To investigate the role of NETs in RA, we first obtained clinical samples from well-characterized patient cohorts. All experimental protocols were approved by the Ethics Committee of Nanjing Drum Tower Hospital (reference number 2021-544-01). Written informed consent was obtained from all patients. RA patients were recruited from the Department of Rheumatology and Immunology at Nanjing Drum Tower Hospital and fulfilled either the 2010 American College of Rheumatology (ACR)/European League Against Rheumatism (EULAR) classification criteria or the 1987 ACR classification criteria (32, 33). Inclusion criteria required patients to be aged 18 years or older. Exclusion criteria included concomitant rheumatic diseases, severe infections, malignancy, pregnancy, and lactation. Peripheral blood was collected from RA patients and age-and-sex matched healthy controls (HC). Human synovial tissues were obtained from RA patients and age-and-sex matched osteoarthritis (OA) patients undergoing knee replacement surgery at our institution.

To extend our in vitro findings to an in vivo setting, we utilized a murine model of inflammatory arthritis. Eight-week male DBA1 mice were purchased from Chares River. All animal experiments were performed under an institutionally approved protocol for the use of animal research. Collagen-induced arthritis (CIA) was induced as previously described (34). Briefly, bovine type II collagen (Chondrex) was dissolved at 4 °C overnight in acetic acid and emulsified with an equal volume of complete or incomplete Freund’s adjuvant (Chondrex). Mice were immunized with 100μg of type II collagen by intradermal injection at the base of the tail on day 1 and day 21. To assess the specific contribution of NETs to disease progression, we administered peptidyl arginine deiminase 4 inhibitor GSK 484 (Cayman) after disease onset (day 28), when arthritis had become established (arthritis score ≥2). CIA mice received daily intraperitoneal injections of either 200ul saline or GSK484 (4 mg/kg) for 14 days. Healthy DBA1 mice served as negative control. On day 42, mice were euthanized, and blood, bone marrow and ankle joints were collected for subsequent studies. Arthritis scoring was performed twice weekly according to the previous protocol (35).

To generate NETs for subsequent functional assays, neutrophils were isolated from peripheral blood of RA patients or HC using Polymorphprep (Axi-Shield) following the manufacturer’s instructions (36). Neutrophils were suspended in RPMI-1640 (Gibco) containing 3% fetal bovine serum (FBS) and seeded into 6-well plates (1×10⁶cells/well). After incubation at 37 °C and 5% CO₂ for 4h under non-interventional conditions, NETs were digested with 10U/mL of micrococcal nuclease (Thermo Fisher) for 15min at 37 °C. Supernatants containing NETs fragments were collected and centrifuged at 1,000g for 10min to remove cellular debris. Separated NETs were transferred to fresh tubes and stored at -80 °C until further use.

To confirm and quantify NETs formation, we employed both immunofluorescence microscopy and flow cytometry. Neutrophils isolated from RA or HC peripheral blood were cultured into 24-well plates (1×10⁵cells/well) and incubated undisturbed at 37 °C and 5% CO₂ for 4h. Cells were subsequently fixed with 4% paraformaldehyde for 15min, followed by blocking with 2% bovine serum albumin (BSA) for 1h. Cells were incubated overnight at 4 °C with the following antibodies: anti-myeloperoxidase (MPO) mAb (Thermo Fisher) or anti-neutrophil elastase (NE) mAb (Thermo Fisher). DNA was stained using 4′,6-diamidino-2-phenylindole (DAPI) (Beyotime Biotechnology). Neutrophils isolated from mouse blood or bone marrow were cultured as above, and then were stained with anti-MPO mAb (Thermo Fisher). NETs were observed by a confocal fluorescence microscope (Olympus), and two independent individuals manually quantified neutrophils and NETs. For flow cytometric analysis, RA or HC derived neutrophils (8×10³ cells/well) were seeded into 96-well plates and incubated in a humidified incubator at 37 °C with 5% CO₂ for 4h, without any intervention. Following incubation, cells were stained with SYTOX Green, a membrane-impermeable DNA-binding dye, in accordance with the manufacturer’s instructions (Invitrogen). Cellular debris was removed by filtration through a mesh, and the remaining cells were subsequently analyzed using the Fortessa flow cytometer (BD Biosciences).

Synovial tissues from RA patients were minced into small pieces and digested in 0.2% of type I collagen (Sigma) dissolved in DMEM (Gibco) for 10h at 37°C in 5% CO₂. Digestion was terminated by adding DMEM with 15% FBS. The cell suspension was filtered through a 100μm cell strainer (Falcon) and resuspended in DMEM with 15% FBS. Cells were cultured for 24 hours to allow adherence, after which the medium was changed to remove non-adherent cells, and the FBS concentration was reduced to 10%. The medium was changed every 2 or 3 days and the cells were passaged at 80% confluence. FLS at passage 3–6 were used for subsequent experiments.

To determine whether NETs influence FLS proliferation, we performed CCK8 assays. RA FLS (1×10⁴ cells/well) were seeded in 96-well plates overnight. Cells were then cultured with RA or HC NETs at various concentrations of 0, 5, 50, and 500ug/ml for 24h, 48h or 72h. After incubation, 10μl of CCK8 reagent (Biosharp) was added to each well and incubated at 37 °C for 2h. The absorbance at 450nm was measured to calculate cell viability. To assess the effect of NETs on FLS migration, we conducted wound healing assays. RA FLS were seeded in 6-well plates at a density of 1×10⁶ cells/well. Once confluent, a uniform scratch was created using a pipette tip, followed by washing with PBS to remove detached cells. DMEM supplemented with 5% FBS and either RA or HC NETs (50μg/mL), or control medium without NETs, was added to the wells. Photographs of the scratch were taken at baseline (0h) and 24h post-treatment. Scratch distance and area were measured using image analysis software. Each condition was tested in three independent experiments.

To explore the molecular mechanisms underlying NETs’ effects on FLS, we performed transcriptome analysis. RA FLS were treated with RA or HC NETs (50ug/ml) for 24h, with untreated FLS serving as controls. Total RNA was isolated and detected for quality control. Library construction, quality inspection, and sequencing were performed by Novogene Bioinformatics Technology Co., Ltd. using the Illumina NovaSeq 6000 system. Differential expression analysis (three biological replicates per condition) was performed using the DESeq2 package. Transcripts with P values of <0.05 and absolute values of log2 (fold change) >1 were assigned as differentially expressed transcripts.

To validate key findings from RNA sequencing, we performed RT-qPCR. Total RNA was extracted from RA FLS or mouse synovium tissues with TRIzol reagent (Invitrogen). cDNA was synthesized from 1μg RNA using RT Mix II (Vazyme Biotech). Real-time PCR was performed using SYBR Green Master Mix (Vazyme Biotech) with the following primers: hGAPDH: 5′-GCACCGTCAAGGCTGAGAAC-3′ (forward), 5′-GCACCGTCAAGGCTGAGAAC-3′(reverse); hGP1BA: 5′-ACCATCCTGGTGTCTGCCACAA-3′ (forward), 5′-ACGGAGCTTTGGTGGCTGATCA-3′ (reverse); hZNF564: 5′-ATGGGAAGACCAGAGCATTGAAG-3′ (forward), 5′-TGACTGAAGGCTTCTCCACATTG-3′ (reverse); hADGRG2: 5′-CTGGTCAGACAATGGCTGCTCT-3′(forward), 5′-CAGAGCCATCATTTGAGCAGGC-3′(reverse); mGAPDH: 5′-CATCACTGCCACCCAGAAGACTG-3′(forward), 5′-ATGCCAGTGAGCTTCCCGTTCAG-3′(reverse); mGP1BA: 5′-GATGTGCCAACTTGGACAATGCG-3′(forward), 5′-CTTGACCTCAGTTCTTGTGGCAG-3′ (reverse). The StepOnePlus system (Applied Biosystems) was used for real-time PCR. Data were analyzed using the 2^-ΔΔCt method, normalized to GAPHD.

To establish a causal relationship between identified genes and NETs’ effects, we performed gene knockdown experiments. Commercially available siRNA targeting GP1BA and control reagents (siNC) were purchased from Invitrogen and transduced at a concentration of 0.2μM using Opti-MEM and Lipofectamine2000. After transduction, RA FLS were stimulated with RA NETs (50ug/ml) for 24h, followed by proliferation and migration assays. RA FLS without NETs stimulation were used as blank control.

Proteins were extracted from human or mouse synovium tissues with RIPA buffer containing phosphatase and protease inhibitors and were separated on 10% acrylamide/bis-SDS gels (Epizyme) and transferred onto polyvinylidene fluoride membranes (Millipore). After blocking with 5% BSA (Sinopharm) for 1h at room temperature, membranes were incubated with primary antibodies against GAPDH (CST, 1:2000) or GPIbα (Proteintech, 1:2000) overnight at 4°C. Membranes were then incubated in secondary antibodies (CST, 1:4000) for 1h at room temperature. After repeated washing, bands were scanned by the Tanon-5200 chemiluminescent imaging system and semi-quantified with ImageJ Software.

To examine pathological changes and protein expression in situ, we performed histological staining. Sections of human synovium or mouse ankle joints were deparaffinized and hydrated with gradient alcohol. For hematoxylin and eosin (H&E) staining, sections were stained with H&E, dehydrated, and mounted with neutral gum. Histopathological changes were observed and recorded using an optical microscope. For immunohistochemistry, tissues were fixed with 4% paraformaldehyde overnight, dehydrated through graded ethanol, cleared, and embedded in paraffin. Sections were stained with antibodies against GPIbα, ZNF564 or ADGRG2 (all from Proteintech, 1:1000). Staining results were observed under an optical microscope.

To visualize protein localization and co-expression, we performed immunofluorescence staining on paraffin-embedded mouse ankle joints. Sections were deparaffinized, antigen-repaired using sodium citrate buffer, and blocked with immunofluorescence blocking buffer (CST). Sections were incubated overnight at 4°C with primary antibodies against GPIbα, Vimentin and Ki-67 (all from Proteintech). After PBS washing, secondary antibodies (Jackson ImmunoResearch) were added. Sections were incubated at 37°C for 50min in the dark, counterstained with DAPI for 10min at room temperature. and sealed with anti-fluorescence quenching tablets. Images were acquired using a confocal microscope (Nikon).

Graphpad Prism 8.0 software was used for statistical analysis. Data were expressed as mean ± standard deviation (SD). Paired data from two groups with non-normal distribution were analyzed by the non-parametric Wilcoxon signed-rank test. Multiple group comparisons of normally distributed data were conducted using an analysis of variance (ANOVA). Paired data from multiple groups with non-normal distribution were compared by the Friedman test. p < 0.05 was considered statistically significant. p-values: ns, p≥0.05; *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001.

To determine the levels of NETs formation in RA patients, neutrophils isolated from the peripheral blood of RA patients and healthy controls (HC) were stained with myeloperoxidase (MPO) and neutrophil elastase (NE). NETs were identified by the extracellular co-localization of MPO or NE with DNA, as visualized by DAPI staining (37). The results demonstrated that neutrophils from RA patients produced significantly more NETs than those from HC, consistent with previous studies (38) (Figures 1A, B). Furthermore, flow cytometry using SYTOX Green, a membrane-impermeable DNA-binding dye, revealed a higher mean fluorescence intensity (MFI) in RA neutrophils, further confirming enhanced NETs formation in RA patients. (Figures 1C, D).

To assess the effects of NETs on RA FLS proliferation, RA FLS were treated with varying concentrations of RA NETs (5, 50, or 500 µg/ml) for 24h, 48h and 72h, followed by CCK8 assay to evaluate cell proliferation. RA NETs at 50 and 500 µg/ml significantly promoted FLS proliferation, whereas 5 µg/mL RA NETs and all concentrations of HC NETs showed no notable effect (Figures 2A, B). Based on these results, 50 µg/ml RA NETs for 24h was selected for subsequent experiments (Figure 2C). In wound healing assays, RA NETs significantly enhanced FLS migration compared to HC NETs (Figures 2D, E). Collectively, these findings suggest that RA NETs directly promote RA FLS proliferation and migration.

To further elucidate the molecular mechanisms underlying RA NETs-induced FLS invasion, RNA sequencing was performed on RA FLS treated with RA NETs, HC NETs, or control medium (Figure 3A). A total of 13 upregulated and 5 downregulated genes were identified in RA FLS treated with RA NETs (Figure 3B). Based on differential gene expression analysis and relevant literature review, three candidate genes—GP1BA, ZNF564, and ADGRG2— were selected for further validation due to their potential roles in FLS proliferation and migration (39–41) (Figure 3C). RT-qPCR showed that only GP1BA expression was significantly increased upon RA NETs stimulation, whereas ZNF564 and ADGRG2 showed minor or no changes (Figure 3D). Immunohistochemistry further demonstrated that GPIbα expression was significantly elevated in RA synovial tissues compared to OA patients, whereas ZNF564 and ADGRG2 showed no notable differences between the two groups (Figure 3E). Western blot analysis also confirmed upregulated GPIbα expression in RA synovial tissues (Figure 3F). Altogether, our findings suggest that NETs may enhance FLS invasiveness through upregulation of GPIbα.

To verify the biological function of GP1BA in NETs-mediated FLS invasion, GP1BA was knocked down in RA FLS using siRNA. Our data showed that RA NETs significantly enhanced the migration and proliferation of RA FLS, whereas this effect was largely abolished in GP1BA-silenced cells (Figure 4). These findings indicate that GP1BA plays a pivotal regulatory role in RA NETs-mediated FLS invasion.

To further validate the function of NETs in vivo, a collagen-induced arthritis (CIA) model was established, and mice were treated with GSK484, a PAD4 inhibitor that blocks NET formation. Compared to CIA mice, GSK484-treated mice exhibited a gradual increase in body weight and a progressive reduction in arthritis scores (Figures 5A, B). Histopathological analysis revealed that GSK484 treatment led to a marked reduction in inflammatory cell infiltration within the synovial tissue and significant alleviation of joint damage (Figure 5C). Moreover, neutrophils isolated from the peripheral blood and bone marrow of CIA mice released more MPO and formed more NETs than those from healthy DBA/1 mice. This enhanced NETs formation was markedly suppressed by GSK484 treatment (Figures 5D-G). Taken together, these findings suggest that excessive NETs formation contributes to synovitis and joint injury in CIA.

In CIA synovium, Ki-67 expression in Vimentin+ FLS was significantly elevated, while GSK484 treatment markedly reduced Ki-67 levels, confirming that PAD4 inhibitor suppressed FLS invasiveness in vivo. (Figure 6A). Immunohistochemistry and immunofluorescence staining showed that GPIbα expression was significantly higher in CIA synovium than in healthy DBA/1 mice, and GSK484 treatment effectively reduced its expression, (Figure 6B). In addition, RT-qPCR and western blot analyses confirmed that GSK484 downregulated GPIbα at both the mRNA and protein levels in CIA synovium (Figures 6C, D). Overall, these findings indicate that NETs inhibition in CIA could alleviate FLS invasive behavior accompanied with downregulation of GPIbα.