Neutrophil extracellular traps exacerbate Th1-mediated autoimmune responses in rheumatoid arthritis by promoting DC maturation

Enhanced spontaneous NETs formation in PMNs from RA patients and CIA bone marrow

NETs have been demonstrated to be a prominent feature in RA polymorphonuclear cells (PMNs) 11. Thus, we sought to assess NETosis in a cohort of RA patients with high inflammatory burden (clinical characteristics are in Table 1). We found that RA PMNs display increased propensity to spontaneously form NETs compared to control PMNs isolated from healthy donors (Fig. 1A and B). Specifically, a significant increase in NET formation in RF (+) as compared to RF (–) RA patients was demonstrated (Supporting Information Fig. 1A). We next assessed whether RA inflammatory milieu could drive NET formation. In order to assess the capacity of soluble factors to induce NETosis, PMNs were isolated from healthy donors and treated with 2% serum or SF isolated from RA patients or healthy serum. It was found that both RA serum and SF are potent inducers of NETosis in healthy PMNs (Fig. 1C). Furthermore, serum or SF derived from RF (+) patients induced higher NETosis as compared to the serum or SF from RF (–) RA patients (Supporting Information Fig. 1B and C). Collectively, the RA inflammatory burden is associated with the increased rate of spontaneous NET release. In addition, the existence of autoantibodies further augments NETosis, which is in accordance with previous reports 11.

To delineate the role of NETs in the adaptive immune responses, we first sought to establish whether aberrant NETosis also occurs in CIA mouse model. Thus, we found that BM PMNs from CIA diseased mice (d35–45 post collagen injection) had higher numbers of spontaneously released NETs, characterized as DNA structures complexed with neutrophil elastase (NE) compared to BM PMNs isolated from naïve mice (Fig. 1D). Moreover, costaining of DNA with anticitrullinated H3 revealed that externalized NETs contained citrullinated epitopes (Supporting Information Fig. 2).

Next, we asked whether the inflammatory environment of CIA could induce NET formation. To this end, BM PMNs from naïve DBA/1 mice were treated with 2% CIA or naïve mouse plasma. Interestingly, CIA plasma induced significantly elevated NET release compared to naïve plasma that failed to induce NETosis, as evidenced by the colocalization of DNA with NE (Fig. 1E). Collectively, these results provide evidence for an increased NET formation by PMNs in CIA model and corroborate human data that NETs are driven by CIA inflammatory milieu.

Cl-amidine treatment during CIA reduces clinical disease activity and joint destruction

PAD4 inhibition has been shown to prevent NETs 15, 16 and thus Cl-amidine is considered as an inhibitor for NET formation. Cl-amidine is a highly specific inhibitor of PAD4 and related intracellular and extracellular PAD isozymes 19. Indeed, in vitro Cl-amidine treatment prevented hypercitrullination of histone H3 in naïve PMNs as shown by Western blot (Supporting Information Fig. 3). Given that citrullination (catalyzed by PAD4) is also an important mechanism during RA pathogenesis, we sought to investigate the phenotypic and immunological impact of PAD4 inhibition—as an inhibitor of NETosis—in the CIA model.

We initially examined the effect of Cl-amidine administration in BM-derived CIA-NETs. To address this, we administered Cl-amidine intraperitoneally (i.p., or DMSO as vehicle control) daily from d21 through d45 after priming with bovine type II collagen (bCII, Fig. 2A). We found that NETs release was significantly decreased in CIA-Cl-amidine administered mice compared to CIA-DMSO animals as assessed by the costaining of DNA with NE (Fig. 2B). Importantly, CIA-Cl-amidine administered mice experienced significantly decreased severity and disease onset as compared to vehicle control administered mice. Specifically, mice receiving daily Cl-amidine exhibited >50% decrease in clinical disease score on days 25–32 compared to mice administered vehicle control (CIA-DMSO, Fig. 2C). Representative histological analyses of proximal knee joints were performed in two mice per group at the end of each experiment, at d32 and 39 post immunization. Specifically, in the proximal knee joint severe arthritis with pannus formation was observed in CIA-DMSO mice (Fig. 2D upper panel and lower panel; grade 4), as compared to moderate arthritis in CIA-Cl-amidine group (Fig. 2D upper panel and lower panel; grade 1). In addition, CIA-Cl-amidine-treated mice exhibited decreased intensity of inflammation and better integrity of bone structure in the proximal joint compared to the CIA-DMSO group (Fig. 2D). Collectively, these results suggest that Cl-amidine treatment reduces NET formation by BM PMNs, attenuates clinical disease severity, and decreases joint inflammation during the effector phase of the disease in CIA model.

Another hallmark of RA pathogenesis and CIA is the generation of autoantibodies. To this end, plasma from bCII-injected mice treated with Cl-amidine or vehicle control (DMSO) was analyzed for the presence of total IgG Ab to bCII. Notably, Cl-amidine treatment significantly attenuated the anti-bovine CII IgG Abs levels compared with DMSO vehicle control at the end point of each experiment (d32 to d45 post immunization, Fig. 2E). These results demonstrate that Cl-amidine treatment decreases the humoral response to bCII. In summary, the above findings provide evidence that Cl-amidine administration in the CIA model inhibits NETosis and attenuates disease phenotype and arthritis progression.

Cl-amidine treatment diminishes IFN-γ-producing Th1 cells and attenuates DC maturation

Next, we examined whether the attenuated disease pathology in the CIA after Cl-amidine administration could be attributed in decreased Th1 and/or Th17 immune responses. To address this, flow cytometry was performed on draining LN (dLN) cells (dLNCs) from CIA mice treated with Cl-amidine daily for 12 days after the first immunization with CII. Interestingly, the frequency of CD4⁺ T cells was decreased in CIA-Cl-amidine administered mice compared to CIA-vehicle control mice (Fig. 3A) and this was accompanied by significantly decreased frequencies of IFN-γ-expressing CD4⁺ T cells by dLNCs in CIA-Cl-amidine. IL-17-expressing CD4⁺ T cells were also decreased but did not reach statistical significance (Fig. 3A).

T-cell activation, polarization, and proliferation in vivo require Ag presentation by professional APCs. DCs serve as the best candidate that upon maturation provide the necessary costimulatory molecules and secrete proinflammatory cytokines to instruct T-cell responses 20, 21. Therefore, we examined whether Cl-amidine administration in CIA could also affect DC maturation. Interestingly, it was found a significantly decreased expression of MHC class II (MHC-II) and CD86 costimulatory molecules expressed by CD11c⁺ DCs isolated from dLNs upon Cl-amidine treatment compared to DMSO vehicle group (Fig. 3B). Taken together, these data demonstrate that treatment of CIA mice with Cl-amidine attenuated the Th1 immune responses and the maturation of DCs in the dLNs.

CIA-derived NETs increase the maturation of myeloid DCs

To provide direct evidence for a role of NETs in DC maturation, we generated BM-derived DCs (BMDCs) from naïve DBA/1 and treated them with supernatants containing CIA-NETs (Fig. 4A) or control supernatants. We first examined the expression of costimulatory molecules on BMDCs (CD11c⁺) in the presence or absence of CIA-NETs. We observed a significant upregulation of both CD80 and CD86 costimulatory molecules in the presence of CIA-NETs as compared to control (Fig. 4B). Moreover, assessment of proinflammatory cytokines in culture supernatants revealed a vast increase of IL-6 by DCs in the presence of NETs (Fig. 4C). TNF-α production was also increased but did not reach statistical significance (data not shown). These results indicate that NETs are capable of directly activating DCs as shown by the induction of costimulatory molecules and proinflammatory cytokine.

NETs potentiate the immunostimulatory effect of BMDCs in shaping Ag-specific Th1 immune responses

To assess the functional importance of our findings, we applied the OT-II mouse model. We performed coculture experiments of CIA-NET-treated ovalbumin (OVA) pulsed BMDCs with OT-II dLNCs as shown in Figure 5A. To this end, WT BMDCs were pulsed with OVA in the presence of CIA-NETs or control supernatants. Forty-eight hours later, the intracellular IL-17 and IFN-γ production by CD4⁺ T cells was assessed. Interestingly, CIA-NET-treated OVA-pulsed BMDCs significantly induced IFN-γ production by CD4⁺ OT-II-LNCs as compared with control-treated OVA-pulsed BMDCs (Fig. 5B). Notably, no significant difference was observed in IL-17 production by CD4⁺ T cells found in OT-II-dLNCs either in the presence of CIA-NETs or control (Fig. 5B). These findings provide evidence for the increased capacity of NET-treated DCs to promote autoreactive Th1 immune responses.

Human RA-NETs enhance the maturation of monocyte-differentiated DCs

Since NETs are a prominent feature in RA, we sought to examine whether RA-NETs could induce the maturation of human DCs. To this end, we exposed human monocyte-differentiated DCs (moDCs) to RA-NETs and compared their effect to that from supernatants derived from control PMNs. RA-NETs increased the expression of HLA-DR and CD86 costimulatory molecule by DCs as compared to control (Fig. 6A). Moreover, assessment of proinflammatory cytokines in culture supernatants revealed that RA-NETs significantly increased IL-6 and TNF-α cytokine release as compared to control supernatants (Fig. 6B). Overall, our data provide evidence for an immunostimulatory role of human RA-NETs in APCs.

Human subjects

Peripheral blood samples were obtained from patients diagnosed with RA according to the 1987 American College of Rheumatology (ACR) criteria 42, followed by the Rheumatology Clinic of the University Hospital of Crete. SF was collected from active RA patients, centrifuged at 800 × g for 15 min at 4°C and supernatants were stored at –80°C until used. Serum was also isolated from patients or healthy donors. The study was approved by the Ethics Committee of the University Hospital of Heraklion, University of Crete School of Medicine, and all participants gave informed consent.

Reagents

Fluorescent-conjugated monoclonal antibodies to CD80 (16-10A1), CD86 (GL-1), CD11c (N418), CD4 (RM4-5), CD25 (pc61), and CD44 (IM7) were all from Biolegend; Ly6G (1A8) and IFN-γ (XMG1.2) from BD Pharmingen; MHC-II (M5/114.15.2) from Miltenyi; IL-17 (eBio17B7) and brefeldin A from eBioscience. Dulbecco's modified Eagle's medium (DMEM-glutamax, Gibco), RPMI-1640 (Gibco), fetal bovine serum, penicillin (100 U/mL), and streptomycin (100 μg/mL), all were from Gibco, Carlsbad, California, USA; PMA, ionomycin, saponin, and A23187 calcium ionophore were from Sigma-Aldrich. Murine PMNs were isolated with Percoll (Sigma) and cultured in RPMI-1640 supplemented with l-glutamine, 2% bovine serum albumin (Gibco), and 10 mM HEPES (Gibco). PBS tablets were from Gibco.

For immunofluorescence the following primary antibodies were used: polyclonal anti-NE Ab (Abcam), polyclonal anti-human myeloperoxidase (anti-MPO, Dako), anticitrullinated H3 (Abcam), CF488-labeled anti-rabbit Ab (Biotium), CF555-conjugated goat anti-rabbit IgG (H + L) Ab (Biotium) and DAPI (Sigma).

Mice

Male DBA/1J and C57BL/6 (B6) mice (7- to 12-week-old males) were obtained from the SPF facility of the Institute of Molecular Biology and Biotechnology (IMBB Heraklion Crete, Greece). OT-II TCR transgenic mice were obtained from Biomedical Research Foundation of the Academy of Athens. All procedures were in accordance to institutional guidelines and were approved by the Greek Federal Veterinary Office.

CIA induction, treatment groups, and scoring system

For induction of CIA, DBA/1J mice were injected intradermally at the base of the tail with 100 μL complete Freund's adjuvant (Sigma), containing 200 μg bCII (MD Biosciences) and 250 μg inactivated Mycobacterium tuberculosis (H37Ra; Difco) on d0, followed by a boost immunization on d21 43. Each experimental group consisted of eight mice that were treated daily after the d21 booster injection through sacrifice on d35–45 with one of three interventions: no injection, PBS/DMSO (vehicle control), or 10 mg/kg Cl-amidine (kindly provided by P. R. Thompson). All injections were given i.p. and doses were calculated for the average weight of the group (20 g). Mice were scored five times a week from d21 according to the following scoring key: 0 points if no visible clinical symptoms, 1 point per swollen toe (regardless of whether this includes one or two joints (exclusive digit I in both front paws), 5 points for involvement of knuckles (metatarsal/metacarpal), or 5 points for involvement of wrist (tarsal/carpal), giving a total score of 9 and 10 per paw for front and hind paws, respectively. Blood was collected by retro-orbital aspiration at the end of the study and the isolated plasma was stored at –80°C for later batch processing. At d35–45, all animals were sacrificed by anesthesia with diethyl-ester and cervical dislocation.

Histological examination

Tissue was fixed in 4% buffered formalin solution, routinely processed, and embedded in paraffin. Bone was previously decalcified in Schaefer solution. Four micrometer thickness sections were stained with hematoxylin & eosin (H&E) and evaluated by a Nikon Eclipse E-400 light microscope. Histological grading of arthritis of the proximal limb joint (knees) from each animal was performed according to the following five-scale grading system: 0 = normal histology of the knee, 1 = normal synovium or mild synovial reaction with scarce inflammatory cells, 2 = moderate synovial reaction and moderate density of inflammatory cells, 3 = severe synovial hyperplasia and dense population of inflammatory cells, 4 = pannus formation associated with articular cartilage erosion. Histopathology pictures were captured using a Nikon Digital sight, DS-SM, photographic system.

Measurement of type II collagen-specific antibodies in plasma

Plasma samples were collected at the peak of the disease and the levels of anti-CII IgG were measured by ELISA as described 43. Horseradish peroxidase conjugated anti-mouse immunoglobulin IgG (Millipore) was used as secondary antibody and 3,3’,5,5’-tetramethylbenzidine as substrate.

Isolation of PMNs

Human and mouse PMNs were isolated as previously described 44, 45. Viability of human PMNs was measured 99% by trypan blue dye exclusion and mouse BM PMNs were more than 85% Ly6G-positive by flow cytometry.

Generation, assessment, and quantification of NETs (human, mouse)

Primary human PMNs (healthy or RA) or mouse PMNs from bCII-injected or control mice were seeded onto coverslips coated with 10% poly-l-lysine (Sigma-Aldrich). Primary human PMNs were cultured for 3 h at 37°C, 5% CO₂ and murine BM-derived PMNs incubated for 4 h at 37°C, 5% CO₂. Cells were fixed with 4% parafolmadehyde, blocked with 5% BSA, and were permeabilized with 0.5% Triton X-100. DNA was stained with 300 nM DAPI. Protein staining was with anti-NE or with anti-MPO primary Ab for 1 h at RT, followed by 1-h incubation (RT) with CF488-labeled or with CF555-conjugated secondary Ab. Three washes with 0.5% BSA/PBS were performed in between all stainings. After staining, coverslips were mounted on Mowiol 4–88 (Sigma-Aldrich) and were observed under confocal microscope (Leica SP8). NETs were manually quantified by two blinded observers. Decondensed nuclei (stained with DAPI), which also stained positively for NE or MPO, were considered NETs. The percentage of NETs was calculated as the average of at least five fields, normalized to the total number of cells. DNA was quantified with a PicoGreen dsDNA kit (Invitrogen).

For NETs isolation, human RA PMNs (1.5 × 10⁶) or murine CIA BM-derived PMNs (2 × 10⁶) were seeded in 6-well tissue culture plates and cultured for 3 or 4 h, respectively. Wells were carefully washed once with prewarmed medium and supernatants were vigorously collected and spun at 50 × g for 5 min at 4°C to remove intact cells. Thus, the supernatant obtained after centrifugation (RA-NETs or CIA-NETs) were lyophilized resuspended in the appropriate volume of medium (3× condensed) and were used for the coculture experiments.

Generation of BMDCs and DC-NETs coculture experiments

DCs were generated from BM progenitors of DBA1/J mice and B6 mice 46 and RBC were lysed with NH₄Cl. On days 0, 3, 6, and 8, cultures were supplemented with fresh complete DMEM glutamax containing 20% X63Ag8 supernatant (derived from a murine granulocyte-macrophage colony-stimulating factor (GM-CSF) secreting cell line; kindly provided by B. Stockinger, National Institute of Medical Research, London, UK 47. After 9 days of culture, nonadherent cells were harvested and purity was assessed based on CD11c expression by flow cytometry. DCs were subsequently cultured in the presence or absence of 5% of concentrated supernatants derived from bCII-injected (CIA-NETs) or naïve (control) mice. Culture supernatants and cells were collected 18 h after stimulation and cytokine levels were determined by ELISA. Cells were acquired after surface staining on a FACS Calibur (BD Biosciences).

Assessment of OVA-specific T-cell responses and intracellular cytokine staining

LNs from OT-II transgenic mice were excised and single-cell suspensions (4 × 10⁵ cells/ 200 μL/well) were subsequently cultured in complete DMEM medium in the presence or absence of DCs (10⁵ cells) previously pulsed with OVA endotoxin free (20 μg/mL, Hyglos) and/or 5% of concentrated CIA-NETs or control supernatants 48. Forty-four hours later LNCs were restimulated with PMA (50 ng/mL) and ionomycin (2 μg/mL) for 6 h at 37°C, 5% CO_2, and brefeldin A (10 μg/mL) was added for the last 2 h. Following staining of surface markers, cells were fixed with parafolmadehyde and permeabilized with 0.5% saponin buffer 49. Cells were then incubated with IL-17 and IFN-γ mAbs and acquired on FACS Calibur.

Generation of moDCs and moDC-NETs coculture experiments

Monocytes were positively selected from human peripheral blood of healthy volunteers using CD14 MicroBeads (Miltenyi Biotec) according to the manufacturer's instructions and cells were cultured in complete RPMI 1640 medium for 5 days with the addition of human GM-CSF and recombinant IL-4 (both from Peprotech) at d0 (100 ng/mL) and d3 (50 ng/mL). After 5 days of culture, nonadherent cells were harvested and moDCs were subsequently cultured in complete RPMI medium in flat-bottom 96-well plates at a density of 4 × 10⁵ cells/200 μL/well in the presence or absence of 5% of concentrated supernatants derived from RA (RA-NETs) or healthy PMNs (control). Culture supernatants and cells were collected 18 h after stimulation. Cytokine levels were determined by ELISA and the cells were acquired after surface staining on a FACS Calibur.

Western blot analysis

Control BM PMNs were lysed on ice in SDS lysis buffer (2% SDS, 62.5 mM Tris pH 6.8, 5% 2-mercaptoethanol, 10% glycerol) supplemented with Complete Protease inhibitor cocktail (Roche) and centrifuged at 13 000 × g for 30 min at 4°C. Whole-cell lysates (40 μg protein) were subjected to SDS-PAGE electrophoresis on 15% gels and then transferred to an Immobilon-P^sq membrane (Millipore). Membranes were blocked with 5% BSA in TBST and then incubated with anti-MPO (1:200) as loading control and anticitrullinated H3 (1:100). Detection was performed using HRP-linked Abs (Cell Signaling Technology) and enhanced chemiluminescent (ECL) detection reagents (Amersham Biosciences).

ELISA

Detection of mouse IL-6 and TNF-α Duo set (R&D systems) and human IL-6 and TNF-α (eBioscience) in culture supernatants harvested at the indicated time was performed by sandwich ELISA following the manufacturer's recommendations. Light absorbance at 450 nm was measured using the ELx800 Biotek.

Flow cytometry

Cells were stained for extracellular markers for 20 min at 4°C in PBS/5% FBS. Cells were acquired on a FACS Calibur and analyzed using the FlowJo software (Tree Star).

Statistics

Statistical analysis was performed with Prism software version 6.01 (GraphPad Software). Student's t-test (two-tailed, 95% confidence interval) was used, unless indicated otherwise. p-Values less than 0.05 were considered statistically significant.