Blood samples were obtained from 14 healthy donors (HDs) and 13 SLE patients. Patients 1–10 developed SLE at pediatric age and were on long term follow-up within Rheumatology Unit of Giannina Gaslini Institute, Genoa; patients 11–13 received SLE diagnosis at adult age and were on follow-up at S. Martino Hospital, Genoa. In both cases disease diagnosis was made according to the American College Rheumatology Systemic Lupus classification criteria as revised by the Systemic Lupus International Collaborating Clinics (SLICC).

Informed written consent was obtained from each individual before participation, upon approval of the local Ethic Committee. Clinical features are shown in Supplementary Table 1.

Polymorphonuclear cells (PMNs) were isolated from fresh heparinized blood samples of patients with LN, using the Ficoll/Dextran separation method (14) and plated onto 24-well plastic dishes at the concentration of 3 × 10⁶ cells/well. NETosis was induced by 30 nM Phorbol Myristate Acetate (PMA) stimulation, as previously described (27). Cells were then incubated with 15 U/ml Micrococcal DNase (Cayman, Ann Arbor, MI, USA) for 30 min at 37°C and the reaction was stopped by the addition of 5 mM EDTA. Samples were cleared from cellular debris by 13,000 rpm centrifugation. The molecular size of DNA fragments, obtained by DNase digestion, was assessed by 2% agarose gel electrophoresis. DNA and protein contents of NET extracts were determined by means of Nanodrop measurements and Bradford assay, respectively. Samples were then stored in aliquots at −80°C.

Peripheral blood mononuclear cells (PBMCs) were obtained from HDs buffy coats or from heparinized blood samples of Lupus patients, through Lymphoprep (Sentinel Diagnostic, Milan, Italy) density gradient centrifugation. B cells were purified from PBMCs by positive selection with CD19+ Microbead Isolation Kit (Miltenyi Biotec Bergish Gladbach, Germany). CD19+ cells were plated onto 48-well plastic dishes, at the concentration of 5 × 10⁶ cells/well in RPMI 1640 medium (BioWhittaker, Lonza, Belgium), containing 10% Bovine Serum (FCS, Defined, Hyclone, Euroclone, Milan, Italy), 2 mM l-Glutamine, 10 mg/ml non-essential amino acids, 1 mM pyruvate, 50 U/ml penicillin, 50 U/ml streptomycin (Gibco, Grand Island, NY, USA), 5 × 10⁻² M 2-mercaptoethanol (Sigma-Aldrich, St. Louis, USA) and supplemented with 4,000 U/ml Interleukin 2 (Proleukin, Chiron Corp., Emeryville, CA, USA), 2.5 μg/ml CpG2006 (5′-tcgtcgttttgtcgttttgtcgtt-3′; TIB Molbiol, Genoa, Italy) and 1.5 μg/ml F(ab')₂ anti human IgM, IgA, IgG (Jackson Immunoresearch Europe, West Grove, PA, USA). Cells were cultured overnight; NET extracts (3 μg/ml protein concentration) were added after 16 h and were incubated for 2 h at 37°C. Following 2 washes in PBS, cells were stained with FITC-conjugated anti-human CD20 monoclonal antibody (Miltenyi Biotec) and plated onto polylysine coated high-quality 96-well clear bottom black plates, suitable for confocal microscopy. After fixation and permeabilization with BD Cytofix Cytoperm Buffers (BD Biosciences, San Jose, CA, USA), according to the manufacturer's instructions, cells were further stained with an Alexa 647-conjugated anti-human Myeloperoxidase (MPO) monoclonal antibody (Dako, Agilents Technologies, Santa Clara, CA, USA) that specifically evidenced the cellular binding and localization of NET fragments (25). Cells were then washed, and cell nuclei were counterstained with Hoechst 33342. Imaging was performed using Opera Phenix (PerkinElmer, Waltham, MA, USA) high-content screening system. Wells were imaged in confocal mode, using a 40X water-immersion objective. Alexa 647 signal was laser-excited at 640 nm and the emission wavelengths were collected between 650 and 760 nm. Excitation and emission wavelengths for visualization of Hoechst 33342 signal were 405 and between 435 and 480 nm, respectively.

Naïve B cells were enriched by cell sorting (BD-FACSAria instrument, BD Biosciences, San Jose, CA, USA), following the staining with Pe-Cy7-conjugated anti-CD19, FITC-conjugated anti-CD24 and APC-conjugated anti-CD38 monoclonal antibodies (Biolegend, San Diego, CA, USA). To identify the population of interest, we used the gating strategy described in Supplementary Figure 1. The sorted naïve B cell population had a purity of around 99.5%. IgD and CD27 expression on our gated CD19+ CD24- CD38^low evidenced the presence of a minimal contamination of memory B cells (CD27+ IgD-) that were ranging from 6.6 to 11.4% (data not shown).

Enriched naïve B cells, from SLE patients and HDs, were labeled with 0.5 μM 5-(and-6-)-carboxyfluorescein diacetate, succinimidyl ester (CFSE) (Molecular Probes, Eugene, OR,) for 8 min at room temperature, plated onto 96 multi-well plates and activated with the following stimuli: 2.5 μg/ml CpG2006, 5 μg/ml, F(ab')₂ anti human IgM, IgA, IgG (Jackson Immunoresearch Europe, West Grove, PA, USA) and 4,000 U/ml IL-2 (Proleukin, Chiron Corp., Emeryville, CA). LN-NET extracts were added to cultures on day 5, at the concentration of 0,5 μg/ml protein.

At day 7 cells were harvested, washed in PBS and labeled with CD19 PE-Cy7, CD27 APC-Cy7, and CD38 APC for 30 min at +4°C; proliferation rate (analyzed by CFSE diluition assay) and plasmablasts (CD27+, CD38^high) percentages were determined in viable (propidium iodide negative) CD19⁺ cells, using a BD-FACSCanto Flow Cytometer and Kaluza Software (Beckman Coulter Life Sciences, IN, USA).

Soluble IgG levels were determined, by ELISA assay (28), in culture supernatants collected at day 7 from enriched naïve B cells, in the same culture conditions described above. Briefly, Dynatech M129A ELISA plates (Corning Costar, Glendale, AZ, USA) were coated with 10 μg/ml goat anti-human IgG or anti-human IgG2 antibodies (Southern Biotech, Birmingham, AL, USA), diluted in 0.1 M Phosphate Buffer (pH 9.6), and left overnight at 4°C. Plates were then washed with PBS + 0.05% Tween and saturated for 2 h with PBS + 10% FCS at room temperature (RT). Serial dilutions of cell supernatants and of IgG isotype standards were incubated for 2 h at RT. Plates were washed four times with PBS + 0.05% Tween and incubated with 0.5 μg/ml sheep HRP-conjugated anti-human IgG1 or anti-IgG2 secondary antibodies for 2 h at RT. After 4 washes with PBS + 0.05% Tween, the reaction was developed with 75 μl of 3,3′,5,5′-tetramethylbenzidine (TMB) substrate (Sigma-Aldrich, St. Louis, MO, USA) and it was stopped after ~5 min by 2 N Sulfuric Acid. Absorbance values were read at 450 nm using a microplate reader (Eppendorf, Milan, Italy).

Ex vivo T-bet expression was evaluated on CD19+ cells, obtained from previously isolated frozen PBMCs of 4 healthy controls and 3 patients with LN. Cells were stained with FITC-conjugated Live/Dead Fixable Dye (Molecular Probes, Eugene, OR, USA) for 15 min at 4°C in the dark, and subsequently labeled with PE-Cy7-conjugated anti-human CD19 Antibody (Biolegend). After fixation and permeabilization, cells were stained with PerCP-Cy5.5-conjugated anti-human T-bet (Thermofisher, MA, USA) for 30 min at RT. Intracellular T-bet expression was measured in alive cells (detected as FITC^dim) by flow cytometry, using a BD-FACS Canto instrument, and it was analyzed with Kaluza. The effect of LN-NETs on T-bet expression was evaluated in enriched naïve B lymphocytes, derived from HDs. Cells were cultured in complete medium and stimulated with 2.5 μg/ml CpG2006 and 1.5 μg/ml F(ab')₂ anti human IgM, IgA, IgG, or only with NET extracts, for up to 48 h, T-bet expression was detected after 48 h stimulation (T48) and compared with the baseline (T0), prior the addition of stimuli. Negative isotype controls were also included in each experiment.

For statistical analysis, GraphPad Prism (Graph Pad software, La Jolla, CA, USA.) software was used. Mann-Whitney U test was used for comparing groups. Statistical significance was defined as P-value < 0.05.

In order to assess if NETs are internalized by B lymphocytes, that is a prerequisite for activating intracellular signaling pathways, CD20+ B cells, cultured in the presence of anti-Ig and CpG2006 were exposed to NET fragments, obtained by PMA-stimulated neutrophils. As assessed by agarose gel electrophoresis, the molecular size of NET fragments ranged from 100 to 200 kB (Supplementary Figure 2), accordingly to previous observations (20). Myeloperoxidase staining was used to visualize NETs (28). As shown in Figure 1, NETs uptake was evidenced by the presence of intracellular small granules, detected only in viable CD20+ B cells: this pattern was not observed in CD20+ B cells, that were not exposed to NETs.

In a preliminary context, we tested the ability of our enriched HD and SLE naïve B cells to proliferate and differentiate into plasmablasts, in response to canonical agents, that are known to trigger class switch recombination (CSR); in parallel, we evaluated the influence of LN-NET exposition on naïve B cells activation. Flow cytometric analysis revealed that, following stimuli, enriched naïve B cells significantly proliferate and differentiate into CD27+ CD38^high plasmablasts; the rate of cell proliferation and the percentage of CD27+ CD38^high cells were not modified by NET addition (Supplementary Figure 3).

We analyzed the effect of LN-NET on different immunoglobulin production from enriched, stimulated naïve B cells. We found that soluble IgG2, after LN-NET addition, were significantly increased in SLE naïve B cells supernatants; conversely, this increment was not observed for IgG1 and IgG3 (Figure 2A), as well as IgM and IgA levels (Figure 2B), neither in SLE, nor in normal B cells. We also did not detect a significant increase in IgG2 production by normal naïve B cells, following NET exposition: therefore, the ability of LN-NET to raise soluble IgG2 levels appears to be specific for SLE B cells, By analyzing the responsiveness of individual patients, we did not observe a significant correlation with disease severity (evaluated by SLEDAI score) on IgG2 induction by LN-NET (Supplementary Table 1): for this reason, it is conceivable that LN-NETs exert a pathogenic function in the early phases of autoimmunity, irrespectively of the disease evolution.

Since T-bet is a master regulator of autoreactive B cell differentiation in murine models and is able to stimulate IgG2 class switch (24), we investigated whether its constitutive expression is altered in SLE patients. We analyzed the amount of CD19+ B cells expressing T-bet, that reached a maximum level in SLE patients (>90% of T-bet expressing cells) and was higher than normal controls (Figures 3A,B). We postulated that the elevated T-bet levels, observed in SLE patients, could result from the in vivo chronic stimulation of NET and hence we investigated if normal naïve B cells might express T-bet protein to the same extent as SLE B cells, when exposed to LN-NET. We therefore tested the expression of T-bet, before (T0) and after 48 h (T48) of NET stimulation; as a positive control, cells were stimulated with CpG2006 and anti-Ig, powerful inducers of T-bet expression. The percentage of T-bet expressing cells was not elevated in normal controls (Figure 3C; mean = 31.18%); as expected, CpG2006 and anti-Ig induced a robust proliferation coupled with high increase in T-bet expression (mean: 89.9%; p = 0.0076). Notably, the addition of NETs from LN patients was sufficient to markedly increase the number of T-bet expressing cells (mean = 73.4%; p = 0.08), despite a small effect on cellular proliferation (Figures 3C,D). Consistently, the amount of T-bet protein, measured by Mean Fluorescence Intensity (MFI) detection, was significantly (p = 0.0002) induced after stimulation with LN-NET and showed a 30% increase compared to its baseline (T0) expression (Figures 3E,F).

Taken together, these results show that NET from LN patients exerts a direct and specific effect on IgG2 isotype switch in SLE naïve B cells, and also point to T-bet as a putative mediator in this mechanism.