Machine Learning-Based Identification of Molecular Signatures in PTOA Cell Subtypes via Single-Cell Transcriptomics

Machine Learning-Based Identification of Molecular Signatures in PTOA Cell Subtypes via Single-Cell Transcriptomics | 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 Machine Learning-Based Identification of Molecular Signatures in PTOA Cell Subtypes via Single-Cell Transcriptomics DuJiang Yang, Gaowen Gong, Junjie Chen, Jiafeng song, Zhijun Ye, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8655656/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 14 Apr, 2026 Read the published version in Stem Cell Reviews and Reports → Version 1 posted 9 You are reading this latest preprint version Abstract Objectives Osteoarthritis (OA) is the most prevalent joint disorder, whereas post-traumatic osteoarthritis (PTOA) denotes a form of arthritis that arises secondary to acute joint injury. Methods As no curative therapy for PTOA currently exists, prevention and early intervention constitute critical avenues of investigation. Among the various injuries predisposing to PTOA, anterior cruciate ligament (ACL) rupture is among the most prototypical. Our study systematically analyzed dynamic RNA sequencing data derived from murine knee joint tissues after ACL rupture. Granulocyte subpopulations were systematically classified and annotated, revealing subgroups related to PTOA progression. Furthermore, co-expressed gene modules positively correlated with OA granulocytes were identified and selected via high-dimensional weighted gene co-expression network analysis (hdWGCNA). Utilizing machine learning (ML)-based methodologies, a predictive model was constructed and validated through nomogram models, calibration curves, and decision curves. The prognostic utility of characteristic genes in these OA granulocyte subtypes was also investigated. Additionally, immune infiltration analysis (IIA) was carried out to visualize immune cell infiltration and explore the relationships between key genes and immune cells, and qRT-PCR was performed for the results. Results In summary, this study identified the distinctive molecular and biological characteristics of granulocyte subtypes and applied ML algorithms to predict diagnostic biomarkers specific to PTOA. Conclusions These findings hold promise for improving targeted predictive capabilities for the disease, with the ultimate goal of interrupting the vicious cycle of inflammation and mechanical abnormalities prior to the onset of irreversible joint damage. PTOA Single-Cell Analysis Granulocytes Bioinformatics ML IIA Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Key-points Predicting diagnostic markers based on the temporal dynamics of immune cell accumulation in mouse knee osteoarthritis. Understanding the role of various immune cells in joint degeneration or joint repair after injury is helpful for improving therapeutic strategies for treating OA. Using machine learning (ML)-based methodologies, a predictive model was constructed and validated through nomogram models, calibration curves, and decision curves. 1. Introduction 1.1 Overview of Post-traumatic osteoarthritis (PTOA) PTOA constitutes a clinically significant and etiologically distinct subtype of osteoarthritis (OA), directly precipitated by acute intra-articular mechanical insults, including ligamentous rupture, meniscal injury, or intra-articular fracture. PTOA is pathologically characterized by progressive degradation of articular cartilage, synovial inflammation, and aberrant remodeling of the subchondral bone, thereby contributing substantially to the increasing global burden of OA. With over 300 million individuals affected worldwide, PTOA accounts for a considerable subset of cases. Although its pathogenesis is more clearly delineated than that of primary OA, it nonetheless involves a comparably intricate pathological cascade [ 1 , 2 ]. The core pathophysiological mechanism is widely regarded as a profound disruption of the homeostatic equilibrium between extracellular matrix synthesis and degradation in articular cartilage and subchondral bone, initiated by multifactorial, multi-stage processes activated after injury [ 3 ]. The hallmark pathological manifestations of PTOA encompass progressive loss of articular cartilage, persistent synovitis, subchondral bone sclerosis, and osteophyte formation, culminating in chronic pain and functional disability. Recent investigations emphasize that aberrant subchondral bone remodeling is not merely a late-stage secondary event but is initiated at the early phase following joint trauma. This process involves abnormal bone turnover and irregular resorption, mediated by dysfunctional osteoclast and osteoblast activity, and is accompanied by microenvironmental perturbations like marrow fibrosis, aberrant angiogenesis, and neural invasion [ 4 – 6 ]. These changes create a self-perpetuating microenvironment conducive to joint degeneration. The initiating insult in PTOA, joint trauma, provokes a stress-induced innate and autoimmune response, mediated by the release of damage-associated molecular patterns (DAMPs). This response leads to a surge in pro-inflammatory cytokines (e.g., IL-1β, TNF-α, IL-6) and a concurrent decline in anti-inflammatory mediators (e.g., IL-10), generating profound metabolic disequilibrium. The resulting imbalance accelerates cartilage catabolism, suppresses anabolic repair, and directly promotes pathological subchondral bone remodeling [ 7 – 9 ]. Within this complex network, diverse cell types actively participate, including infiltrating innate immune cells (granulocytes, natural killer cells, monocytes/macrophages) and resident stromal cells (osteocytes, osteoblasts). Among these, granulocytes, particularly granulocytes, exert a pivotal influence in post-traumatic bone remodeling. Granulocytes mediate their effects via the release of granular proteases (e.g., granulocyte elastase, MMP-9) and formation of granulocyte extracellular traps (NETs). These processes accelerate bone matrix mineralization, regulate collagen fiber reorganization, modulate matrix protein phosphorylation, remodel the Haversian canal system and cortical bone, and foster hydroxyapatite crystal deposition, collectively driving PTOA progression [ 10 , 11 ]. Given these critical functions, granulocytes and their effector mediators have emerged as potential prognostic biomarkers and promising therapeutic targets in PTOA. Moreover, accumulating evidence indicates that granulocytes are a heterogeneous population; specific subsets (e.g., CD16⁺CD62L^low aged granulocytes or low-density granulocytes) exert distinct, and sometimes opposing, effects at different stages of PTOA. These subsets modulate the local immune microenvironment (e.g., T-cell polarization, osteoclastogenesis) with precision and actively promote calcified bone matrix deposition [ 12 ]. Therefore, precise identification and functional characterization of pathogenic versus protective granulocyte subsets are imperative for the development of novel immunomodulatory and targeted therapeutic strategies tailored to the heterogeneous clinical phenotypes of PTOA. 1.2 Single-Cell RNA Sequencing (scRNA-seq) scRNA-seq has been employed to define the comprehensive gene expression profiles of individual cells [ 13 , 14 ]. This technology has been widely utilized to dissect cellular heterogeneity that was previously obscured in bulk populations, thereby advancing insights into both physiological and pathological processes [ 15 ]. Through this high-resolution approach, the transcriptomic features of granulocytes in normal and osteoarthritic tissues have been delineated. Nevertheless, a comprehensive understanding of granulocyte subpopulation composition, gene expression programs, and molecular functions in osteoarthritic tissues remains limited [ 16 ]. Furthermore, the clinical relevance and prognostic implications of granulocyte subsets in PTOA pathogenesis have yet to be elucidated. 1.3 Machine Learning (ML) Validation ML represents a transformative paradigm in data science, capable of autonomously constructing predictive models by identifying key features within complex, high-dimensional datasets [ 17 ]. In comparison to conventional statistical methods, ML approaches offer marked advantages in analyzing large-scale biological data and have demonstrated significant value in the diagnosis and management of various diseases, including OA [ 18 ]. This study applied three complementary ML algorithms to identify robust biomarkers of PTOA. First, the least absolute shrinkage and selection operator (LASSO) logistic regression, a penalized linear modeling approach, was employed to select disease-specific feature genes by minimizing prediction error via cross-validation, thereby enhancing interpretability [ 19 ]. Second, support vector machine recursive feature elimination (SVM-RFE), an approach particularly effective in microarray analysis, was utilized to iteratively identify optimal feature combinations for classification by leveraging nonlinear discriminative capacity [ 20 – 22 ]. Third, the random forest (RF) algorithm was applied for its ability to model complex, nonlinear interactions among variables. These approaches mitigate the risk of model overfitting and, when integrated with bioinformatic analyses, substantially improve the diagnostic accuracy and translational utility of identified core genes [ 23 , 24 ]. Therefore, employing complementary ML strategies to define PTOA biomarkers and construct reliable predictive models holds significant promise for advancing personalized diagnosis and treatment. scRNA-seq datasets were retrieved from the Gene Expression Omnibus (GEO) ( http://www.ncbi.nlm.nih.gov/geo/ ), encompassing transcriptional profiles of normal and post-traumatic knee tissues in C57BL/6J mice during the critical window of 0–15 days following non-invasive anterior cruciate ligament (ACL) rupture. These datasets have been rigorously validated in prior literature [ 25 – 27 ]. The choice of a non-surgical mechanical ACL rupture model is particularly advantageous, as it closely replicates the biomechanical insult of human knee trauma while avoiding the confounding inflammatory responses induced by surgical intervention, thereby enhancing clinical translatability [ 28 , 29 ]. The C57BL/6J strain was selected for its well-defined genetic background and established use in OA research, ensuring reproducibility and facilitating comparison with an extensive body of prior work. Focusing on the acute phase (0–15 days post-injury) is strategically critical, as this period captures the earliest molecular and cellular events, including acute inflammation, early extracellular matrix degradation, and initiation of repair signaling, that drive PTOA pathogenesis. Elucidating this early cascade is essential for identifying upstream drivers of disease before irreversible joint damage ensues, thereby informing strategies for timely intervention [ 30 , 31 ]. 2. Materials and Methods 2.1. Data Collection and Processing The scRNA-seq dataset GSE200843 was retrieved from the GEO to investigate granulocyte heterogeneity in normal and PTOA human knee tissues [ 32 ]. Datasets GSE26475 and validation set GSE112641 were used for predictive model construction and validation [ 33 , 34 ] (Supplementary Table 5). Raw count data were processed via Seurat (v4.4.0) in R. Quality control was performed by filtering out cells with fewer than 500 or more than 6,000 detected genes, as well as cells in which the percentage of mitochondrial transcripts exceeded 25%. Normalization, variance stabilization, and scaling of raw counts were conducted utilizing the SCTransform function. To mitigate technical batch effects, data integration was performed utilizing Harmony (v1.2.1). Principal component analysis (PCA) was applied to the integrated dataset for initial dimensionality reduction. A shared nearest neighbor (SNN) graph was constructed utilizing the FindNeighbors function, followed by cell clustering with the FindClusters function, which implements the Louvain algorithm for modularity optimization. Cell clusters were visualized in two dimensions utilizing Uniform Manifold Approximation and Projection (UMAP). Each cluster was annotated into specific cell types based on the expression of well-established canonical marker genes. DEGs between conditions (PTOA vs. normal) and cluster-specific marker genes were identified utilizing the FindMarkers function in Seurat. A similar analytical pipeline was applied for sub-clustering analysis of granulocyte populations. 2.2. Pseudotime Trajectory and Intercellular Communication Analysis To reconstruct the developmental trajectory of granulocytes, pseudotemporal ordering analysis was enabled by Monocle (v2.32.0) with the DDRTree reduction method and default parameters. Dynamic expression patterns of key genes along the inferred trajectory were visualized utilizing the plot_pseudotime_heatmap function. Intercellular communication networks were systematically mapped via the CellChat toolbox (v1.6.1), analyzing ligand-receptor interactions based on the expression of a curated database of signaling pathways. 2.3. Functional Enrichment Analysis DEGs for specific cell subpopulations were identified via Seurat’s FindMarkers function, applying thresholds of absolute log2 fold-change (|log2FC|) > 1 and adjusted p-value (FDR) < 0.05. GO biological process enrichment analysis and single-sample GSEA (ssGSEA) were performed via clusterProfiler (v4.12.6) to interpret the functional themes of the DEGs. 2.4. hdWGCNA To identify granulocyte-associated gene modules correlated with PTOA, high-dimensional weighted gene co-expression network analysis (hdWGCNA) was carried out using R (v1.73). Meta-cells for each sample and cell cluster were first constructed utilizing the MetacellsByGroups function, aggregating data from 50 cells per meta-cell to reduce sparsity. The standard hdWGCNA pipeline was then executed sequentially on the granulocyte subset, encompassing the following key steps: TestSoftPowers to determine the appropriate soft-thresholding power, ConstructNetwork to build the co-expression network and identify modules, ModuleEigengenes to calculate MEs, ModuleConnectivity to compute intramodular connectivity (kME), and RunModuleUMAP for visualization. 2.5. Construction of ML-Based Prediction Models for PTOA Granulocyte-Associated Genes To evaluate the predictive potential of PTOA granulocyte-associated genes identified by hdWGCNA in PTOA progression, three ML algorithms, LASSO, RF, and SVM, were applied to further identify hub genes (Supplementary Table 6) [ 35 ]. The predictive performance of these models was assessed utilizing ROC curves generated with “timeROC” in R v0.4 in dataset GSE26475 and validation set GSE112641. 2.6. IIA CIBERSORT, which estimated relative subsets of RNA transcripts to infer cell type composition, was used to predict the relative abundance of 20 infiltrating immune cell types. Boxplots were generated to compare immune cell composition between the normal and disease groups. Additionally, Spearman correlation analysis was performed to explore the relation of the expression levels of key genes to the abundance of immune cells. The correlations were visualized utilizing a lollipop plot generated via “ggplot2” 3.5.1. 2.7. PTOA Mouse Model 12 healthy 10-week-old male C57BL/6 mice, weighing 26.5 ± 1.7 g, were obtained from the Experimental Animal Center of Southwest Medical University (License No.: SCXK [Chuan] 2024-0046). Animals were acclimated for one week under standard housing conditions (room temperature 21–25°C; relative humidity 45%-65%) with unrestricted activity and free access to food and water. All experimental procedures were approved by the Institutional Animal Ethics Committee of Southwest Medical University. Following one week of acclimatization, mice were randomized by a random-number method to either the sham-operation group (n = 6) or the anterior cruciate ligament transection (ACLT) group (n = 6). PTOA induction in the model group was performed using the ACLT procedure as described previously [ 36 ]. General anesthesia was administered via intraperitoneal injection of 1% sodium pentobarbital (40 mg/kg). With the knee joint fixed, the skin was prepared and disinfected. A medial parapatellar incision was made, and tissues were dissected layer by layer to expose the joint cavity. The anterior cruciate ligament (ACL) was fully exposed and transected using microsurgical scissors. After repositioning the patella, a drawer test was performed to confirm a complete ACL rupture. The incision was closed in layers and disinfected again. In the sham-operation group, the joint cavity was exposed without ACL transection. At 12 weeks after the operation, mice were euthanized, and the right knee joints were harvested for anatomical examination. All procedures strictly adhered to applicable guidelines for animal experimentation in mice. 2.8 RT-PCR Validation of Key Genes Approximately 0.02 g of synovial tissue from the right hind-knee joints of mice in the disease and sham-operation groups was collected. Total RNA was extracted using the Trizol method after thorough homogenization, lysis, centrifugation, and purification. RNA was reverse-transcribed into cDNA using the SweScript All-in-One SuperMix. Quantitative real-time PCR was performed using Universal Blue SYBR Green qPCR Master Mix on a CFX Connect Real-Time System (Bio-Rad). Target gene expression was normalized to the reference gene GAPDH and calculated through the ΔΔCT method. All procedures were repeated to ensure the accuracy and reliability of the results. Primer sequences were: Mpp7 (forward: 5′-TCCAGAACAAGCCACCAAACA-3′; reverse: 5′-CACAGAGTCTTCCTCGTCATCAA-3′); Tgfbi (forward: 5′-CTGTTGCCGAAACCGACATC-3′; reverse: 5′-CAGGGGCAAGTCGCATAGAA-3′); GAPDH (reference gene) (forward: 5′-CTGGAGAAACCTGCCAAGTATG-3′; reverse: 5′-GGTGGAAGAATGGGAGTTGCT-3′). 2.9. Statistical Analysis All biological experiments included at least three biological replicates. For measurement data expressed as mean ± standard error of the mean (SEM), independent-sample t-tests were performed. All statistical analyses and data visualizations were conducted via R 4.4.1. The Spearman correlation coefficient was employed to investigate associations between continuous variables. The differences between the two groups were determined via the Wilcoxon signed-rank test. p < 0.05 denoted statistical significance. Statistical notations were as follows: * for p < 0.05, ** for p < 0.01, *** for p < 0.001, **** for p 0.05. 3. Results 3.1 scRNA-seq Reveals the Transcriptomic Landscape of granulocytes in Normal and PTOA Tissues To investigate the cellular composition and diversity in normal and PTOA tissues, scRNA-seq data from mouse PTOA tissue were collected and analyzed. After quality control was implemented, batch effects were mitigated and cell types were annotated; the samples were classified into OA and normal groups [ 37 ]. Nine major cell populations were identified: granulocytes, macrophages, fibroblasts, erythrocytes, monocytes, B, T, endothelial, NK cells, and others (Figs. 1 A and 1 B). Figure 1 C illustrates the temporal dynamics of each cell population across the analyzed samples. Notably, the proportion of granulocyte clusters increased over time in the OA group in comparison to the normal group (Supplementary Table 1). Furthermore, differentially expressed genes (DEGs) were analyzed, and the expression of lineage-specific markers was examined across the nine major cell clusters (Figs. 1 D and 1 E). Unsupervised hierarchical subclustering was performed on granulocytes isolated from both PTOA and normal knee tissues to further elucidate their heterogeneity within the articular microenvironment. This analysis identified five transcriptionally distinct granulocyte subpopulations (Fig. 2 A). Comparative quantification of their abundances revealed two clusters exhibiting statistically significant temporal alterations in PTOA relative to normal samples: one cluster was markedly expanded, whereas the other was reduced under disease conditions. These were subsequently designated as OA-associated granulocytes (OA granulocytes) and non-OA-associated granulocytes (NOA granulocytes), respectively (Figs. 2 A, 2 B). The identity of each cluster was corroborated by the expression of unique marker gene signatures, visualized via feature plots (Fig. 2 C). The transcriptional programs characterizing these subtypes were subsequently delineated by computing relative gene expression scores across individual cells, followed by unsupervised clustering, which resolved distinct co-expression modules (Fig. 2 D). Genes within each module were further grouped according to similarity in expression dynamics, yielding five principal patterns potentially related to discrete biological processes. Notably, genes upregulated in the OA granulocyte cluster were significantly enriched in pathways directly implicated in the pathogenesis of OA, including cytokine receptor activity and cytokine receptor binding (Fig. 2 D). In parallel, gene set enrichment analysis (GSEA) demonstrated a significant negative correlation between the OA granulocyte signature and gene sets related to inflammatory response activation (Fig. 2 E). The foregoing findings highlight a pivotal role of OA granulocytes in promoting the initiation and progression of OA through aberrant cytokine signaling, rather than through canonical inflammatory pathways. 3.2. Pseudotemporal Trajectory Analysis of PTOA Granulocyte Subtypes during Arthritis Progression To elucidate the cellular origins and developmental trajectories of granulocytes related to PTOA during disease progression, pseudotime trajectory analysis was applied to granulocyte subpopulations. The analysis positioned granulocyte cluster 1, together with non-OA granulocyte clusters, near the origin of the trajectory. In contrast, OA granulocytes were predominantly distributed at the branchpoint of trajectory 1 and extended toward the termini of trajectories 2 and 3 (Figs. 3 A- 3 C). To identify genes critical to arthritis progression, dynamic gene expression changes along the pseudotemporal axis were examined. This revealed a set of genes exhibiting significant variation during the maturation of OA granulocytes, including Cd44, a recognized marker of chondrocytes, Cebpb, Il1r2, Mcl1, and Plk3 (Figs. 3 D, 3 E). The top 50 variable genes were subsequently clustered according to pseudotemporal expression patterns, followed by functional enrichment analyses for each cluster. Five distinct gene clusters with unique temporal expression profiles were identified (Fig. 3 F). Specifically, genes in cluster 1 displayed high expression from early to mid pseudotime, whereas clusters 3, 4, and 5 reached peak expression during the late phase, indicating stage-specific functional roles in granulocyte differentiation in PTOA. 3.3. Intercellular Communication Analysis of Granulocyte Subpopulations in PTOA The availability of single-cell datasets provides an excellent opportunity to explore the intercellular communication mediated through ligand-receptor interactions. The intercellular communication network involving granulocyte subpopulations and other cell types in both post-traumatic arthritic and normal joint tissues was examined via CellChat. Overall, OA granulocytes exhibited robust communication with other cell types during the pathogenesis of post-traumatic arthritis (Fig. 4 A). OA granulocytes displayed higher interaction intensity, both qualitatively and quantitatively, in comparison to other granulocyte populations (Fig. 4 B). Notably, OA granulocyte clusters engaged in direct interactions with other granulocyte subtypes through adhesion ligand-receptor pairs like Cxcl2/Cxcr2, Ccl6/Ccr1, and Thbs1/Cd36 (Figs. 4 C, 4 D). Moreover, our analysis revealed the upregulation of pro-inflammatory signals, including CD52, SELPLG, and CXCL, in the communication between OA granulocytes and other granulocyte subsets (Figs. 4 E- 4 G). Collectively, our findings suggest that OA granulocytes establish a reciprocal interaction network with other cell types, promoting mutual support and functional maintenance. 3.4 Identification of Co-expressed Gene Modules Correlated with Granulocytes in PTOA via hdWGCNA hdWGCNA, a comprehensive framework for co-expression network analysis in scRNA-seq data, was employed to identify co-expressed gene modules and explore their functional roles in PTOA-associated granulocytes. An ideal soft-threshold power of 7 was used to build a scale-free co-expression network (Fig. 5 A). 12 distinct gene co-expression modules were identified. Specifically, the yellow, brown, and black ones were highly activated, primarily in OA granulocytes (Figs. 5 B, 5 C, 5 D). Further analysis was conducted to examine the correlations between each module (Figs. 5 E- 5 F). Figure 5 G displays the top 10 hub genes from the three highly activated modules. 3.5. Construction of PTOA Granulocyte-Related Predictive Models Based on ML Subsequently, the hub genes of three PTOA granulocyte-associated modules identified through hdWGCNA, DEGs of OA and NOA granulocytes, and the DEGs from the external dataset GSE26475 were intersected to predict the onset and progression of PTOA (Fig. 6 A; Supplementary Tables 2–4). Three ML algorithms were further applied to the external dataset (GSE26475) to identify granulocyte-related hub genes related to PTOA. Through the LASSO regression algorithm, five key genes closely related to prognosis were identified (Fig. 6 B- 6 C). SVM-RFE analysis yielded prediction error curves and cross-validation error rates, with lower values indicating reduced error rates (Fig. 6 D). In addition, RF analysis ranked all genes according to importance and highlighted the top 10 candidates (Fig. 6 E). The overlapping genes obtained from the three ML approaches were visualized utilizing a Venn diagram (Fig. 6 F). Expression levels of four genes were then displayed with box plots and ROC curves (Figs. 6 G- 6 H). Validation utilizing the independent dataset GSE112641 revealed that Tgfbi and Mpp7 achieved AUC values greater than 0.7, suggesting diagnostic value (Figs. 6 I- 6 J). Finally, a nomogram was constructed to estimate the probability of disease occurrence based on these two genes, and calibration curves confirmed the accuracy of the predictive model (Figs. 6 K- 6 L). The foregoing findings indicate that two key genes, Tgfbi and Mpp7, are strongly related to prognosis in the PTOA population. 3.6. Immune Infiltration Analysis (IIA) of Granulocyte-Related Prognostic Features in PTOA An assessment of immune cell infiltration in PTOA was carried out. Given that the synovial microenvironment influences the prognosis of patients with PTOA and that immune cells are pivotal within this microenvironment, our study included a detailed analysis of immune cell infiltration in synovial tissue. A stacked bar chart revealed the composition of 20 immune cell types (Fig. 7 A). Violin plots demonstrated significantly higher infiltration levels of activated M1 macrophages in the PTOA group, whereas naive B cell infiltration was reduced (Fig. 7 B). IIA and correlation analyses further showed that naive B cells were significantly correlated with Mpp7 (p < 0.001) and Tgfbi (p < 0.001). Moreover, M1 macrophages exhibited significant correlations with Mpp7 (p < 0.05) and Tgfbi (p < 0.01) (Figs. 7 C- 7 D). 3.7. Expression of Key Genes in In Vitro Experiments To validate the expression of Mpp7 and Tgfbi in the synovium of PTOA, a PTOA animal model was established, and RT-PCR analysis was performed. The expression of Mpp7 was lower in the PTOA model group than in the control group (t = 21.46, p < 0.0002) (Fig. 8 A). In contrast, Tgfbi mRNA levels were significantly elevated in the PTOA model group compared with controls (t = 22.16, p = 0.0014) (Fig. 8 B), consistent with the trend identified in our analyses. 4. Discussion Advances in scRNA-seq have profoundly transformed our ability to dissect cellular heterogeneity within complex tissues. Leveraging this technology, the granulocyte lineage in OA joints was analyzed to identify two principal granulocyte subpopulations with distinct disease associations. These subsets were designated as OA granulocytes and NOA (non-OA) granulocytes, reflecting their dynamic distribution in PTOA tissues versus normal tissues. Subsequent in-depth characterization of their molecular features and regulatory pathways, coupled with bioinformatic analyses, identified two hub genes, Tgfbi and Mpp7, as critical mediators. Granulocytes in OA extend beyond mere bystanders of inflammation, encompassing complex, stage-specific roles in disease pathogenesis. Our study identified MMP8, Ngp, Cxcl2, S100a8, and S100a9 as core-enriched and DEGs within PTOA granulocytes. During the acute post-injury phase of OA, granulocytes are rapidly recruited to the joint via a finely orchestrated chemokine cascade, primarily mediated by Cxcl1, Cxcl2, and their receptor Cxcr2. Their activation triggers a robust yet frequently deleterious response, characterized by the release of serine proteases (granulocyte elastase, proteinase 3), matrix metalloproteinases (MMPs) (MMP8, MMP9), and reactive oxygen species, which directly contribute to cartilage matrix degradation and synovitis [ 38 – 40 ]. Emerging single-cell transcriptomic evidence reveals substantial heterogeneity among infiltrating granulocytes, indicating the coexistence of functionally distinct subpopulations, like a pro-inflammatory (N1-like) phenotype marked by high S100a8/a9 and Cxcr2 expression [ 41 ], suggesting their potential role in immune modulation within the arthritic joint microenvironment. Through extensive characterization of chondrocyte heterogeneity in arthritic tissue, a unique granulocyte subpopulation predominantly present in joint cartilage was successfully identified. The enriched molecular functions of this subset, including endopeptidase regulatory activity, chemokine receptor binding, and broader peptidase regulatory activity, are critically dysregulated in OA pathogenesis. These pathways act in an interconnected manner to drive disease progression [ 42 ]. Specifically, proteins with chemokine receptor binding functions are essential for the initial recruitment and activation of immune cells, like granulocytes and macrophages, into the synovial joint. This sustained inflammatory influx establishes a foundation for the destructive phase, which is primarily mediated by two additional pathways. Endopeptidase regulators and broader peptidase activities directly govern intra-articular proteolytic cascades, modulating the activity and function of key enzymes, including MMPs and aggrecanases (ADAMTS), which are responsible for catastrophic degradation of cartilage extracellular matrix components, particularly collagen and proteoglycans [ 43 , 44 ]. Therefore, the aberrant expression of genes within these three functional categories establishes a vicious cycle: chemokine-driven inflammation promotes the expression and activation of peptidases, whose uncontrolled activity degrades the matrix and releases bioactive fragments, further amplifying chemokine production and chronic inflammation. Among these, Cd44, Cebpb, Il1r2, Mcl1, and Plk3 emerged as the most significantly altered genes in OA granulocytes within arthritic cartilage tissues, playing pivotal roles in OA pathogenesis [ 45 ]. Our findings indicated that OA granulocytes had increased expression of genes related to inflammatory and reparative processes. For instance, potential anti-inflammatory agents like PCA interact with chondrocytes via Cd44, a differentiation cluster on the cell surface. CEBPB, encoding CCAAT/enhancer-binding protein β (C/EBPβ), is a transcription factor and one of the strongest trans-activators in chondrocytes, playing a crucial role in endochondral ossification [ 46 ]. Additionally, IL-1β, widely recognized as a key cytokine in OA progression, binds to its membrane receptor IL-1R1, activating transcription factors like MAPK and NF-κB, thereby inducing inflammatory mediators [ 47 ]. Conversely, NOA granulocyte abundance demonstrated an inverse correlation with disease progression. Intercellular communication analysis of key ligand-receptor pairs, including Cxcl2/Cxcr2, Ccl6/Ccr1, and Thbs1/Cd36, as well as genes like CD52, SELPLG, and members of the CXCL family, revealed a complex cell-cell signaling network driving inflammation in OA. The Cxcl2/Cxcr2 axis serves as a principal driver of granulocyte recruitment and activation. Following joint injury, synovial cells and macrophages secrete Cxcl2, which binds to Cxcr2 on granulocytes, triggering their influx into the joint and amplifying the acute inflammatory response. Simultaneously, the Ccl6/Ccr1 pathway facilitates the recruitment of monocytes and macrophages, further sustaining a pro-inflammatory synovial microenvironment. The Thbs1/Cd36 interaction introduces a critical regulatory dimension; while Thbs1 modulates cell adhesion and angiogenesis, its binding to Cd36 on immune cells may influence inflammatory signaling and the clearance of apoptotic debris, a process often dysregulated in chronic OA [ 48 – 50 ]. Moreover, the involvement of CD52 (a mediator of lymphocyte activation), SELPLG (PSGL-1, essential for leukocyte tethering and rolling on endothelium), and other CXCL chemokines underscores the multifaceted nature of this communication network, encompassing both innate and adaptive immune cells. The cooperative action of these pathways establishes a vicious cycle of leukocyte recruitment, persistent synovitis, and cartilage degradation, positioning them as central mediators of PTOA pathology and potential therapeutic targets. Based on these findings, our study employed three distinct algorithmic models, support vector machine (SVM), LASSO, and RF, for feature selection, with external datasets used for validation. Receiver operating characteristic (ROC) curve analysis confirmed the accuracy of the results. Tgfbi and Mpp7 were significantly differentially expressed. IIA revealed that Tgfbi, an extracellular matrix protein induced by TGF-β signaling, constitutes a critical risk factor; its pathological overexpression in chondrocytes leads to abnormal accumulation within the cartilage matrix. Mpp7, conversely, appears to act as a protective factor, although its association with B cells requires further investigation. Finally, a nomogram was constructed to enable quantitative prediction of disease risk. Overall, utilizing bioinformatics approaches, our study identified these hub genes and established a predictive model capable of effectively forecasting the prognosis of patients with PTOA, providing valuable insights for disease prediction and intervention. Nevertheless, the study has certain limitations. Although public datasets offer valuable resources, the sample sizes remain relatively small. While supportive evidence for our findings exists in previously published studies, the markers were derived from our analyses. Future studies will further investigate the expression patterns of hub genes across different disease stages and explore factors influencing their expression, thereby contributing to the early diagnosis and treatment of PTOA. 5. Conclusions In this study, Tgfbi and Mpp7 were identified as potential diagnostic biomarkers for PTOA. Overall, these findings enhance the understanding of granulocyte heterogeneity within PTOA tissues. The identification of unique molecular and biological features of granulocyte subtypes, together with the prognostic utility of characteristic genes, provides valuable insights into the pathogenesis of PTOA. Declarations Clinical trial number Not applicable. Declaration of Artificial Intelligence The author declare that no artificial intelligence was used at the time of writing the paper. Ethics approval and consent to participate The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. All experimental procedures were approved by the Institutional Animal Ethics Committee of Southwest Medical University. Consent for publication Not applicable. Declaration of interests The authors declare that they have no competing interests. Funding This work was supported by the Study on the Efficacy and Mechanism of Shaoyang Shenggu Formula in Treating Early to Mid-Stage Knee Osteoarthritis Based on a "Blood-Entry Components-Targets-Pathways" Network (25NSFSC1423). Data statement The datasets used and analysed during the current study are available from the corresponding author on reasonable request. Author Contributions All authors contributed to the study conception and design. Writing - original draft preparation: Gaowen Gong; Dujiang Yang ; Writing - review and editing: Gaowen Gong; Dujiang Yang; Junjie Chen ; Conceptualization: Gaowen Gong; Dujiang Yang; Junjie Chen ; Methodology: Gaowen Gong; Dujiang Yangi ; Formal analysis and investigation: Gaowen Gong; Dujiang Yang; Jiafeng Song ; Funding acquisition: Guoyou Wang; Resources: Guoyou Wang; Supervision: Guoyou Wang, and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript. Acknowledgments Not applicable. References Drugs for Osteoarthritis (2021). Jama . ;325(6):581–582. doi: 10.1001/jama.2020.8395 . Katz, J. N., Arant, K. R., & Loeser, R. F. (2021). Diagnosis and Treatment of Hip and Knee Osteoarthritis: A Review. Jama , 325 (6), 568–578. 10.1001/jama.2020.22171 Slomski, A. (2020). Six Weeks of Prednisolone Reduced Hand Osteoarthritis Pain. Jama , 323 (4), 301. 10.1001/jama.2019.20918 Jeon, O. H., Kim, C., Laberge, R. M., Demaria, M., Rathod, S., Vasserot, A. P., et al. (2017). Local clearance of senescent cells attenuates the development of post-traumatic osteoarthritis and creates a pro-regenerative environment. Nature Medicine , 23 (6), 775–781. 10.1038/nm.4324 Schäfer, N., & Grässel, S. (2022). Targeted therapy for osteoarthritis: progress and pitfalls. Nature Medicine , 28 (12), 2473–2475. 10.1038/s41591-022-02057-x Baker, M. C., Sheth, K., Lu, R., Lu, D., von Kaeppler, E. P., Bhat, A., et al. (2023). Increased risk of osteoarthritis in patients with atopic disease. Annals Of The Rheumatic Diseases , 82 (6), 866–872. 10.1136/ard-2022-223640 Kelly, P. N. (2017). Targeting senescence to combat osteoarthritis. Science , 356 (6338), 595–596. 10.1126/science.356.6338.595-b Sun, Y., Wu, Q., Dai, K., You, Y., & Jiang, W. (2021). Generating 3D-cultured organoids for pre-clinical modeling and treatment of degenerative joint disease. Signal Transduct Target Ther , 6 (1), 380. 10.1038/s41392-021-00675-4 Nesic, D., Whiteside, R., Brittberg, M., Wendt, D., Martin, I., & Mainil-Varlet, P. (2006). Cartilage tissue engineering for degenerative joint disease. Advanced Drug Delivery Reviews , 58 (2), 300–322. 10.1016/j.addr.2006.01.012 Slomski, A. (2022). Surgical Procedures for Ankle Osteoarthritis Produce Similar Results. Jama , 328 (24), 2387. 10.1001/jama.2022.20054 Centor, R. M., & Perez, P. L. (2023). Web Exclusive. Annals On Call - Exercise for Knee Osteoarthritis: Is More Better? Annals Of Internal Medicine , 176 (3), eA220006. 10.7326/a22-0006 Metcalfe, D., Perry, D. C., Claireaux, H. A., Simel, D. L., Zogg, C. K., & Costa, M. L. (2019). Does This Patient Have Hip Osteoarthritis? The Rational Clinical Examination Systematic Review. Jama , 322 (23), 2323–2333. 10.1001/jama.2019.19413 Van de Sande, B., Lee, J. S., Mutasa-Gottgens, E., Naughton, B., Bacon, W., Manning, J., et al. (2023). Applications of single-cell RNA sequencing in drug discovery and development. Nature Reviews. Drug Discovery , 22 (6), 496–520. 10.1038/s41573-023-00688-4 Han, J., DePinho, R. A., & Maitra, A. (2021). Single-cell RNA sequencing in pancreatic cancer. Nature Reviews. Gastroenterology & Hepatology , 18 (7), 451–452. 10.1038/s41575-021-00471-z Poran, A., Nötzel, C., Aly, O., Mencia-Trinchant, N., Harris, C. T., Guzman, M. L., et al. (2017). Single-cell RNA sequencing reveals a signature of sexual commitment in malaria parasites. Nature , 551 (7678), 95–99. 10.1038/nature24280 Zhong, S., Zhang, S., Fan, X., Wu, Q., Yan, L., Dong, J., et al. (2018). A single-cell RNA-seq survey of the developmental landscape of the human prefrontal cortex. Nature , 555 (7697), 524–528. 10.1038/nature25980 Rinehart, N. I., Saunthwal, R. K., Wellauer, J., Zahrt, A. F., Schlemper, L., Shved, A. S., et al. (2023). A machine-learning tool to predict substrate-adaptive conditions for Pd-catalyzed C-N couplings. Science , 381 (6661), 965–972. 10.1126/science.adg2114 Jayakrishnan, T. T., Sangwan, N., Barot, S. V., Farha, N., Mariam, A., Xiang, S., et al. (2024). Multi-omics machine learning to study host-microbiome interactions in early-onset colorectal cancer. NPJ Precis Oncol , 8 (1), 146. 10.1038/s41698-024-00647-1 Xu, J., Liang, C., & Li, J. (2022). A signal recognition particle-related joint model of LASSO regression, SVM-RFE and artificial neural network for the diagnosis of systemic sclerosis-associated pulmonary hypertension. Frontiers In Genetics , 13 , 1078200. 10.3389/fgene.2022.1078200 Zihan, C. L. Y., Zong-wei, H., & Sufang, Q. (2023). 910P Machine learning-based survival prediction model for postoperative parotid mucoepidermoid carcinoma. ANNALS OF ONCOLOGY , 34 (0923–7534), S578. Tateishi, A. H. H., Takasawa, K., Kouno, N., Mizuno, T., Okubo, Y., et al. (2023). Development and validation of a prediction model for the recurrence of stage 1 EGFR mutation positive NSCLC in patients using machine learning with WES-based gene sets. JOURNAL OF CLINICAL ONCOLOGY , 41 (0732-183X), 8563. He, B. Y. P., Taghavi, M. M., Secondus-Samuel, S., Albaugh, J., Garey, J. S., et al. (2023). A machine learning model to predict treatment initiation among new patients in a community oncology network. JOURNAL OF CLINICAL ONCOLOGY , 41 (0732-183X), e13539. Aleksiuk, V., Baleisis, J., Kirdaite, G., Uzieliene, I., Denkovskij, J., Bernotas, P., et al. (2023). Evaluation of Cartilage Integrity Following Administration of Oral and Intraarticular Nifedipine in a Murine Model of Osteoarthritis. Biomedicines , 11 (9). 10.3390/biomedicines11092443 Park, I. H. J., Hwang, H., & Kim, H. (2018). Evaluation of cartilage degeneration and osteoarthritis pain on female and male mouse model of osteoarthritis. Osteoarthritis and Cartilage , 26 (1063–4584), S360. Joshi, N., Yan, J., Dang, M., Slaughter, K., Wang, Y., Wu, D., et al. (2025). A mechanically resilient soft hydrogel improves drug delivery for treating post-traumatic osteoarthritis in physically active joints. Proc Natl Acad Sci U S A , 122 (14), e2409729122. 10.1073/pnas.2409729122 Van Zeeland, E. M., Kassel, B., Montoya, T., Saviola, A. J., Burton, L. H., Easley, J. T., et al. (2025). Structural, functional, and proteomic based sex differences in murine post-traumatic osteoarthritis development following mechanical anterior cruciate ligament injury. Osteoarthritis Cartilage , 33 (8), 992–1006. 10.1016/j.joca.2025.05.004 Leite, C. B. G., Fricke, H. P., Tavares, L. P., Nshimiyimana, R., Mekhail, J., Kilgallen, E., et al. (2025). Maresin 1-LGR6 axis mitigates inflammation and posttraumatic osteoarthritis after transection of the anterior cruciate ligament in mice. Osteoarthritis Cartilage , 33 (7), 861–873. 10.1016/j.joca.2025.03.005 Takahata, K., Lin, Y. Y., Osipov, B., Arakawa, K., Enomoto, S., Christiansen, B. A., et al. (2025). Concurrent joint contact in anterior cruciate ligament injury induces cartilage micro-injury and subchondral bone sclerosis, resulting in knee osteoarthritis. Osteoarthritis Cartilage . 10.1016/j.joca.2025.06.007 Ai, H., Dou, C., Wu, Y., Zhang, D., Zhang, Z., Zhang, C., et al. (2025). Osteoclast-derived apoptotic bodies accelerate the pathological progression of osteoarthritis via disturbing subchondral bone remodeling. J Orthop Translat , 51 , 108–118. 10.1016/j.jot.2025.01.004 Che, X., Zhang, C., Zhuo, Y., Li, H., Ding, X., Wu, G., et al. (2025). Serum TSP-1 is a useful biomarker in severity assessment and the diagnosis of osteoarthritis. J Transl Med , 23 (1), 987. 10.1186/s12967-025-07022-z Khan, N. M., Kaiser, J. M., Chihab, S., Eng, T., & Drissi, H. (2025). Dimethyl Fumarate attenuates synovial inflammation, reduces nociception, and inhibits the development of post-traumatic osteoarthritis. Biomedicine & Pharmacotherapy , 183 , 117865. 10.1016/j.biopha.2025.117865 Sebastian, A., Hum, N. R., McCool, J. L., Wilson, S. P., Murugesh, D. K., Martin, K. A., et al. (2022). Single-cell RNA-Seq reveals changes in immune landscape in post-traumatic osteoarthritis. Frontiers In Immunology , 13 , 938075. 10.3389/fimmu.2022.938075 Burleigh, A., Chanalaris, A., Gardiner, M. D., Driscoll, C., Boruc, O., Saklatvala, J., et al. (2012). Joint immobilization prevents murine osteoarthritis and reveals the highly mechanosensitive nature of protease expression in vivo. Arthritis And Rheumatism , 64 (7), 2278–2288. 10.1002/art.34420 Sebastian, A., Chang, J. C., Mendez, M. E., Murugesh, D. K., Hatsell, S., Economides, A. N., et al. (2018). Comparative Transcriptomics Identifies Novel Genes and Pathways Involved in Post-Traumatic Osteoarthritis Development and Progression. International Journal Of Molecular Sciences , 19 (9). 10.3390/ijms19092657 Um, I. H., Scott-Hayward, L., MacKenzie, M., Tan, P. H., Kanesvaran, R., Choudhury, Y., et al. (2019). 978P - Algorithms derived from quantitative pathology can be a gatekeeper in patient selection for clinical trials in localised clear cell renal cell carcinoma (ccRCC). Annals of Oncology , 30 , v397. 10.1093/annonc/mdz249.074 Glasson, S. S., Blanchet, T. J., & Morris, E. A. (2007). The surgical destabilization of the medial meniscus (DMM) model of osteoarthritis in the 129/SvEv mouse. Osteoarthritis Cartilage , 15 (9), 1061–1069. 10.1016/j.joca.2007.03.006 Hautz, T., Salcher, S., Fodor, M., Sturm, G., Ebner, S., Mair, A., et al. (2023). Immune cell dynamics deconvoluted by single-cell RNA sequencing in normothermic machine perfusion of the liver. Nature Communications , 14 (1), 2285. 10.1038/s41467-023-37674-8 Greven, D. E., Cohen, E. S., Gerlag, D. M., Campbell, J., Woods, J., Davis, N., et al. (2015). Preclinical characterisation of the GM-CSF receptor as a therapeutic target in rheumatoid arthritis. Annals Of The Rheumatic Diseases , 74 (10), 1924–1930. 10.1136/annrheumdis-2014-205234 Crotti, C., Agape, E., Becciolini, A., Biggioggero, M., & Favalli, E. G. (2019). Targeting Granulocyte-Monocyte Colony-Stimulating Factor Signaling in Rheumatoid Arthritis: Future Prospects. Drugs , 79 (16), 1741–1755. 10.1007/s40265-019-01192-z Caxaria, S., Kouvatsos, N., Eldridge, S. E., Alvarez-Fallas, M., Thorup, A. S., Cici, D., et al. (2023). Disease modification and symptom relief in osteoarthritis using a mutated GCP-2/CXCL6 chemokine. Embo Molecular Medicine , 15 (1), e16218. 10.15252/emmm.202216218 Zhang, J., Li, K., & Qiu, X. (2024). Exploring causal correlations between inflammatory cytokines and knee osteoarthritis: a two-sample Mendelian randomization. Frontiers In Immunology , 15 , 1362012. 10.3389/fimmu.2024.1362012 Seco-Calvo, J., Sánchez-Herráez, S., Casis, L., Valdivia, A., Perez-Urzelai, I., Gil, J., et al. (2020). Synovial fluid peptidase activity as a biomarker for knee osteoarthritis clinical progression. Bone Joint Res , 9 (11), 789–797. 10.1302/2046-3758.911.Bjr-2020-0022.R2 Li, T., Peng, J., Li, Q., Shu, Y., Zhu, P., & Hao, L. (2022). The Mechanism and Role of ADAMTS Protein Family in Osteoarthritis. Biomolecules , 12 (7). 10.3390/biom12070959 Hurley, M. M., Coffin, J. D., Doetschman, T., Valera, C., Clarke, K., & Xiao, L. (2022). FGF receptor inhibitor BGJ398 partially rescues osteoarthritis-like phenotype in older high molecular weight FGF2 transgenic mice via multiple mechanisms. Scientific Reports , 12 (1), 15968. 10.1038/s41598-022-20269-6 Hsiao, Y. W., Chi, J. Y., Li, C. F., Chen, L. Y., Chen, Y. T., Liang, H. Y., et al. (2022). Disruption of the pentraxin 3/CD44 interaction as an efficient therapy for triple-negative breast cancers. Clin Transl Med , 12 (1), e724. 10.1002/ctm2.724 Armanmehr, A., Jafari Khamirani, H., Zoghi, S., & Dianatpour, M. (2022). Analysis of DYRK1B, PPARG, and CEBPB Expression Patterns in Adipose-Derived Stem Cells from Patients Carrying DYRK1B R102C and Healthy Individuals During Adipogenesis. Metabolic Syndrome And Related Disorders , 20 (10), 576–583. 10.1089/met.2021.0140 Panaroni, C., Fulzele, K., Soucy, R., Siu, K., Mukaihara, K., Huang, C., et al. (2018). Arachidonic Acid Induces Ferroptosis-Mediated Cell-Death in Multiple Myeloma. Blood , 132 , 4498. 10.1182/blood-2018-99-118482 Yu, E., Zhang, M., Xu, G., Liu, X., & Yan, J. (2023). Consensus cluster analysis of apoptosis-related genes in patients with osteoarthritis and their correlation with immune cell infiltration. Frontiers In Immunology , 14 , 1202758. 10.3389/fimmu.2023.1202758 Lu, S., & Fang, C. (2024). Isosakuranetin inhibits subchondral osteoclastogenesis for attenuating osteoarthritis via suppressing NF-κB/CXCL2 axis. International Immunopharmacology , 143 (Pt 1), 113321. 10.1016/j.intimp.2024.113321 Bi, J., Li, Q., Yang, Z., Cai, L., Lv, T., Yang, X., et al. (2021). CXCL2 Impairs Functions of Bone Marrow Mesenchymal Stem Cells and Can Serve as a Serum Marker in High-Fat Diet-Fed Rats. Front Cell Dev Biol , 9 , 687942. 10.3389/fcell.2021.687942 Additional Declarations No competing interests reported. Supplementary Files supplement.zip Cite Share Download PDF Status: Published Journal Publication published 14 Apr, 2026 Read the published version in Stem Cell Reviews and Reports → Version 1 posted Editorial decision: Revision requested 27 Mar, 2026 Reviews received at journal 27 Mar, 2026 Reviewers agreed at journal 26 Mar, 2026 Reviewers agreed at journal 09 Feb, 2026 Reviewers agreed at journal 23 Jan, 2026 Reviewers invited by journal 23 Jan, 2026 Editor assigned by journal 22 Jan, 2026 Submission checks completed at journal 22 Jan, 2026 First submitted to journal 21 Jan, 2026 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-8655656","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":580708007,"identity":"61bca3be-8198-474e-a908-5a6055c39047","order_by":0,"name":"DuJiang Yang","email":"","orcid":"","institution":"Chengdu University of Traditional Chinese medicine,No. 37, Shierqiao Road, Chengdu, Sichuan Province, 610075, P.R. China","correspondingAuthor":false,"prefix":"","firstName":"DuJiang","middleName":"","lastName":"Yang","suffix":""},{"id":580708008,"identity":"50a90009-f4b2-4436-bdf5-e61076e63aaf","order_by":1,"name":"Gaowen Gong","email":"","orcid":"","institution":"Department of Joint Surgery, The Affiliated Traditional Chinese Medicine Hospital of Southwest Medical University, Luzhou, Sichuan 646000, China.","correspondingAuthor":false,"prefix":"","firstName":"Gaowen","middleName":"","lastName":"Gong","suffix":""},{"id":580708009,"identity":"334ba7e1-905c-47ec-bfd3-51bbc373563f","order_by":2,"name":"Junjie Chen","email":"","orcid":"","institution":"Chengdu University of Traditional Chinese medicine,No. 37, Shierqiao Road, Chengdu, Sichuan Province, 610075, P.R. China","correspondingAuthor":false,"prefix":"","firstName":"Junjie","middleName":"","lastName":"Chen","suffix":""},{"id":580708010,"identity":"50c3372e-60e6-4bb5-ac5c-d5e7a2119f61","order_by":3,"name":"Jiafeng song","email":"","orcid":"","institution":"Department of Joint Surgery, The Affiliated Traditional Chinese Medicine Hospital of Southwest Medical University, Luzhou, Sichuan 646000, China.","correspondingAuthor":false,"prefix":"","firstName":"Jiafeng","middleName":"","lastName":"song","suffix":""},{"id":580708011,"identity":"d3203fa8-6566-4d9d-b838-c00ea9135520","order_by":4,"name":"Zhijun Ye","email":"","orcid":"","institution":"Department of Joint Surgery, The Affiliated Traditional Chinese Medicine Hospital of Southwest Medical University, Luzhou, Sichuan 646000, China.","correspondingAuthor":false,"prefix":"","firstName":"Zhijun","middleName":"","lastName":"Ye","suffix":""},{"id":580708012,"identity":"56919583-4c30-47e6-8ed9-92c9f2412424","order_by":5,"name":"Shuang Wang","email":"","orcid":"","institution":"Department of Joint Surgery, The Affiliated Traditional Chinese Medicine Hospital of Southwest Medical University, Luzhou, Sichuan 646000, China.","correspondingAuthor":false,"prefix":"","firstName":"Shuang","middleName":"","lastName":"Wang","suffix":""},{"id":580708013,"identity":"8827bc35-9bc9-4629-bff8-1338b671d335","order_by":6,"name":"Guoyou Wang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA/0lEQVRIiWNgGAWjYLCDAx8qJOTY2NsPEK2D8eCMMxbGfDxnEojWwnyYt60icZ6EgwFeZQbHzx5+wfDrsJzB8R6DwzxnJNLbJBgSGH5UbMOt5UxemgVj32FjgzNnDA7OqZDIbZNuPMDYc+Y2Ti1mB3LMDBh7DiduuJFjcODNGaAWmQMJzIxteLScf4OkhbdNIp1NIsEAv5YbOcYPGH5AtBwEakkgqMX+xhszBsaGdGPJM8cKgIEsYdgGDOSD+Pwi2Z9j/IHhj7Uc3/HmzR8+VNTJy7e3H3zwowK3FiBgk/7bxsCgcABJ6AAOpTDADLSFgUG+gYCyUTAKRsEoGLkAAFj4Y5ZEjC6PAAAAAElFTkSuQmCC","orcid":"","institution":"Department of Joint Surgery, The Affiliated Traditional Chinese Medicine Hospital of Southwest Medical University, Luzhou, Sichuan 646000, China.","correspondingAuthor":true,"prefix":"","firstName":"Guoyou","middleName":"","lastName":"Wang","suffix":""}],"badges":[],"createdAt":"2026-01-21 05:56:33","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8655656/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8655656/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1007/s12015-026-11121-9","type":"published","date":"2026-04-14T15:57:56+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":101332228,"identity":"0e85b400-8d1a-4009-be62-be1cd145cb4d","added_by":"auto","created_at":"2026-01-28 14:56:45","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":39056642,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eThe molecular characterization of cellular heterogeneity in mouse PTOA tissue utilizing the GSE200843 dataset.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) The cells from all five time-point samples are visualized on UMAP plots and colored according to their origin from either the normal or OA groups. (B) Nine major cellular clusters across the five groups are distinguished by different colors. (C) The distribution of distinct cell populations in normal and arthritic tissues. (D) LogFC visualization presenting the top ten genes in each cell type, demonstrating the differential analysis comparing normal and arthritic knee tissues. (E) A heatmap for the representative DEGs of every cell group.\u003c/p\u003e","description":"","filename":"Figure1.png","url":"https://assets-eu.researchsquare.com/files/rs-8655656/v1/e3a4e9ce7c8de06871e61284.png"},{"id":101332320,"identity":"cebfa0b2-2025-4d72-bb23-748358840b35","added_by":"auto","created_at":"2026-01-28 14:57:01","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":17674844,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003escRNA-seq analysis for the heterogeneity of granulocyte subtypes in PTOA.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) Further identification of five distinct subtypes of granulocytes in PTOA and normal joint cartilage cells. Each granulocyte subcluster is visualized through a color-coded UMAP plot. (B) The cell proportions of granulocyte subclusters in normal and PTOA joint cartilage tissues over time. (C) LogFC visualization of the top ten genes in every subcluster, displaying the comparison between normal osteoarticular and PTOA soft tissues after differential analysis. (D) Left panel: This series of plots illustrates the dynamic expression patterns of representative DEGs within every granulocyte population. Middle panel: A heatmap shows the representative DEGs of each cell cluster. Right panel: Representative enriched Gene Ontology (GO) terms for every cluster. (E) The GSEA enrichment plot displays the representative signaling pathways upregulated in OA granulocytes compared with other granulocyte subtypes.\u003c/p\u003e","description":"","filename":"Figure2.png","url":"https://assets-eu.researchsquare.com/files/rs-8655656/v1/5bf4b5014ba907482c3ba874.png"},{"id":101332224,"identity":"7079e40d-bd04-46e5-97dc-803a5ad0a18e","added_by":"auto","created_at":"2026-01-28 14:56:44","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":29705617,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTrajectory Analysis of Granulocyte Subpopulations\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A-C) Single-cell analysis for trajectory inference of granulocyte subclusters, with cells colored according to the cluster. The developmental trajectory of granulocytes is color-coded by state (A), related cell subgroups (B), and pseudotime (C). Scatter plots (D, E) for the expression of selected genes across subclusters and states as pseudotime progresses. (F) A heatmap for the dynamic changes in gene expression across various cell clusters.\u003c/p\u003e","description":"","filename":"Figure3.png","url":"https://assets-eu.researchsquare.com/files/rs-8655656/v1/82015871a2064be7ab8d1c7d.png"},{"id":101332219,"identity":"36c7a8a7-17f8-4d1b-aef5-fb64da96659b","added_by":"auto","created_at":"2026-01-28 14:56:39","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":23994748,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eIntercellular Communication Analysis of Granulocyte Subpopulations in PTOA\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A, B) Circular plots illustrating the quantity and intensity of ligand-receptor interactions between different cell populations. (C, D) The cell-cell ligand-receptor and cytokine-related pathway networks highlight how OA Cluster 0 and NOA Cluster 2 interact with other cell populations. (E-G) Heatmaps depicting the significance of the inferred intercellular communication networks for CD52 (E), SELPLG (F), and CXCL signaling (G).\u003c/p\u003e","description":"","filename":"Figure4.png","url":"https://assets-eu.researchsquare.com/files/rs-8655656/v1/12cc8a83e020e95cc8b44fef.png"},{"id":101332183,"identity":"265af7a8-8746-42ce-b62c-1572a254857a","added_by":"auto","created_at":"2026-01-28 14:56:33","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":8177244,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eIdentification of Potential PTOA-Related Genes in OA Granulocytes Through hdWGCNA\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) A scale-free network is constructed with a soft threshold power of 7. The left panel illustrates the influence of soft threshold power on the scale-free topology fitting index, while the right panel demonstrates the average network connectivity at varied weight coefficients. (B) hdWGCNA for assessing module activity across different granulocyte clusters. (C) The UMAP plot, similar to that in panel 2A, is colored according to the module eigengenes (MEs) of the 12 co-expressed gene modules. (D) Through hdWGCNA, highly variable genes are classified into 12 modules. (E) The UMAP plot presents the co-expression network of 12 gene modules. Node size corresponds to the kME values, and nodes are colored based on the co-expression module assignments. The two central genes of each module are annotated. (F) Matrix plot depicting the inter-module relationship by displaying the correlation between module feature genes. (G) Three PTOA granulocyte-related gene modules were identified, and the top hub genes are presented according to the hdWGCNA pipeline.\u003c/p\u003e","description":"","filename":"Figure5.png","url":"https://assets-eu.researchsquare.com/files/rs-8655656/v1/cb47d8441e9325b558e42eeb.png"},{"id":101332216,"identity":"bd7242d9-a975-4397-ba1f-b72eafd9b9d1","added_by":"auto","created_at":"2026-01-28 14:56:39","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":2907263,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eIdentification of Central Genes in PTOA via ML Techniques\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) Venn diagram depicting 799 module genes identified by WGCNA analysis, 350 DEGs identified in the training set, and 4,231 DEGs between OA and NOA granulocytes.\u003c/p\u003e\n\u003cp\u003e(B, C) Feature gene selection utilizing LASSO Cox regression analysis.\u003c/p\u003e\n\u003cp\u003e(D) Error rate visualization utilizing the SVM algorithm.\u003c/p\u003e\n\u003cp\u003e(E) Gene ranking performed by RF analysis.\u003c/p\u003e\n\u003cp\u003e(F) Venn diagram illustrating overlapping genes identified by the three algorithms.\u003c/p\u003e\n\u003cp\u003e(G-J) Boxplots and ROC curves of four candidate biomarkers in the training set (GSE26475) and validation set (GSE112641).\u003c/p\u003e\n\u003cp\u003e(K, L) Construction of a nomogram based on the expression of feature genes. Each gene is assigned a score on the scale, and the cumulative score corresponds to the predicted probability of PTOA.\u003c/p\u003e","description":"","filename":"Figure6.png","url":"https://assets-eu.researchsquare.com/files/rs-8655656/v1/19b27f63a3fe4ba7544e78cb.png"},{"id":101332164,"identity":"ad766025-2492-449b-9baa-88486188bb86","added_by":"auto","created_at":"2026-01-28 14:56:23","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":2605646,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eIIA\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A)Bar charts illustrating immune cell infiltration. (B) Violin plots depicting the proportions of 20 distinct immune cell types and functional states between PTOA and normal samples. (C-D) Analysis of links of hub genes to immune cells.\u003c/p\u003e","description":"","filename":"Figure7.png","url":"https://assets-eu.researchsquare.com/files/rs-8655656/v1/3d7e373ee4f32ddaf112f52a.png"},{"id":101332222,"identity":"21a284b7-03b3-417a-81cd-192be6278302","added_by":"auto","created_at":"2026-01-28 14:56:43","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":328064,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eValidation of Key Genes\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A, B) RT-PCR expression of Mpp7 and Tgfbi in synovial tissues from osteoarthritic and control mouse knee joints. All experiments were performed in triplicate, and data are presented as mean ± standard error (p \u0026lt; 0.05; ns, not significant). RT-PCR, reverse transcription–polymerase chain reaction.\u003c/p\u003e","description":"","filename":"Figure8.png","url":"https://assets-eu.researchsquare.com/files/rs-8655656/v1/94349ee11ccbec2ba9527d31.png"},{"id":107351053,"identity":"20703c21-4738-430c-a411-8f0b4e26f2e9","added_by":"auto","created_at":"2026-04-20 16:08:32","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":106086916,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8655656/v1/0f023f2e-8218-4e25-82de-b6ad3889c6f1.pdf"},{"id":101332271,"identity":"1dfee96c-499b-4f46-96c6-0c0bc5fe1f8c","added_by":"auto","created_at":"2026-01-28 14:56:49","extension":"zip","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":27934354,"visible":true,"origin":"","legend":"","description":"","filename":"supplement.zip","url":"https://assets-eu.researchsquare.com/files/rs-8655656/v1/e3fb08c762484ecec7ede1cc.zip"}],"financialInterests":"No competing interests reported.","formattedTitle":"Machine Learning-Based Identification of Molecular Signatures in PTOA Cell Subtypes via Single-Cell Transcriptomics","fulltext":[{"header":"Key-points","content":"\u003col\u003e\n \u003cli\u003ePredicting diagnostic markers based on the temporal dynamics of immune cell accumulation in mouse knee osteoarthritis.\u003c/li\u003e\n \u003cli\u003eUnderstanding the role of various immune cells in joint degeneration or joint repair after injury is helpful for improving therapeutic strategies for treating OA.\u003c/li\u003e\n \u003cli\u003eUsing machine learning (ML)-based methodologies, a predictive model was constructed and validated through nomogram models, calibration curves, and decision curves.\u0026nbsp;\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"1. Introduction","content":"\u003cdiv id=\"Sec2\" class=\"Section2\"\u003e \u003ch2\u003e1.1 Overview of Post-traumatic osteoarthritis (PTOA)\u003c/h2\u003e \u003cp\u003ePTOA constitutes a clinically significant and etiologically distinct subtype of osteoarthritis (OA), directly precipitated by acute intra-articular mechanical insults, including ligamentous rupture, meniscal injury, or intra-articular fracture. PTOA is pathologically characterized by progressive degradation of articular cartilage, synovial inflammation, and aberrant remodeling of the subchondral bone, thereby contributing substantially to the increasing global burden of OA. With over 300\u0026nbsp;million individuals affected worldwide, PTOA accounts for a considerable subset of cases. Although its pathogenesis is more clearly delineated than that of primary OA, it nonetheless involves a comparably intricate pathological cascade [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. The core pathophysiological mechanism is widely regarded as a profound disruption of the homeostatic equilibrium between extracellular matrix synthesis and degradation in articular cartilage and subchondral bone, initiated by multifactorial, multi-stage processes activated after injury [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eThe hallmark pathological manifestations of PTOA encompass progressive loss of articular cartilage, persistent synovitis, subchondral bone sclerosis, and osteophyte formation, culminating in chronic pain and functional disability. Recent investigations emphasize that aberrant subchondral bone remodeling is not merely a late-stage secondary event but is initiated at the early phase following joint trauma. This process involves abnormal bone turnover and irregular resorption, mediated by dysfunctional osteoclast and osteoblast activity, and is accompanied by microenvironmental perturbations like marrow fibrosis, aberrant angiogenesis, and neural invasion [\u003cspan additionalcitationids=\"CR5\" citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. These changes create a self-perpetuating microenvironment conducive to joint degeneration. The initiating insult in PTOA, joint trauma, provokes a stress-induced innate and autoimmune response, mediated by the release of damage-associated molecular patterns (DAMPs). This response leads to a surge in pro-inflammatory cytokines (e.g., IL-1β, TNF-α, IL-6) and a concurrent decline in anti-inflammatory mediators (e.g., IL-10), generating profound metabolic disequilibrium. The resulting imbalance accelerates cartilage catabolism, suppresses anabolic repair, and directly promotes pathological subchondral bone remodeling [\u003cspan additionalcitationids=\"CR8\" citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. Within this complex network, diverse cell types actively participate, including infiltrating innate immune cells (granulocytes, natural killer cells, monocytes/macrophages) and resident stromal cells (osteocytes, osteoblasts). Among these, granulocytes, particularly granulocytes, exert a pivotal influence in post-traumatic bone remodeling. Granulocytes mediate their effects via the release of granular proteases (e.g., granulocyte elastase, MMP-9) and formation of granulocyte extracellular traps (NETs). These processes accelerate bone matrix mineralization, regulate collagen fiber reorganization, modulate matrix protein phosphorylation, remodel the Haversian canal system and cortical bone, and foster hydroxyapatite crystal deposition, collectively driving PTOA progression [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. Given these critical functions, granulocytes and their effector mediators have emerged as potential prognostic biomarkers and promising therapeutic targets in PTOA. Moreover, accumulating evidence indicates that granulocytes are a heterogeneous population; specific subsets (e.g., CD16⁺CD62L^low aged granulocytes or low-density granulocytes) exert distinct, and sometimes opposing, effects at different stages of PTOA. These subsets modulate the local immune microenvironment (e.g., T-cell polarization, osteoclastogenesis) with precision and actively promote calcified bone matrix deposition [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. Therefore, precise identification and functional characterization of pathogenic versus protective granulocyte subsets are imperative for the development of novel immunomodulatory and targeted therapeutic strategies tailored to the heterogeneous clinical phenotypes of PTOA.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e1.2 Single-Cell RNA Sequencing (scRNA-seq)\u003c/h2\u003e \u003cp\u003escRNA-seq has been employed to define the comprehensive gene expression profiles of individual cells [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. This technology has been widely utilized to dissect cellular heterogeneity that was previously obscured in bulk populations, thereby advancing insights into both physiological and pathological processes [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. Through this high-resolution approach, the transcriptomic features of granulocytes in normal and osteoarthritic tissues have been delineated. Nevertheless, a comprehensive understanding of granulocyte subpopulation composition, gene expression programs, and molecular functions in osteoarthritic tissues remains limited [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. Furthermore, the clinical relevance and prognostic implications of granulocyte subsets in PTOA pathogenesis have yet to be elucidated.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e1.3 Machine Learning (ML) Validation\u003c/h2\u003e \u003cp\u003eML represents a transformative paradigm in data science, capable of autonomously constructing predictive models by identifying key features within complex, high-dimensional datasets [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. In comparison to conventional statistical methods, ML approaches offer marked advantages in analyzing large-scale biological data and have demonstrated significant value in the diagnosis and management of various diseases, including OA [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. This study applied three complementary ML algorithms to identify robust biomarkers of PTOA. First, the least absolute shrinkage and selection operator (LASSO) logistic regression, a penalized linear modeling approach, was employed to select disease-specific feature genes by minimizing prediction error via cross-validation, thereby enhancing interpretability [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. Second, support vector machine recursive feature elimination (SVM-RFE), an approach particularly effective in microarray analysis, was utilized to iteratively identify optimal feature combinations for classification by leveraging nonlinear discriminative capacity [\u003cspan additionalcitationids=\"CR21\" citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. Third, the random forest (RF) algorithm was applied for its ability to model complex, nonlinear interactions among variables. These approaches mitigate the risk of model overfitting and, when integrated with bioinformatic analyses, substantially improve the diagnostic accuracy and translational utility of identified core genes [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e, \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]. Therefore, employing complementary ML strategies to define PTOA biomarkers and construct reliable predictive models holds significant promise for advancing personalized diagnosis and treatment.\u003c/p\u003e \u003cp\u003escRNA-seq datasets were retrieved from the Gene Expression Omnibus (GEO) (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.ncbi.nlm.nih.gov/geo/\u003c/span\u003e\u003cspan address=\"http://www.ncbi.nlm.nih.gov/geo/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e), encompassing transcriptional profiles of normal and post-traumatic knee tissues in C57BL/6J mice during the critical window of 0\u0026ndash;15 days following non-invasive anterior cruciate ligament (ACL) rupture. These datasets have been rigorously validated in prior literature [\u003cspan additionalcitationids=\"CR26\" citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]. The choice of a non-surgical mechanical ACL rupture model is particularly advantageous, as it closely replicates the biomechanical insult of human knee trauma while avoiding the confounding inflammatory responses induced by surgical intervention, thereby enhancing clinical translatability [\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e, \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]. The C57BL/6J strain was selected for its well-defined genetic background and established use in OA research, ensuring reproducibility and facilitating comparison with an extensive body of prior work. Focusing on the acute phase (0\u0026ndash;15 days post-injury) is strategically critical, as this period captures the earliest molecular and cellular events, including acute inflammation, early extracellular matrix degradation, and initiation of repair signaling, that drive PTOA pathogenesis. Elucidating this early cascade is essential for identifying upstream drivers of disease before irreversible joint damage ensues, thereby informing strategies for timely intervention [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e, \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e].\u003c/p\u003e \u003c/div\u003e"},{"header":"2. Materials and Methods","content":"\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003e2.1. Data Collection and Processing\u003c/h2\u003e \u003cp\u003eThe scRNA-seq dataset GSE200843 was retrieved from the GEO to investigate granulocyte heterogeneity in normal and PTOA human knee tissues [\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]. Datasets GSE26475 and validation set GSE112641 were used for predictive model construction and validation [\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e, \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e] (Supplementary Table\u0026nbsp;5). Raw count data were processed via Seurat (v4.4.0) in R. Quality control was performed by filtering out cells with fewer than 500 or more than 6,000 detected genes, as well as cells in which the percentage of mitochondrial transcripts exceeded 25%. Normalization, variance stabilization, and scaling of raw counts were conducted utilizing the SCTransform function. To mitigate technical batch effects, data integration was performed utilizing Harmony (v1.2.1). Principal component analysis (PCA) was applied to the integrated dataset for initial dimensionality reduction. A shared nearest neighbor (SNN) graph was constructed utilizing the FindNeighbors function, followed by cell clustering with the FindClusters function, which implements the Louvain algorithm for modularity optimization. Cell clusters were visualized in two dimensions utilizing Uniform Manifold Approximation and Projection (UMAP). Each cluster was annotated into specific cell types based on the expression of well-established canonical marker genes. DEGs between conditions (PTOA vs. normal) and cluster-specific marker genes were identified utilizing the FindMarkers function in Seurat. A similar analytical pipeline was applied for sub-clustering analysis of granulocyte populations.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003e2.2. Pseudotime Trajectory and Intercellular Communication Analysis\u003c/h2\u003e \u003cp\u003eTo reconstruct the developmental trajectory of granulocytes, pseudotemporal ordering analysis was enabled by Monocle (v2.32.0) with the DDRTree reduction method and default parameters. Dynamic expression patterns of key genes along the inferred trajectory were visualized utilizing the plot_pseudotime_heatmap function. Intercellular communication networks were systematically mapped via the CellChat toolbox (v1.6.1), analyzing ligand-receptor interactions based on the expression of a curated database of signaling pathways.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003e2.3. Functional Enrichment Analysis\u003c/h2\u003e \u003cp\u003eDEGs for specific cell subpopulations were identified via Seurat\u0026rsquo;s FindMarkers function, applying thresholds of absolute log2 fold-change (|log2FC|)\u0026thinsp;\u0026gt;\u0026thinsp;1 and adjusted p-value (FDR)\u0026thinsp;\u0026lt;\u0026thinsp;0.05. GO biological process enrichment analysis and single-sample GSEA (ssGSEA) were performed via clusterProfiler (v4.12.6) to interpret the functional themes of the DEGs.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003e2.4. hdWGCNA\u003c/h2\u003e \u003cp\u003eTo identify granulocyte-associated gene modules correlated with PTOA, high-dimensional weighted gene co-expression network analysis (hdWGCNA) was carried out using R (v1.73). Meta-cells for each sample and cell cluster were first constructed utilizing the MetacellsByGroups function, aggregating data from 50 cells per meta-cell to reduce sparsity. The standard hdWGCNA pipeline was then executed sequentially on the granulocyte subset, encompassing the following key steps: TestSoftPowers to determine the appropriate soft-thresholding power, ConstructNetwork to build the co-expression network and identify modules, ModuleEigengenes to calculate MEs, ModuleConnectivity to compute intramodular connectivity (kME), and RunModuleUMAP for visualization.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003e2.5. Construction of ML-Based Prediction Models for PTOA Granulocyte-Associated Genes\u003c/h2\u003e \u003cp\u003eTo evaluate the predictive potential of PTOA granulocyte-associated genes identified by hdWGCNA in PTOA progression, three ML algorithms, LASSO, RF, and SVM, were applied to further identify hub genes (Supplementary Table\u0026nbsp;6) [\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e]. The predictive performance of these models was assessed utilizing ROC curves generated with \u0026ldquo;timeROC\u0026rdquo; in R v0.4 in dataset GSE26475 and validation set GSE112641.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003e2.6. IIA\u003c/h2\u003e \u003cp\u003eCIBERSORT, which estimated relative subsets of RNA transcripts to infer cell type composition, was used to predict the relative abundance of 20 infiltrating immune cell types. Boxplots were generated to compare immune cell composition between the normal and disease groups. Additionally, Spearman correlation analysis was performed to explore the relation of the expression levels of key genes to the abundance of immune cells. The correlations were visualized utilizing a lollipop plot generated via \u0026ldquo;ggplot2\u0026rdquo; 3.5.1.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003e2.7. PTOA Mouse Model\u003c/h2\u003e \u003cp\u003e12 healthy 10-week-old male C57BL/6 mice, weighing 26.5\u0026thinsp;\u0026plusmn;\u0026thinsp;1.7 g, were obtained from the Experimental Animal Center of Southwest Medical University (License No.: SCXK [Chuan] 2024-0046). Animals were acclimated for one week under standard housing conditions (room temperature 21\u0026ndash;25\u0026deg;C; relative humidity 45%-65%) with unrestricted activity and free access to food and water. All experimental procedures were approved by the Institutional Animal Ethics Committee of Southwest Medical University.\u003c/p\u003e \u003cp\u003eFollowing one week of acclimatization, mice were randomized by a random-number method to either the sham-operation group (n\u0026thinsp;=\u0026thinsp;6) or the anterior cruciate ligament transection (ACLT) group (n\u0026thinsp;=\u0026thinsp;6). PTOA induction in the model group was performed using the ACLT procedure as described previously [\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e]. General anesthesia was administered via intraperitoneal injection of 1% sodium pentobarbital (40 mg/kg). With the knee joint fixed, the skin was prepared and disinfected. A medial parapatellar incision was made, and tissues were dissected layer by layer to expose the joint cavity. The anterior cruciate ligament (ACL) was fully exposed and transected using microsurgical scissors. After repositioning the patella, a drawer test was performed to confirm a complete ACL rupture. The incision was closed in layers and disinfected again. In the sham-operation group, the joint cavity was exposed without ACL transection. At 12 weeks after the operation, mice were euthanized, and the right knee joints were harvested for anatomical examination. All procedures strictly adhered to applicable guidelines for animal experimentation in mice.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003e2.8 RT-PCR Validation of Key Genes\u003c/h2\u003e \u003cp\u003eApproximately 0.02 g of synovial tissue from the right hind-knee joints of mice in the disease and sham-operation groups was collected. Total RNA was extracted using the Trizol method after thorough homogenization, lysis, centrifugation, and purification. RNA was reverse-transcribed into cDNA using the SweScript All-in-One SuperMix. Quantitative real-time PCR was performed using Universal Blue SYBR Green qPCR Master Mix on a CFX Connect Real-Time System (Bio-Rad). Target gene expression was normalized to the reference gene GAPDH and calculated through the ΔΔCT method. All procedures were repeated to ensure the accuracy and reliability of the results.\u003c/p\u003e \u003cp\u003ePrimer sequences were: Mpp7 (forward: 5\u0026prime;-TCCAGAACAAGCCACCAAACA-3\u0026prime;; reverse: 5\u0026prime;-CACAGAGTCTTCCTCGTCATCAA-3\u0026prime;); Tgfbi (forward: 5\u0026prime;-CTGTTGCCGAAACCGACATC-3\u0026prime;; reverse: 5\u0026prime;-CAGGGGCAAGTCGCATAGAA-3\u0026prime;); GAPDH (reference gene) (forward: 5\u0026prime;-CTGGAGAAACCTGCCAAGTATG-3\u0026prime;; reverse: 5\u0026prime;-GGTGGAAGAATGGGAGTTGCT-3\u0026prime;).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003e2.9. Statistical Analysis\u003c/h2\u003e \u003cp\u003eAll biological experiments included at least three biological replicates. For measurement data expressed as mean\u0026thinsp;\u0026plusmn;\u0026thinsp;standard error of the mean (SEM), independent-sample t-tests were performed. All statistical analyses and data visualizations were conducted via R 4.4.1. The Spearman correlation coefficient was employed to investigate associations between continuous variables. The differences between the two groups were determined via the Wilcoxon signed-rank test. p\u0026thinsp;\u0026lt;\u0026thinsp;0.05 denoted statistical significance. Statistical notations were as follows: * for p\u0026thinsp;\u0026lt;\u0026thinsp;0.05, ** for p\u0026thinsp;\u0026lt;\u0026thinsp;0.01, *** for p\u0026thinsp;\u0026lt;\u0026thinsp;0.001, **** for p\u0026thinsp;\u0026lt;\u0026thinsp;0.0001, and ns for p\u0026thinsp;\u0026gt;\u0026thinsp;0.05.\u003c/p\u003e \u003c/div\u003e"},{"header":"3. Results","content":"\u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003e3.1 scRNA-seq Reveals the Transcriptomic Landscape of granulocytes in Normal and PTOA Tissues\u003c/h2\u003e \u003cp\u003eTo investigate the cellular composition and diversity in normal and PTOA tissues, scRNA-seq data from mouse PTOA tissue were collected and analyzed. After quality control was implemented, batch effects were mitigated and cell types were annotated; the samples were classified into OA and normal groups [\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e]. Nine major cell populations were identified: granulocytes, macrophages, fibroblasts, erythrocytes, monocytes, B, T, endothelial, NK cells, and others (Figs.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eA and \u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eB). Figure\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eC illustrates the temporal dynamics of each cell population across the analyzed samples. Notably, the proportion of granulocyte clusters increased over time in the OA group in comparison to the normal group (Supplementary Table\u0026nbsp;1). Furthermore, differentially expressed genes (DEGs) were analyzed, and the expression of lineage-specific markers was examined across the nine major cell clusters (Figs.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eD and \u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eE).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eUnsupervised hierarchical subclustering was performed on granulocytes isolated from both PTOA and normal knee tissues to further elucidate their heterogeneity within the articular microenvironment. This analysis identified five transcriptionally distinct granulocyte subpopulations (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA). Comparative quantification of their abundances revealed two clusters exhibiting statistically significant temporal alterations in PTOA relative to normal samples: one cluster was markedly expanded, whereas the other was reduced under disease conditions. These were subsequently designated as OA-associated granulocytes (OA granulocytes) and non-OA-associated granulocytes (NOA granulocytes), respectively (Figs.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA, \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB). The identity of each cluster was corroborated by the expression of unique marker gene signatures, visualized via feature plots (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eC). The transcriptional programs characterizing these subtypes were subsequently delineated by computing relative gene expression scores across individual cells, followed by unsupervised clustering, which resolved distinct co-expression modules (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eD). Genes within each module were further grouped according to similarity in expression dynamics, yielding five principal patterns potentially related to discrete biological processes. Notably, genes upregulated in the OA granulocyte cluster were significantly enriched in pathways directly implicated in the pathogenesis of OA, including cytokine receptor activity and cytokine receptor binding (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eD). In parallel, gene set enrichment analysis (GSEA) demonstrated a significant negative correlation between the OA granulocyte signature and gene sets related to inflammatory response activation (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eE). The foregoing findings highlight a pivotal role of OA granulocytes in promoting the initiation and progression of OA through aberrant cytokine signaling, rather than through canonical inflammatory pathways.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003e3.2. Pseudotemporal Trajectory Analysis of PTOA Granulocyte Subtypes during Arthritis Progression\u003c/h2\u003e \u003cp\u003eTo elucidate the cellular origins and developmental trajectories of granulocytes related to PTOA during disease progression, pseudotime trajectory analysis was applied to granulocyte subpopulations. The analysis positioned granulocyte cluster 1, together with non-OA granulocyte clusters, near the origin of the trajectory. In contrast, OA granulocytes were predominantly distributed at the branchpoint of trajectory 1 and extended toward the termini of trajectories 2 and 3 (Figs.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eA-\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eC). To identify genes critical to arthritis progression, dynamic gene expression changes along the pseudotemporal axis were examined. This revealed a set of genes exhibiting significant variation during the maturation of OA granulocytes, including Cd44, a recognized marker of chondrocytes, Cebpb, Il1r2, Mcl1, and Plk3 (Figs.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eD, \u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eE). The top 50 variable genes were subsequently clustered according to pseudotemporal expression patterns, followed by functional enrichment analyses for each cluster. Five distinct gene clusters with unique temporal expression profiles were identified (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eF). Specifically, genes in cluster 1 displayed high expression from early to mid pseudotime, whereas clusters 3, 4, and 5 reached peak expression during the late phase, indicating stage-specific functional roles in granulocyte differentiation in PTOA.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003e3.3. Intercellular Communication Analysis of Granulocyte Subpopulations in PTOA\u003c/h2\u003e \u003cp\u003eThe availability of single-cell datasets provides an excellent opportunity to explore the intercellular communication mediated through ligand-receptor interactions. The intercellular communication network involving granulocyte subpopulations and other cell types in both post-traumatic arthritic and normal joint tissues was examined via CellChat. Overall, OA granulocytes exhibited robust communication with other cell types during the pathogenesis of post-traumatic arthritis (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eA). OA granulocytes displayed higher interaction intensity, both qualitatively and quantitatively, in comparison to other granulocyte populations (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eB). Notably, OA granulocyte clusters engaged in direct interactions with other granulocyte subtypes through adhesion ligand-receptor pairs like Cxcl2/Cxcr2, Ccl6/Ccr1, and Thbs1/Cd36 (Figs.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eC, \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eD). Moreover, our analysis revealed the upregulation of pro-inflammatory signals, including CD52, SELPLG, and CXCL, in the communication between OA granulocytes and other granulocyte subsets (Figs.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eE-\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eG). Collectively, our findings suggest that OA granulocytes establish a reciprocal interaction network with other cell types, promoting mutual support and functional maintenance.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003e3.4 Identification of Co-expressed Gene Modules Correlated with Granulocytes in PTOA via hdWGCNA\u003c/h2\u003e \u003cp\u003ehdWGCNA, a comprehensive framework for co-expression network analysis in scRNA-seq data, was employed to identify co-expressed gene modules and explore their functional roles in PTOA-associated granulocytes. An ideal soft-threshold power of 7 was used to build a scale-free co-expression network (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eA). 12 distinct gene co-expression modules were identified. Specifically, the yellow, brown, and black ones were highly activated, primarily in OA granulocytes (Figs.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eB, \u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eC, \u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eD). Further analysis was conducted to examine the correlations between each module (Figs.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eE-\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eF). Figure\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eG displays the top 10 hub genes from the three highly activated modules.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003e3.5. Construction of PTOA Granulocyte-Related Predictive Models Based on ML\u003c/h2\u003e \u003cp\u003eSubsequently, the hub genes of three PTOA granulocyte-associated modules identified through hdWGCNA, DEGs of OA and NOA granulocytes, and the DEGs from the external dataset GSE26475 were intersected to predict the onset and progression of PTOA (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eA; Supplementary Tables\u0026nbsp;2\u0026ndash;4). Three ML algorithms were further applied to the external dataset (GSE26475) to identify granulocyte-related hub genes related to PTOA. Through the LASSO regression algorithm, five key genes closely related to prognosis were identified (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eB-\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eC). SVM-RFE analysis yielded prediction error curves and cross-validation error rates, with lower values indicating reduced error rates (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eD). In addition, RF analysis ranked all genes according to importance and highlighted the top 10 candidates (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eE). The overlapping genes obtained from the three ML approaches were visualized utilizing a Venn diagram (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eF). Expression levels of four genes were then displayed with box plots and ROC curves (Figs.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eG-\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eH). Validation utilizing the independent dataset GSE112641 revealed that Tgfbi and Mpp7 achieved AUC values greater than 0.7, suggesting diagnostic value (Figs.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eI-\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eJ). Finally, a nomogram was constructed to estimate the probability of disease occurrence based on these two genes, and calibration curves confirmed the accuracy of the predictive model (Figs.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eK-\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eL). The foregoing findings indicate that two key genes, Tgfbi and Mpp7, are strongly related to prognosis in the PTOA population.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003e3.6. Immune Infiltration Analysis (IIA) of Granulocyte-Related Prognostic Features in PTOA\u003c/h2\u003e \u003cp\u003eAn assessment of immune cell infiltration in PTOA was carried out. Given that the synovial microenvironment influences the prognosis of patients with PTOA and that immune cells are pivotal within this microenvironment, our study included a detailed analysis of immune cell infiltration in synovial tissue. A stacked bar chart revealed the composition of 20 immune cell types (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eA). Violin plots demonstrated significantly higher infiltration levels of activated M1 macrophages in the PTOA group, whereas naive B cell infiltration was reduced (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eB). IIA and correlation analyses further showed that naive B cells were significantly correlated with Mpp7 (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and Tgfbi (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Moreover, M1 macrophages exhibited significant correlations with Mpp7 (p\u0026thinsp;\u0026lt;\u0026thinsp;0.05) and Tgfbi (p\u0026thinsp;\u0026lt;\u0026thinsp;0.01) (Figs.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eC-\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eD).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003e3.7. Expression of Key Genes in In Vitro Experiments\u003c/h2\u003e \u003cp\u003eTo validate the expression of Mpp7 and Tgfbi in the synovium of PTOA, a PTOA animal model was established, and RT-PCR analysis was performed. The expression of Mpp7 was lower in the PTOA model group than in the control group (t\u0026thinsp;=\u0026thinsp;21.46, p\u0026thinsp;\u0026lt;\u0026thinsp;0.0002) (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eA). In contrast, Tgfbi mRNA levels were significantly elevated in the PTOA model group compared with controls (t\u0026thinsp;=\u0026thinsp;22.16, p\u0026thinsp;=\u0026thinsp;0.0014) (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003eB), consistent with the trend identified in our analyses.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"4. Discussion","content":"\u003cp\u003eAdvances in scRNA-seq have profoundly transformed our ability to dissect cellular heterogeneity within complex tissues. Leveraging this technology, the granulocyte lineage in OA joints was analyzed to identify two principal granulocyte subpopulations with distinct disease associations. These subsets were designated as OA granulocytes and NOA (non-OA) granulocytes, reflecting their dynamic distribution in PTOA tissues versus normal tissues. Subsequent in-depth characterization of their molecular features and regulatory pathways, coupled with bioinformatic analyses, identified two hub genes, Tgfbi and Mpp7, as critical mediators.\u003c/p\u003e \u003cp\u003eGranulocytes in OA extend beyond mere bystanders of inflammation, encompassing complex, stage-specific roles in disease pathogenesis. Our study identified MMP8, Ngp, Cxcl2, S100a8, and S100a9 as core-enriched and DEGs within PTOA granulocytes. During the acute post-injury phase of OA, granulocytes are rapidly recruited to the joint via a finely orchestrated chemokine cascade, primarily mediated by Cxcl1, Cxcl2, and their receptor Cxcr2. Their activation triggers a robust yet frequently deleterious response, characterized by the release of serine proteases (granulocyte elastase, proteinase 3), matrix metalloproteinases (MMPs) (MMP8, MMP9), and reactive oxygen species, which directly contribute to cartilage matrix degradation and synovitis [\u003cspan additionalcitationids=\"CR39\" citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e]. Emerging single-cell transcriptomic evidence reveals substantial heterogeneity among infiltrating granulocytes, indicating the coexistence of functionally distinct subpopulations, like a pro-inflammatory (N1-like) phenotype marked by high S100a8/a9 and Cxcr2 expression [\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e], suggesting their potential role in immune modulation within the arthritic joint microenvironment.\u003c/p\u003e \u003cp\u003eThrough extensive characterization of chondrocyte heterogeneity in arthritic tissue, a unique granulocyte subpopulation predominantly present in joint cartilage was successfully identified. The enriched molecular functions of this subset, including endopeptidase regulatory activity, chemokine receptor binding, and broader peptidase regulatory activity, are critically dysregulated in OA pathogenesis. These pathways act in an interconnected manner to drive disease progression [\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e]. Specifically, proteins with chemokine receptor binding functions are essential for the initial recruitment and activation of immune cells, like granulocytes and macrophages, into the synovial joint. This sustained inflammatory influx establishes a foundation for the destructive phase, which is primarily mediated by two additional pathways. Endopeptidase regulators and broader peptidase activities directly govern intra-articular proteolytic cascades, modulating the activity and function of key enzymes, including MMPs and aggrecanases (ADAMTS), which are responsible for catastrophic degradation of cartilage extracellular matrix components, particularly collagen and proteoglycans [\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e, \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e]. Therefore, the aberrant expression of genes within these three functional categories establishes a vicious cycle: chemokine-driven inflammation promotes the expression and activation of peptidases, whose uncontrolled activity degrades the matrix and releases bioactive fragments, further amplifying chemokine production and chronic inflammation. Among these, Cd44, Cebpb, Il1r2, Mcl1, and Plk3 emerged as the most significantly altered genes in OA granulocytes within arthritic cartilage tissues, playing pivotal roles in OA pathogenesis [\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e]. Our findings indicated that OA granulocytes had increased expression of genes related to inflammatory and reparative processes. For instance, potential anti-inflammatory agents like PCA interact with chondrocytes via Cd44, a differentiation cluster on the cell surface. CEBPB, encoding CCAAT/enhancer-binding protein β (C/EBPβ), is a transcription factor and one of the strongest trans-activators in chondrocytes, playing a crucial role in endochondral ossification [\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e]. Additionally, IL-1β, widely recognized as a key cytokine in OA progression, binds to its membrane receptor IL-1R1, activating transcription factors like MAPK and NF-κB, thereby inducing inflammatory mediators [\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e]. Conversely, NOA granulocyte abundance demonstrated an inverse correlation with disease progression. Intercellular communication analysis of key ligand-receptor pairs, including Cxcl2/Cxcr2, Ccl6/Ccr1, and Thbs1/Cd36, as well as genes like CD52, SELPLG, and members of the CXCL family, revealed a complex cell-cell signaling network driving inflammation in OA. The Cxcl2/Cxcr2 axis serves as a principal driver of granulocyte recruitment and activation. Following joint injury, synovial cells and macrophages secrete Cxcl2, which binds to Cxcr2 on granulocytes, triggering their influx into the joint and amplifying the acute inflammatory response. Simultaneously, the Ccl6/Ccr1 pathway facilitates the recruitment of monocytes and macrophages, further sustaining a pro-inflammatory synovial microenvironment. The Thbs1/Cd36 interaction introduces a critical regulatory dimension; while Thbs1 modulates cell adhesion and angiogenesis, its binding to Cd36 on immune cells may influence inflammatory signaling and the clearance of apoptotic debris, a process often dysregulated in chronic OA [\u003cspan additionalcitationids=\"CR49\" citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e]. Moreover, the involvement of CD52 (a mediator of lymphocyte activation), SELPLG (PSGL-1, essential for leukocyte tethering and rolling on endothelium), and other CXCL chemokines underscores the multifaceted nature of this communication network, encompassing both innate and adaptive immune cells. The cooperative action of these pathways establishes a vicious cycle of leukocyte recruitment, persistent synovitis, and cartilage degradation, positioning them as central mediators of PTOA pathology and potential therapeutic targets.\u003c/p\u003e \u003cp\u003eBased on these findings, our study employed three distinct algorithmic models, support vector machine (SVM), LASSO, and RF, for feature selection, with external datasets used for validation. Receiver operating characteristic (ROC) curve analysis confirmed the accuracy of the results. Tgfbi and Mpp7 were significantly differentially expressed. IIA revealed that Tgfbi, an extracellular matrix protein induced by TGF-β signaling, constitutes a critical risk factor; its pathological overexpression in chondrocytes leads to abnormal accumulation within the cartilage matrix. Mpp7, conversely, appears to act as a protective factor, although its association with B cells requires further investigation. Finally, a nomogram was constructed to enable quantitative prediction of disease risk.\u003c/p\u003e \u003cp\u003eOverall, utilizing bioinformatics approaches, our study identified these hub genes and established a predictive model capable of effectively forecasting the prognosis of patients with PTOA, providing valuable insights for disease prediction and intervention. Nevertheless, the study has certain limitations. Although public datasets offer valuable resources, the sample sizes remain relatively small. While supportive evidence for our findings exists in previously published studies, the markers were derived from our analyses. Future studies will further investigate the expression patterns of hub genes across different disease stages and explore factors influencing their expression, thereby contributing to the early diagnosis and treatment of PTOA.\u003c/p\u003e"},{"header":"5. Conclusions","content":"\u003cp\u003eIn this study, Tgfbi and Mpp7 were identified as potential diagnostic biomarkers for PTOA. Overall, these findings enhance the understanding of granulocyte heterogeneity within PTOA tissues. The identification of unique molecular and biological features of granulocyte subtypes, together with the prognostic utility of characteristic genes, provides valuable insights into the pathogenesis of PTOA.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eClinical trial number\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDeclaration of Artificial Intelligence\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe author declare that no artificial intelligence was used at the time of writing the paper.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. All experimental procedures were approved by the Institutional Animal Ethics Committee of Southwest Medical University.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDeclaration of interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was supported by the Study on the Efficacy and Mechanism of Shaoyang Shenggu Formula in Treating Early to Mid-Stage Knee Osteoarthritis Based on a \u0026quot;Blood-Entry Components-Targets-Pathways\u0026quot; Network (25NSFSC1423).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets used and analysed during the current study are available from the corresponding author on reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor Contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll authors contributed to the study conception and design.\u0026nbsp;\u003cstrong\u003eWriting - original draft preparation:\u0026nbsp;\u003c/strong\u003eGaowen Gong; Dujiang Yang ;\u003cstrong\u003e\u0026nbsp;Writing - review and editing:\u003c/strong\u003eGaowen Gong; Dujiang Yang; Junjie Chen\u003cstrong\u003e;\u003c/strong\u003e \u003cstrong\u003eConceptualization:\u0026nbsp;\u003c/strong\u003eGaowen Gong; Dujiang Yang; Junjie Chen \u003cstrong\u003e; Methodology:\u0026nbsp;\u003c/strong\u003eGaowen Gong; Dujiang Yangi\u003cstrong\u003e; Formal analysis and investigation:\u0026nbsp;\u003c/strong\u003eGaowen Gong; Dujiang Yang; Jiafeng Song\u003cstrong\u003e; Funding acquisition:\u0026nbsp;\u003c/strong\u003eGuoyou Wang;\u003cstrong\u003e\u0026nbsp;Resources:\u0026nbsp;\u003c/strong\u003eGuoyou Wang;\u003cstrong\u003e\u0026nbsp;Supervision:\u003c/strong\u003eGuoyou Wang, and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgments\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eDrugs for Osteoarthritis (2021). \u003cem\u003eJama\u003c/em\u003e. ;325(6):581\u0026ndash;582. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1001/jama.2020.8395\u003c/span\u003e\u003cspan address=\"10.1001/jama.2020.8395\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKatz, J. N., Arant, K. R., \u0026amp; Loeser, R. F. (2021). Diagnosis and Treatment of Hip and Knee Osteoarthritis: A Review. \u003cem\u003eJama\u003c/em\u003e, \u003cem\u003e325\u003c/em\u003e(6), 568\u0026ndash;578. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1001/jama.2020.22171\u003c/span\u003e\u003cspan address=\"10.1001/jama.2020.22171\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSlomski, A. (2020). Six Weeks of Prednisolone Reduced Hand Osteoarthritis Pain. \u003cem\u003eJama\u003c/em\u003e, \u003cem\u003e323\u003c/em\u003e(4), 301. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1001/jama.2019.20918\u003c/span\u003e\u003cspan address=\"10.1001/jama.2019.20918\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJeon, O. H., Kim, C., Laberge, R. M., Demaria, M., Rathod, S., Vasserot, A. P., et al. (2017). Local clearance of senescent cells attenuates the development of post-traumatic osteoarthritis and creates a pro-regenerative environment. \u003cem\u003eNature Medicine\u003c/em\u003e, \u003cem\u003e23\u003c/em\u003e(6), 775\u0026ndash;781. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/nm.4324\u003c/span\u003e\u003cspan address=\"10.1038/nm.4324\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSch\u0026auml;fer, N., \u0026amp; Gr\u0026auml;ssel, S. (2022). Targeted therapy for osteoarthritis: progress and pitfalls. \u003cem\u003eNature Medicine\u003c/em\u003e, \u003cem\u003e28\u003c/em\u003e(12), 2473\u0026ndash;2475. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41591-022-02057-x\u003c/span\u003e\u003cspan address=\"10.1038/s41591-022-02057-x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBaker, M. C., Sheth, K., Lu, R., Lu, D., von Kaeppler, E. P., Bhat, A., et al. (2023). Increased risk of osteoarthritis in patients with atopic disease. \u003cem\u003eAnnals Of The Rheumatic Diseases\u003c/em\u003e, \u003cem\u003e82\u003c/em\u003e(6), 866\u0026ndash;872. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1136/ard-2022-223640\u003c/span\u003e\u003cspan address=\"10.1136/ard-2022-223640\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKelly, P. N. (2017). Targeting senescence to combat osteoarthritis. \u003cem\u003eScience\u003c/em\u003e, \u003cem\u003e356\u003c/em\u003e(6338), 595\u0026ndash;596. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1126/science.356.6338.595-b\u003c/span\u003e\u003cspan address=\"10.1126/science.356.6338.595-b\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSun, Y., Wu, Q., Dai, K., You, Y., \u0026amp; Jiang, W. (2021). Generating 3D-cultured organoids for pre-clinical modeling and treatment of degenerative joint disease. \u003cem\u003eSignal Transduct Target Ther\u003c/em\u003e, \u003cem\u003e6\u003c/em\u003e(1), 380. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41392-021-00675-4\u003c/span\u003e\u003cspan address=\"10.1038/s41392-021-00675-4\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNesic, D., Whiteside, R., Brittberg, M., Wendt, D., Martin, I., \u0026amp; Mainil-Varlet, P. (2006). Cartilage tissue engineering for degenerative joint disease. \u003cem\u003eAdvanced Drug Delivery Reviews\u003c/em\u003e, \u003cem\u003e58\u003c/em\u003e(2), 300\u0026ndash;322. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.addr.2006.01.012\u003c/span\u003e\u003cspan address=\"10.1016/j.addr.2006.01.012\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSlomski, A. (2022). Surgical Procedures for Ankle Osteoarthritis Produce Similar Results. \u003cem\u003eJama\u003c/em\u003e, \u003cem\u003e328\u003c/em\u003e(24), 2387. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1001/jama.2022.20054\u003c/span\u003e\u003cspan address=\"10.1001/jama.2022.20054\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCentor, R. M., \u0026amp; Perez, P. L. (2023). Web Exclusive. Annals On Call - Exercise for Knee Osteoarthritis: Is More Better? \u003cem\u003eAnnals Of Internal Medicine\u003c/em\u003e, \u003cem\u003e176\u003c/em\u003e(3), eA220006. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.7326/a22-0006\u003c/span\u003e\u003cspan address=\"10.7326/a22-0006\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMetcalfe, D., Perry, D. C., Claireaux, H. A., Simel, D. L., Zogg, C. K., \u0026amp; Costa, M. L. (2019). Does This Patient Have Hip Osteoarthritis? The Rational Clinical Examination Systematic Review. \u003cem\u003eJama\u003c/em\u003e, \u003cem\u003e322\u003c/em\u003e(23), 2323\u0026ndash;2333. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1001/jama.2019.19413\u003c/span\u003e\u003cspan address=\"10.1001/jama.2019.19413\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVan de Sande, B., Lee, J. S., Mutasa-Gottgens, E., Naughton, B., Bacon, W., Manning, J., et al. (2023). Applications of single-cell RNA sequencing in drug discovery and development. \u003cem\u003eNature Reviews. Drug Discovery\u003c/em\u003e, \u003cem\u003e22\u003c/em\u003e(6), 496\u0026ndash;520. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41573-023-00688-4\u003c/span\u003e\u003cspan address=\"10.1038/s41573-023-00688-4\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHan, J., DePinho, R. A., \u0026amp; Maitra, A. (2021). Single-cell RNA sequencing in pancreatic cancer. \u003cem\u003eNature Reviews. Gastroenterology \u0026amp; Hepatology\u003c/em\u003e, \u003cem\u003e18\u003c/em\u003e(7), 451\u0026ndash;452. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41575-021-00471-z\u003c/span\u003e\u003cspan address=\"10.1038/s41575-021-00471-z\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePoran, A., N\u0026ouml;tzel, C., Aly, O., Mencia-Trinchant, N., Harris, C. T., Guzman, M. L., et al. (2017). Single-cell RNA sequencing reveals a signature of sexual commitment in malaria parasites. \u003cem\u003eNature\u003c/em\u003e, \u003cem\u003e551\u003c/em\u003e(7678), 95\u0026ndash;99. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/nature24280\u003c/span\u003e\u003cspan address=\"10.1038/nature24280\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhong, S., Zhang, S., Fan, X., Wu, Q., Yan, L., Dong, J., et al. (2018). A single-cell RNA-seq survey of the developmental landscape of the human prefrontal cortex. \u003cem\u003eNature\u003c/em\u003e, \u003cem\u003e555\u003c/em\u003e(7697), 524\u0026ndash;528. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/nature25980\u003c/span\u003e\u003cspan address=\"10.1038/nature25980\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRinehart, N. I., Saunthwal, R. K., Wellauer, J., Zahrt, A. F., Schlemper, L., Shved, A. S., et al. (2023). A machine-learning tool to predict substrate-adaptive conditions for Pd-catalyzed C-N couplings. \u003cem\u003eScience\u003c/em\u003e, \u003cem\u003e381\u003c/em\u003e(6661), 965\u0026ndash;972. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1126/science.adg2114\u003c/span\u003e\u003cspan address=\"10.1126/science.adg2114\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJayakrishnan, T. T., Sangwan, N., Barot, S. V., Farha, N., Mariam, A., Xiang, S., et al. (2024). Multi-omics machine learning to study host-microbiome interactions in early-onset colorectal cancer. \u003cem\u003eNPJ Precis Oncol\u003c/em\u003e, \u003cem\u003e8\u003c/em\u003e(1), 146. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41698-024-00647-1\u003c/span\u003e\u003cspan address=\"10.1038/s41698-024-00647-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXu, J., Liang, C., \u0026amp; Li, J. (2022). A signal recognition particle-related joint model of LASSO regression, SVM-RFE and artificial neural network for the diagnosis of systemic sclerosis-associated pulmonary hypertension. \u003cem\u003eFrontiers In Genetics\u003c/em\u003e, \u003cem\u003e13\u003c/em\u003e, 1078200. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fgene.2022.1078200\u003c/span\u003e\u003cspan address=\"10.3389/fgene.2022.1078200\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZihan, C. L. Y., Zong-wei, H., \u0026amp; Sufang, Q. (2023). 910P Machine learning-based survival prediction model for postoperative parotid mucoepidermoid carcinoma. \u003cem\u003eANNALS OF ONCOLOGY\u003c/em\u003e, \u003cem\u003e34\u003c/em\u003e(0923\u0026ndash;7534), S578.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTateishi, A. H. H., Takasawa, K., Kouno, N., Mizuno, T., Okubo, Y., et al. (2023). Development and validation of a prediction model for the recurrence of stage 1 EGFR mutation positive NSCLC in patients using machine learning with WES-based gene sets. \u003cem\u003eJOURNAL OF CLINICAL ONCOLOGY\u003c/em\u003e, \u003cem\u003e41\u003c/em\u003e(0732-183X), 8563.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHe, B. Y. P., Taghavi, M. M., Secondus-Samuel, S., Albaugh, J., Garey, J. S., et al. (2023). A machine learning model to predict treatment initiation among new patients in a community oncology network. \u003cem\u003eJOURNAL OF CLINICAL ONCOLOGY\u003c/em\u003e, \u003cem\u003e41\u003c/em\u003e(0732-183X), e13539.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAleksiuk, V., Baleisis, J., Kirdaite, G., Uzieliene, I., Denkovskij, J., Bernotas, P., et al. (2023). Evaluation of Cartilage Integrity Following Administration of Oral and Intraarticular Nifedipine in a Murine Model of Osteoarthritis. \u003cem\u003eBiomedicines\u003c/em\u003e, \u003cem\u003e11\u003c/em\u003e(9). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/biomedicines11092443\u003c/span\u003e\u003cspan address=\"10.3390/biomedicines11092443\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePark, I. H. J., Hwang, H., \u0026amp; Kim, H. (2018). Evaluation of cartilage degeneration and osteoarthritis pain on female and male mouse model of osteoarthritis. \u003cem\u003eOsteoarthritis and Cartilage\u003c/em\u003e, \u003cem\u003e26\u003c/em\u003e(1063\u0026ndash;4584), S360.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJoshi, N., Yan, J., Dang, M., Slaughter, K., Wang, Y., Wu, D., et al. (2025). A mechanically resilient soft hydrogel improves drug delivery for treating post-traumatic osteoarthritis in physically active joints. \u003cem\u003eProc Natl Acad Sci U S A\u003c/em\u003e, \u003cem\u003e122\u003c/em\u003e(14), e2409729122. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1073/pnas.2409729122\u003c/span\u003e\u003cspan address=\"10.1073/pnas.2409729122\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVan Zeeland, E. M., Kassel, B., Montoya, T., Saviola, A. J., Burton, L. H., Easley, J. T., et al. (2025). Structural, functional, and proteomic based sex differences in murine post-traumatic osteoarthritis development following mechanical anterior cruciate ligament injury. \u003cem\u003eOsteoarthritis Cartilage\u003c/em\u003e, \u003cem\u003e33\u003c/em\u003e(8), 992\u0026ndash;1006. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.joca.2025.05.004\u003c/span\u003e\u003cspan address=\"10.1016/j.joca.2025.05.004\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLeite, C. B. G., Fricke, H. P., Tavares, L. P., Nshimiyimana, R., Mekhail, J., Kilgallen, E., et al. (2025). Maresin 1-LGR6 axis mitigates inflammation and posttraumatic osteoarthritis after transection of the anterior cruciate ligament in mice. \u003cem\u003eOsteoarthritis Cartilage\u003c/em\u003e, \u003cem\u003e33\u003c/em\u003e(7), 861\u0026ndash;873. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.joca.2025.03.005\u003c/span\u003e\u003cspan address=\"10.1016/j.joca.2025.03.005\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTakahata, K., Lin, Y. Y., Osipov, B., Arakawa, K., Enomoto, S., Christiansen, B. A., et al. (2025). Concurrent joint contact in anterior cruciate ligament injury induces cartilage micro-injury and subchondral bone sclerosis, resulting in knee osteoarthritis. \u003cem\u003eOsteoarthritis Cartilage\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.joca.2025.06.007\u003c/span\u003e\u003cspan address=\"10.1016/j.joca.2025.06.007\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAi, H., Dou, C., Wu, Y., Zhang, D., Zhang, Z., Zhang, C., et al. (2025). Osteoclast-derived apoptotic bodies accelerate the pathological progression of osteoarthritis via disturbing subchondral bone remodeling. \u003cem\u003eJ Orthop Translat\u003c/em\u003e, \u003cem\u003e51\u003c/em\u003e, 108\u0026ndash;118. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.jot.2025.01.004\u003c/span\u003e\u003cspan address=\"10.1016/j.jot.2025.01.004\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChe, X., Zhang, C., Zhuo, Y., Li, H., Ding, X., Wu, G., et al. (2025). Serum TSP-1 is a useful biomarker in severity assessment and the diagnosis of osteoarthritis. \u003cem\u003eJ Transl Med\u003c/em\u003e, \u003cem\u003e23\u003c/em\u003e(1), 987. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12967-025-07022-z\u003c/span\u003e\u003cspan address=\"10.1186/s12967-025-07022-z\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKhan, N. M., Kaiser, J. M., Chihab, S., Eng, T., \u0026amp; Drissi, H. (2025). Dimethyl Fumarate attenuates synovial inflammation, reduces nociception, and inhibits the development of post-traumatic osteoarthritis. \u003cem\u003eBiomedicine \u0026amp; Pharmacotherapy\u003c/em\u003e, \u003cem\u003e183\u003c/em\u003e, 117865. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.biopha.2025.117865\u003c/span\u003e\u003cspan address=\"10.1016/j.biopha.2025.117865\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSebastian, A., Hum, N. R., McCool, J. L., Wilson, S. P., Murugesh, D. K., Martin, K. A., et al. (2022). Single-cell RNA-Seq reveals changes in immune landscape in post-traumatic osteoarthritis. \u003cem\u003eFrontiers In Immunology\u003c/em\u003e, \u003cem\u003e13\u003c/em\u003e, 938075. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fimmu.2022.938075\u003c/span\u003e\u003cspan address=\"10.3389/fimmu.2022.938075\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBurleigh, A., Chanalaris, A., Gardiner, M. D., Driscoll, C., Boruc, O., Saklatvala, J., et al. (2012). Joint immobilization prevents murine osteoarthritis and reveals the highly mechanosensitive nature of protease expression in vivo. \u003cem\u003eArthritis And Rheumatism\u003c/em\u003e, \u003cem\u003e64\u003c/em\u003e(7), 2278\u0026ndash;2288. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1002/art.34420\u003c/span\u003e\u003cspan address=\"10.1002/art.34420\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSebastian, A., Chang, J. C., Mendez, M. E., Murugesh, D. K., Hatsell, S., Economides, A. N., et al. (2018). Comparative Transcriptomics Identifies Novel Genes and Pathways Involved in Post-Traumatic Osteoarthritis Development and Progression. \u003cem\u003eInternational Journal Of Molecular Sciences\u003c/em\u003e, \u003cem\u003e19\u003c/em\u003e(9). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/ijms19092657\u003c/span\u003e\u003cspan address=\"10.3390/ijms19092657\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUm, I. H., Scott-Hayward, L., MacKenzie, M., Tan, P. H., Kanesvaran, R., Choudhury, Y., et al. (2019). 978P - Algorithms derived from quantitative pathology can be a gatekeeper in patient selection for clinical trials in localised clear cell renal cell carcinoma (ccRCC). \u003cem\u003eAnnals of Oncology\u003c/em\u003e, \u003cem\u003e30\u003c/em\u003e, v397. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/annonc/mdz249.074\u003c/span\u003e\u003cspan address=\"10.1093/annonc/mdz249.074\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGlasson, S. S., Blanchet, T. J., \u0026amp; Morris, E. A. (2007). The surgical destabilization of the medial meniscus (DMM) model of osteoarthritis in the 129/SvEv mouse. \u003cem\u003eOsteoarthritis Cartilage\u003c/em\u003e, \u003cem\u003e15\u003c/em\u003e(9), 1061\u0026ndash;1069. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.joca.2007.03.006\u003c/span\u003e\u003cspan address=\"10.1016/j.joca.2007.03.006\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHautz, T., Salcher, S., Fodor, M., Sturm, G., Ebner, S., Mair, A., et al. (2023). Immune cell dynamics deconvoluted by single-cell RNA sequencing in normothermic machine perfusion of the liver. \u003cem\u003eNature Communications\u003c/em\u003e, \u003cem\u003e14\u003c/em\u003e(1), 2285. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41467-023-37674-8\u003c/span\u003e\u003cspan address=\"10.1038/s41467-023-37674-8\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGreven, D. E., Cohen, E. S., Gerlag, D. M., Campbell, J., Woods, J., Davis, N., et al. (2015). Preclinical characterisation of the GM-CSF receptor as a therapeutic target in rheumatoid arthritis. \u003cem\u003eAnnals Of The Rheumatic Diseases\u003c/em\u003e, \u003cem\u003e74\u003c/em\u003e(10), 1924\u0026ndash;1930. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1136/annrheumdis-2014-205234\u003c/span\u003e\u003cspan address=\"10.1136/annrheumdis-2014-205234\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCrotti, C., Agape, E., Becciolini, A., Biggioggero, M., \u0026amp; Favalli, E. G. (2019). Targeting Granulocyte-Monocyte Colony-Stimulating Factor Signaling in Rheumatoid Arthritis: Future Prospects. \u003cem\u003eDrugs\u003c/em\u003e, \u003cem\u003e79\u003c/em\u003e(16), 1741\u0026ndash;1755. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s40265-019-01192-z\u003c/span\u003e\u003cspan address=\"10.1007/s40265-019-01192-z\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCaxaria, S., Kouvatsos, N., Eldridge, S. E., Alvarez-Fallas, M., Thorup, A. S., Cici, D., et al. (2023). Disease modification and symptom relief in osteoarthritis using a mutated GCP-2/CXCL6 chemokine. \u003cem\u003eEmbo Molecular Medicine\u003c/em\u003e, \u003cem\u003e15\u003c/em\u003e(1), e16218. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.15252/emmm.202216218\u003c/span\u003e\u003cspan address=\"10.15252/emmm.202216218\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang, J., Li, K., \u0026amp; Qiu, X. (2024). Exploring causal correlations between inflammatory cytokines and knee osteoarthritis: a two-sample Mendelian randomization. \u003cem\u003eFrontiers In Immunology\u003c/em\u003e, \u003cem\u003e15\u003c/em\u003e, 1362012. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fimmu.2024.1362012\u003c/span\u003e\u003cspan address=\"10.3389/fimmu.2024.1362012\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSeco-Calvo, J., S\u0026aacute;nchez-Herr\u0026aacute;ez, S., Casis, L., Valdivia, A., Perez-Urzelai, I., Gil, J., et al. (2020). Synovial fluid peptidase activity as a biomarker for knee osteoarthritis clinical progression. \u003cem\u003eBone Joint Res\u003c/em\u003e, \u003cem\u003e9\u003c/em\u003e(11), 789\u0026ndash;797. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1302/2046-3758.911.Bjr-2020-0022.R2\u003c/span\u003e\u003cspan address=\"10.1302/2046-3758.911.Bjr-2020-0022.R2\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi, T., Peng, J., Li, Q., Shu, Y., Zhu, P., \u0026amp; Hao, L. (2022). The Mechanism and Role of ADAMTS Protein Family in Osteoarthritis. \u003cem\u003eBiomolecules\u003c/em\u003e, \u003cem\u003e12\u003c/em\u003e(7). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/biom12070959\u003c/span\u003e\u003cspan address=\"10.3390/biom12070959\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHurley, M. M., Coffin, J. D., Doetschman, T., Valera, C., Clarke, K., \u0026amp; Xiao, L. (2022). FGF receptor inhibitor BGJ398 partially rescues osteoarthritis-like phenotype in older high molecular weight FGF2 transgenic mice via multiple mechanisms. \u003cem\u003eScientific Reports\u003c/em\u003e, \u003cem\u003e12\u003c/em\u003e(1), 15968. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41598-022-20269-6\u003c/span\u003e\u003cspan address=\"10.1038/s41598-022-20269-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHsiao, Y. W., Chi, J. Y., Li, C. F., Chen, L. Y., Chen, Y. T., Liang, H. Y., et al. (2022). Disruption of the pentraxin 3/CD44 interaction as an efficient therapy for triple-negative breast cancers. \u003cem\u003eClin Transl Med\u003c/em\u003e, \u003cem\u003e12\u003c/em\u003e(1), e724. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1002/ctm2.724\u003c/span\u003e\u003cspan address=\"10.1002/ctm2.724\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eArmanmehr, A., Jafari Khamirani, H., Zoghi, S., \u0026amp; Dianatpour, M. (2022). Analysis of DYRK1B, PPARG, and CEBPB Expression Patterns in Adipose-Derived Stem Cells from Patients Carrying DYRK1B R102C and Healthy Individuals During Adipogenesis. \u003cem\u003eMetabolic Syndrome And Related Disorders\u003c/em\u003e, \u003cem\u003e20\u003c/em\u003e(10), 576\u0026ndash;583. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1089/met.2021.0140\u003c/span\u003e\u003cspan address=\"10.1089/met.2021.0140\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePanaroni, C., Fulzele, K., Soucy, R., Siu, K., Mukaihara, K., Huang, C., et al. (2018). Arachidonic Acid Induces Ferroptosis-Mediated Cell-Death in Multiple Myeloma. \u003cem\u003eBlood\u003c/em\u003e, \u003cem\u003e132\u003c/em\u003e, 4498. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1182/blood-2018-99-118482\u003c/span\u003e\u003cspan address=\"10.1182/blood-2018-99-118482\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYu, E., Zhang, M., Xu, G., Liu, X., \u0026amp; Yan, J. (2023). Consensus cluster analysis of apoptosis-related genes in patients with osteoarthritis and their correlation with immune cell infiltration. \u003cem\u003eFrontiers In Immunology\u003c/em\u003e, \u003cem\u003e14\u003c/em\u003e, 1202758. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fimmu.2023.1202758\u003c/span\u003e\u003cspan address=\"10.3389/fimmu.2023.1202758\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLu, S., \u0026amp; Fang, C. (2024). Isosakuranetin inhibits subchondral osteoclastogenesis for attenuating osteoarthritis via suppressing NF-κB/CXCL2 axis. \u003cem\u003eInternational Immunopharmacology\u003c/em\u003e, \u003cem\u003e143\u003c/em\u003e(Pt 1), 113321. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.intimp.2024.113321\u003c/span\u003e\u003cspan address=\"10.1016/j.intimp.2024.113321\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBi, J., Li, Q., Yang, Z., Cai, L., Lv, T., Yang, X., et al. (2021). CXCL2 Impairs Functions of Bone Marrow Mesenchymal Stem Cells and Can Serve as a Serum Marker in High-Fat Diet-Fed Rats. \u003cem\u003eFront Cell Dev Biol\u003c/em\u003e, \u003cem\u003e9\u003c/em\u003e, 687942. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fcell.2021.687942\u003c/span\u003e\u003cspan address=\"10.3389/fcell.2021.687942\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"stem-cell-reviews-and-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"stcr","sideBox":"Learn more about [Stem Cell Reviews and Reports](https://www.springer.com/journal/12015)","snPcode":"12015","submissionUrl":"https://submission.nature.com/new-submission/12015/3","title":"Stem Cell Reviews and Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"PTOA, Single-Cell Analysis, Granulocytes, Bioinformatics, ML, IIA","lastPublishedDoi":"10.21203/rs.3.rs-8655656/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8655656/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eObjectives\u003c/h2\u003e \u003cp\u003eOsteoarthritis (OA) is the most prevalent joint disorder, whereas post-traumatic osteoarthritis (PTOA) denotes a form of arthritis that arises secondary to acute joint injury.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eAs no curative therapy for PTOA currently exists, prevention and early intervention constitute critical avenues of investigation. Among the various injuries predisposing to PTOA, anterior cruciate ligament (ACL) rupture is among the most prototypical. Our study systematically analyzed dynamic RNA sequencing data derived from murine knee joint tissues after ACL rupture. Granulocyte subpopulations were systematically classified and annotated, revealing subgroups related to PTOA progression. Furthermore, co-expressed gene modules positively correlated with OA granulocytes were identified and selected via high-dimensional weighted gene co-expression network analysis (hdWGCNA). Utilizing machine learning (ML)-based methodologies, a predictive model was constructed and validated through nomogram models, calibration curves, and decision curves. The prognostic utility of characteristic genes in these OA granulocyte subtypes was also investigated. Additionally, immune infiltration analysis (IIA) was carried out to visualize immune cell infiltration and explore the relationships between key genes and immune cells, and qRT-PCR was performed for the results.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eIn summary, this study identified the distinctive molecular and biological characteristics of granulocyte subtypes and applied ML algorithms to predict diagnostic biomarkers specific to PTOA.\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eThese findings hold promise for improving targeted predictive capabilities for the disease, with the ultimate goal of interrupting the vicious cycle of inflammation and mechanical abnormalities prior to the onset of irreversible joint damage.\u003c/p\u003e","manuscriptTitle":"Machine Learning-Based Identification of Molecular Signatures in PTOA Cell Subtypes via Single-Cell Transcriptomics","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-01-28 14:54:05","doi":"10.21203/rs.3.rs-8655656/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2026-03-27T12:43:07+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-03-27T08:31:54+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"296836524595464868028053915531917821978","date":"2026-03-26T10:51:21+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"118415339034939672975197454674273069988","date":"2026-02-09T11:39:39+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"5140183023812567111506168500658305370","date":"2026-01-23T17:02:09+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-01-23T16:52:27+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-01-22T06:47:45+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-01-22T06:45:28+00:00","index":"","fulltext":""},{"type":"submitted","content":"Stem Cell Reviews and Reports","date":"2026-01-21T05:38:21+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"stem-cell-reviews-and-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"stcr","sideBox":"Learn more about [Stem Cell Reviews and Reports](https://www.springer.com/journal/12015)","snPcode":"12015","submissionUrl":"https://submission.nature.com/new-submission/12015/3","title":"Stem Cell Reviews and Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"526dc74c-c19f-47b1-9b81-6a676d882fff","owner":[],"postedDate":"January 28th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2026-04-20T16:05:12+00:00","versionOfRecord":{"articleIdentity":"rs-8655656","link":"https://doi.org/10.1007/s12015-026-11121-9","journal":{"identity":"stem-cell-reviews-and-reports","isVorOnly":false,"title":"Stem Cell Reviews and Reports"},"publishedOn":"2026-04-14 15:57:56","publishedOnDateReadable":"April 14th, 2026"},"versionCreatedAt":"2026-01-28 14:54:05","video":"","vorDoi":"10.1007/s12015-026-11121-9","vorDoiUrl":"https://doi.org/10.1007/s12015-026-11121-9","workflowStages":[]},"version":"v1","identity":"rs-8655656","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8655656","identity":"rs-8655656","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}