In this non-randomized observational single-center study, we enrolled 42 patients with GPA (active stage n=21 or remission n=21) and a control group of 21 age- and sex-matched healthy volunteers (HC) ( Table 1 ). GPA was diagnosed according to ACR 1990 (5) but including the 2012 Revised Chappel Hill Consensus Nomenclature (6). All GPA patients were positive for anti-PR3 IgG antibodies and negative for anti-MPO IgG antibodies. Disease activity was evaluated using the Birmingham Vasculitis Activity Score (BVAS, version 3) and organ damage was ascertained using the Vasculitis Damage Index (VDI). Remission was defined as lack of symptoms of active disease (BVAS = 0). Exacerbation of the disease was identified by the presence of new or reemerging signs and symptoms of vasculitis (confirmed by the BVAS > 6), requiring intensification of immunosuppressive therapy according to EULAR recommendations (7). Patients with cancer, kidney failure requiring dialysis, or infection were excluded from this study. Blood samples from all patients during the active stage of GPA were collected before the onset of intensive treatment with high-dose corticosteroids or immunosuppressants. In the group of patients with remission of GPA, 19 patients (90%) were on remission maintenance treatment, 19 patients (90%) received oral glucocorticosteroids (up to 8 mg/day), and 7 patients (33%) were on disease-modifying anti-rheumatic drugs (3 on methotrexate and 4 on azathioprine). Peripheral blood samples were collected without anticoagulant or with sodium citrate anticoagulant using a blood collection system (Venous Blood S-Monovette, Sarstedt, Germany) between October 2016 and August 2019. Serum and plasma were separated using standard laboratory procedures, aliquoted, and frozen at -80° C for further analyses. Laboratory tests (CBC, CRP, anti-PR3/anti-MPO IgG level) were performed on all study participants at the time of blood sampling. Written informed consent was obtained from all study participants and the study protocol was accepted by the ethics committee of Jagiellonian University.

Neutrophils were isolated from anticoagulated blood using a negative magnetic separation kit. (EasySep Human Neutrophil Enrichment Kit, STEMCELL Technologies Inc., Canada). Isolated cells were suspended in Hanks’ balanced salts solution (HBSS) with calcium and magnesium-containing 2% fetal bovine serum (FBS, Sigma-Aldrich Chemical Co., USA) for the spontaneous DNA release assay or suspended in lysis buffer for targeted phosphoproteomics measurement.

Quantification of spontaneously released dsDNA by isolated neutrophils was performed in 10x10⁵cells cultured for 3 h in HBSS (250 µL, with calcium and magnesium and 2% FBS) at 37°C, 5% CO₂. Next, DNase I (2.5 U/ml, EURx, Gdańsk, Poland) was added and the reaction was stopped after 10 minutes by the addition of EDTA (pH 8.0, Sigma-Aldrich Chemical Co, USA) to the final 2.5 mM concentration. After sedimentation of cells (190 g, 5 min, 4°C), the concentration of oligonucleotides in the supernatant was measured by fluorometry (Quant-iT Pico Green dsDNA Assay, Thermo Fisher Scientific, USA) in Qubit 3 fluorimeter and expressed as relative fluorescence units (RFU).

Levels of selected phosphoproteins: Akt^(Ser473), BAD^(Ser136), GSK-3α/β^(Ser21/Ser9), IRS-1^{(Ser636/Ser639)}, mTOR^(Ser2448), p70 S6 kinase^(Thr389), PTEN^(Ser380), S6 ribosomal protein ^{(Ser235/Ser236)}), and ULK1 ^(Ser555) were measured using a commercially available Bio-Plex Pro™ Cell Signaling Akt Panel (8-plex, Bio-Rad, USA) in Luminex instrument or ELISA (PathScan^® Phospho-ULK1, Cell Signaling Technology, USA). Briefly, the total cellular protein was isolated from neutrophils (5x10⁶ cells) with the lysis buffer according to the manufacturer’s protocol, and 10 µg of the protein lysate per well was used for the Luminex assay or ELISA.

Total circulating cell-free DNA was isolated from serum using the phenol/chloroform method (8). In brief, 500 µL of serum was spiked with 0.34 ng of an internal standard of DNA (plasmid pGEM-3Zf(+), Promega, Madison, USA) to compensate for errors in sample processing, diluted with 1 mL of ultrapure water, and extracted four times with the same volume of phenol/chloroform mixture. DNA was recovered by precipitation with 1:10 v/v 3M sodium acetate and 2:1 v/v cold 96% ethanol. After rinsing with 70% ethanol and drying DNA, the pellet was resuspended in 20 µL of water. Circulating cell-free mitochondrial (mtDNA) and nuclear DNA (nDNA) were measured separately by quantitative PCR (qPCR, 7900HT real-time PCR system, Applied Biosystems, Foster City, USA). Details of qPCR reactions and primer sequences for genomic DNA (Alu repeats target) and mitochondrial 16s RNA gene (mtDNA79) were as previously described (5). Quantification of the internal standard was performed using M13 primers (Promega). The quantification cycles of mtDNA and nDNA were corrected for the internal standard and calculated as relative expression (RE) with the use of the 2^-ΔCt formula.

Circulating complexes of neutrophil myeloperoxidase with DNA were measured using ELISA as previously described (9). Briefly, 96-well ELISA plates were coated with anti-MPO antibody (5µg/mL, AbD Serotec, USA), and a Death plus EIA kit (DNA-Death plus, Roche Switzerland) was used as a source of all buffers and of a secondary, HRP-conjugated antibody. Raw results were presented as mean optical densities (OD).

Profiles of circulating cytokines were compared between active GPA and HC using Proteome Profiler Human XL Cytokine Array Kit (Bio-Techne, Minneapolis, USA). Aliquots of 100 µL of serum from five participants randomly selected from each group were pooled, and then diluted to the final volume of 1.5 mL and used for overnight incubation of membranes spotted with capture antibodies. After washing, the membranes were incubated with detection antibodies and developed with a chemiluminescent reagent. Images were collected using C-DiGit Blot Scanner (Li-Cor Bioscience, Lincoln USA). Optical densities of the detection spots ( Supplementary Figure 1 ) were used to select proteins subsequently measured in the serum of all study participants by a custom Luminex multiplex assay (Bio-Techne). Quantified plasma levels of IGFBP-2, IGFBP-3, TFF-3, DPP-IV, basigin (CD147), and CHI3L1 were interpreted from the calibration curves and presented in pg/mL or ng/mL.

Statistical calculus was done using GraphPad Prism 9.0 software (GraphPad Software Inc., San Diego, USA). Descriptive statistics was presented as a mean and standard deviation. Differences between the groups were tested by one-way analysis of variance (ANOVA) with Tukey’s post hoc test. The goodness of fit for the canonical discriminant function was evaluated by Wilks’s Lambda. Type I statistical error P < 0.05 was considered significant.

We observed elevated dsDNA release from non-stimulated neutrophils isolated from patients with active GPA (active: 11070 ± 1233, remission: 7941 ± 1177, HC: 7996 ± 771.7 RFU, p<0.05) ( Figure 1A ). A similar finding was notable for circulating mtDNA (active: 0.01860 ± 0.003, remission: 0.004 ± 0.002 and HC: 0.0006 ± 0.0005 RE, p<0.05) ( Figure 1B ) and DNA-MPO complexes (active: 1.058 ± 0.35, remission: 0.5 ± 0.2, HC: 0.3 ± 0.11 OD, p<0.05) ( Figure 1C ). However, no differences between studied groups were observed in nuclear DNA (nDNA) release (not shown).

Levels of phosphorylated mTOR^(Ser2448), PTEN^(Ser380), and ULK1^(Ser555) were significantly lower in neutrophils from patients with GPA. This was more apparent for mTOR^(Ser2448) levels (active 4 ± 0.5, remission: 15.2 ± 0.6, HC: 35 ± 2.9 MFI, p<0.05) ( Figure 1D ) but also significant for PTEN^(Ser380) (active: 3.11 ± 0.7, remission: 6.6 ± 1.4, HC: 7.2 ± 1.5 MFI, p<0.05) ( Figure 1D ) and ULK1^(Ser555)(active: 0.28 ± 0.1, remission:0.44 ± 0.15, HC: 0.99 ± 0.55, p<0.05) ( Figure 1D ). In patients with active disease, a significant elevation of phosphorylated GSK-3α/β^(Ser21/Ser9) was detected (active: 181 ± 47.4, remission: 102.5 ± 19.12, HC: 116.1 ± 20.83 MFI, p<0.05) ( Figure 1D ). No differences between the study groups were observed in five phosphoproteins measured: Akt^(Ser473), BAD^(Ser136), IRS-1^{(Ser636/Ser639)}, p70 S6 kinase^(Thr389), and S6 ribosomal protein^{(Ser235/Ser236)}) (not shown).

Using a semiquantitative protein array, we performed a screening for differentially expressed serum proteins in patients with active GPA vs. healthy controls. The results suggested altered circulating levels of IGFBP-2, IGFBP-3, TFF-3, DPP-IV, CD147, and CHI3L1 ( Supplementary Figure 1 ). Measurements of the selected proteins in all the participants of the study revealed that in the active stage of GPA, levels of IGFBP-2 were the highest (active: 210 ± 30, remission: 140 ± 15 and HC: 85 ± 8 ng/mL, p<0.05) ( Figure 2A ). Significant elevations also characterized TFF-3 (active: 1.6 ± 0.7, remission: 1.9 ± 0.3 and HC: 0.7 ± 0.2 ng/mL, p<0.05), CD147 (active: 5.2 ± 1.3, remission: 2.0 ± 0.7 and HC: 1.7 ± 0.4 ng/mL, p<0.05), and CHI3L1 (active: 42.5 ± 6, remission: 17 ± 4.5 and HC: 11 ± 5 ng/mL, p<0.05) ( Figure 2A ). However, DPP-IV levels were the lowest in the serum of patients with the active stage of GPA (active: 4 ± 0.5, remission: 6.8 ± 0.7 ng/mL and HC: 6.3 ± 1.5 ng/mL, p<0.05).

Principal component analysis of differentially expressed phosphoproteins and serum proteins allowed for an efficient (98% correct) classification of all participants of the study into active or remission GPA or HC. Only a single participant from the healthy control group was incorrectly classified as GPA at remission (Wilks’s Lambda V = 0.067, P < 0.001, Figure 2B ). A discrimination matrix was tested using a jackknife subsampling and was correct for 92% of participants. PCA vectors explained 66% of the variance and had canonical correlations of 0.92 and 0.75. Positive correlations were observed between biomarkers of NETosis (circulating mtDNA, DNA-MPO complexes levels, and dsDNA level) and the circulating proteins (IGFBP-2, TFF-3, CD147, and CHI3L1) or phosphorylated GSK-3 α/β. Phosphorylated mTOR and PTEN correlated negatively ( Figure 2C ) with NETosis biomarkers or circulating proteins. A very similar pattern of correlation matrix characterized BVAS, which correlated positively with dsDNA, mtDNA, DNA-MPO complexes, IGFBP-2, CD147, CHI3L1, and phosphorylated GSK-3 α/β, and negatively with phosphorylated mTOR and PTEN ( Figure 2D ).