Abstract To identify the dynamics of neutrophil autoimmunity, here we focus on anti-neutrophil cytoplasmic antibody (ANCA)-associated vasculitis and perform single-cell transcriptome and surface proteome analyses on peripheral white blood cells from patients with new-onset microscopic polyangiitis (MPA). Compared with controls, two neutrophil populations, immature neutrophils and neutrophils with type II interferon signature genes (Neu_T2ISG), are increased in patients with MPA. Trajectory and cell–cell interaction analyses identify Neu_T2ISG as a subset that differentiates from mature neutrophils upon stimulation with IFN-γ and TNF, which synergize to induce myeloperoxidase and Fcγ receptors expression on the neutrophil cell surface and promote ANCA–induced neutrophil extracellular trap formation. Case-by-case analysis indicates that patients with a high proportion of the Neu_T2ISG subset are associated with persistent vasculitis symptoms. A larger cohort analysis shows that serum IFN-γ levels at disease onset correlate with susceptibility to disease relapse. Our findings thus identify neutrophil diversity at the single cell level and implicate a biomarker for predicting relapse in small vessel vasculitis. Subject terms: Neutrophils, Vasculitis syndromes, Interferons, Systems analysis __________________________________________________________________ Neutrophils are early mediators for inflammation, but their functions in autoimmunity is still unclear. By using single cell analyses to compare patients with microscopic polyangiitis (MPA) and controls, here the authors find increased IFN-II-related neutrophils induced by IFN-γ and TNF signaling as a relapse risk marker for small vessel vasculitis. Introduction Neutrophils play crucial roles in the innate immune system by engulfing pathogens^[62]1, producing inflammatory cytokines^[63]2, and forming neutrophil extracellular traps (NETs)^[64]3,[65]4. Traditionally considered homogeneous and short-lived, advances in single-cell RNA sequencing (scRNA-seq) have revealed individual variability among these cells. Studies have shown that neutrophils from healthy individuals can be divided into three groups based on gene expression: immature neutrophils, mature neutrophils, and neutrophils characterized by interferon signature genes (ISGs)^[66]5. Granulocyte colony stimulating factor (G-CSF)-induced myelopoiesis leads to a marked increase in immature neutrophils and ISG-high neutrophil subsets^[67]6. In sepsis, the recruitment of IL1R2^+ immature neutrophils is associated with worse patient outcomes^[68]7. In addition, recent studies have highlighted the diversity of neutrophils in cancer, showing their pro-angiogenic^[69]8 or anti-tumor antigen-presenting^[70]9 functions. Despite these important findings, research has largely focused on understanding neutrophil homeostasis and responses to non-self factors, such as pathogens, drugs, or tumors. In the field of autoimmune rheumatic diseases, most single-cell analyses have been performed on peripheral blood mononuclear cells (PBMCs) and diseased tissues, leaving the diversity of neutrophils poorly understood. Furthermore, challenges remain in translating single-cell findings from patient samples into clinical practice and developing personalized treatment strategies. To explore neutrophil dynamics in autoimmune diseases and bridge the single-cell analysis to the bedside, this study considered three key factors: the appropriate disease target, the timing of sample collection, and the recruitment of treatment-naïve patients. First, anti-neutrophil cytoplasmic antibody (ANCA)-associated vasculitis is an appropriate research target due to the central role of abnormal neutrophil activation in its pathogenesis. Microscopic polyangiitis (MPA), a typical form of ANCA-associated vasculitis, is characterized by autoantibodies against myeloperoxidase (MPO). MPO-ANCA can directly activate neutrophils independently of other cellular interactions^[71]10,[72]11. This activation typically occurs when MPO and MPO-ANCA interact on the surface of primed neutrophils, leading to excessive activation through Fcγ receptors (FcγRs) and rapid induction of inflammation^[73]12,[74]13. Second, new-onset patients were recruited. When vasculitis becomes chronic, irreversible organ damage with fibrotic changes is established, and patients’ daily lives are significantly disrupted despite intensive immunosuppressive therapy. Focusing on the early stages of vasculitis is crucial for understanding autoimmune dysregulation in neutrophils and identifying specific subpopulations that contribute to pathogenesis. Finally, when targeting peripheral neutrophils for single-cell analyses, their sensitivity to treatment effects should be considered. Steroids, commonly used to treat autoimmune diseases, significantly affect neutrophil recruitment from the bone marrow^[75]14. In addition, immunosuppressive drugs often reduce the number and function of peripheral neutrophils^[76]15. Therefore, it is essential to analyze patient samples prior to therapeutic intervention to accurately capture neutrophil behavior. In this study, 179,664 peripheral white blood cells from patients with new-onset MPA and healthy donors are collected for single-cell transcriptome and surface proteome analyses. The results indicate a significant increase in a subset of neutrophils characterized by type II ISG expression in MPA. This unique subset is associated with susceptibility to NET formation and persistent vasculitis symptoms. Our work elucidates neutrophil dynamics at the single cell level and suggests a potential biomarker for predicting relapse in MPA. Results Abseq analysis of white blood cells from healthy donors and patients with newly diagnosed, treatment-naïve MPA Whole white blood cells were collected from six patients with MPA and seven healthy donors. All patients had new-onset disease, and blood samples were collected prior to the induction of immunosuppressive therapy. The clinical characteristics of each patient with MPA are provided in Supplementary Table [77]1. The isolated cells were analyzed on a BD Rhapsody® platform, and single-cell transcriptomes and expression of surface proteins were simultaneously obtained using Abseq (Fig. [78]1a). A total of 179,664 cells were collected for analysis, and cell populations were manually annotated on the basis of gene expression and surface protein profiles. The white blood cells were categorized into 15 populations, which included the following four granulocyte subsets: Neutrophil_1, Neutrophil_2, Eosinophil/Basophil, and Myelocyte. Uniform manifold and projection (UMAP) plots of white blood cells from healthy donors (n = 7) and patients with MPA (n = 6) are shown in Fig. [79]1b. Highly expressed genes and surface proteins in each cell population are shown in Fig. [80]1c and Fig. [81]1d, respectively. UMAP plots of white blood cells obtained from a pool of 13 samples (Supplementary Fig. [82]1a) and from individual participants (Supplementary Fig. [83]1b) are also shown. The ratio of each subset to the total number of white blood cells was calculated. The proportion of the Neutrophil_1 population, the predominant granulocyte subset, was significantly higher in patients with MPA, while the proportions of the Neutrophil_2 and Myelocyte populations were similar between the two groups. The proportion of the Eosinophil/Basophil population was significantly lower in patients with MPA (Fig. [84]1e). The proportions of most PBMC subsets were significantly lower in patients with MPA, while the proportions of the CD14^+ monocyte and Plasmablast/Plasmacell subsets were comparable between the two groups (Fig. [85]1e). To exclude the relative effect of increased granulocytes, the PBMC population was extracted, and the proportion of each subset within the total PBMC population was calculated. As expected, the proportions of the CD14^+ monocyte and Plasmablast/Plasmacell subsets were significantly higher in patients with MPA (Supplementary Fig. [86]2). Next, we conducted differential abundance analysis using Milo, a cluster-free and unbiased analysis that identifies differences between cell populations by assigning cells to partially overlapping neighborhoods on a k-NN graph^[87]16. As a result, the significant increase in cellular proportions (α < 0.05) in patients with MPA compared to healthy controls was detected in the Neutrophil_1 and CD14MONO subsets (Fig. [88]1f and Fig. [89]1g). Fig. 1. Comprehensive single-cell profiling of white blood cells from healthy donors and treatment-naïve patients with new-onset MPA. [90]Fig. 1 [91]Open in a new tab a Overview of the experimental workflow. MPA microscopic polyangiitis, HD healthy donors. b UMAP plots showing Abseq data of 105,258 white blood cells derived from healthy donors (HD, n = 7, left) and 74,406 white blood cells derived from patients with MPA (n = 6, right). Fifteen cellular clusters were identified. Mono; monocytes, Eosino/Baso; eosinophils/basophils, cDC; classical dendritic cells, pDC; plasmacytoid dendritic cells, T[CM]; central memory T cells, T[EM]; effector memory T cells, CTL; cytotoxic T lymphocytes, NK; natural killer cells, PBPC; plasmablasts/plasma cells. Balloon plot showing highly expressed genes c and surface proteins d in each cell population. Balloon color indicates the averaged scaled expression of the indicated genes/proteins. Balloon size indicates the percentage of cells expressing the indicated genes/proteins. e Percentage of each cell subpopulation relative to the total number of white blood cells derived from healthy donors (n = 7, blue dots) or patients with MPA (n = 6, red dots). Values are means with SEMs, and nominal P-values were calculated using the two-sided unpaired t-test. f Neighborhood graph of white blood cells using Milo differential abundance testing. Nodes represent neighborhoods from the white blood cell population. Colors indicate the log[2]-fold difference between healthy donors and patients with MPA for the nodes that represent significantly enriched neighborhoods (α < 0.05). Neighborhoods increased in patients with MPA are shown in red. Neighborhoods decreased in patients with MPA are shown in blue. g Beeswarm and box plots showing the distribution of log[2]-fold differences in neighborhoods between cell-type clusters. Colors are represented similarly to f. Box plots show median and interquartile range (IQR); the lower and upper hinges correspond to the first and third quartiles. The upper whisker extends from the hinge to the largest value no further than 1.5*IQR from the hinge. The lower whisker extends from the hinge to the smallest value at most 1.5*IQR from the hinge. All Neighbors outside the box range are plotted individually. The number of neighborhoods for each cell population is shown. We previously reported a unique CD14^+ monocyte population with an activated status related to the pathogenesis of MPA using Cellular Indexing of Transcriptomes and Epitopes by Sequencing (CITE-Seq)^[92]17. In the current Abseq analysis, monocytes were divided into five subsets: activated CD14^+ monocytes (CD14Mono_Activated), CD14^+ monocytes characterized by VCAN gene expression (CD14Mono_VCAN), CD14^+ monocytes characterized by HLA gene expression (CD14Mono_HLA), CD16^+ monocytes (CD16Mono), and classical dendritic cells (cDC) (Supplementary Fig. [93]3a). Highly expressed genes and surface proteins in each cell population are shown in Supplementary Fig. [94]3b and Supplementary Fig. [95]3c, respectively. The proportion of the CD14Mono_Activated subset was significantly higher in patients with MPA (Supplementary Fig. [96]3d). Differential abundance analysis showed that significantly upregulated neighbors (α < 0.05) in patients with MPA compared to healthy controls were particularly enriched in the CD14Mono_Activated subset (Supplementary Fig. [97]3e and [98]f). These findings are consistent with our previous observations^[99]17. Neutrophil subset diversity in MPA at the single-cell level To further elucidate the diversity of neutrophils, we divided the cells into seven subsets based on the RNA expression of known marker genes: myelocytes (Myelocyte), immature neutrophils characterized by PADI4 and MME gene expression (Neu_Immature), mature neutrophils characterized by intermediate gene expression (Neu_Mature), aged neutrophils characterized by CXCR4 and TNFAIP2 gene expression (Neu_Aged), neutrophils characterized by type I interferon signature gene expression (Neu_T1ISG), neutrophils characterized by type II interferon signature gene expression (Neu_T2ISG), and long-lived neutrophils (Neu_Longlived) (Fig. [100]2a). Specifically, the Neutrophil_1 subset shown in Fig. [101]1b corresponds to the Neu_Immature, Neu_Mature, Neu_Aged, Neu_T1ISG, and Neu_T2ISG subsets, while the Neutrophil_2 subset corresponds to the Neu_Longlived subset (Supplementary Fig. [102]4a). Highly expressed genes and representative surface proteins in each cell population are shown in Fig. [103]2b and [104]c, respectively. All surface proteins detected by Abseq analysis in each cell population are shown in Supplementary Fig. [105]4b. UMAP plots of neutrophils obtained from a pool of 13 samples (Supplementary Fig. [106]4c) and from individual study participants (Supplementary Fig. [107]4d) are also shown. Fig. 2. Altered composition and abundance of neutrophil subpopulations in MPA. [108]Fig. 2 [109]Open in a new tab a UMAP plots showing Abseq data of 65,301 neutrophils derived from healthy donors (HD, n = 7, left) and 59,021 neutrophils derived from patients with MPA (n = 6, right). Seven cellular clusters were identified: myelocytes (Myelocyte), immature neutrophils (Neu_Immature), mature neutrophils (Neu_Mature), aged neutrophils (Neu_Aged), neutrophils characterized by type I interferon signature gene expression (Neu_T1ISG), neutrophils characterized by type II interferon signature gene expression (Neu_T2ISG), and long-lived neutrophils (Neu_Longlived). Balloon plot showing highly expressed genes b and surface proteins c in each cell population. Balloon color indicates the averaged scaled expression of the indicated genes/proteins. Balloon size indicates the percentage of cells expressing the indicated genes/proteins. d Percentage of each cellular subpopulation relative to the total number of neutrophils derived from healthy donors (n = 7, blue dots) or patients with MPA (n = 6, red dots). Values are means with SEMs, and nominal P-values were calculated using the two-sided unpaired t-test. e Neighborhood graph of neutrophils using Milo differential abundance testing. Nodes represent neighborhoods from the neutrophil population. Colors indicate in a similar fashion in Fig. [110]1f. f Beeswarm and box plots of neutrophils based on Milo differential abundance testing. Colors are represented similarly to e. Box plots show median and interquartile range (IQR); the lower and upper hinges correspond to the first and third quartiles. The upper whisker extends from the hinge to the largest value no further than 1.5*IQR from the hinge. The lower whisker extends from the hinge to the smallest value at most 1.5*IQR from the hinge. All Neighbors outside the box range are plotted individually. The number of neighborhoods for each cell population is shown. The similarities between each subset and previously reported neutrophil groups were confirmed by module scoring using scRNA-seq data. The gene expression profile of the Myelocyte subset and some parts of the Neu_Immature subset aligned with that of low-density neutrophils (LDN), which are typically found in inflammatory and cancer pathologies^[111]6 (Supplementary Fig. [112]5a). Normal-density neutrophils include both steady-state neutrophils and those activated by cytokine stimulation^[113]18. Module scoring showed that neutrophils activated by interferon (IFN)-α stimulation^[114]6 aligned with the Neu_T1ISG subset, while those activated by IFN-γ stimulation^[115]6 aligned with the Neu_T2ISG subset (Supplementary Fig. [116]5a). Neutrophils that survive long-term in the presence of proinflammatory cytokines^[117]19 showed a high module score concordance with the Neu_Longlived subset (Supplementary Fig. [118]5a). Additional module score analysis revealed that neutrophils from patients treated with G-CSF or those with pancreatic invasive ductal adenocarcinoma gave rise to various subsets of mature neutrophils, including those resembling the Neu_T1ISG and Neu_T2ISG subsets (Supplementary Fig. [119]5b). The characteristics of neutrophils from hematopoietic stem cell transplantation patients closely resembled those of the Neu_T1ISG and Neu_T2ISG subsets, suggesting that neutrophils under stress-induced activation and those from MPA patients share commonalities, particularly in the interferon response of their mature subsets (Supplementary Fig. [120]5b). Next, the ratio of each subset to the total number of neutrophils was calculated. The proportions of the Neu_Immature and Neu_T2ISG subsets were significantly increased in patients with MPA, whereas the proportion of the Neu_mature subset was decreased (Fig. [121]2d). Differential abundance analysis using Milo confirmed that the Neu_T2ISG subset showed the highest compositional change of each neighborhood between patients with MPA and healthy donors (median log[2] fold change: 1.70) (Fig. [122]2e and [123]f). Differential gene expression and functional characteristics of neutrophil subsets associated with MPA pathogenesis Out of a total of 2846 genes detected in more than 10% of all neutrophils, 59 genes were identified as differentially expressed genes (DEGs) in comparisons between neutrophils from MPA patients and healthy donors, including 28 genes associated with IFN-γ signaling (Fig. [124]3a). The lists of DEGs (average log2 fold change > 0.25) and interferon response genes are shown in Supplementary Table [125]2a and [126]b. In addition, focusing on the Neu_Mature subset, which constitutes the majority of neutrophils, 22 IFN-γ response genes were still identified as DEGs characteristic of MPA patients (Supplementary Table [127]2c). These data suggest that the upregulation of IFN-γ response genes is broadly distributed across neutrophil subsets, representing a key feature of MPA patients. Fig. 3. Functional characteristics of neutrophil subpopulations in MPA. [128]Fig. 3 [129]Open in a new tab a Venn diagram showing differentially expressed genes in comparisons between neutrophils from MPA patients and healthy donors. Out of 2846 expressed genes (with expression rate > 0.1), 59 genes were upregulated (average log2 fold change > 0.25) in MPA patients including 28 genes associated with IFN-γ signaling. Pathway enrichment analysis of genes expressed in the Neu_Immature b and Neu_T2ISG c subset, performed using Enrichr. P-values for each pathway were calculated using the Benjamini–Hochberg method. d Heatmap showing the gene expression enriched in pathways related to neutrophil functions. In each neutrophil subset, the averaged gene expression in healthy donors (HD) and patients with MPA (n = 6) is shown. e Feature plot showing the expression of representative genes related to type II interferon signature genes (GBP1/GBP5), degranulation (MPO/MMP9), neutrophil extracellular trap (NET) formation (PADI4/ITGAM), and Fc-gamma receptors (FCGR1A/FCGR2A/FCGR3A/FCGR3B). Genes detected as differentially expressed genes (DEGs) in comparison of neutrophils from MPA patients to those from healthy donors (average log2 fold change > 0.25) are shown in red. To explore the pathogenic properties of the Neu_T2ISG subset, we performed pathway analysis of DEGs comparing this subset with total neutrophils. IFN signaling, cytokine signaling in the immune system, and IFN-γ signaling pathways were enriched in the Neu_T2ISG subset (Fig. [130]3b), whereas the neutrophil degranulation pathway was enriched in the Neu_Immature subset (Fig. [131]3c). Next, we examined gene expression in each neutrophil subset related to functional pathways that contribute to the pathogenesis of MPA. Existing databases were used to assess the genes related to the IFN-γ signature^[132]20, degranulation^[133]20, NET formation^[134]21, and FcγR expression^[135]22. As expected, IFN-γ signature genes were enriched in the Neu_T2ISG subset (Fig. [136]3d). GBP1 and GBP5, which are representative IFN-γ response genes, were highly distributed in the Neu_T2ISG subset (Fig. [137]3e). Genes related to degranulation and NET formation were particularly enriched in the Myelocyte subset and moderately enriched in the Neu_Immature subset, with few enrichment differences between groups (Fig. [138]3d). Consistently, the genes related to early neutrophil activation, MPO and MMP9, and the NET-related genes, PADI4 and ITGAM, were mainly distributed in the Myelocyte and Neu_Immature subsets (Fig. [139]3e). Genes related to FcγR expression were broadly expressed in most neutrophil subsets (Fig. [140]3d). Specifically, FCGR1A, a member of the IFN-γ response genes, was highly distributed in Neu_T2ISG and was detected as a DEG in patients with MPA. FCGR2A, FCGR3A, and FCGR3B were broadly expressed in all neutrophils and were not detected as DEGs (Fig. [141]3e). Thus, the Neu_Immature subset exhibited moderate enrichment of degranulation- and NET-related genes, yet these genes were not identified as DEGs, suggesting that simply increasing the number of Neu_Immature is characteristic of MPA. On the other hand, the Neu_T2ISG subset is characterized by the IFN-γ response genes, many of which were identified as DEGs. These data suggest that each neutrophil subpopulation has distinct characteristics and plays a role in the pathogenesis of MPA. Gene expression profiles and cellular interactions characterizing the Neu_T2ISG subset We next investigated the lineage of differentiation into Neu_T2ISG using pseudo-time trajectory mapping. The root node was set to Neu_Immature. We found that the Neu_T2ISG subset deviated from the Neu_Mature subset in the developmental trajectory (Fig. [142]4a and [143]b). Furthermore, DEGs in the Neu_T2ISG subset showed the highest expression at the end of the trajectory, indicating that this was a terminally differentiated subset (Fig. [144]4c). Since we conducted a single cell analysis of all peripheral white blood cells in this study, we next aimed to identify cellular interactions that influence the emergence of the Neu_T2ISG subset. Cell–cell interaction analysis was conducted with each white blood cell subset as the signal sender and the Neu_T2ISG subset as the signal receiver. The analysis revealed that GBP family genes, which are representative of the Neu_T2ISG subset, were upregulated by IFN-γ from CD8^+ T cells. In addition, TNF from CD14^+ monocytes also broadly affected Neu_T2ISG subset gene expression (Fig. [145]4d). The Neu_Immature subset was strongly influenced by IL-15 and CSF1 from neutrophils and by TGFB1 from monocytes (Supplementary Fig. [146]6). Fig. 4. Neutrophil dynamics and cellular Interactions in peripheral white blood cells. [147]Fig. 4 [148]Open in a new tab a Trajectory analysis showing the possible cell lineages of neutrophils. Black lines projected on the UMAP indicate the trajectory from the Neu_Immature subset. b UMAP plot marked with the inferred pseudotime of each cell calculated by Monocle3. Black lines projected on the UMAP indicate the trajectory from the Neu_Immature subset. The convergent differentiation of each subset over pseudotime is shown. c Heatmap showing the expression changes of highly expressed genes in each population over pseudotime. The convergent differentiation of each subset is tracked. d Circos plot showing cell–cell interactions between whole white blood cells and Neu_T2ISG neutrophils. Each cluster of white blood cells were set as the “sender,” the Neu_T2ISG subset was set as the “receiver”. For the receiver cell population, differentially expressed gene (DEG) analysis was performed to identify highly expressed genes. e Flow cytometry of Fcγ receptors (FcγRs) 1, 2, 3, and MPO cell surface expression by peripheral neutrophils. The expression of each cell surface molecule was assessed immediately after sample collection (Pre-culture: light gray–filled histogram). Subsequently, cells were incubated for 16 hours with standard media (Untreated: dark gray–filled histogram), 100 U/mL IFN-γ (IFN-γ Treated; red-filled histogram), 10 ng/mL TNF (TNF treated; blue-filled histogram), or a combination of these cytokines (IFN-γ + TNF treated; purple-filled histogram). Single-cell analysis revealed that the Neu_T2ISG was characterized by IFN-γ response genes including FCGR1A (Fig. [149]3e), and this subset developed under the influence of IFN-γ and TNF (Fig. [150]4d). Therefore, we next investigated the effects of IFN-γ and TNF on neutrophil cell surface molecules. Freshly isolated neutrophils from healthy donors were stimulated with IFN-γ and TNF and changes in FcγR cell surface expression were analyzed. As expected, significant upregulation of cell surface FcγR1 was observed in neutrophils after IFN-γ stimulation. In addition, IFN-γ stimulation maintained the cell surface expression of FcγR2 and FcγR3, which are normally observed to decrease over time in the absence of stimulation (Fig. [151]4e). In contrast, cell surface MPO molecules were not induced by IFN-γ priming but were induced by TNF priming. Consistent with these results, the combination of IFN-γ and TNF priming led to the simultaneous exposure of MPO and FcγRs on the neutrophil cell surface (Fig. [152]4e). These data suggest that the Neu_T2ISG subset represents a terminally differentiated population characterized by surface exposure of pathogenic molecules, potentially driving neutrophil-mediated inflammation in MPA. Synergistic effects of IFN-γ and TNF priming on neutrophil activation by MPO-ANCA It is well known that MPO-ANCA binds to MPO expressed on the cell surface of primed neutrophils and induces reactive oxygen species (ROS) production and NET formation through cross-linking of FcγRs^[153]12,[154]13. Therefore, we assessed MPO-ANCA–induced ROS production and NET formation after priming neutrophils with IFN-γ and TNF. Both IFN-γ and TNF priming significantly prolonged neutrophil survival (Fig. [155]5a). Upon subsequent stimulation with MPO-ANCA, TNF priming could enhance ROS production and NET formation, which was further enhanced by concurrent IFN-γ priming (Fig. [156]5b–[157]d). Next, we purified serum IgG from MPA patients (patient IgG) and evaluated NET formation following stimulation with these immunoglobulins. As expected, priming with both IFN-γ and TNF induced significantly enhanced NET formation in response to patient IgG compared to priming with TNF alone (Fig. [158]5e). Fig. 5. Functional impact of IFN-γ and TNF priming on neutrophil activation in vitro. [159]Fig. 5 [160]Open in a new tab a Percentage of viable neutrophils after priming with IFN-γ, TNF, or their combination. Values are means with SEMs, and nominal P-values were calculated using the two-sided unpaired t-test. b Reactive oxygen species (ROS) production of neutrophils isolated from healthy donors. TNF–primed cells were treated with or without IFN-γ for 16 hours, followed by stimulation with anti-MPO antibody (10 µg/mL) for 1 hour. ROS levels were monitored every minute for a total of 60 cycles. Representative traces show the means with SEMs from a triplicate experiment. P-value was calculated using the two-sided Wilcoxon matched-pairs signed rank test. c SYTOX green immunofluorescence analysis to detect neutrophil extracellular trap (NET) formation. TNF–primed cells were treated with or without IFN-γ for 16 hours, followed by stimulation with anti-MPO antibody (10 µg/mL) for 4 hours to induce NET formation. Scale bar: 100 µm. The area of SYTOX Green–positive cells per field (green area) was calculated against the area of DAPI-positive neutrophils (blue area) in the same fields for quantitative analysis of NET formation induced by anti-MPO antibody d and purified IgG from MPA patients e. Five randomly selected fields, 400–500 neutrophils per field were counted and averaged in each experiment. Values are means with SEMs, and P-values were calculated using the two-sided unpaired t-test d and e. f UMAP plots showing scRNA-seq data of neutrophils isolated from healthy donors and stimulated with IFN-γ + TNF, TNF alone, IFN-γ alone, or left unstimulated. g Balloon plot showing highly expressed genes in each cell cluster. Balloon color indicates the averaged scaled expression of the indicated genes. Balloon size indicates the percentage of cells expressing the indicated genes. h Sankey plot showing the correspondence between the stimulation condition and each cell cluster. i Module scoring analysis of each cell cluster. Module scores were calculated using the highly expressed genes in each cluster. Each cell was colored according to the scaled module score. The cells with high module score are indicated by dashed circle. To further investigate the characteristics of the TNF and IFN-γ-primed neutrophil subset, we conducted additional scRNA-seq analyses on in vitro stimulated neutrophils (Fig. [161]5f and [162]g). Stimulation with a combination of TNF and IFN-γ induced a uniform subset (Cluster1), resembling Neu_T2ISG (Fig. [163]5h and [164]i). In contrast, in vitro stimulation with TNF or IFN-γ alone gave rise to subsets with the distinct characteristics, which resembling Neu_Longlived (Cluster2) or Neu_T1ISG/Neu_T2ISG mixed (Cluster3) (Fig. [165]5h and [166]i). These findings suggest that priming with a combination of TNF and IFN-γ induces a uniquely primed neutrophil subset, which is highly sensitive to MPO-ANCA stimulation and readily triggers ROS production and NET formation. Clinical impact of IFN-γ on neutrophils in the pathogenesis of MPA To detect case-specific differences in IFN-γ response gene expression between patients with MPA, a heat map was created to show the individual expression patterns of the DEGs identified in MPA patients (Fig. [167]6a). Based on 59 DEGs (Supplementary Table [168]2a), genes with similar expression variation were clustered hierarchically. Gene cluster 1 was predominantly composed of type I interferon signature genes. Genes associated with granulopoiesis, including S100 family genes and NAIP^[169]7, were included in gene cluster 2. The characteristic genes in the Neu_T2ISG subset, including GBP1, GBP5, and FCGR1A, were included in gene cluster 3. As shown in Fig. [170]6a, three out of six patients, specifically MPA-4, MPA-5, and MPA-6, exhibited significantly higher expression of gene clusters 2 and 3, suggesting an enriched phenotype with myelopoiesis and IFN-γ response genes. Consistently, these three patients exhibited an elevated proportion of the Neu_T2ISG subset (MPA-4: 15.4%, MPA-5: 11.1%, MPA-6: 16.4%) in comparison with the other cases (MPA-1: 6.7%, MPA-2: 8.8%, MPA-3: 9.6%). When examining only the expression of type II ISGs, the differences between these three patients (MPA-4, MPA-5, and MPA-6) and the remaining patients were more evident in heat maps (Supplementary Fig. [171]7a). The expression distribution of genes characteristic of the Neu_Immature subset did not differ among cases (Supplementary Fig. [172]7b). Fig. 6. Clinical implications of IFN-γ on aberrant neutrophil priming. [173]Fig. 6 [174]Open in a new tab a Case-sensitive heatmap showing the scaled expression of 59 differentially expressed genes. Gene clusters 1, 2, and 3 were generated using a hierarchical clustering method. The HD (healthy donor) column shows the expression in seven study participants. In the MPA group, each column represents one patient with MPA (n = 6). Disease duration indicates the number of months from the onset of the first vasculitis symptom to the time of sample collection. Treatment response indicates BVAS-based disease status 6 months after treatment induction. b Correlations between serum IFN-γ concentrations and representative clinical parameters. Correlation coefficients (r; Spearman’s rho) and associated P-values were calculated using the two-sided Spearman’s rank correlation test. MPO; myeloperoxidase, ANCA; anti-neutrophil cytoplasmic antibody, CRP; C-reactive protein, Alb; albumin, Hb; hemoglobin. c Serum IFN-γ concentrations in healthy donors (HD, n = 7), patients with new-onset MPA (n = 24), and in under-treated patients (n = 13). d Serum IFN-γ concentrations of new-onset patients who relapsed (n = 9) or not relapsed (n = 15) after treatment. e Receiver Operating Characteristic (ROC) curve for predicting relapse of MPA from serum IFN-γ, MPO-ANCA and CRP levels. Values are means with SEMs, and P-values are calculated using a two-sided Unpaired t-test for c and d. From a clinical point of view, MPA-4, MPA-5, and MPA-6 had a long duration between the onset of vasculitis symptoms and initiation of treatment (5 months, 7 months, and 13 months, respectively). They also exhibited persistent vasculitis symptoms (defined by BVAS > 0) following treatment (Fig. [175]6a). These data indicate that the neutrophil subset characterized by IFN-γ response genes is implicated in refractory vasculitis. Therefore, we expanded the study sample size using stored sera and measured serum IFN-γ concentrations in 37 MPA patients: 24 new-onset patients and 13 under treatment (Supplementary Table. [176]3). Serum IFN-γ levels showed significant positive correlations with MPO-ANCA levels and C-reactive protein (CRP) levels, and negative correlations with titers of hemoglobin and albumin (Fig. [177]6b). Serum IFN-γ concentrations were significantly higher in new-onset patients compared to those under treatment (Fig. [178]6c). In particular, those concentrations at the onset of vasculitis were significantly higher in patients who subsequently experienced relapse during the clinical course; in this study, among 24 new-onset patients assessed, the top six patients with the highest serum IFN-γ concentrations all experienced relapses (Fig. [179]6d). Finally, to provide prognostic insights from our cohort, we constructed a receiver operating characteristic (ROC) curve for predicting relapse from serum concentrations of IFN-γ, MPO-ANCA and CRP in newly diagnosed MPA patients. This ROC curve can predict the risk of relapse before initiation of immunosuppressive treatment with a sensitivity of 93% and a specificity of 78% (Fig. [180]6e). These data indicate that Neu_T2ISG is a characteristic subset in patients with longer disease duration and treatment-resistant symptoms. More than just an inflammatory indicator, high serum IFN-γ levels at the onset of MPA represent the risk of relapse. Taken together, high-resolution single-cell analysis of neutrophils from patients with new-onset MPA revealed that Neu_T2ISG represents a uniquely primed subset. Neu_T2ISG emerges upon stimulation with IFN-γ and TNF, which enhances MPO-ANCA–induced NET formation beyond TNF priming alone through increased and sustained cell surface FcγR expression. Clinically, serum IFN-γ levels serve as a biomarker to predict disease relapse (Fig. [181]7). Fig. 7. Graphical scheme of this study. [182]Fig. 7 [183]Open in a new tab Single-cell transcriptome and surface proteome analyses using Abseq on 179,664 peripheral blood cells from treatment-naïve patients with new-onset MPA (n = 6) and healthy donors (n = 7). MPA is characterized by increased proportions of neutrophils, activated CD14^+ monocytes, and plasmablasts/plasma cells. Neutrophils are classified into seven subsets, two of which are significantly increased in MPA patients: immature neutrophils (Neu_Immature) and neutrophils characterized by type II interferon signature genes (Neu_T2ISG). The Neu_T2ISG subset differentiates from the Neu_Mature subset upon stimulation with IFN-γ and TNF. The synergistic effect of IFN-γ and TNF enhances surface MPO and FcγR expression in neutrophils and promotes NET formation upon ANCA stimulation. High serum IFN-γ levels at the onset of MPA indicate susceptibility to disease relapse. MPA microscopic polyangiitis, MPO myeloperoxidase, FcγR Fcγ receptor, ANCA anti-neutrophil cytoplasmic antibody, NET neutrophil extracellular trap. Discussion Neutrophil priming is a critical process that enhances the functional capabilities, allowing them to respond more effectively to pathogens and other stimuli^[184]23,[185]24. However, dysregulation of neutrophil priming can play an important role in the pathogenesis of autoimmune diseases: ANCA-associated vasculitis is a clear example^[186]25–[187]27. Neutrophil priming is also involved in the pathogenesis in other autoimmune diseases. In rheumatoid arthritis (RA), primed neutrophils migrate into the synovial fluid and contribute to joint inflammation and destruction^[188]28. Similarly, in systemic lupus erythematosus (SLE), primed neutrophils can form NETs that expose autoantigens and accerate autoimmunity^[189]29. Distinguishing various neutrophil states using conventional surface marker phenotyping has been challenging. In this study, single-cell transcriptome and surface antigen analyses of neutrophils from treatment-naïve patients with new-onset MPA revealed seven subtypes of neutrophils. High expression levels of CD15 and HLA-DR were observed in the Myelocyte subset, while CD62L was highly expressed in the Neu_Immature subset, suggesting these markers may be indicative of these specific subsets. However, no specific surface markers were identified to definitively isolate the remaining neutrophil subsets. From a transcriptomic perspective, a characteristic feature of neutrophils from MPA patients is the increased proportion of immature neutrophils (Neu_Immature) and a uniquely primed neutrophil subset, neutrophils characterized by type II ISGs (Neu_T2ISG). The Neu_Immature subset contains low-density neutrophils, which is typically observed in acute inflammatory conditions and in autoimmune diseases such as RA^[190]30, SLE^[191]31, and ANCA-associated vasculitis^[192]32–[193]34. Therefore, an increase in the Neu_Immature subset is expected given that MPA is characterized by inflammation with abnormal neutrophil activation. Of note, genes involved in degranulation or NET formation were not identified as DEGs when comparing MPA patients with healthy donors. These data suggest that the numerical increase of Neu_Immature cells, rather than their enhanced function in inducing degranulation or NETs, may be involved in MPA pathogenesis. The Neu_T2ISG subset is mainly characterized by IFN-γ response genes. A similar neutrophil subpopulation has been reported in patients following hematopoietic stem cell transplantation and those receiving G-CSF therapy^[194]6, suggesting a potential link with myelopoiesis. Our findings consistently demonstrated elevated expression of myelopoiesis-related genes, including S100 family genes and NAIP, in patients enriched for the Neu_T2ISG subset. ANCA-associated vasculitis is known to be triggered by preceding infections^[195]25. Given that patients with a longer duration after the onset of vasculitis symptoms have a higher proportion of the Neu_T2ISG subset, chronic exposure to pathogens may induce continuous myelopoiesis and contribute to an increase in longer-lived subsets such as Neu_T2ISG. Interestingly, IFN-γ, which normally acts antagonistically to myelopoiesis^[196]35, may lead to the development of pathogenic neutrophil subsets in MPA. In regard to type I IFN, we observed no significant differences in the Neu_T1ISG population between MPA patients and healthy donors. This finding is consistent with previous reports indicating that systemic type I IFN responses are not major drivers of AAV pathogenesis^[197]36. However, our previous study identified a subset of monocytes in MPA patients with characteristics of type I IFN activation^[198]17. Similarly, in the current study, a few patients (e.g., MPA-3 and MPA-4) exhibited mildly elevated type I IFN signaling in neutrophils. These findings indicate that while type I IFN signaling in neutrophils may not be a predominant factor, it could still play a role in disease pathology in a specific subset of MPA patients. Trajectory analysis revealed interesting details of the Neu_T2ISG subset during neutrophil differentiation. As neutrophils differentiate from immature to mature, then aged, and finally long-lived^[199]19, the Neu_T2ISG subset branches off from the Neu_Mature subset. A previous study showed that in an analysis of bulk cell populations, a neutrophil phenotype enriched in the inflamed joints of RA patients was characterized by the aged and IFN-γ signatures^[200]37. However, a key marker of the aged neutrophil, CXCR4, is partially suppressed by IFN-γ. Considering this discrepancy, both aged and IFN-γ signatures may not coexist in a single neutrophil population. Our single-cell trajectory analysis clearly showed that upon stimulation with IFN-γ, mature neutrophils deviate from the typical trajectory toward the aged phenotype, resulting in their presence as an independent Neu_T2ISG population. Cell–cell interaction analysis further suggested that the formation of the Neu_T2ISG subset involves IFN-γ from CD8^+ T cells and TNF from CD14^+ monocytes; importantly, IFN-γ elicits a variety of responses from neutrophils such as increased ROS production and induction of antigen presentation^[201]38, and TNF has long been recognized for its role in neutrophil priming^[202]39. In this study, the synergy between these factors enhances surface FcγR expression on neutrophils, prolongs cell survival, and promotes NET formation upon ANCA stimulation. An increased population of this uniquely primed neutrophil subset may contribute to the pathogenesis. Furthermore, we found serum IFN-γ levels at the onset of MPA reflect susceptibility to disease relapse. While MPA has a high relapse rate, there are no good biomarkers to predict relapse^[203]40; measurement of IFN-γ may be a strategy to determine the intensity of treatment for each case. Considering that an excessive increase in serum IFN-γ can trigger an over-primed neutrophil state by inducing cell surface MPO and FcγR expression, appropriate modulation of IFN-γ may be a novel strategy for the treatment of MPA. This study has several limitations. First, all patients recruited were Japanese, and potential racial differences were not evaluated. Second, the neutrophil profiles of other disease control groups were not analyzed, leaving it unclear whether the unique neutrophil profile identified in this study is specific to MPA. Importantly, a previous study demonstrated that a Th1 cytokine pattern, particularly IFN-γ expression, predominates in granulomatosis with polyangiitis (GPA)^[204]41. Although this study focuses on MPA and does not include GPA patients, the IFN-γ signaling pathway may represent a shared feature in the pathophysiology of other forms of ANCA-associated vasculitis. In conclusion, single-cell analysis identified that two distinct neutrophil populations are involved in MPA pathogenesis. A numerical increase in the Neu_Immature subset may reflect the inflammatory status, whereas a uniquely primed Neu_T2ISG subset induces excessive neutrophil priming and NET formation upon ANCA stimulation. Patients with high numbers of Neu_T2ISG are associated with persistent vasculitis symptoms and elevated serum IFN-γ levels can predict disease relapse. Our findings highlight the diversity of neutrophils at the single cell level and provide insights into a novel biomarker for ANCA-associated vasculitis. Future comprehensive time series and inflamed tissue analyses are expected to elucidate the perturbation of these pathogenic neutrophil subsets. Methods Study participants This study was conducted in accordance with the Declaration of Helsinki and with approval from the Ethical Review Board Osaka University Hospital (No. 855). Samples were obtained after study participants provided written informed consent. Consent was also obtained to publish information such as age, sex, the name of the medical center, and the diagnosis. The sex of each study participant was determined based on self-report. Study participants were not compensated. Patients with ANCA-associated vasculitis were diagnosed as having MPA according to the 2022 American College of Rheumatology (ACR) / European Alliance of Associations for Rheumatology (EULAR) classification criteria^[205]42,[206]43. The diagnosis was verified by at least two rheumatologists. Definition of disease activity status The Birmingham Vasculitis Activity Score (BVAS) 2008 version 3 was used to rate MPA disease activity. The definitions for “New”, “Remission”, “Worse”, and “Persistent” were based on established BVAS criteria; “New” was defined as the appearance of newly developed disease manifestations in BVAS. “Remission” was defined as a BVAS score of 0. “Worse” was defined as a worsening of existing disease activity within 4 weeks, accompanied by an increase in the BVAS score. “Persistent” was defined as a state in which the BVAS score remained ≥ 1 for 4 weeks or longer, with no changes in pre-existing symptoms or signs and no evidence of new disease activity. In this study, “Relapse” was defined as a transition from a “remission” or “persistent” state to a “worse” state. The clinical status of each patient was recorded, and the time to relapse was calculated from the initiation of induction therapy, with December 31, 2023, serving as the study endpoint. Patient profiles Six patients with MPA (three females; average age, 74 years) and seven healthy donors (four females; average age, 67 years) were recruited for scRNA-seq experiments. All patients with MPA had new-onset disease and had not received any immunosuppressive therapy. All patients were admitted to Osaka University Hospital and underwent a comprehensive assessment to rule out potential vasculitis mimics, including infectious diseases and neoplastic lesions, before applying the 2022 ACR/EULAR MPA classification criteria. Patient MPA1 presented with retinal vasculitis via fundoscopic examination. Patient MPA2 presented with systemic symptoms and pauci-immune glomerulonephritis. Patients MPA3 and MPA5 presented with multiple mononeuropathy. Patient MPA4 presented with systemic symptoms and extensive pachymeningitis. Patient MPA6 presented with systemic symptoms, progressive interstitial pneumonia, and aseptic recurrent bilateral otitis media. We previously recruited patients MPA1, MPA3, MPA4, and MPA6 for CITE-Seq analysis on PBMCs^[207]17. For this study, the blood samples were processed on a different platform with a suitable pipeline for granulocytes. Thirty-seven patients with MPA (20 females; average age, 73 years) were additionally recruited to evaluate clinical and laboratory parameters. Serum and blood cells preparation Whole blood (3.5 mL) was collected in Vacutainer SST II tubes (BD Diagnostics, Cat. No. 365920). Tubes were centrifuged for 10 minutes at 1200 × g. The resultant supernatant was collected as serum and stored at −80 °C. For white blood cell collection, whole blood (2 mL) was collected in a Na-EDTA blood collection tube (Terumo, Cat. No. VP-NA052K). White blood cells were separated using Polymorphprep (Serumwerk, Cat. No. 1895). Single-cell library construction White blood cells were attached with DNA-barcoded antibodies for antibody sequencing using AbSeq Immune Discovery Panel (BD, Cat. No. 625970). Information on the antibodies used for antibody sequencing is shown in Supplementary Table [208]4. Single-cell suspensions were processed using the BD Rhapsody Express System (BD Biosciences). The libraries were constructed with WTA Reagent Kits (BD Biosciences, Cat. No. 665915). Briefly, up to 10000 labeled live cells per sample were separately loaded into the BD Rhapsody platform without sample mixing to create a barcoded cDNA library for individual cells. Data quality control was performed using the Bioanalyzer system (Agilent). Individual libraries were pooled for sequencing on the HiSeq 2500 or Novaseq 6000 platform (Illumina) to measure expression of genes and surface proteins from over 10000 white blood cells. Sequence information is summarized in Supplementary Table [209]5. Analysis of Abseq data Raw FASTQ files were matched to the GRCh38 reference genome using the BD Rhapsody Sequence Analysis Pipeline (version 1.12). Molecule counts were adjusted using distribution-based error correction (DBEC) and presented as expression data table files to establish a Seurat object. The Seurat R package (version 4.2.0) was used for data quality control, scaling, transformation, clustering, dimensionality reduction, differential expression analysis, and visualization. A total of 179,664 cells out of 199,050 putative cells were selected for further analysis using unique molecular identifiers. Low-quality cells with nFeature_RNA values below 200 were removed. Data were normalized and scaled using the SCTransform function. Data from all samples were integrated utilizing reciprocal PCA. The RunUMAP function was used for UMAP dimensional reduction with 25 precomputed principal component analysis (PCA) dimensions. A nearest-neighbor graph using the 25 dimensions of the PCA reduction was computed using the FindNeighbors function, followed by clustering using the FindClusters function. The newly generated UMAP was visualized using the DimPlot function. Each cluster was manually annotated using gene expression and protein data. Platelets and erythrocytes were removed from the analysis, and doublets were manually removed using cell-surface protein data. Differential abundance analysis using scRNA-seq data Differential abundance analysis of patients with MPA and healthy donors was performed using miloR (version 3.15) to detect sets of cells that were differentially abundant in the two groups by modeling counts of cells in the neighborhoods of a k-nearest neighbor (KNN) graph^[210]16. The buildGraph function was first used to construct a KNN graph on the basis of precomputed supervised PCA with k = 10, using 30 principal components (d = 30). Next, the makeNhoods function was used to assign cells to neighborhoods on the basis of their connectivity over the KNN graph. For computational efficiency, 10% of cells were subsampled for white blood cells, neutrophils or monocytes. To test for differential abundance, Milo was used to fit a negative binomial generalized linear model to the counts for each neighborhood, accounting for different numbers of cells across samples using Trimmed Mean of M-values (TMM) normalization. Age was included as a covariate in the testNhoods function. The log[2] fold change of the number of cells between two conditions in each neighborhood was used for visualization. Pathway analysis Gene set enrichment analysis for highly expressed genes in the Neu_immature and Neu_T2ISG subsets was performed using Enrichr^[211]44. Reactome 2022^[212]18 was used as the dataset, and adjusted P-values for the pathway were calculated by the Benjamini–Hochberg method. Module scoring analysis Gene scores were visualized using the Featureplot function of Seurat on the basis of cell-based scores, which were calculated using the AddModuleScore function. The LDN signature score was calculated on the basis of genes highly expressed in LDNs isolated from peripheral blood^[213]6. Type1-ISG and Type2-ISG signature scores were calculated on the basis of genes upregulated in developing neutrophils treated ex vivo with IFN-β and IFN-γ, respectively^[214]6. The long-lived signature score was calculated on the basis of genes upregulated after 48-h culture with GM-CSF, TNF, and IL-4 priming^[215]19. Pseudotime analysis Pseudotime analysis of patients with MPA was performed using Monocle3 (version 1.0.0)^[216]45 to reconstruct possible neutrophil cell lineages. The previously annotated Seurat object was imported into Monocle3. The learn_cells function was used to fit a principal graph and the graph was plotted using the UMAP coordinates. Then the Order_cells function was used to calculate the pseudotime of each cell. The Neu_Immature subset was identified as the root cell state. Expression changes of highly expressed genes in each population over pseudotime are shown in heatmaps using the Heatmap function. Cell–cell interaction analysis Cell–cell interaction analysis of patients with MPA was performed using Nichenet (version 1.1.1)^[217]46 to predict active cell–cell interaction with the Neu_Immature or Neu_T2ISG populations. Each cluster from the Seurat object of white blood cells was set as the “sender,” and the Neu_Immature subset or Neu_T2ISG subset was set as the “receiver.” For the receiver cell population, a DEG analysis was performed to identify genes highly expressed in the Neu_Immature or Neu_T2ISG populations. Then the predict_ligand_activities function was used to infer active ligand–target links. Circos plots were generated to visualize links between ligands on white blood cells and receptors on neutrophil clusters. scRNA-seq and flow cytometry of in vitro stimulated neutrophils Whole blood (5 mL) from healthy donors was collected into a Na-EDTA blood collection tube. Neutrophils were separated using Polymorphprep and used for cell culture and fluorescence-activated cell sorting analysis. Neutrophils (1 ×10^3 cells/well) were cultured in 96-well plates (Thermo Fisher Scientific, cat. No. 167008) in RPMI-1640 with or without 100 U/mL IFN-γ (R&D, Cat. No. 10291-TA) and 5 ng/mL TNF (R&D, Cat. No. 285-IF/CF) supplementation for 16 h in a CO[2] incubator at 37 °C. Stimulated neutrophils were stained with Single-Cell Multiplexing Kit (BD Biosciences, Cat. No. 633781), then mixed and subjected to single-cell analysis. Remaining cells were washed twice using phosphate-buffered saline and incubated for 30 min at 4 °C with an antibody cocktail in stain buffer with Live/Dead (BioLegend, Cat. No.77184), anti-CD15 antibody (BD Bioscience, Cat. No. 562371), anti-MPO antibody (Dako, Cat. No. F0714), anti-FcγR1 antibody (BD Bioscience, Cat. No. 561188), anti-FcγR2 antibody (BD Bioscience, Cat. No. 755300), and anti-FcγR3 antibody (BD Bioscience, Cat. No. 563785). For analysis of surface proteins, gating of singlet, live, CD15^+ neutrophils was performed on FACSAria as described in Supplementary Fig. [218]8. The dilution ratios of the antibodies used are shown in Supplementary Table [219]6. Detection and quantitation of reactive oxygen species (ROS) and neutrophil extracellular traps (NETs) Neutrophils were seeded onto 96-well plates at a density of 1 × 10^5 cells per well and cultured in RPMI-1640 containing TNF (5 ng/mL), with or without IFN-γ (100 U/mL). Cells were incubated for 16 h in a CO[2] incubator at 37 °C. Primed neutrophils were then stimulated with anti-MPO antibody (Hycult Biotech, Cat. No. HM1135) for indicated hours. Extracellular ROS were quantified using the Fluorimetric Hydrogen Peroxide Assay Kit (Sigma-Aldrich, Cat No. MAK165). A master mix was applied at 50 µL per well, and fluorescence was measured every minute for one hour at an excitation wavelength of 540 nm and an emission wavelength of 590 nm using a microplate reader preheated to 37 °C. For NET detection, SYTOX green (Invitrogen, Cat No. S7020) was added to non-fixed live cells so that only extracellular DNA would be detected. Hoechst 33342 (Invitrogen, Cat. No. H3570) was added to detect the total number of cells. Fluorescence images were acquired on a BZ-X700 microscope (Keyence). Quantitative analysis of NET formation was performed using the BZ-X700 “Hybrid cell count system,” an algorithm for accurate quantitation of images of cultured cells. In the indicated experiment, serum IgG purification from MPA patients was performed using the Melon gel IgG purification kit (Thermo Fisher, Cat No. 45206). The dilution ratios of the antibodies used are shown in Supplementary Table [220]6. Measurement of serum IFN-γ levels Frozen serum samples from patients with MPA and healthy donors were thawed and diluted at a 1:4 ratio with Sample Diluent HB (BioRad). Serum IFN-γ concentrations were measured using Bio-Plex Pro Human Inflammation Assays (BioRad, Cat. No. 171AL001M) with the Bio-Plex 200 System (BioRad). ROC curve The ROC curve for relapse prediction was constructed using pROC package (v1.18.0). We used the glm function and calculated the coefficient in generalized linear model (GLM) to create the combination ROC curve. The cut-off value was determined using Youden’s index. Reporting summary Further information on research design is available in the [221]Nature Portfolio Reporting Summary linked to this article. Supplementary information [222]Supplementary Information^ (1.9MB, pdf) [223]Reporting Summary^ (187.9KB, pdf) [224]Transparent Peer Review file^ (1.2MB, pdf) Source data [225]Source Data^ (56.7KB, zip) Acknowledgements