Cellular Adhesion Is a Controlling Factor in Neutrophil Extracellular Trap Formation Induced by Anti-Neutrophil Cytoplasmic Antibodies

Anti-neutrophil cytoplasmic Abs (ANCAs) occur during several forms of vasculitis, leading to the term ANCA-associated vasculitis (AAV) (1). AAV is characterized by damage to tissues and small vessels, particularly in the respiratory tract and kidneys, and it can be clinically classified as granulomatosis with polyangiitis (GPA), microscopic polyangiitis, or eosinophilic GPA. Although the prevalence of AAV is low, it has a significant impact on quality of life and mortality (2, 3) and a substantial economic burden (4). ANCAs consist of anti-myeloperoxidase (MPO) or anti-proteinase 3 (PR3) Abs, and these activate TNF-α–primed neutrophils, allowing them to adhere to endothelial cells (5), degranulate (6, 7), and damage the endothelium (8), with a range of pathological consequences in different tissues. More recently, the ability of neutrophils to form neutrophil extracellular traps (NETs) has been implicated in AAV pathogenesis (9). This unique cellular response results in the unraveling of nuclear DNA, mixing of nuclear and cytoplasmic components, and finally release of large clouds of DNA peppered with various proteins such as histones and proteases (10). Although originally described as a potent weapon to immobilize and kill invading pathogens, NETs are now known to have a range of negative pathological effects, including damaging endothelial cells (11). NETs may contribute to AAV directly through vascular damage, or, because ANCA Ags are abundant on NETs, may lead to Ag presentation and enhanced production of ANCAs, creating a positive feedback loop (12).

Although neutrophil attachment to the endothelium is known to play a critical part in AAV, the role of adhesion in ANCA-induced NET formation has not been addressed. Indeed, studies investigating NETs almost exclusively involve the initial attachment of neutrophils to a surface to facilitate later analysis of NETs. Exceptions to this include two recent studies. The first demonstrated that LPS-induced NET formation requires substrate attachment, whereas PMA-induced NETs do not (13). The second study showed that substrate stiffness and coating dictate the ability of mouse-derived neutrophils to spontaneously form NETs (14). Other than the role of substrate attachment, studies have also indicated that specific adhesion receptors and ligands can be critical for NET formation. Examples include MAC-1, thought to be involved in LPS-induced NETs when attached to poly-l-lysine (15), P-selectin, shown to trigger NET formation in adhered murine neutrophils through P-selectin glycoprotein ligand-1 (PSGL-1) (16), and fibronectin, shown to be a critical ligand in NETs induced during attachment to β-glucan (17, 18). These findings suggest that the control of neutrophil adhesion may offer a unique approach to mitigate pathological NET formation in diseases such as AAV. The therapeutic utility of targeting cell adhesion is increasingly being recognized, and adhesive molecules such as integrins and focal adhesion kinase (FAK) are being investigated as targets in a range of diseases and conditions such as cancer, inflammatory bowel disease, and multiple sclerosis, with some compounds already on the market and many others undergoing clinical trials (19–21).

In this study, we investigated the influence of adhesion on the ability of neutrophils to form NETs in response to ANCAs. We found that NET formation in response to ANCAs, either commercially produced anti-MPO Ab or as total IgG isolated directly from AAV patient serum, was highly reliant on attachment, which could be blocked using Abs for β₂ integrins, or inhibitors of the focal adhesion pathway. We demonstrate the critical role of adhesion molecules in NET formation during AAV and suggest that this can be exploited as a therapeutic pathway.

Human samples were obtained after informed consent was provided in accordance with the Declaration of Helsinki and with approval from the Ethical Review Board of the Graduate School of Medicine, Osaka University, Osaka, Japan (nos. T19204 and 11122-5). Patients with AAV were diagnosed as having microscopic polyangiitis or GPA, according to the criteria of the Research Committees of the Japanese Ministry of Health, Labor and Welfare. GPA was also diagnosed according to the American College of Rheumatology classification criteria. Disease activity score of AAV was estimated by Birmingham vasculitis activity score version 3. IgG was isolated from healthy control or AAV patient serum using a Melon gel IgG spin purification kit (Thermo Fisher Scientific, Waltham, MA).

To mitigate neutrophil adherence during preparation we performed isolation within 30 min of blood collection, used low-attachment LabWare, and kept isolated neutrophils for a maximum of 1 h before use, maintaining them in suspension. Neutrophils were isolated from blood using an EasySep direct human neutrophil isolation kit (STEMCELL Technologies, Vancouver, BC, Canada), with 0.05% human serum albumin (HSA, Sigma-Aldrich, St. Louis, MO) added to the isolation buffer. Isolated neutrophils were washed in HBSS (Thermo Fisher Scientific) with 0.05% HSA.

For fluorescence microscopy, neutrophils in DMEM/F-12 with 15 mM HEPES and 0.05% HSA, without phenol red (Thermo Fisher Scientific), were added to either tissue culture–treated plates coated with 10 µg/ml human plasma fibronectin (Sigma-Aldrich) or Corning Costar ultra-low attachment plates (Sigma-Aldrich), which are coated with a proprietary hydrogel designed to minimize cell attachment. Cells were stimulated with 5 ng/ml TNF-α (PeproTech, Rocky Hill, NJ) followed by 5 µg/ml anti-MPO Abs (A0398 polyclonal, Agilent Technologies, Santa Clara, CA) or 40 µg/ml isolated serum IgG, and incubated for 4 h at 37°C with 5% CO₂. In some cases, cells were stimulated with 100 nM PMA without TNF-α. Cells were stained with 4 µM Hoechst 33342 (Sigma-Aldrich) as a membrane-permeable DNA stain, and 500 nM Sytox Green (Thermo Fisher Scientific) as a membrane-impermeable DNA stain, and imaged without washing steps using either a ToxInsight (Thermo Fisher Scientific) or CQ1 (Yokogawa, Tokyo, Japan) fluorescence microscope with a ×10 objective. In some cases, neutrophils were preincubated with 20 µM FAK inhibitor-1 (Merck, Darmstadt, Germany), 1 µM Src inhibitor-1 (Merck), 2.5 µM Src inhibitor PP1 (Cayman Chemical, Ann Arbor, MI), 10 µM actin inhibitor cytochalasin B from Helminthosporium dematioideum (Nacalai Tesque, Kyoto, Japan), 200 nM PI3K inhibitor copanlisib (Cayman Chemical), 25 µM MEK inhibitor U0126 (Nacalai Tesque), 100 µM reactive oxygen species (ROS) scavenger pyrocatechol (Sigma-Aldrich), 10 µM NADPH oxidase inhibitor DPI (Sigma-Aldrich), 20 µM NE inhibitor GW 311616A (Axon Medchem, the Netherlands), or 10 µg/ml azide-free blocking Abs for integrin β₁ (CD29, clone TS2/16), β₂ (CD18, clone TS1/18), β₃ (CD61, clone 23C6), α₅ (CD49e, clone NKI-SAM-1), α_L (CD11a, clone HI111), α_M (CD11b, clone M1/70), α_X (CD11c, clone 3.9) (all BioLegend, San Diego, CA), or α₉ (CD49i, clone Y9A2, Merck) for 30 min at 37°C before stimulation, with tube inversion every 10 min to minimize cell settling, before being aliquoted into plate wells for the incubation step.

Fluorescent microscopy images were analyzed using CellProfiler v4.0.6 (22) using a custom pipeline. Hoechst stains all DNA, whereas Sytox Green will be excluded from healthy cells and will only stain DNA within dead cells, or DNA present in NETs. Sytox Green staining was brighter than Hoechst staining, which was especially important for defining diffuse NET structures. Therefore, these images were combined, using each pixel’s maximum value from the two channels, creating a grayscale image of merged fluorescence that represents the overall distribution of DNA. This image was used to identify objects using the Otsu thresholding approach. Sytox Green staining intensity and area have been previously used to classify healthy cells and NETs, and in this study we used a similar approach (23). We measured two parameters for each object. The object area, defined as the number of pixels in the object, which we refer to as “DNA area,” and Sytox Green median intensity, which is obtained by overlaying the object masks onto the original Sytox Green image and measuring the median pixel intensity for each object. Object measurements were exported into one CSV (comma-separated values) file for each well of the experiment. CSV files were converted to an FCS (flow cytometry standard) format, and FlowJo v10.7.1 was used for further analysis. DNA area was plotted against Sytox Green median intensity for each object, and three populations were defined: healthy neutrophils, DNA area (low) Sytox Green (low); Sytox Green–positive neutrophils (non-NETs), DNA area (low) Sytox Green (high); and NETs, DNA area (high) Sytox Green (high).

Imaging flow cytometry of NETs was performed as previously described (23). Briefly, neutrophils in DMEM/F-12 with 15 mM HEPES and 0.05% HSA, without phenol red, and with 1.6 µM Hoechst 33342 (Sigma-Aldrich) and 25 nM Sytox Green (Thermo Fisher Scientific) were incubated for 4 h at 37°C with 5% CO₂ with stimulants in low-attachment 1.5-ml tubes (Prokeep, Watson Biolab, Kobe, Japan) with caps open. Ten microliters of 90% v/v Percoll in DMEM/F-12 was layered under cells during incubation to minimize cell aggregation and interactions with the tube walls and create a three-dimensional culture environment for the cells. After incubation, cells were fixed in 1% paraformaldehyde (Sigma-Aldrich) for 10 min at room temperature, then gently resuspended and analyzed as is on an ImageStream^X Mk II instrument (Luminex, Austin, TX) using a ×60 objective. Events with a bright-field area or Sytox Green area >20 μm² were collected. Compensation was performed using single-stained controls.

Imaging flow cytometry images were processed using IDEAS v6.2 (Luminex) and FlowJo as described previously (23). Briefly, the “Object(M01,Ch01,Tight) (Brightfield)” and “Morphology(M02,Ch02) (Sytox Green)” masks were used to identify cell and DNA boundaries, and the feature “Intensity_Morphology(M02,Ch02) And Not Object(M01,Ch01,Tight)_Ch02” was calculated and defined as “exDNA (intensity).” We retained the default nomenclature used by the IDEAS software, which indicates the feature measured (“Intensity”), masking strategy (“Morphology” and “Object, Tight”), fluorescence channels for which the masks are calculated (“Ch02” and “Ch01”), and finally the fluorescence image on which the mask is overlaid and the feature calculated from (“Ch02”). NETs were classified as the “exDNA (intensity)” (high), “Intensity_MC_Ch07 (HO (Intensity))” (high) population.

Isolated neutrophils were treated and prepared as described for NET measurements. ROS release was monitored for 3 h with measurements every 2 min using a fluorometric hydrogen peroxide assay kit (Sigma-Aldrich) and a GloMax microplate reader (Promega, Madison, WI).

Isolated neutrophils were stimulated with 5 ng/ml TNF-α, 5 µg/ml anti-MPO Abs, a combination of these, or left untreated. After a 30-min incubation at 37°C with gentle agitation every 10 min, Fc receptors were blocked with human TruStain FcX (BioLegend), and cells were stained with 1.5 µg/ml Brilliant Violet 605 anti-CD16 (clone 3G8) and 1 µg/ml allophycocyanin anti-CD18 (clone 1B4/CD18) (BioLegend). Cell data were collected using an Attune flow cytometer (Thermo Fisher Scientific) and analyzed with FlowJo. Single neutrophils were gated based on forward scatter/side scatter and CD16-positive staining. The mean fluorescence intensity of CD18 on neutrophils was then measured.

Data were analyzed by two-way ANOVA with interactions and post hoc Tukey tests using R v3.6.1 (24). A p value <0.05 was considered significant. Data are presented as the mean with individual data points or SE indicated.

To investigate the role of adhesion in NET formation in ANCA vasculitis we stimulated neutrophils using anti-MPO Abs. Peripheral blood neutrophils from healthy donors were stimulated with TNF-α and allowed to settle on either fibronectin-coated or commercially manufactured ultra-low attachment surface. The latter surface consists of a covalently bonded hydrogel proven to prevent neutrophil attachment, and we confirmed reduced neutrophil spreading on this surface with bright-field imaging (Fig. 1A). Anti-MPO Abs readily induced NETs on the fibronectin surface, whereas almost no NET formation occurred for neutrophils on the low-attachment surface (Fig. 1B). This was not the case for PMA stimulation, which induced NETs in both conditions. We tested collagen-coated, ICAM-1–coated, and uncoated tissue culture plastic, which resulted in similar levels of NET formation, indicating that this effect is not limited to fibronectin (Supplemental Fig. 1).

To quantify NET formation, we measured two parameters, DNA area and intensity of Sytox Green staining (SYG intensity) (Fig. 1C). We plotted these two parameters and defined NETs as the DNA area (high) and SYG intensity (high) population. The SYG intensity (high) and DNA area (low) population were excluded from NET quantification, as these cells more closely resemble necrotic cells without the release of DNA beyond the cell membrane. Approximately 36% of anti-MPO–stimulated neutrophils formed NETs when incubated on a fibronectin-coated surface, whereas only 7% formed NETs on the low-attachment surface (Fig. 1D, 1E). For comparison, we found that neutrophils formed 69 and 55% NETs on these surfaces when stimulated with PMA. To further investigate the role of adhesion, we stimulated neutrophils in solution and analyzed them by imaging flow cytometry using a method we have described previously (23). Under these conditions, no NET formation could be detected with anti-MPO stimulation, whereas PMA stimulation induced abundant NETs as expected (Fig. 1F, Supplemental Fig. 2A). Neutrophils treated with an isotype control Ab did not produce NETs, suggesting that any potential stimulation from the Fc region of the anti-MPO Ab in triggering NETs is unlikely (Supplemental Fig. 2B).

Given the apparent importance of cell attachment in anti-MPO–induced NET formation, we next investigated the focal adhesion signaling pathway. In neutrophils preincubated with FAK inhibitor I, NET formation reverted to baseline levels (Fig. 2A–C). FAK activation is facilitated by phosphorylation by Src kinases. As expected, inhibiting these kinases with Src inhibitor 1 and PP1 also reduced NET formation significantly, although PP1 inhibition was more effective. Actin polymerization is initiated by focal adhesions and allows for cell spreading and migration, as well as signaling at the focal adhesion junction through mechanical forces on FAK and talin. Activation of these enzymes has widespread effects on cell processes, with a major target being PI3K. Inhibition of actin polymerization by cytochalasin B, and inhibiting PI3K both completely prevented NET formation in response to anti-MPO. Overall, these results indicate a critical role for the focal adhesion pathway in MPO-induced NET formation.

To provide a link between the focal adhesion pathway and NET formation, we considered the role of NADPH oxidase and ROS. Neutrophils incubated with the ROS scavenger pyrocatechol and the NADPH oxidase inhibitor DPI failed to produce NETs in response to anti-MPO (Fig. 3A, 3B), indicating a ROS-dependent pathway of NET formation. The MEK-ERK pathway can be activated through PI3K and the focal adhesion pathway to phosphorylate and activate NADPH oxidase (25). Consistent with this, the MEK inhibitor U0126 blocked NET formation. Downstream of ROS, neutrophil elastase (NE) is a critical enzyme involved in NET formation, and inhibiting this enzyme also blocked NET formation. We next measured ROS release in the context of anti-MPO stimulation with or without these inhibitors. ROS increased significantly with TNF-α and anti-MPO stimulation but remained at close to baseline levels with TNF-α alone (Fig. 3C). Importantly, inhibition of FAK almost completely prevented ROS release, even under anti-MPO stimulation, indicating a combination of anti-MPO stimulation and activation of the focal adhesion pathway is required before ROS production occurs. Downstream of FAK, Src kinase and MEK inhibition partially reduced ROS, whereas PI3K inhibition completely blocked ROS release. Inhibition of NADPH oxidase also prevented ROS release as expected. NE activation is thought to occur downstream of ROS production. Consistent with this, NE inhibition had little effect on ROS release. Inhibition of actin polymerization by cytochalasin B enhanced ROS production, in line with previous studies demonstrating that this facilitates NADPH oxidase activation (26). Taken together, these results indicate a pathway in which attachment primes cells for ROS production leading to activation of NE and eventually NET release.

We next investigated which neutrophil adhesion molecules could be responsible for attachment and activation of the focal adhesion pathway using blocking Abs. We focused on integrin molecules known to be expressed on human neutrophils, that is, integrins β₁, β₂, β₃, α₅, α₉, α_L, α_M, and α_X. Of all the integrins we tested, only blocking integrin β₂ reduced NET formation significantly (Fig. 4A, 4B), indicating that the engagement of this integrin is critical for NET formation in response to anti-MPO Abs. Furthermore, the expression of integrin β₂ was significantly increased with TNF-α priming and anti-MPO stimulation, and synergistically in combination, as indicated by mean fluorescence intensity of neutrophils stained with anti–integrin β2 measured by flow cytometry (Fig. 4C). We next tested whether β₂ receptor ligation and crosslinking could prime neutrophils for NET formation, negating the need for cell attachment. However, neutrophils incubated on low attachment plates with crosslinking of β₁, β₂, β₃, α₅, α₉, α_L, α_M, α_X, or a combination of all remained unresponsive to anti-MPO in terms of NET formation, indicating that additional factors such as engagement of other receptors, cell spreading, or mechanical force signaling are required (Supplemental Fig. 3).

To confirm that our results with anti-MPO Ab stimulation could be reproduced using more clinically relevant stimuli, we collected serum from patients with ANCA vasculitis, who had not yet received immunosuppressive treatments, and isolated total IgG Abs. Neutrophils stimulated with patient IgG produced significant NET formation, whereas those treated with IgG isolated from healthy controls did not (Fig. 5). In line with our results using anti-MPO Abs, NET formation was highly dependent on cell adhesion, with almost no NETs present on the low-attachment surface. NET formation was not correlated with clinical parameters such as Birmingham vasculitis activity score or serum levels of MPO-ANCAs, but it was positively correlated with the serum levels of C-reactive protein, suggesing that the capacity of inducing NETs is associated with the severity of systemic inflammation (Supplemental Fig. 4). Consistent with results using commercial anti-MPO Abs, ANCA vasculitis patient sourced IgG-induced NETs were prevented using inhibitors for FAK, Src kinase, actin polymerization, and PI3K (Fig. 5). Overall, our results highlight a critical role for neutrophil attachment mediated by β₂ integrins, the focal adhesion pathway, and ROS in the formation of NETs in response to ANCAs, as summarized in (Fig. 6.

This study reveals an underappreciated role for neutrophil attachment in dictating NET formation in response to ANCAs both in the form of commercial anti-MPO Abs and isolated IgG from AAV patients. NET formation was mediated by the focal adhesion pathway and ROS production, and required integrin β₂, with other major neutrophil integrins being dispensable.

Neutrophil adhesion is known to play an important role in the pathology of AAV (27–29). Inflammatory cytokines such as TNF-α are known pathological factors during AAV (30, 31), and along with complement C5a and ANCAs, these reduce neutrophil deformability (32, 33) and encourage neutrophil aggregation (34), thereby promoting interactions with the vessel walls, particularly in the small capillaries, which are the main site of AAV-associated pathology (35). Although neutrophil responses to adhesion such as ROS release and degranulation have been well studied (36), there are comparatively few studies that have addressed the role of substrate attachment in NET formation. One study found that adhesion was required for LPS-induced NET formation but was dispensable when cells were stimulated with PMA (13), whereas a second study found that substrate stiffness correlated with spontaneous NET formation in mice (14). Other studies have investigated specific adhesion receptors and ligands, but have not compared unattached versus attached neutrophils (15–18). In the case of AAV, the role of adhesion in NET formation was previously unknown. Preventing neutrophil adhesion and measuring NET formation in suspension is extremely difficult due to the delicate nature of NETs. Although several conventional flow cytometry approaches claim to identify NETs in suspension, these have not attempted to distinguish NETs from other types of cell death (37–39), and their ability to measure NETs has been questioned (40). More advanced imaging flow cytometry techniques have also claimed to identify NETs (41, 42); however, these failed to show the clouds or strings of DNA typically considered NETs. None of these studies addressed potential differences in NET formation when adhered versus in suspension. We recently developed a protocol that maintains NETs in suspension using density control, followed by analysis using imaging flow cytometry, and demonstrated the ability to image fully formed NETs identified as large clouds of DNA (23). In the present study, we used this approach to demonstrate that unlike PMA-induced NET formation, ANCA-induced NET formation is completely blocked in the absence of adhesion. We further corroborated this using ultra-low attachment surfaces, on which neutrophils also failed to produce NETs in response to ANCAs.

Engagement of integrins during neutrophil attachment initiates the focal adhesion pathway. Central to this pathway is FAK, whose activation is mediated through phosphorylation by Src kinase (43–45). In our study, we demonstrate that both FAK and Src are required for ROS production and NET formation. In the case of Src, we used two different inhibitors, Src inhibitor 1 and PP1, providing 40 and 38% reduction in ROS, and 48 and 90% reduction in NETs, respectively. PP1 is known to cause partial off-target inhibition of p38 MAPKs (46), which are reportedly involved in NET formation in response to PMA (47). This dual-target inhibition of PP1 may explain its increased effect on NET formation compared with the more specific Src inhibitor 1. PI3K activation occurs downstream of FAK (48, 49), and we show that inhibitors of PI3K also block both ROS and NET formation. Although PI3K is a central kinase involved in many cellular processes, this finding nevertheless adds weight to the importance of the FAK pathway in NET formation and may help explain previously reported results for the involvement of PI3K in NET formation, such as in response to parasites, microcrystals, and PMA (50, 51). An important consequence of integrin engagement and FAK activation is cell remodeling and actin polymerization. Actin provides a link to the focal adhesion site, allowing the cell to move and stretch, actions that can be detected through force-sensitive proteins at the focal adhesion junction, such as talin (52). The lack of NET formation in neutrophils in which we crosslinked integrin receptors to mimic adhesion conditions provides evidence for the importance of mechanical signaling; however, we were unable to prevent ROS production through inhibition of actin polymerization. It is known that formation of the NADPH complex is facilitated by disrupting the cytoskeleton (26), and therefore it seems likely that this effect overwhelms any evidence for a role of actin polymerization in triggering ROS through mechanical sensing. We did not explore this possibility thoroughly, and further studies would be required before confirming this point. Finally, we emphasize that an important limitation of this study is the specificity of inhibitors. The FAK inhibitor we used is reported to be highly specific (53); however, unexpected off-target effects are possible.

ROS production is a clear candidate linking cell attachment to NET formation. ROS has been implicated in many pathways of NET formation, most commonly through NADPH oxidase activity (54). Indeed, this was the case in our study, with NETs completely dependent on both ROS and NADPH oxidase. There are a large variety of pathways that can activate NADPH oxidase, including many pathways initiated through focal adhesion, and it is well known that neutrophil ROS activity is strongly influenced by adhesion. In this study, we focused on MEK, an enzyme recently reported to be required for PMA-induced NET formation (55), as it is known to be activated downstream of PI3K. This proved to be the case for our study too, with ANCA-induced ROS largely prevented, and NET formation completely blocked by MEK inhibition. Finally, NADPH oxidase ROS production leads to release and activation of granule proteins such as NE, which is well known to be a critical enzyme involved in NET formation (56, 57), and we found was also required in our study.

Inflammation or damage to the endothelium triggers expression of adhesion molecules such as ICAM-1 and deposition of plasma fibronectin (58–60), which provide attachment sites for normally benign neutrophils (61, 62). Integrins, particularly β₂ integrins such as MAC-1 (α_Mβ₂) and LFA-1 (α_Lβ₂), provide attachment points to these sites (63, 64). Both ANCAs and TNF-α are known to increase expression of β₂ integrins on the neutrophil surface (65, 66), and this was the case in our study. It is plausible that a combination of cytokine priming and neutrophil attachment is required to overcome inhibitory mechanisms and result in NETs in response to ANCAs. In support of this, we found that when we blocked β₂ integrins, ANCA-induced NET formation was reduced to nearly background levels. This is consistent with a previous study reporting that β₂ integrin is required for neutrophil respiratory burst in response to ANCAs (67). In a clinical setting, patients with active AAV show increased expression of both β₁ and β₂ integrins (68). Although β₁ integrins have been reported to enhance neutrophil attachment to endothelial cells, specifically α₅β₁ (69) and α₉β₁ (70), in our case we found no effect of blocking these. Likewise, we found no evidence for the involvement of β₃ integrins. We emphasize that our study was limited in examining only the interaction of β₂ integrins with fibronectin. The situation in vivo is considerably more complex, with numerous integrin–ligand binding interactions occurring depending on the local microenvironment. Indeed, we found that attachment to a wide range of surfaces was sufficient to allow NET formation in response to ANCAs.

Overall, our results help to clarify the complex feedback cycle of neutrophil activation in AAV. Low-level neutrophil activation induced by TNF-α and ANCAs results in increased surface β₂ integrins, which promote neutrophil aggregation (65, 66, 71), encouraging physical trapping of aggregates in capillaries. Neutrophil arrest in small capillaries is enhanced by adhesion mediated by endothelial receptor expression and deposition of microparticles and plasma fibronectin (59, 60, 72). Adhesion of neutrophils further promotes activation and endothelial inflammation through degranulation and oxidative burst (36), and in the present study we demonstrate that it is also required for NET formation in the presence of ANCAs. NET formation contributes to this feedback cycle in two ways. NETs contain ANCA Ags and have been shown to enhance their immunogenicity, resulting in increased ANCA production (12). NETs have also been shown to damage endothelial cells (11), which would enhance the deposition of plasma fibronectin and increase the expression of endothelial receptors. Thus, our study postulates a central role of neutrophil adhesion in ANCA pathogenesis and highlights the potential utility of therapies aimed at controlling or preventing pathological neutrophil adhesion.

The authors have no financial conflicts of interest.