear Models via Coordinate Descent. J STAT SOFTW 33(1):1–22 Sanz H, Valim C, Vegas E, Oller JM, Reverter F (2018) SVM-RFE: selection and visualization of the most relevant features through non-linear kernels. BMC Bioinformatics 19(1):432 Pontisso I, Ornelas-Guevara R, Chevet E, Combettes L, Dupont G (2024) Gradual ER calcium depletion induces a progressive and reversible UPR signaling. PNAS Nexus 3(6):pgae229 Weichert M, Guirao-Abad J, Aimanianda V, Krishnan K, Grisham C, Snyder P, Sheehan A, Abbu RR, Liu H, Filler SG et al (2020) Functional Coupling between the Unfolded Protein Response and Endoplasmic Reticulum/Golgi Ca(2+)-ATPases Promotes Stress Tolerance, Cell Wall Biosynthesis, and Virulence of Aspergillus fumigatus. MBIO 11(3) Zhang Y, Li F, Liu L, Jiang H, Hu H, Du X, Ge X, Cao J, Wang Y (2019) Salinomycin triggers endoplasmic reticulum stress through ATP2A3 upregulation in PC-3 cells. BMC Cancer 19(1):381 Tao Z, Yang W, Zhu W, Wang L, Li KY, Guan X, Su Y (2024) A neural-related gene risk score for head and neck squamous cell carcinoma. ORAL DIS 30(2):477–491 Chemaly ER, Troncone L, Lebeche D (2018) SERCA control of cell death and survival. Cell Calcium 69:46–61 Yang X, Lv L, Zhang Y, Zhang Z, Zeng S, Zhang X, Wang Q, Dorf M, Li S, Fu B (2025) ATP2A2 regulates STING1/MITA-driven signal transduction including selective autophagy. AUTOPHAGY :1–16 Zhang Z, Zhang C (2025) Regulation of cGAS-STING signalling and its diversity of cellular outcomes. NAT REV IMMUNOL 25(6):425–444 Hong Z, Mei J, Li C, Bai G, Maimaiti M, Hu H, Yu W, Sun L, Zhang L, Cheng D et al (2021) STING inhibitors target the cyclic dinucleotide binding pocket. P NATL ACAD SCI USA 118(24) Marciniak SJ, Chambers JE, Ron D (2022) Pharmacological targeting of endoplasmic reticulum stress in disease. NAT REV DRUG DISCOV 21(2):115–140 Chen X, Shi C, He M, Xiong S, Xia X (2023) Endoplasmic reticulum stress: molecular mechanism and therapeutic targets. SIGNAL TRANSDUCT TAR 8(1):352 Kondo N, Kuroda T, Kobayashi D (2021) Cytokine Networks in the Pathogenesis of Rheumatoid Arthritis. INT J MOL SCI 22(20) Jang S, Kwon E, Lee JJ (2022) Rheumatoid Arthritis: Pathogenic Roles of Diverse Immune Cells. INT J MOL SCI 23(2) Yu J, Wang H, Ding M, Zhao X, Zhang X (2025) Current and Future Landscape of SERCAs' Functions in Non-Excitatory Cells and Diseases. BioEssays :e70029 Alivernini S, Firestein GS, McInnes IB (2022) The pathogenesis of rheumatoid arthritis. Immunity 55(12):2255–2270 Zhang B, Xu P, Ablasser A (2025) Regulation of the cGAS-STING Pathway. ANNU REV IMMUNOL 43(1):667–692 Zhu Q, Zhou H (2024) The role of cGAS-STING signaling in rheumatoid arthritis: from pathogenesis to therapeutic targets. FRONT IMMUNOL 15:1466023 Additional Declarations The authors declare no competing interests. Supplementary Files SupplementaryFigure1.tiff Supplementary Figure 1 SupplementaryFigure2.tiff Supplementary Figure 2 SupplementaryFigureLegends.docx Supplementary Figure legends SupplementaryTable1.docx Supplementary tbale Cite Share Download PDF Status: Posted Version 1 posted 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-8165027","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":548158504,"identity":"980357a2-0fc2-45b9-a374-051acd7ebdc0","order_by":0,"name":"Yujie Cai","email":"","orcid":"","institution":"Department of Rheumatology and Immunology, Nanfang Hospital, Southern Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yujie","middleName":"","lastName":"Cai","suffix":""},{"id":548158505,"identity":"fa1cbb0d-42d4-4f16-825e-4b0cb18410e2","order_by":1,"name":"Donghong Guo","email":"","orcid":"","institution":"Department of Orthopedics, Nanfang Hospital, Southern Medical University","correspondingAuthor":false,"prefix":"","firstName":"Donghong","middleName":"","lastName":"Guo","suffix":""},{"id":548158506,"identity":"ea47c538-9ce1-4d96-85d2-1fdd935b5780","order_by":2,"name":"Songyuan Zheng","email":"","orcid":"","institution":"Department of Rheumatology and Immunology, Nanfang Hospital, Southern Medical University","correspondingAuthor":false,"prefix":"","firstName":"Songyuan","middleName":"","lastName":"Zheng","suffix":""},{"id":548158507,"identity":"fe9ae2f6-4b5f-4670-9d20-a3e73e300115","order_by":3,"name":"Dingji Zhu","email":"","orcid":"","institution":"Department of Rheumatology and Immunology, Nanfang Hospital, Southern Medical University","correspondingAuthor":false,"prefix":"","firstName":"Dingji","middleName":"","lastName":"Zhu","suffix":""},{"id":548158508,"identity":"df6b0e08-a0d9-44b7-bf1b-cf50a525caee","order_by":4,"name":"Jinjun Zhao","email":"","orcid":"","institution":"Department of Rheumatology and Immunology, Nanfang Hospital, Southern Medical University","correspondingAuthor":false,"prefix":"","firstName":"Jinjun","middleName":"","lastName":"Zhao","suffix":""},{"id":548158509,"identity":"3a66ca67-7b8b-4781-a265-9caadf16eeab","order_by":5,"name":"Weinan Lai","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA0UlEQVRIiWNgGAWjYBACAwbGBiBpY8cP4TMTq6UgLVmygXgtIPDhMOOGA0RrYT/cJl1gwMxsfP50mgRDhXViA/vZA/i18CS2Sc8wYOMzu5G7TYLhTHpiA09eAgGHJbbd5jHgYTa7wbtNgrHtcGKDBI8Bfi38D0FaJBg3958FavlHjBYJsC0GjBsYgA5jbCBKy8P23zwGCckSN3I3WyQcSzdu48nBr8W+P/2xMc+f/3b8/Wc33vhQYy3bz34GvxZUkADEbCSoHwWjYBSMglGAAwAABP5Amtugc1wAAAAASUVORK5CYII=","orcid":"","institution":"Department of Rheumatology and Immunology, Nanfang Hospital, Southern Medical University","correspondingAuthor":true,"prefix":"","firstName":"Weinan","middleName":"","lastName":"Lai","suffix":""}],"badges":[],"createdAt":"2025-11-20 13:13:20","currentVersionCode":1,"declarations":{"humanSubjects":true,"vertebrateSubjects":false,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":true,"humanSubjectConsent":true,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":false},"doi":"10.21203/rs.3.rs-8165027/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8165027/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":96566658,"identity":"fb8074d9-91f4-4212-a31d-b14a742c72e5","added_by":"auto","created_at":"2025-11-23 15:53:40","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":96412,"visible":true,"origin":"","legend":"","description":"","filename":"Manuscript.docx","url":"https://assets-eu.researchsquare.com/files/rs-8165027/v1/87affac23f33a9e45309b946.docx"},{"id":96566659,"identity":"cdd75aa1-ef37-4498-b1b3-e2e143af3c5c","added_by":"auto","created_at":"2025-11-23 15:53:40","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":342,"visible":true,"origin":"","legend":"","description":"","filename":"rs8165027.json","url":"https://assets-eu.researchsquare.com/files/rs-8165027/v1/952fcc8916d013b524d9da05.json"},{"id":96604831,"identity":"027db7b1-6788-4714-85c8-e96c2285c11c","added_by":"auto","created_at":"2025-11-24 09:15:11","extension":"xml","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":112263,"visible":true,"origin":"","legend":"","description":"","filename":"rs81650270enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-8165027/v1/3ed03aedf7a5d0431d0a081a.xml"},{"id":96566672,"identity":"ae538058-dd36-4238-b537-d6b9b540fc9d","added_by":"auto","created_at":"2025-11-23 15:53:41","extension":"xml","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":109028,"visible":true,"origin":"","legend":"","description":"","filename":"rs81650270structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-8165027/v1/1b434edfeeda843086527269.xml"},{"id":96566673,"identity":"b069f7ab-5fd5-46be-9979-e938cdfbb657","added_by":"auto","created_at":"2025-11-23 15:53:41","extension":"html","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":121129,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-8165027/v1/d86d08d1da029b55e4de0355.html"},{"id":96566661,"identity":"c3c9626c-611c-4a62-a0f1-bbd508da2df0","added_by":"auto","created_at":"2025-11-23 15:53:40","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":763148,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eDifferential expression and functional analysis\u003c/strong\u003e \u003cstrong\u003ebetween RA and healthy controls.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) Data processing and methodology pipeline. (B-C) Principal Component Analysis (PCA) plots of the three integrated GEO datasets before (B) and after (C) batch correction. (D) Volcano plot of the differentially expressed genes (DEGs) between RA patients and healthy controls (HC). (E) Heatmap showing the top 50 DEGs between RA and HC. (F) Gene Ontology (GO) enrichment analysis of DEGs. (G) Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways enrichment analysis of DEGs.\u003c/p\u003e","description":"","filename":"Figure1.png","url":"https://assets-eu.researchsquare.com/files/rs-8165027/v1/eed7611e171da7c08f0c7a8e.png"},{"id":96566663,"identity":"3f2d6d26-fbd7-4300-9e44-13e627a66e45","added_by":"auto","created_at":"2025-11-23 15:53:40","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":892962,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eIdentification of ERSRGs and molecular subtypes\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) GSEA plot showing enrichment of response to ER stress pathway across differentially expressed genes (DEGs).\u003cstrong\u003e \u003c/strong\u003e(B) Venn diagram to identify overlapping genes between DEGs and ER stress.\u003cstrong\u003e \u003c/strong\u003e(C)\u003cstrong\u003e \u003c/strong\u003eHeatmap showing the expression pattern of overlapping ER stress-related genes (ERSRGs) in RA and healthy controls (HC).\u003cstrong\u003e \u003c/strong\u003e(D-E)\u003cstrong\u003e \u003c/strong\u003eConsensus and CDF matrix plots showing optimal number of clusters based on the expression of 17 ERSRGs. (F) Heatmap of expression pattern of ERSRGs in two clusters.\u003cstrong\u003e \u003c/strong\u003e(G) Violin plots showing the expression level of each ERSRG in the two clusters. (H)\u003cstrong\u003e \u003c/strong\u003eGene Ontology (GO) enrichment analysis of DEGs.\u003cstrong\u003e \u003c/strong\u003e(I)\u003cstrong\u003e \u003c/strong\u003eKyoto Encyclopedia of Genes and Genomes (KEGG) pathways enrichment analysis of DEGs. Wilcoxon Rank-Sum Test.*\u003cem\u003ep \u003c/em\u003e\u0026lt; 0.05; **\u003cem\u003ep \u003c/em\u003e\u0026lt; 0.01; ***\u003cem\u003ep\u003c/em\u003e \u0026lt; 0.001.\u003c/p\u003e","description":"","filename":"Figure2.png","url":"https://assets-eu.researchsquare.com/files/rs-8165027/v1/c8b0f050d75cfca6c0f75172.png"},{"id":96604673,"identity":"947e339c-d6ac-4066-8367-0f746904a900","added_by":"auto","created_at":"2025-11-24 09:14:33","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":452718,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eIdentification of ER stress diagnostic biomarkers\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A-B) The least absolute shrinkage and selection operator (LASSO) algorithm shows the optimal coefficient based on the ERSRGs. (C-D) The Support Vector Machine-Recursive Feature Elimination (SVM-RFE) algorithm identified eight genes (achieving a maximum precision of 0.881 and a minimum RMSE of 0.119) as the most predictive features. (E-F) The random forest (RF) algorithm identified five genes with an importance score greater than 2. (G) Venn diagrams show 3 diagnostic feature biomarkers using LASSO, SVM- RFE, and RF algorithms. (H-J) Receiver operating characteristic (ROC) curves and the area under the curve (AUC) values demonstrating the diagnostic performance of ATP2A3 (H), MARCKS (I), and UBE2J1 (J).\u003c/p\u003e","description":"","filename":"Figure3.png","url":"https://assets-eu.researchsquare.com/files/rs-8165027/v1/86ca539c4e65535507ea619d.png"},{"id":96566668,"identity":"def824a8-82e5-46f7-a6e1-3eaa2c1252b8","added_by":"auto","created_at":"2025-11-23 15:53:41","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":829776,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFunctional features of ER stress diagnostic biomarkers.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A-B) Protein-protein interaction network and Gene Ontology (GO) enrichment analysis of ATP2A3. (C-D) Protein-protein interaction network and GO enrichment analysis of MARCKS. (E-F) Protein-protein interaction network and GO enrichment analysis of UBE2J1. (G-H) Principal Component Analysis (PCA) plots of the integrated validation datasets before (G) and after (H) batch correction. (I) Violin plots showing the expression difference of ATP2A3,\u003cstrong\u003e \u003c/strong\u003eMARCKS and UBE2J1 between RA and healthy controls (HC). (J) Receiver operating characteristic (ROC) curves demonstrating the diagnostic performance of ATP2A3, MARCKS, and UBE2J1 in the validation cohort. Wilcoxon Rank-Sum Test. *\u003cem\u003ep \u003c/em\u003e\u0026lt; 0.05; **\u003cem\u003ep\u003c/em\u003e \u0026lt; 0.01; ***\u003cem\u003ep \u003c/em\u003e\u0026lt; 0.001.\u003c/p\u003e","description":"","filename":"Figure4.png","url":"https://assets-eu.researchsquare.com/files/rs-8165027/v1/d622d0724398034483643c0b.png"},{"id":96604827,"identity":"957ba500-fa37-4a9b-813b-d56ca4f98f99","added_by":"auto","created_at":"2025-11-24 09:15:10","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":662553,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eATP2A3 promotes inflammatory cytokine secretion in RA FLS.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) Immunohistochemistry (IHC) images of ATP2A3, MARCKS and UBE2J1 in synovial tissues from RA patients and healthy controls (HC). Scale bar, 50 μm. (B) Quantification of IHC staining for ATP2A3, MARCKS, and UBE2J1 in RA patients (n=9) and HC (n=3). (Mann-Whitney U test; data are presented as mean ± SEM; *\u003cem\u003ep\u003c/em\u003e \u0026lt; 0.05, **\u003cem\u003ep\u003c/em\u003e \u0026lt; 0.01). (C) Western blot analysis of ATP2A3 expression in RA-FLS transfected with ATP2A3 siRNA or adenovirus expressing ATP2A3. (D) ELISA was used to analyze IFNα, TNFα, IL-6, IL-8, IL-1β, CXCL10 level in RA-FLS transfected with ATP2A3 siRNA. (E) ELISA was used to analyze IFNα, TNFα, IL-6, IL-8, IL-1β, CXCL10 level in RA-FLS transfected with adenovirus expressing ATP2A3. Student t test. Data are presented as mean standard deviation (SD). *\u003cem\u003ep\u003c/em\u003e\u0026lt; 0.05 **\u003cem\u003ep\u003c/em\u003e \u0026lt; 0.01.\u003c/p\u003e","description":"","filename":"Figure5.png","url":"https://assets-eu.researchsquare.com/files/rs-8165027/v1/d0dacfb074b5414f6c9ebee1.png"},{"id":96566670,"identity":"f9b2b1b7-a62d-4e72-89f5-abfa0358bf04","added_by":"auto","created_at":"2025-11-23 15:53:41","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":383795,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eATP2A3 promotes inflammatory cytokine secretion by activating the STING signaling pathway.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) Western blot analysis of the STING signaling pathway proteins (p-STING, STING, p-TBK1, TBK1, p-IRF3, IRF3, p-p65, p65) in RA-FLS following transfection with ATP2A3 siRNA or an ATP2A3 overexpressing adenovirus. (B) Western blot analysis of the STING signaling pathway proteins (p-STING, STING, p-TBK1, TBK1, p-IRF3, IRF3, p-p65, p65) in RA-FLS following ATP2A3 overexpression and pretreatment with the STING inhibitor SN-011 (1μM for 3 hours). (C-H) Secreted levels of IFNα, TNFα, IL-6, IL-8, IL-1β, and CXCL10 in the culture supernatant were measured by ELISA in RA-FLS following ATP2A3 overexpression and pretreatment with SN-011 (1μM for 3 hours). Student t test. Data are presented as mean standard deviation (SD). *\u003cem\u003ep \u003c/em\u003e\u0026lt; 0.05 **\u003cem\u003ep\u003c/em\u003e \u0026lt; 0.01.\u003c/p\u003e","description":"","filename":"Figure6.png","url":"https://assets-eu.researchsquare.com/files/rs-8165027/v1/ead971b225f798f97b262acb.png"},{"id":96608046,"identity":"df64b671-47c3-4812-905b-3157c1e8f456","added_by":"auto","created_at":"2025-11-24 09:28:12","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":4656564,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8165027/v1/fc8d3878-78cd-4cae-8522-c3dbe48f09ca.pdf"},{"id":96566664,"identity":"ae812833-0c3e-43e8-96c4-31179114667a","added_by":"auto","created_at":"2025-11-23 15:53:40","extension":"tiff","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":455072,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Figure 1\u003c/p\u003e","description":"","filename":"SupplementaryFigure1.tiff","url":"https://assets-eu.researchsquare.com/files/rs-8165027/v1/9c0e654e1a3f4548cf1cd0fc.tiff"},{"id":96566666,"identity":"e2af80a0-868c-4f34-bb16-97f479bc7b65","added_by":"auto","created_at":"2025-11-23 15:53:40","extension":"tiff","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":518754,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Figure 2\u003c/p\u003e","description":"","filename":"SupplementaryFigure2.tiff","url":"https://assets-eu.researchsquare.com/files/rs-8165027/v1/ff4e465d31f47f9ab465c602.tiff"},{"id":96605279,"identity":"5875dbf1-f884-4c4d-9f81-8b18661ad776","added_by":"auto","created_at":"2025-11-24 09:22:00","extension":"docx","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":16812,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Figure \u0026nbsp;legends\u003c/p\u003e","description":"","filename":"SupplementaryFigureLegends.docx","url":"https://assets-eu.researchsquare.com/files/rs-8165027/v1/865a132e262a53513bb0903e.docx"},{"id":96566667,"identity":"c50cfcca-9103-406c-9347-460503cd7e1e","added_by":"auto","created_at":"2025-11-23 15:53:41","extension":"docx","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":15705,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary tbale\u003c/p\u003e","description":"","filename":"SupplementaryTable1.docx","url":"https://assets-eu.researchsquare.com/files/rs-8165027/v1/fbfe8150112c20dbfca41d5f.docx"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003e\u003cstrong\u003eATP2A3 promotes inflammatory cytokine secretion in rheumatoid arthritis fibroblast-like synoviocytes via activation of the STING signaling pathway\u003c/strong\u003e\u003c/p\u003e","fulltext":[{"header":"Introduction","content":"\u003cp\u003eRheumatoid Arthritis (RA) is a chronic autoimmune disease, characterized by persistent inflammation of the synovial membrane, leading to joint destruction, cartilage loss, and bone erosion [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. This progressive joint damage often results in significant disability and a reduced quality of life [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. Moreover, beyond joint involvement, RA can also affect extra-articular systems, highlighting its systemic nature [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. Despite advances in therapeutic options, including disease-modifying anti-rheumatic drugs (DMARDs), RA continues to place a substantial physical and economic burden on patients and healthcare systems worldwide [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. Therefore, early diagnosis and personalized treatment plans remain critical for improving outcomes and reducing disease-related complications.\u003c/p\u003e\u003cp\u003eEndoplasmic reticulum (ER) stress plays an important role in RA pathology that links cellular stress to immune dysfunction [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. ER stress occurs when the ER's protein-folding capacity is overwhelmed, activating the unfolded protein response (UPR) to restore homeostasis [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. In RA, reactive oxygen species (ROS), hypoxia, nutrient deprivation, and inflammatory cytokines contributes to chronic ER stress [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. This persistent ER stress enhances the survival and proliferation of fibroblast-like synoviocytes (FLS), driving pannus formation and joint damage [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. Furthermore, the UPR exacerbates inflammation by increasing cytokine production and altering immune cell activity, amplifying systemic and local immune dysregulation in RA [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. This highlights ER stress and the consequent cellular response as a key contributor to pathology and a potential therapeutic target. Despite this potential, the role of ER stress in RA remains underexplored, particularly through the lens of bioinformatics, which is increasingly leveraged to uncover prognostic and predictive biomarkers at the gene level [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eFLS, which have an aggressive phenotype, play an important role in the pathological processes of RA [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. FLS not only produce the extracellular matrix but also secrete inflammatory cytokines and proteases that contribute to disease pathogenesis and perpetuation [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. For example, FLS secrete inflammatory cytokines such as IL-6 to recruit immune cells, thereby promoting joint inflammation [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. Moreover, ITGA5\u0026thinsp;+\u0026thinsp;synovial fibroblasts secrete TGF-β1 to induce the differentiation of naive CD4\u0026thinsp;+\u0026thinsp;T cells into CXCL13\u003csup\u003ehi\u003c/sup\u003e PD-1\u003csup\u003ehi\u003c/sup\u003e peripheral helper T cells, thereby remodeling the pro-inflammatory microenvironment and facilitating the progression of RA [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. Although several studies have indicated that ER stress is involved in the activation of FLS, the molecular mechanisms by which ER stress induces this activation are largely unknown.\u003c/p\u003e\u003cp\u003eIn this study, bioinformatics approaches were used to identify ER stress response-related genes (ERSRGs) in synovial tissues of RA patients using Gene Expression Omnibus (GEO) database datasets. Machine learning algorithms were applied to refine and prioritize key biomarkers. Pathway enrichment analyses further explored their functional roles. We identified ATPase Sarcoplasmic/Endoplasmic Reticulum Ca2\u0026thinsp;+\u0026thinsp;Transporting 3 (ATP2A3) as a potential diagnostic biomarker of ER stress in RA. ATP2A3 can promote the secretion of inflammatory cytokines in RA FLS through activation of the stimulator of interferon genes (STING) signaling pathway.\u003c/p\u003e"},{"header":"Methods \u0026 Materials","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003eDatasets Collection and Preprocessing\u003c/h2\u003e\u003cp\u003eSix publicly available gene expression datasets (GSE55235, GSE55457, GSE77298, GSE1919, GSE236924, and GSE89408) were retrieved from the GEO database (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.ncbi.nlm.nih.gov/geo/\u003c/span\u003e\u003cspan address=\"https://www.ncbi.nlm.nih.gov/geo/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e).\u003c/span\u003e Among these, GSE55235, GSE55457, and GSE77298, which contained balanced numbers of RA and healthy control samples, were designated as the training set. The remaining datasets (GSE1919, GSE236924, and GSE89408) were reserved for external validation. Details of the sample sizes and platforms used in each dataset are summarized in Supplementary Table\u0026nbsp;1.\u003c/p\u003e\u003cp\u003eThe datasets originated from various microarray platforms: GSE55235 and GSE55457 utilized the GPL96 ([HG-U133A] Affymetrix Human Genome U133A Array), GSE77298 and GSE236924 were based on GPL570 ([HG-U133_Plus_2] Affymetrix Human Genome U133 Plus 2.0 Array), and GSE1919 used GPL91 ([HG_U95A] Affymetrix Human Genome U95A Array). GSE89408 was generated using high-throughput sequencing on the GPL11154 (Illumina HiSeq 2000) platform.\u003c/p\u003e\u003cp\u003eRaw gene expression data were preprocessed using custom Perl scripts to extract and annotate expression matrices. To ensure comparability, only the intersection of common genes across all datasets was retained. For each dataset, probes were averaged to gene-level expression using the avereps function, and sample names were standardized to include dataset identifiers. To correct for platform and batch-related variability, expression values were normalized using the limma package in R, and batch effects were adjusted using the ComBat algorithm implemented in the sva package. Each sample was assigned a batch label based on its dataset of origin. The ComBat function applied empirical Bayes methods to model batch effects, assuming parametric prior distributions and preserving biological variation.\u003c/p\u003e\u003cp\u003ePrincipal component analysis (PCA) was conducted before and after ComBat adjustment to assess data structure and batch correction efficacy. The PCA was performed using the prcomp function on the transposed expression matrix. Sample clustering was visualized with ggpubr::ggscatter, displaying 95% confidence ellipses for each batch. Prior to correction, clear batch-specific clustering was observed. After ComBat correction, samples distributed more homogenously, indicating successful mitigation of batch effects.\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eDifferential expression analysis\u003c/h3\u003e\n\u003cp\u003eDifferential expression analysis was conducted using the limma R package to identify differentially expressed genes (DEGs) between risk subgroups. The criteria for defining the DEGs was log fold change (logFC)\u0026thinsp;=\u0026thinsp;0.585 and false discovery rate (FDR)\u0026thinsp;\u0026lt;\u0026thinsp;0.05.\u003c/p\u003e\n\u003ch3\u003eEndoplasmic reticulum stress response-related genes\u003c/h3\u003e\n\u003cp\u003eThe ERSRGs were sourced from the Gene Ontology Consortium pathway \u0026ldquo;GOBP_RESPONSE_TO_ENDOPLASMIC_RETICULUM_STRESS\u0026rdquo; available through the Molecular Signatures Database (MSigDB) (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.gsea-msigdb.org/gsea/msigdb\u003c/span\u003e\u003cspan address=\"https://www.gsea-msigdb.org/gsea/msigdb\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e).\u003c/span\u003e This gene set contains a total of 267 ERSRGs, which were extracted for further analysis.\u003c/p\u003e\n\u003ch3\u003ePathway enrichment analysis\u003c/h3\u003e\n\u003cp\u003eFunctional annotation of the identified DEGs was performed through Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses to investigate their biological roles and associated pathways. These analyses were conducted using the \"clusterProfiler\" R package [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. Additionally, GO enrichment information for individual ER stress biomarkers was retrieved from the STRING database using its \"Analysis\" module (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://string-db.org/\u003c/span\u003e\u003cspan address=\"https://string-db.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e).\u003c/span\u003e\u003c/p\u003e\n\u003ch3\u003eProtein-protein interaction analysis\u003c/h3\u003e\n\u003cp\u003eThe interaction network of ER stress biomarkers associated with RA was analyzed using the STRING database (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://string-db.org/\u003c/span\u003e\u003cspan address=\"https://string-db.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e)\u003c/span\u003e [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. A medium confidence threshold (score\u0026thinsp;\u0026ge;\u0026thinsp;4.00) was applied to ensure reliable interactions. Active interaction sources included evidence from text mining, experimental data, curated databases, co-expression patterns, genomic neighborhood, gene fusion events, and co-occurrence. This comprehensive analysis facilitated the identification of functional connections and potential regulatory networks involving ER stress biomarkers.\u003c/p\u003e\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\u003ch2\u003eConsensus clustering\u003c/h2\u003e\u003cp\u003eTo stratify patients based on the expression patterns of ERSRGs, an unsupervised clustering approach using consensus clustering was applied. This method systematically evaluated patient groupings across a range of cluster numbers (k), varying from 2 to 11. The optimal cluster number was determined by assessing the consensus stability, utilizing cumulative distribution function (CDF) plots to identify the value of k that offered the most robust clustering solution. The analysis was conducted using the Consensus ClusterPlus R package, set to 1000 iterations to ensure reliability and reproducibility of the clustering results [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e].\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eDiscovery of ER stress biomarkers using machine learning algorithms\u003c/h3\u003e\n\u003cp\u003eTo ensure robust identification of ER stress biomarkers in RA, we employed a multi-model machine learning framework using Least Absolute Shrinkage and Selection Operator (LASSO) regression, Support Vector Machine-Recursive Feature Elimination (SVM-RFE), and Random Forest (RF). Each algorithm incorporated appropriate cross-validation, model diagnostics, and feature selection techniques to prevent overfitting and maximize generalizability.\u003c/p\u003e\n\u003ch3\u003eModel Selection, Assumptions, and Diagnostics\u003c/h3\u003e\n\u003cp\u003eLASSO regression was conducted using the glmnet R package [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. The gene expression matrix was transposed such that rows represented samples and columns represented genes, with binary labels assigned as control\u0026thinsp;=\u0026thinsp;0 and RA\u0026thinsp;=\u0026thinsp;1. The optimal penalty parameter (lambda) was selected by 10-fold cross-validation using cv.glmnet, minimizing binomial deviance. Genes with non-zero coefficients at lambda.min were retained. To quantify diagnostic performance and uncertainty, we computed sample-wise risk scores as the weighted sum of selected gene expressions and coefficients, then evaluated model performance using Recursive operator curve (ROC) and bootstrapped AUC (n\u0026thinsp;=\u0026thinsp;1000) via the pROC R package.\u003c/p\u003e\u003cp\u003eSVM-RFE was performed using the caret and e1071 packages with an RBF kernel (svmRadial) [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]. A 10-fold cross-validation strategy was applied using rfeControl, evaluating model accuracy across feature subsets of increasing size. The optimal feature set was selected based on maximal cross-validated accuracy. Genes selected through this procedure were exported, and diagnostic validity was assessed via ROC-AUC analysis using predicted labels from the final model. This approach captures non-linear relationships and is particularly suitable for high-dimensional biological data.\u003c/p\u003e\u003cp\u003eRF analysis was implemented using the randomForest package with 500 decision trees. The expression matrix was again transposed, and group labels were extracted from sample names. The optimal number of trees was identified by selecting the point of minimum error from the out-of-bag (OOB) error rate. Feature importance was measured using the Mean Decrease Gini index. Genes were ranked accordingly, and the top 30 most informative features were retained. A variable importance plot was generated to visualize key contributors to classification. Expression profiles of these top genes were exported for downstream validation and visualization.\u003c/p\u003e\u003cp\u003eTogether, these three complementary approaches enabled robust and reproducible feature selection. LASSO ensured sparse linear modeling with embedded regularization; SVM-RFE captured non-linear margins with recursive feature filtering; and Random Forest leveraged ensemble learning and internal validation via OOB error estimation. Final candidate biomarkers were identified by intersecting outputs across all three methods using a Venn diagram approach, ensuring selection of consistently predictive features across distinct algorithms.\u003c/p\u003e\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\u003ch2\u003eRecursive operator curve analysis\u003c/h2\u003e\u003cp\u003eThe diagnostic performance of the selected biomarkers was evaluated using ROC analysis, with AUC values calculated via the pROC R package. This multi-algorithm approach ensured reliable identification of biomarkers with strong diagnostic potential for ERSRGs-RA.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\u003ch2\u003eClinical specimens\u003c/h2\u003e\u003cp\u003eWe collected synovial tissue samples from 9 RA patients and 3 healthy controls between July 2023 and August 2024 at the Department of Orthopedic Surgery, Nanfang Hospital, Southern Medical University. Informed consent was obtained from the patients and healthy controls, and approval was granted by the internal review and ethics boards of the Southern Medical University.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\u003ch2\u003eImmunohistochemistry\u003c/h2\u003e\u003cp\u003eImmunohistochemistry (IHC) analysis was conducted on 4 \u0026micro;m thick sections of formalin-fixed, paraffin-embedded synovial tissue from RA patients and healthy controls. Tissue sections were deparaffinized with xylene and ethanol, followed by antigen retrieval using citrate buffer (pH 6.0) through microwave heating. Endogenous peroxidase activity was quenched with 0.3% hydrogen peroxide, and the sections were blocked with 5% bovine serum albumin in phosphate-buffered saline (PBS). Primary antibodies for ATP2A3 (Proteintech, #13619-1-AP, rabbit, 1:50), MARCKS (Proteintech, #10004-2-Ig, rabbit, 1:50), and UBE2J1 (Affinity, #MG805029, rabbit, 1:100) were applied, and samples were incubated overnight at 4\u0026deg;C. Biotinylated secondary antibodies were then applied at room temperature, and staining was visualized using a 3,5-diaminobenzidine (DAB) substrate with hematoxylin counterstaining. Staining intensity was scored on a 0\u0026ndash;3 scale (0: negative, 1: weak, 2: moderate, 3: strong), and positive cell frequency on a 0\u0026ndash;4 scale (0: \u0026lt;5%, 1: 5\u0026ndash;25%, 2: 26\u0026ndash;50%, 3: 51\u0026ndash;75%, 4: \u0026gt;75%). The IHC score was calculated as the product of intensity and frequency. For heterogeneous tissues, individual regions were scored and combined for the final result.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec14\" class=\"Section2\"\u003e\u003ch2\u003eCulture of fibroblast-like synoviocytes\u003c/h2\u003e\u003cp\u003eThe fresh human synovial tissues were digested to isolate FLS, which were subsequently cultured in Dulbecco's Modified Eagle Medium (DMEM; Thermo Fisher Scientific, Waltham, MA) containing 10% fetal bovine serum (FBS; Gibco, Grand Island, NY). Cells from passages four to six were employed in all experiments. Homogeneity and purity (\u0026gt;\u0026thinsp;98%) of the FLS population were confirmed via vimentin-specific immunofluorescence staining.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec15\" class=\"Section2\"\u003e\u003ch2\u003eWestern blot analysis\u003c/h2\u003e\u003cp\u003ePrimary antibodies anti-STING, anti-p-STING, anti-IRF3, anti-p-IRF3, anti-TBK1, anti-p-TBK1, anti-p65, anti-p-p65 were purchased from Cell Signaling Technology (Beverly, MA, USA). Primary antibody anti-β-actin were obtained from Protein Tech Group (Rosemont, IL). Briefly, RA-FLS were lysed in ice-cold radioimmunoprecipitation assay (RIPA) buffer (BestBio, Shanghai). Proteins were resolved by 10% sodium dodecyl sulfate\u0026ndash;polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto polyvinylidene difluoride (PVDF) membranes (Millipore, USA). Immunoreactive bands were quantified using Quantity One software (Bio-Rad).\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec16\" class=\"Section2\"\u003e\u003ch2\u003esiRNA and adenovirus transfection\u003c/h2\u003e\u003cp\u003eThe siRNAs target for ATP2A3 were purchased from RiboBio (Guangzhou, China). Adenovirus vector containing the coding sequences for ATP2A3 were obtain form Genepharma (Shanghai, China). The transfection process was performed using Lipofectamine 3000 (Hanbio Biotechnology, Shanghai, China) according to the manufacturer\u0026rsquo;s protocol. The expression level of ATP2A3 were confirmed by Western Blot Analysis.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec17\" class=\"Section2\"\u003e\u003ch2\u003eEnzyme-linked immunosorbent assay (ELISA)\u003c/h2\u003e\u003cp\u003eThe levels of TNFα, IFNα, CXCL10, IL-1β, IL-6, IL-8 in the supernatants of RA-FLS were detected by Enzyme-linked immunosorbent assay (ELISA). The ELISA kits were purchased from Boster Biological Technology (Pleasanton, CA, USA). Cells were seeded into 6-well plates and grown to 80% confluence. They were then cultured in fresh serum-free media, and supernatant was harvested at 24 hours. Measurements were according to the manufacturer's instructions.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec18\" class=\"Section2\"\u003e\u003ch2\u003eStatistical analysis\u003c/h2\u003e\u003cp\u003eStatistical analyses were conducted using R software (v4.0.3, \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.r-project.org\u003c/span\u003e\u003cspan address=\"http://www.r-project.org\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). Data are presented as the means standard deviation and were analyzed using a Student t test. For comparisons of gene expression/enrichment/IHC scores between two groups, the non-parametric Wilcoxon Rank-Sum Test (equivalent to the Mann-Whitney U Test) was applied. Correlation analyses were performed using either Spearman's rank/Pearson's correlation tests. \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05 was considered statistical significance.\u003c/p\u003e\u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec20\" class=\"Section2\"\u003e\u003ch2\u003eDifferential expression and functional analysis in RA\u003c/h2\u003e\u003cp\u003eThe research strategy and methodology employed in this study are illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eA. Key steps included: performing differential expression analysis; identifying diagnostic and predictive genes associated with response to ER stress; constructing molecular subtypes; conducting pathway enrichment and validating findings through external datasets; and experimental approaches.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eTranscriptomic data from RA patients were obtained from three GEO microarray datasets (GSE55235, GSE55457, GSE77298). After merging the datasets and performing batch correction, a meta-cohort was created. Principal component analysis (PCA) before batch correction revealed distinct clustering of samples by dataset (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eB). Post-ComBat PCA showed a significant reduction in inter-dataset variance, indicating successful harmonization (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eC). Differential expression analysis was conducted between RA patients (n\u0026thinsp;=\u0026thinsp;39) and healthy controls (n\u0026thinsp;=\u0026thinsp;27), identifying 1433 DEGs, including 872 upregulated and 561 downregulated genes based on the criteria of logFC\u0026thinsp;\u0026ge;\u0026thinsp;0.585 and FDR\u0026thinsp;\u0026le;\u0026thinsp;0.05 (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eD). A heatmap displayed the top 50 DEGs, highlighting distinct expression patterns (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eE).\u003c/p\u003e\u003cp\u003ePathway enrichment analysis revealed significant functional implications of the DEGs. The enriched pathways highlighted key mechanisms underlying RA pathology (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eF-G). These include immune response activation (e.g., Th17 cell differentiation, leukocyte adhesion, chemokine signaling), osteoclast differentiation, and cell adhesion molecules, all of which contribute to chronic inflammation and tissue damage in RA. Additionally, pathways such as Epstein-Barr virus infection and hematopoietic cell lineage suggest a link between viral triggers, immune dysregulation, and disease progression in RA.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec21\" class=\"Section2\"\u003e\u003ch2\u003eIdentify ERSRGs and ER stress-based molecular subtypes in RA\u003c/h2\u003e\u003cp\u003eThe enrichment of the ERSRGs pathway obtained from Gene Ontology Consortium (GOBP_RESPONSE_TO_ENDOPLASMIC_RETICULUM_STRESS) corresponded to the upregulated DEGs in RA (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA). An intersection of DEGs (n\u0026thinsp;=\u0026thinsp;1433) with ERSRGs (n\u0026thinsp;=\u0026thinsp;267) revealed 17 overlapping ERSRGs (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB). The heatmap shows their expression patterns, highlighting 13 genes upregulated and 4 downregulated in RA (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eC). This result underscores the involvement of ERSRGs in RA pathology.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eAs these 17 ERSRGs were significantly differentially expressed in RA, we employed consensus clustering to obtain molecular subtypes of RA and identify the implications of ER stress. The RA patients were categorized into two subgroups based on the expression of these 17 genes as indicated by the observation of consensus matrix and CDF matrix at K\u0026thinsp;=\u0026thinsp;2 (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eD-E). The 39 RA samples were stratified into two clusters: Cluster A (n\u0026thinsp;=\u0026thinsp;25) and Cluster B (n\u0026thinsp;=\u0026thinsp;14). The heatmap illustrates the similarity in the expression of upregulated and downregulated genes between the clusters to the expression pattern in control versus RA patients (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eF). Cluster B exhibited elevated expression of JUN, PPP1R15A, ATF3, and PIK3R1\u0026mdash;genes that were also highly expressed in healthy controls. This suggests that the ER stress profile in Cluster B patients may share some features with that of healthy individuals. Analysis of the features driving cluster separation identified 12 key genes: 2 that were downregulated (JUN, PPP1R15A) and 10 that were upregulated (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eG). Pathway enrichment analysis of the DEGs between the clusters also highlighted similar functional differences seen in control versus RA, with enhanced enrichment of immune response activation and inflammatory functions (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eH-I).\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec22\" class=\"Section2\"\u003e\u003ch2\u003eIdentification of ER stress diagnostic biomarkers\u003c/h2\u003e\u003cp\u003eTo explore the potential diagnostic role of ERSRGs in distinguishing RA patients from healthy individuals, we employed a robust multi-algorithm approach. This involved three machine learning techniques: LASSO regression, SVM-RFE, and RF. LASSO regression selected 10 genes from an initial pool of 17 candidates (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eA-B). SVM-RFE identified 8 ERSRGs with high precision (0.881) and minimal root mean square error (RMSE\u0026thinsp;=\u0026thinsp;0.119) (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eC-D). Random Forest analysis ranked five genes based on an importance score threshold of \u0026gt;\u0026thinsp;2 (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eE-F). Integration of these results revealed three key genes\u0026mdash;ATP2A3, MARCKS (Myristoylated Alanine-Rich C-Kinase Substrate), and UBE2J1 (ubiquitin conjugating enzyme E2 J1)\u0026mdash;as significant biomarkers (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eG). ROC curve analysis validated their diagnostic accuracy, with each gene achieving an AUC greater than 0.80 (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eH-J). These results underscore their potential in differentiating RA patients from healthy individuals.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cdiv id=\"Sec23\" class=\"Section3\"\u003e\u003ch2\u003eFunctional and immunological relevance of ER stress biomarkers\u003c/h2\u003e\u003cp\u003e To enhance our understanding of the functional roles of these three ERSRGs biomarkers, we first identified their protein-protein interaction networks using the STRING database, followed by the pathway enrichment analysis. GO analysis of the ATP2A3 and MARCKS networks indicated their critical role in calcium homeostasis with slightly opposite functions. ATP2A3 was associated with positive regulation of calcium transmembrane transport (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eA-B) and MARCKS was associated with negative regulation of calcium ion export across plasma membrane and depletion of calcium ion (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eC-D). Calcium is essential for maintaining ER homeostasis, and a significant depletion of Ca2\u0026thinsp;+\u0026thinsp;in the ER serves as a potent inducer of ER stress [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. Lastly, UBE2J1 network was enriched in ER-associated degradation (ERAD) pathway, essential for managing misfolded proteins under ER stress (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eE-F). All three biomarkers are critical for ER stress and cellular response to ER stress.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eThe association of these genes with immune cell infiltration varied: ATP2A3 was linked to increased activated CD4 memory T cells and plasma cells (Supplementary Fig.\u0026nbsp;1A); MARCKS was associated with neutrophils, plasma cells, and CD8 T cells (Supplementary Fig.\u0026nbsp;1B); while UBE2J1 was associated with plasma cells and gamma delta T cells (Supplementary Fig.\u0026nbsp;1C). These findings suggest that each gene may influence different immune cell subsets, contributing to the inflammatory microenvironment characteristic of RA.\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Sec24\" class=\"Section2\"\u003e\u003ch2\u003eExternal validation of ER stress diagnostic biomarkers\u003c/h2\u003e\u003cp\u003eTo validate the diagnostic utility of ER stress biomarkers for RA, we analyzed their expression across three independent cohorts from the GEO database. The datasets included GSE236924 (N\u0026thinsp;=\u0026thinsp;43; 7 healthy controls, 36 RA), GSE1919 (N\u0026thinsp;=\u0026thinsp;8; 4 healthy controls, 4 RA), and GSE89408 (N\u0026thinsp;=\u0026thinsp;179; 27 healthy controls, 152 RA). The datasets were combined following normalization and batch effect correction. PCA conducted prior to batch correction demonstrated clear clustering of samples according to their originating datasets (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eG). After applying the ComBat method, PCA revealed a marked decrease in inter-dataset variability, reflecting effective data harmonization (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eH). ATP2A3, MARCKS, and UBE2J1 were significantly upregulated in RA samples compared to healthy controls (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eI).\u003c/p\u003e\u003cp\u003eROC curve analysis for the meta-cohort revealed that ATP2A3, MARCKS, and UBE2J1 achieved AUC values of 0.650, 0.874, and 0.958, respectively, demonstrating their potential diagnostic value (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eJ). In individual datasets, MARCKS consistently showed upregulation across all three cohorts, with AUC values ranging from 0.666 to 1.000 (Supplementary Fig.\u0026nbsp;2A-C). UBE2J1 was significantly upregulated in two cohorts, achieving AUC values between 0.800 and 0.983 (Supplementary Fig.\u0026nbsp;2A-C). ATP2A3 displayed upregulation in one cohort, with AUC values ranging from 0.560 to 0.909 (Supplementary Fig.\u0026nbsp;2A-C). These findings underscore the strong diagnostic potential of MARCKS and UBE2J1 as biomarkers for RA, while ATP2A3 demonstrated a more variable but noteworthy performance. This validation supports the clinical relevance of these biomarkers in distinguishing RA from healthy individuals.\u003c/p\u003e\u003cdiv id=\"Sec25\" class=\"Section3\"\u003e\u003ch2\u003eATP2A3 promotes inflammatory cytokine secretion in RA FLS\u003c/h2\u003e\u003cp\u003eWe performed IHC on synovial tissue samples from nine RA patients and three healthy controls to validate the diagnostic utility of ER stress biomarkers. We found that the protein expression of ATP2A3, MARCKS, and UBE2J1 was significantly increased in RA synovial tissues compared to healthy controls (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eA-B). These results suggested that ATP2A3, MARCKS, and UBE2J1 are involvement in ER stress pathways and RA pathology. Among these ER stress diagnostic biomarkers, ATP2A3 has attracted our attention. ATP2A3 belongs to the sarco/endoplasmic reticulum Ca2-ATPase (SERCA) family, which can be induced by ER stress [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. It plays an important role in maintaining intracellular Ca2\u0026thinsp;+\u0026thinsp;homeostasis and ER stress-associated apoptosis [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]. Moreover, it has been reported that ATP2A3 involved in T lymphocyte activation and cytokine secretion in head and neck squamous cell carcinoma [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]. We suspected that ATP2A3 might be involved in inflammatory cytokine secretion in RA FLS. Therefore, we selected ATP2A3 for further investigation.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eTo investigate the role of ATP2A3 in the inflammatory response of RA FLS, we inhibited its expression using siRNA and overexpressed it with an adenoviral vector (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eC). We then evaluated the effect of ATP2A3 modulation on inflammatory cytokine secretion. The results showed that inhibition of ATP2A3 significantly decreased the secretion of inflammatory cytokines, including TNFα, IFNα, CXCL10, IL-1β, IL-6, and IL-8, in RA FLS (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eD). Conversely, overexpression of ATP2A3 significantly upregulated the secretion of these cytokines (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eE). These results indicated that ATP2A3 promotes inflammatory cytokine secretion in RA FLS.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec26\" class=\"Section3\"\u003e\u003ch2\u003eATP2A3 promotes inflammatory cytokine secretion by activating the STING signaling\u003c/h2\u003e\u003cp\u003eThe SERCA family, which includes ATP2A1, ATP2A2, and ATP2A3, is composed of members that possess conserved structures and exhibit analogous functions [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]. Previous studies have indicated that ATP2A2 directly interacts with STING and regulates its phosphorylation, thereby activating STING-induced inflammatory response [\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]. Given this functional conservation within the SERCA family, we hypothesized that ATP2A3 regulates STING phosphorylation, thereby activating STING-induced inflammatory response. Upon activation, STING recruits and activates the kinases TANK-binding kinase 1 (TBK1) and inhibitor of kappa B kinase (IKK); this leads to the phosphorylation of interferon regulatory factor 3 (IRF3) and nuclear factor kappa-B (NF-κB), thereby driving the expression of downstream inflammatory cytokines [\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]. Consistently, we found that inhibition of ATP2A3 significantly decreased the phosphorylation of STING, TBK1, IRF3, p65 (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eA). In contrast, overexpression of ATP2A3 enhanced the phosphorylation of STING, TBK1, IRF3, and p65 (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eA). These results indicate that ATP2A3 enhances STING signaling pathway activation.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eTo investigate whether the effect of ATP2A3 on inflammatory cytokine secretion is dependent on STING signaling in RA FLS, we overexpressed ATP2A3 and then treated the cells with the STING inhibitor SN-011. SN-011 is a STING inhibitor that can effectively inhibit STING phosphorylation [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]. We found that SN-011 blocked the phosphorylation of STING, TBK1, IRF3, and p65 induced by ATP2A3 overexpression (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eB). Importantly, the inhibitor also blocked the upregulation of inflammatory cytokines (TNFα, IFNα, CXCL10, IL-1β, IL-6, and IL-8) induced by ATP2A3 overexpression (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eC-H). These results demonstrate that ATP2A3 promotes inflammatory cytokine secretion by activating the STING signaling pathway in RA FLS.\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eER stress contributes to RA by disrupting cellular homeostasis and amplifying immune dysfunction [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. In this study, we explored the role of ERSRGs in RA. Seventeen ERSRGs were identified as significantly dysregulated in RA, enabling consensus clustering to classify patients into two distinct subtypes with unique functional and immunological characteristics. Through machine learning, three key biomarkers\u0026mdash;ATP2A3, MARCKS, and UBE2J1\u0026mdash;were identified. Importantly, we found that ATP2A3 promotes the secretion of inflammatory cytokines in RA FLS. Mechanistically, ATP2A3 enhances the phosphorylation of STING, thereby activating the STING signaling pathway and leading to increased inflammatory cytokines production. This work provides novel insights into the molecular underpinnings of RA and highlights ATP2A3 as a potential target for therapeutic intervention.\u003c/p\u003e\u003cp\u003eER stress is implicated in a broad spectrum of diseases, including cancer, cardiovascular conditions, neurodegenerative disorders, and autoimmune diseases [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e, \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]. In RA, the synovial microenvironment\u0026mdash;characterized by hypoxia, nutrient deprivation, and an abundance of pro-inflammatory cytokines\u0026mdash;provides conditions that could trigger ER stress [\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e, \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]. A detailed examination of ER stress could provide critical insights into the molecular mechanisms driving RA and identify new diagnostic biomarkers as well as target for RA therapy. We identified 17 dysregulated ERSRGs in RA. Essentially, the upregulation of 13 ERSRGs, including the three ER stress biomarkers (ATP2A3, MARCKS, and UBE2J1) and downregulation of 4 ERSRGs (JUN, ATF3, PIK3R1 and PPP1R15A) differentiated the RA from healthy controls. Molecular heterogeneity was evident when two molecular subtypes based on the expression pattern of these 17 ERSRGs were obtained. Interestingly, 12 of the 17 ERSRGs (10 upregulated and 2 downregulated) contributed to the cluster differentiation with rather consistent expression pattern. Future studies are warranted to determine whether these subtypes correlate with distinct clinical manifestations or treatment responses. Altogether, our findings highlight that ER stress induces specific alterations in ERSRGs, which likely contribute to RA pathogenesis.\u003c/p\u003e\u003cp\u003eAmong the upregulated ERSRGs, ATP2A3, MARCKS, and UBE2J1 were identified as the ER stress biomarkers with enhanced diagnostic potential for RA. ATP2A3 was specifically selected for further investigation. ATP2A3 is an important member of the SERCA family and can be induced by ER stress [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. Previous study indicated that ATP2A3 plays a critical role in calcium homeostasis by catalyzing ATP hydrolysis to transport calcium from the cytosol into the sarcoplasmic reticulum lumen, facilitating processes such as muscle excitation and contraction [\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e]. But, the role of ATP2A3 in RA is largely unknown. Here, we identified that ATP2A3 enhances inflammatory cytokine production by activating the STING signaling pathway in RA FLS. To the best of our knowledge, this is the first study to identify ATP2A3 as a key gene that regulates inflammatory cytokine secretion in RA FLS. Our study highlights that the ER stress biomarker ATP2A3 is a key regulator in the inflammatory activation of RA FLS. However, the roles of other ER stress biomarkers, such as MARCKS and UBE2J1, in RA require further investigation in future studies.\u003c/p\u003e\u003cp\u003eRA is characterized by inflammation of the joint synovium and the infiltration of multiple immune cells [\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e]. The STING signaling pathway, which is widely expressed in both immune and non-immune cells, plays a pivotal role in regulating inflammatory responses [\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e]. Numerous studies have demonstrated the implication of the STING signaling pathway in RA pathogenesis, highlighting its potential as a therapeutic target [\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e]. In present study, we observed a significant upregulation of ATP2A3 in RA synovial tissues compared to healthy controls. Importantly, we discovered that ATP2A3 enhances STING phosphorylation, thereby activating the STING signaling pathway and its downstream effectors, IRF3 and NF-κB. This cascade ultimately leads to increased secretion of inflammatory cytokines from RA-FLS. This study improves our understanding of the mechanisms underlying the activation of the inflammatory response in RA and offers a promising target for RA treatment. ER stress biomarker ATP2A3 serves as a link between ER stress, the STING signaling pathway, and inflammation of the joint synovium of RA.\u003c/p\u003e\u003cp\u003eHowever, this study also has several limitations that should be acknowledged. First, the integration of multiple GEO datasets introduces heterogeneity due to differences in platforms, sample processing, and patient populations, which may affect the consistency of the findings. Although we applied the ComBat algorithm to correct for batch effects and assessed correction using PCA, residual inter-dataset variability may still persist. Second, the lack of prospective validation in independent clinical cohorts limits the generalizability and clinical translation of the identified biomarkers. Third, the exploratory nature of our analysis, primarily driven by machine learning algorithms, raises the possibility of overfitting despite cross-validation and robust feature selection methods. Additionally, the limited availability of clinical metadata across datasets restricted stratified analyses and precluded the assessment of biomarker associations with disease subtypes or treatment response.\u003c/p\u003e"},{"header":"Conclusions","content":"\u003cp\u003eThis study provides valuable insights into the role of ERSRGs in RA. We establish ATP2A3 as a promising diagnostic biomarker for ER stress in RA and demonstrate that it promotes the secretion of inflammatory cytokines in RA FLS through STING pathway activation. ATP2A3 represents a potential therapeutic target for RA treatment.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cdiv class=\"DefinitionList\"\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003e\u003cb\u003eRA\u003c/b\u003e\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eRheumatoid arthritis\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003e\u003cb\u003eER\u003c/b\u003e\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eEndoplasmic reticulum\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003e\u003cb\u003eERSRGs\u003c/b\u003e\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eER stress-related genes\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003e\u003cb\u003eATP2A3\u003c/b\u003e\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eATPase Sarcoplasmic/Endoplasmic Reticulum Ca2\u0026thinsp;+\u0026thinsp;Transporting 3\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003e\u003cb\u003eMARCKS\u003c/b\u003e\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eMyristoylated Alanine-Rich C-Kinase Substrate\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003e\u003cb\u003eUBE2J1\u003c/b\u003e\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eUbiquitin conjugating enzyme E2J1\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003e\u003cb\u003eLASSO\u003c/b\u003e\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eLeast Absolute Shrinkage and Selection Operator\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003e\u003cb\u003eSVM-RFE\u003c/b\u003e\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eSupport Vector Machine-Recursive Feature Elimination\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003e\u003cb\u003eRF\u003c/b\u003e\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eRandom Forest\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003e\u003cb\u003eIHC\u003c/b\u003e\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eImmunohistochemistry\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003e\u003cb\u003eDMARDs\u003c/b\u003e\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eDisease-modifying anti-rheumatic drugs\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003e\u003cb\u003eFLS\u003c/b\u003e\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eFibroblast-like synoviocytes\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003e\u003cb\u003eACPAs\u003c/b\u003e\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eAnti-citrullinated protein antibodies\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003e\u003cb\u003eUPR\u003c/b\u003e\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eUnfolded protein response\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003e\u003cb\u003eROS\u003c/b\u003e\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eReactive oxygen species\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003e\u003cb\u003ePCA\u003c/b\u003e\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003ePrincipal component analysis\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003e\u003cb\u003eGO\u003c/b\u003e\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eGene Ontology\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003e\u003cb\u003eROC\u003c/b\u003e\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eRecursive operator curve\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003e\u003cb\u003eDEGs\u003c/b\u003e\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eDifferentially expressed genes\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003e\u003cb\u003eMSigDB\u003c/b\u003e\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eMolecular Signatures Database\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003e\u003cb\u003eERAD\u003c/b\u003e\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eER-associated degradation\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003e\u003cb\u003eJUN\u003c/b\u003e\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eJun proto-oncogene\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003e\u003cb\u003ePPP1R15A\u003c/b\u003e\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eProtein phosphatase 1 regulatory subunit 15A\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003e\u003cb\u003eATF3\u003c/b\u003e\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eActivating transcription factor 3\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003e\u003cb\u003ePIK3R1\u003c/b\u003e\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003ePhosphoinositide-3-kinase regulatory subunit 1\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003e\u003cb\u003eELISA\u003c/b\u003e\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eEnzyme-linked immunosorbent assay\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003e\u003cb\u003eSTING\u003c/b\u003e\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eStimulator of interferon genes\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003e\u003cb\u003eTBK1\u003c/b\u003e\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eTANK binding kinase 1\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003e\u003cb\u003eIRF3\u003c/b\u003e\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003einterferon regulatory factor 3\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003e\u003cb\u003eSERCA\u003c/b\u003e\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eSarco/endoplasmic reticulum Ca2-ATPase\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003e\u003cb\u003eKEGG\u003c/b\u003e\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eKyoto Encyclopedia of Genes and Genomes.\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe studies involving human participants were reviewed and approved by the Internal Review and Ethics Boards of the Nanfang Hospital, Southern Medical University. The patients provided written informed consent to participate in the study. The study adhered to the ethical principles outlined in the Helsinki Declaration.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eClinical trial number\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAuthors declare no conflict of interest\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was supported by the Natural Science Foundation of Guangdong Province of China [2023A1515111036] and the President Foundation of Nanfang Hospital, Southern Medical University [grant number 2023B042].\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026apos; contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eJ.Z. and W.L. conceived and designed the study. Y.C. and D.G. drafted the manuscript and created the figure. S.Z. and D.Z. performed IHC staining and Western blot analysis. All authors reviewed and approved the final version of the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe thank the patients with RA and healthy donors who contributed to this study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of Data and Materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets used in this study are available from the GEO database (https://www.ncbi.nlm.nih.gov/geo/). Raw data from our experiments can be requested from the corresponding author, Weinan Lai, upon reasonable request.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eDi Matteo A, Bathon JM, Emery P (2023) Rheumatoid arthritis. Lancet 402(10416):2019\u0026ndash;2033\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGravallese EM, Firestein GS (2023) Rheumatoid Arthritis - Common Origins, Divergent Mechanisms. NEW ENGL J MED 388(6):529\u0026ndash;542\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eConforti A, Di Cola I, Pavlych V, Ruscitti P, Berardicurti O, Ursini F, Giacomelli R, Cipriani P (2021) Beyond the joints, the extra-articular manifestations in rheumatoid arthritis. AUTOIMMUN REV 20(2):102735\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePrasad P, Verma S, Surbhi, Ganguly NK, Chaturvedi V, Mittal SA (2023) Rheumatoid arthritis: advances in treatment strategies. MOL CELL BIOCHEM 478(1):69\u0026ndash;88\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKonzett V, Aletaha D (2024) Management strategies in rheumatoid arthritis. NAT REV RHEUMATOL 20(12):760\u0026ndash;769\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRahmati M, Moosavi MA, McDermott MF (2018) ER Stress: A Therapeutic Target in Rheumatoid Arthritis? TRENDS PHARMACOL SCI 39(7):610\u0026ndash;623\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAcosta-Alvear D, Harnoss JM, Walter P, Ashkenazi A (2024) Homeostasis control in health and disease by the unfolded protein response. NAT REV MOL CELL BIO\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWiseman RL, Mesgarzadeh JS, Hendershot LM (2022) Reshaping endoplasmic reticulum quality control through the unfolded protein response. MOL CELL 82(8):1477\u0026ndash;1491\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMiglioranza Scavuzzi B, Holoshitz J: Endoplasmic Reticulum Stress, Oxidative Stress, and, Diseases R (2022) \u003cem\u003eANTIOXIDANTS-BASEL\u003c/em\u003e 11(7)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAo Q, Hu H, Huang Y (2024) Ferroptosis and endoplasmic reticulum stress in rheumatoid arthritis. FRONT IMMUNOL 15:1438803\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eNavid F, Colbert RA (2017) Causes and consequences of endoplasmic reticulum stress in rheumatic disease. NAT REV RHEUMATOL 13(1):25\u0026ndash;40\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLi J, Cui J, Wu L, Liu Y, Wang Q (2023) Machine learning and molecular subtype analyses provide insights into PANoptosis-associated genes in rheumatoid arthritis. ARTHRITIS RES THER 25(1):233\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eQian H, Deng C, Chen S, Zhang X, He Y, Lan J, Wang A, Shi G, Liu Y (2024) Targeting pathogenic fibroblast-like synoviocyte subsets in rheumatoid arthritis. ARTHRITIS RES THER 26(1):103\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eNygaard G, Firestein GS (2020) Restoring synovial homeostasis in rheumatoid arthritis by targeting fibroblast-like synoviocytes. NAT REV RHEUMATOL 16(6):316\u0026ndash;333\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSymons RA, Colella F, Collins FL, Rafipay AJ, Kania K, McClure JJ, White N, Cunningham I, Ashraf S, Hay E et al (2022) Targeting the IL-6-Yap-Snail signalling axis in synovial fibroblasts ameliorates inflammatory arthritis. ANN RHEUM DIS 81(2):214\u0026ndash;224\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZheng L, Gu M, Li X, Hu X, Chen C, Kang Y, Pan B, Chen W, Xian G, Wu X et al (2025) ITGA5(+) synovial fibroblasts orchestrate proinflammatory niche formation by remodelling the local immune microenvironment in rheumatoid arthritis. ANN RHEUM DIS 84(2):232\u0026ndash;252\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWu T, Hu E, Xu S, Chen M, Guo P, Dai Z, Feng T, Zhou L, Tang W, Zhan L et al (2021) clusterProfiler 4.0: A universal enrichment tool for interpreting omics data. Innov (Camb) 2(3):100141\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSzklarczyk D, Gable AL, Nastou KC, Lyon D, Kirsch R, Pyysalo S, Doncheva NT, Legeay M, Fang T, Bork P et al (2021) The STRING database in 2021: customizable protein-protein networks, and functional characterization of user-uploaded gene/measurement sets. NUCLEIC ACIDS RES 49(D1):D605\u0026ndash;D612\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWilkerson MD, Hayes DN (2010) ConsensusClusterPlus: a class discovery tool with confidence assessments and item tracking. Bioinformatics 26(12):1572\u0026ndash;1573\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eFriedman J, Hastie T, Tibshirani R (2010) Regularization Paths for Generalized Linear Models via Coordinate Descent. J STAT SOFTW 33(1):1\u0026ndash;22\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSanz H, Valim C, Vegas E, Oller JM, Reverter F (2018) SVM-RFE: selection and visualization of the most relevant features through non-linear kernels. BMC Bioinformatics 19(1):432\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePontisso I, Ornelas-Guevara R, Chevet E, Combettes L, Dupont G (2024) Gradual ER calcium depletion induces a progressive and reversible UPR signaling. PNAS Nexus 3(6):pgae229\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWeichert M, Guirao-Abad J, Aimanianda V, Krishnan K, Grisham C, Snyder P, Sheehan A, Abbu RR, Liu H, Filler SG et al (2020) Functional Coupling between the Unfolded Protein Response and Endoplasmic Reticulum/Golgi Ca(2+)-ATPases Promotes Stress Tolerance, Cell Wall Biosynthesis, and Virulence of Aspergillus fumigatus. MBIO 11(3)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhang Y, Li F, Liu L, Jiang H, Hu H, Du X, Ge X, Cao J, Wang Y (2019) Salinomycin triggers endoplasmic reticulum stress through ATP2A3 upregulation in PC-3 cells. BMC Cancer 19(1):381\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTao Z, Yang W, Zhu W, Wang L, Li KY, Guan X, Su Y (2024) A neural-related gene risk score for head and neck squamous cell carcinoma. ORAL DIS 30(2):477\u0026ndash;491\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eChemaly ER, Troncone L, Lebeche D (2018) SERCA control of cell death and survival. Cell Calcium 69:46\u0026ndash;61\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYang X, Lv L, Zhang Y, Zhang Z, Zeng S, Zhang X, Wang Q, Dorf M, Li S, Fu B (2025) ATP2A2 regulates STING1/MITA-driven signal transduction including selective autophagy. AUTOPHAGY :1\u0026ndash;16\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhang Z, Zhang C (2025) Regulation of cGAS-STING signalling and its diversity of cellular outcomes. NAT REV IMMUNOL 25(6):425\u0026ndash;444\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHong Z, Mei J, Li C, Bai G, Maimaiti M, Hu H, Yu W, Sun L, Zhang L, Cheng D et al (2021) STING inhibitors target the cyclic dinucleotide binding pocket. \u003cem\u003eP NATL ACAD SCI USA\u003c/em\u003e 118(24)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMarciniak SJ, Chambers JE, Ron D (2022) Pharmacological targeting of endoplasmic reticulum stress in disease. NAT REV DRUG DISCOV 21(2):115\u0026ndash;140\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eChen X, Shi C, He M, Xiong S, Xia X (2023) Endoplasmic reticulum stress: molecular mechanism and therapeutic targets. SIGNAL TRANSDUCT TAR 8(1):352\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKondo N, Kuroda T, Kobayashi D (2021) Cytokine Networks in the Pathogenesis of Rheumatoid Arthritis. INT J MOL SCI 22(20)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eJang S, Kwon E, Lee JJ (2022) Rheumatoid Arthritis: Pathogenic Roles of Diverse Immune Cells. INT J MOL SCI 23(2)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYu J, Wang H, Ding M, Zhao X, Zhang X (2025) Current and Future Landscape of SERCAs' Functions in Non-Excitatory Cells and Diseases. BioEssays :e70029\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAlivernini S, Firestein GS, McInnes IB (2022) The pathogenesis of rheumatoid arthritis. Immunity 55(12):2255\u0026ndash;2270\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhang B, Xu P, Ablasser A (2025) Regulation of the cGAS-STING Pathway. ANNU REV IMMUNOL 43(1):667\u0026ndash;692\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhu Q, Zhou H (2024) The role of cGAS-STING signaling in rheumatoid arthritis: from pathogenesis to therapeutic targets. FRONT IMMUNOL 15:1466023\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"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":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Rheumatoid arthritis, Endoplasmic reticulum stress, Fibroblast-like synoviocytes, ATPase Sarcoplasmic/Endoplasmic Reticulum Ca2 + Transporting 3","lastPublishedDoi":"10.21203/rs.3.rs-8165027/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8165027/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e\u003cp\u003eRheumatoid arthritis (RA) is an autoimmune disorder characterized by synovial inflammation and progressive bone destruction. Endoplasmic reticulum (ER) stress plays a key role in the pathogenesis of RA. However, the biomarkers of ER stress in RA remain elusive. We aimed to identify key ER stress-related genes (ERSRGs) in RA, evaluate their diagnostic potential, and investigate their functional roles.\u003c/p\u003e\u003ch2\u003eMethods.\u003c/h2\u003e\u003cp\u003eTranscriptomic data from six RA datasets were analyzed using machine learning algorithms. ATPase Sarcoplasmic/Endoplasmic Reticulum Ca2\u0026thinsp;+\u0026thinsp;Transporting 3 (ATP2A3) expression was validated via immunohistochemistry (IHC). Cytokine secretion and signaling pathways in RA fibroblast-like synoviocytes (FLS) were examined using enzyme-linked immunosorbent assay (ELISA) and Western blot analysis, respectively.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e\u003cp\u003eSeventeen ERSRGs were identified by intersecting differentially expressed genes (DEGs) between RA and control samples with ERSRGs from the Gene Ontology Consortium. RA patients were classified into two molecular subtypes based on the expression pattern of these 17 genes. Using machine learning, ATP2A3 emerged as key diagnostic biomarker in RA and was validated in external datasets. IHC further confirmed ATP2A3 was highly expressed in RA synovial tissues. Importantly, we found that ATP2A3 promotes the secretion of inflammatory cytokines in RA FLS. Mechanistically, ATP2A3 enhances the phosphorylation of stimulator of interferon genes (STING), thereby activating the STING signaling pathway and leading to increased inflammatory cytokine secretion.\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e\u003cp\u003eWe establish ATP2A3 as a promising diagnostic biomarker for ER stress in RA and demonstrate that it promotes the secretion of inflammatory cytokines in RA FLS through STING pathway activation.\u003c/p\u003e","manuscriptTitle":"ATP2A3 promotes inflammatory cytokine secretion in rheumatoid arthritis fibroblast-like synoviocytes via activation of the STING signaling pathway","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-11-23 15:53:36","doi":"10.21203/rs.3.rs-8165027/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"9b4b9206-12fe-4bbf-96a4-dd41f303628d","owner":[],"postedDate":"November 23rd, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":58324966,"name":"Cell Communication and Signaling"}],"tags":[],"updatedAt":"2025-11-23T15:53:36+00:00","versionOfRecord":[],"versionCreatedAt":"2025-11-23 15:53:36","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8165027","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8165027","identity":"rs-8165027","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}