Insulin-like growth factor binding protein 4 (IGFBP4)high fibroblasts contribute to extracellular matrix degradation in rheumatoid arthritis

To the Editor: Rheumatoid arthritis (RA) is a chronic autoimmune disease marked by persistent synovial inflammation, cartilage degradation, and bone erosion.[1] Despite treatment advancements, 30% of patients have inadequate responses, suggesting the presence of unexplored pathways.[2] Fibroblast-like synoviocytes (FLSs) maintain joint homeostasis but contribute to inflammation and tissue destruction in RA.[3] Recent single-cell RNA sequencing (scRNA-seq) identified distinct FLS subsets,[4] including thymocyte differentiation antigen 1 positive (THY1)+ and fibroblast activation protein alpha positive (FAPα)+ fibroblasts, which exacerbate inflammation and severe arthritis.[5] While inflammatory FLS subsets have been well-studied, less attention has been given to those involved in extracellular matrix (ECM) remodeling, a crucial factor in RA pathology. The study recruited a total of 207 participants, including 56 RA patients in remission, 56 RA patients with active disease, 35 systemic lupus erythematosus (SLE) patients, 4 osteoarthritis (OA) patients, and 56 healthy controls. All participants met the inclusion criteria and provided informed consent to participate in the study. The ethical approval for this study was granted by the Institutional Review Board (IRB) of the General Hospital of Ningxia Medical University (No. KYLL-2025-0993). Animal care and experimental procedures were approved by the Laboratory Animal Research Board of Ningxia Medical University (No. KYLL-2024-1033). Herein, scRNA-seq and immunofluorescence were used to identify a fibroblast population with high expression of insulin-like growth factor binding protein 4 (IGFBP4hi) in RA synovial tissues. Subsequently, proteomic analysis of RA-FLS culture supernatants was conducted to assess IGFBP4 levels. Moreover, adeno-associated virus (AAV) was used to knockdown Igfbp4 expression in synovial fibroblasts in vivo. The impact on arthritis lesions in the zymosan A (ZYM)-induced arthritic mouse model was evaluated using arthritis scoring, hematoxylin and eosin (H&E) staining and toluidine blue O (TBO) staining, along with the expression of collagen Type IV Alpha 1 Chain (COL4A1), COL4A2, fibronectin 1 (FN1), and matrix metallopeptidase 2 (MMP2). An in vitro RA-FLS model was established using IL-1β stimulation to investigate how IGFBP4 modulates the expression of COL4A1, COL4A2, and FN1. Finally, serum IGFBP4 levels were assessed for their diagnostic potential in RA patients, with healthy individuals and systemic lupus erythematosus (SLE) patients serving as controls. The detailed information regarding the methods, antibody, and primer sequences is presented in the Supplementary Methods and Supplementary Tables 1 and 2, https://links.lww.com/CM9/C885. H&E staining of the synovial tissue from RA patients revealed vascular proliferation, fibrous hyperplasia, and extensive infiltration of lymphocytes and plasma cells. Knee magnetic resonance imaging (MRI) showed signs of joint damage, including medial and lateral meniscal degeneration with outward bulging, an anterior cruciate ligament (ACL) tear, small effusions in the joint cavity and suprapatellar bursa, and pronounced synovial thickening, indicative of active RA inflammation [Supplementary Figure 1A, https://links.lww.com/CM9/C885]. To investigate the cellular heterogeneity in RA synovium, by using MobiDrop platform, scRNA-seq was conducted on synovial tissues from one RA patient and one OA control [Supplementary Table 3, https://links.lww.com/CM9/C885], and 20,077 cells were retained for analysis after filtering. Based on canonical marker genes established,[6] uniform manifold approximation and projection (UMAP) visualization resolved eight cell types, comprising dendritic cells (DCs), B cells, macrophages, endothelial cells, neutrophils, T cells, fibroblasts, and pericytes [Supplementary Figure 1B and C, https://links.lww.com/CM9/C885]. Comparative analysis revealed a substantial expansion of B cells and fibroblasts in RA synovial tissues, whereas macrophages, neutrophils, and T cells were markedly reduced [Supplementary Figure 1D, https://links.lww.com/CM9/C885]. Differentially expressed genes (DEGs) distinguishing each cluster were identified [Supplementary Figure 1E, https://links.lww.com/CM9/C885]. Gene Ontology (GO) enrichment analysis showed that fibroblasts were significantly enriched for ECM degradation processes, including extracellular matrix disassembly and the collagen catabolic process [Supplementary Figure 1F, https://links.lww.com/CM9/C885]. These findings underscore the expansion of ECM-degrading fibroblasts in RA synovial tissues and suggest their critical role in RA pathogenesis. To further investigate transcriptional alterations in ECM-degrading fibroblasts in RA, we performed unsupervised analysis, which revealed two clusters separated by IGFBP4 expression, where IGFBP4hi denotes fibroblasts with IGFBP4 expression above the dataset median and IGFBP4lo denotes those at or below the median. Quantitative analysis revealed a significant expansion of IGFBP4hi fibroblasts in RA synovial tissues compared to OA controls; similar results were also confirmed in the public Single Cell Portal dataset SCP738 [Supplementary Figure 2A–F, https://links.lww.com/CM9/C885]. Pseudotime analysis in RA showed that IGFBP4lo fibroblasts progress toward IGFBP4hi along distinct trajectories [Supplementary Figure 2G, https://links.lww.com/CM9/C885]. Consistent with scRNA-seq, immunofluorescence revealed more IGFBP4+ cells co-expressing cadherin-11 (CDH11) in RA synovium than in OA [Supplementary Figure 2H–I, https://links.lww.com/CM9/C885]. Immunohistochemistry and Western blot further validated IGFBP4 upregulation in RA patients [Supplementary Figure 2J–K, https://links.lww.com/CM9/C885]. Importantly, IGFBP4hi fibroblast proportion was positively correlated with disease activity score 28 (DAS28) [Supplementary Figure 2L, https://links.lww.com/CM9/C885]. GO analysis identified ECM degradation as a primary function of IGFBP4hi fibroblasts [Supplementary Figure 2M, https://links.lww.com/CM9/C885], and monocle pseudotime revealed stepwise downregulation of COL4A1, COL4A2, and FN1 during the IGFBP4lo-to-IGFBP4hi transition [Supplementary Figure 2N, https://links.lww.com/CM9/C885]. Immunofluorescence further showed increased IGFBP4+CDH11+ fibroblasts with low COL4A1, COL4A2, and FN1 in RA compared with OA [Supplementary Figure 2O, https://links.lww.com/CM9/C885]. Single-cell interaction analysis revealed that the IGFBP4hi fibroblasts interact most closely with DCs [Supplementary Figure 2P, https://links.lww.com/CM9/C885]. Moreover, we identified that the immunomodulatory GO terms predominantly enriched in DCs include dendritic cell chemotaxis and activation of innate immune response [Supplementary Figure 2Q, https://links.lww.com/CM9/C885]. Proteomic profiling of RA-FLS and OA-FLS also showed that IGFBP4 was notably upregulated in RA-FLS [Supplementary Figure 2R, https://links.lww.com/CM9/C885]. Protein–protein interaction (PPI) enrichment analysis revealed that IGFBP4 interacts closely with COL4A1, COL4A2, and FN1 [Supplementary Figure 2S, https://links.lww.com/CM9/C885]. Chronic arthritis model was established in SKG mice using ZYM. Adeno-associated virus serotype 9 (AAV9)-shIgfbp4 was injected intra-articularly into the mouse joints to achieve local knockdown of Igfbp4 in vivo [Supplementary Figure 3A, https://links.lww.com/CM9/C885]. In ZYM-treated mice, Igfbp4 knockdown reduced arthritis scores, attenuated synovial inflammatory-cell infiltration and cartilage damage, and was associated with increased expression of COL4A1, COL4A2, and FN1, and decreased expression of MMP2 [Supplementary Figure 3B–D, https://links.lww.com/CM9/C885]. To better understand how IGFBP4 regulates the ECM, we compared IGFBP4hi and IGFBP4lo fibroblasts to identify DEGs, and found that the DEGs are associated with key signaling pathways and ECM remodeling, including IGF1 and IGF1R, components of the phosphoinositide 3-kinase/protein kinase B/mammalian target of rapamycin (PI3K/AKT/mTOR) pathway such as PIK3R3, AKT3, RICTOR, FOXO1, and ECM proteins such as MMP2, COL4A1, COL4A2, FN1, TIMP1, and TIMP3 [Supplementary Figure 4A, https://links.lww.com/CM9/C885]. These expression patterns suggest that IGFBP4 may act through the IGF-I/IGF1R-PI3K-AKT pathway to promote MMP2 activation, thereby driving the degradation of COL4A1, COL4A2, and FN1. We further silenced and overexpressed Igfbp4 in RA-FLS using small interfering RNA (siRNA), si-Igfbp4) and a lentiviral vector (LV), respectively. Among the three tested siRNAs, the third siRNA most effectively suppressed endogenous Igfbp4 expression [Supplementary Figure 4B–D, https://links.lww.com/CM9/C885], therefore, the third siRNA was selected for subsequent experiments. Additionally, LV-mediated overexpression significantly upregulated endogenous Igfbp4 levels in RA-FLS [Supplementary Figure 4E–G, https://links.lww.com/CM9/C885]. IL-1β treatment and/or IGFBP4 overexpression increased the phosphorylation of IGF1R, AKT, and mTOR, as evidenced by higher phospho/total ratios. These treatments also elevated MMP-2 levels and reduced the expression of ECM proteins COL4A1, COL4A2, and FN1 in RA-FLS. Notably, combination with recombinant IGF-I protein further enhanced the phosphorylation of IGF1R, AKT, and mTOR, elevated MMP-2 levels, and reduced the expression of COL4A1, COL4A2, and FN1 [Supplementary Figure 4H, https://links.lww.com/CM9/C885]. In contrast, Igfbp4 knockdown resulted in opposite effects, showing a decrease in the phosphorylation levels of IGF1R, AKT, and mTOR, along with reduced MMP-2 levels and increased expression of COL4A1, COL4A2, and FN1 [Supplementary Figure 4I, https://links.lww.com/CM9/C885]. To determine the correlation between IGFBP4 and RA, serum samples from patients with RA, healthy participants, and patients with SLE, as well as synovial fluid from patients with RA, were collected, and IGFBP4 concentrations were evaluated. Serum IGFBP4 levels were significantly higher in RA patients than in both healthy controls and SLE patients [Supplementary Figure 5A, https://links.lww.com/CM9/C885]. Additionally, patients with active RA showed significantly higher serum IGFBP4 expression than those in remission and healthy controls [Supplementary Figure 5B, https://links.lww.com/CM9/C885]. Moreover, RA patients with active disease exhibited higher synovial fluid IGFBP4 levels than those in remission [Supplementary Figure 5C, https://links.lww.com/CM9/C885]. The clinical characteristics of patients with RA in remission and those with active RA are summarized in Supplementary Table 4, https://links.lww.com/CM9/C885. To assess the diagnostic significance of IGFBP4, univariate and multivariable logistic regression analyses were applied to identify variables independently associated with active RA and found that serum IGFBP4 (odds ratio [OR]: 1.477, 95% confidence interval [CI]: 1.061–2.056, P = 0.021), anti-cyclic citrullinated peptide antibody (anti-CCP) antibody (OR: 1.011, 95% CI: 1.004–1.018, P = 0.002), c-reactive protein (CRP) (OR: 1.049, 95% CI: 1.003–1.097, P = 0.037), erythrocyte sedimentation rate (ESR) (OR: 1.047, 95% CI: 1.004–1.092, P = 0.031), and tender joint count (TJC) (OR: 1.154, 95% CI: 1.055–1.262, P = 0.002) can independently predict active RA [Supplementary Table 5, https://links.lww.com/CM9/C885]. Based on these predictors, we developed a nomogram incorporating serum IGFBP4, anti-CCP antibody, CRP, ESR, and TJC to estimate the probability of active disease in RA patients [Supplementary Figure 5D, https://links.lww.com/CM9/C885]. Importantly, integrating IGFBP4 with anti-CCP antibody, CRP, ESR, and TJC significantly improved the predictive accuracy of the nomogram for identifying active disease in RA patients compared to the model based on these four factors alone [Supplementary Figure 5E–F, https://links.lww.com/CM9/C885]. Specifically, the C-index of the nomogram increased from 0.918 (95% CI: 0.868–0.967) to 0.929 (95% CI: 0.885–0.973) upon inclusion of IGFBP4. Calibration curves demonstrated strong agreement between predicted and observed probabilities, reflecting excellent model calibration [Supplementary Figure 5G, https://links.lww.com/CM9/C885]. Furthermore, decision curve analysis (DCA) indicated that the addition of IGFBP4 provided superior net clinical benefit across a wide range of risk thresholds [Supplementary Figure 5H, https://links.lww.com/CM9/C885]. In this study, we identified IGFBP4hi fibroblasts as a distinct ECM-degradative subset in RA synovium that signals through the IGF-1/IGF1R-PI3K-AKT axis to activate MMP2, leading to degradation of COL4A1, COL4A2, and FN1. Prior work has established that FAPα+ THY1− fibroblasts constitute a synovial-lining, tissue-destructive population enriched for degradative matrix programs.[5] Although their lineage trajectories, spatial niches, and upstream drivers differ, both subsets converge on matrix breakdown.[5] In summary, our results offer a new mechanistic avenue for understanding RA joint damage by delineating extracellular matrix degradation as a key driver. However, certain potential issues require further examination. Specifically, we did not validate our conclusions using synovial fibroblast-specific Igfbp4 knockout mice. Funding This study was supported by grants from the High-Level Talent Recruitment Program (No. 2025BEH04109), the Ningxia Hui autonomous region Key Research and Development Plan (Nos. 2023BEG03014 and 2023BEG02021), the National Natural Science Foundation of China (Nos. 82460020 and 82500103), Special Project of Central Government Guiding Local Science and Technology Development of Ningxia (No. 2024FRD05046), and the Higher Education Research Program of the Ningxia Hui autonomous region’s Department of Education (No. NYG2024161). Conflicts of interest None.