Systemic Activation of Neutrophils by Immune Complexes Is Critical to IgA Vasculitis Free

Immunoglobulin A vasculitis (IgAV) is an inflammation of the walls of small blood vessels, characterized by deposition of IgA1-dominant immune complexes (ICs) around these vessels. IgAV can be skin limited (1), or a systemic disease involving the gastrointestinal tract and kidneys (2). Some pathomechanisms of vessel and tissue damage in IgAV have been shown to resemble IgG-dominant IC vasculitides (e.g., cryoglobulinemic vasculitis, serum sickness) and IC vasculitides observed in experimental models such as the Arthus reaction. The kidney phenotype is histologically indistinguishable from IgA nephropathy, one of the commonest patterns of glomerulonephritis in the world.

In the skin, IgAV and other IC-mediated vasculitides occur in the postcapillary venules (3–5), notably the site of leukocyte transmigration (6). It has been hypothesized from animal models which develop IgG-containing ICs that deposition of ICs in the vascular wall is the major igniting factor of the inflammatory cascade (7), mainly by activating PMNs (8). In vitro studies support this proposition, as IgG (9–11) and to some extent IgA (12) when fixed in solid phase activate PMNs to undergo oxidative burst, degranulation, and NETosis.

However, perivascular deposition of IgA, IgG, and IgM appears to be a mandatory but not sufficient prerequisite for vasculitis because deposits are also found in clinically and histologically normal skin between flares of vasculitis (4), a phenomenon not sufficiently explained so far.

Thus, in contrast to the common theory (13, 14), we alternatively hypothesized that there are additional systemic factors required to orchestrate initial activation of PMNs for destruction of the postcapillary venules in IgAV. In this context we also wanted to understand why PMNs or their cytotoxic agents exert damaging effects on walls of postcapillary venules despite blood flow, and why it occurs exclusively in vasculitides and not in other causes of inflammation.

With regard to the in vivo mechanisms by which PMNs could injure venules, we focus on NETosis (15, 16). Neutrophil extracellular traps (NETs), endowed with cytotoxic proteins, including histones, elastase, and myeloperoxidases (MPOs), are released by PMNs following crosslinking of cell surface Fcγ and Fcα receptors by ICs (9, 12).

To dissect the vasculitis-specific pathomechanism from pathomechanisms common in general inflammation, we performed comparative analysis of tissues from patients with psoriasis, a PMN-rich nonvasculitic dermatosis, where PMNs also exert prominent effector functions.

We revealed that during flares of IgAV there is transient binding of elevated IgA-IC to PMNs in the circulation and that consequent PMN priming promotes secondary adhesion and marked NETosis in proximity to endothelial cells (ECs), resulting in EC damage in vitro and vessel wall injury in vivo.

This need for preactivation of circulating PMNs to become destructive for vessel walls in IgAV is a novel aspect that not only would explain why deposition of ICs alone does not necessarily result in tissue damage, but that could also widen the concept at which point to tackle IgAV and likely other IC-mediated diseases.

Adult patients presenting with IgAV (biopsy-proven leukocytoclastic vasculitis, deposition of IgA, and clinically palpable purpura on dependent body parts) (1) were enrolled in this local ethical committee–approved study (reference nos. 2015-439-f-S and 2016-243-f-S). All patients and controls were enrolled and treated at the Department of Dermatology (University Hospital Muenster, Muenster, Germany) between 2016 and 2020.

Patients had not received treatment (such as glucocorticoids, immunosuppressants, dapsone, or colchicine) prior to blood sampling and tissue biopsy. Tissue biopsies for lesional and nonlesional skin were taken from the legs. All patients had negative urinalysis and microscopy (no hematuria, proteinuria, erythrocyte casts, or dysmorphic RBCs), and none had other clinical signs of glomerulonephritis.

As controls we chose age-matched patients with psoriasis vulgaris and age-matched clinically healthy subjects (control patients [CTRLs]) who had no other relevant diseases or infections and had not received treatment with anti-inflammatory or immunomodulatory medications prior to blood sampling.

For evaluation of biopsies, analysis of serum, and isolation of PMNs, we included a total of 39 patients with (skin-limited) IgAV (18 males, 21 females), 14 patients with psoriasis vulgaris (9 males, 5 females), and 39 CTRLs (15 males, 24 females), aged between 21 and 87 y. For measurement of serum IgA we included serum samples from 33 additional patients with skin-limited IgAV. From this cohort individuals were randomly selected for collection of PMNs, the respective sample numbers of which are given in the figure legends.

Leg tissue biopsies were obtained from early vasculitic lesions (i.e., partial blanching, slight palpable purpura) and from clinically uninvolved skin from patients with IgAV and from lesional skin of psoriasis patients and normal skin of CTRLs.

Skin sections were processed and stained by immunofluorescence for detection of IgA, vessels, PMNs, complement component 3 (C3) and NETosis as previously described (4, 17).

Primary Abs were neutrophil elastase (mouse anti-human, Dako, no. M0752), IgA (rabbit anti-human, Dako, no. F0188), citrullinated histone 3 (H3Cit) (rabbit anti-human, Abcam, no. ab5103), MPO (rabbit anti-human, Dako, no. A0398), CD15 (rabbit anti-human, DCS, no. 01063CD002), CD31 (mouse anti-human, Thermo Fisher Scientific, no. MA5-13188), C3-PE (rabbit anti-human, Thermo Fisher Scientific, no. PA5-21349), and von Willebrand factor (rabbit anti-human, Thermo Fisher Scientific, no. PA5-81122).

Secondary Abs were Alexa Fluor 488 (goat anti-mouse, Life Technologies, no. A11011) or Alexa Fluor 546 (goat anti-rabbit, Life Technologies, no. A11071) mixed with DAPI (Sigma-Aldrich, no. D9542). Stained sections were mounted with fluorescent mounting medium (Dako, no. S3023), stored at 4°C in the dark, and analyzed and imaged with an inverse microscope (Zeiss Axio Observer).

Peripheral blood was collected in EDTA S-Monovette (Sarstedt, no. 02.1066.001), whole blood was centrifuged at 300 × g for 10 min without brake at room temperature, and plasma was collected and stored at −20°C for further studies. PMNs were isolated from the remaining blood cell pellet by Pancoll density gradient centrifugation. The RBC pellet with the granulocyte-rich layer was resuspended with 3% dextran (Roth, no. 9219.2; ratio 1:2 blood to dextran) for erythrocyte sedimentation. Remaining erythrocytes were lysed with hypotonic solution, and PMNs were resuspended in RPMI 1640 medium supplemented with 3% FCS and 2% l-glutamine and counted using a Countess II automated cell counter (Thermo Fisher Scientific, no. AMQAX1000). By virtue of this varied sequence of steps we obtained high percentages (>90%) of viable PMNs.

PMNs were incubated with heat-aggregated serum, non–heat-aggregated serum, isolated total IgA (82 µg/ml), isolated IgA monomer, dimer, polymer, or PMA (50 nM). IgA was isolated as described previously (18); IgA preparations were free of complement due to affinity chromatography and a molecular mass cutoff. All samples and negative controls were mixed with SYTOX Green (5 µM) in RPMI 1640 medium (3% FCS, 2% l-glutamine, 10% fresh human serum, either heat-inactivated or not heat-inactivated). RPMI 1640 medium and plasmid DNA (500 ng/well) served as controls. Samples were transferred into 96-well black plates, and freshly isolated PMNs were added to each well, except internal controls, at a density of 2 × 10⁵ cells/well. PMNs were stimulated at 37°C with 5% CO₂ in the dark. Release of extracellular DNA was determined by measuring fluorescence from triplicates (excitation wavelength 485 nm, emission wavelength 527 nm after 0, 60, 120, and 180 min).

Isolated PMNs were treated with IgA-ICs (72 µg) immediately followed by measurement of extracellular DNA release. To verify the impact of IgA-ICs, the Fcα receptor CD89 (Abcam, no. Ab53357) was blocked. Briefly, 1 µl of CD89 Ab (per 100,000 cells) was added to IgA-IC (72 µg)–stimulated PMNs, again followed by measurement of extracellular DNA release.

In another series of experiments, isolated CTRL PMNs were prestimulated with either IL-6 or IL-18 for 20 min at concentrations that corresponded to the minimum and maximum serum concentrations measured in IgAV patients (see Cytokine measurement), and then exposed to IgA-ICs to induce NET release.

Freshly isolated PMNs were used at a density of 2.5 × 10⁵ cells/ml. Cytospin application was mounted and funnel filled with 200 µl of cell solution. Cells were centrifuged onto microscope slides at 600 rpm for 3 min and dried for 1 h at room temperature. Slides were fixed with methanol for 30 min at 4°C and used for subsequent staining.

NETosis was induced using isolated PMNs as previously described (19).

HUVECs were grown to confluence in 48-well plates, treated with endothelial growth medium (EGM), lysis buffer, NETs from IgAV patients, or CTRL patients for 24 h at 37°C, 5% CO₂. Cells were harvested by trypsinization and stained with trypan blue for counting of live and dead cells using a Countess II automated cell counter (Thermo Fisher Scientific, no. AMQAX1000).

Image acquisition was undertaken as described (20).

Measurements of IL-6 and IL-18 serum concentrations were performed according to the LEGENDplex human inflammation panel 1 (BioLegend) user manual. The other measured cytokines were IL-1β, IL-8, IL-10, IL-12, IL-17A, IL-23, IL-33, INF-α, INF-γ, MCP-1, TNF-α, ICAM-1, VCAM-1, and VEGF.

HUVECs were isolated as described previously (21).

Perfusion channels (μ-Slide Luer; Ibidi, München, Germany) were coated with gelatin (0.5%) for 30 min, washed with EGM, and left overnight in the incubator (37°C, 5% CO₂) for temperature adaptation. Isolated HUVECs were seeded (3 × 10⁶ cells/ml) in the perfusion channels and left for 2 h in the incubator to allow the cells to adhere. Cells were washed with prewarmed EGM and put on a shaker (37°C, 5% CO₂) until confluent.

Adherent HUVECs were stimulated with TNF-α (20 ng/ml, in culture medium) for 4–6 h at 37°C before use. TNF-α was washed away with EGM. Freshly isolated PMNs, erythrocytes, and ICs were perfused (0.8 dyn/cm²) for 45 min in HEPES buffer until PMNs adhered and NETosis occurred. Slides were washed with EGM to wash out erythrocytes and either treated or not with DNase (500 U/ml in EGM, Sigma-Aldrich, no. 11284932001) and incubated for 24 h under flow in an incubator (37°C, 5% CO₂). Multiple slides were used in each experiment, one for determination of NET release and one for cell death determination. Quantitative evaluation was performed by counting cells per microscope field of view at designated time points.

IgA was isolated as described previously (18).

For this specially developed ELISA, 96-well immunoplates were coated with IgA F(ab′)₂ (Jackson ImmunoResearch Laboratories) for at least 24 h at 4°C. Wells were blocked (2% BSA in PBS) for 1 h at room temperature followed by addition of samples overnight at 4°C. The detection Ab, rabbit anti-human IgG-HRP (Dako; 1:2000 in PBS), was applied and incubated for 90 min at room temperature. The ELISA was developed with o-phenylenediamine dihydrochloride (OPD) substrate (Thermo Fisher Scientific) and measured at 492 nm using an automated plate reader. Between steps, plates were washed with washing buffer (PBS/0.3 M NaCl/0.1% Tween 20).

For all experiments, 8- to 12-wk-old female C57BL/6J wild-type mice were reared in the animal facility of the Institute of Immunology (University of Muenster, Muenster, Germany) or purchased from Harlan Laboratories (Indianapolis, IN). All of the mice were healthy and did not display evidence of disease or infection. Experimental procedures were approved by the State Office for Nature, Environment, and Consumer Protection (North Rhine–Westphalia, Germany). Mice were injected i.v. with BSA (2.5 mg in 50 µl of PBS) in the tail and briefly anesthetized with isoflurane for intradermal injection of anti-BSA (148 µg in 20 µl of PBS) into the left ear and 20 µl of PBS in the right ear for controls. Mice were treated i.v. with either DNase (5000 U/g in PBS, 10 µl/g [Sigma-Aldrich no. 11284932001], 2.5 h after induction) or protein arginine deiminase (PAD) inhibitor (150 µM in PBS, 10 µl [Merck Millipore, no. 506282-10MG], 10 min after induction) or PBS as control.

After these procedures, mice were continuously monitored for 6 h and thereafter at 8, 10, 12, and 24 h. To quantify the extent of vasculitis, the number of petechiae in the left ear were counted. After 24 h, animals were sacrificed with CO₂, whole blood was collected by intracardiac puncture, and ears were collected and frozen immediately in liquid nitrogen or embedded in paraffin.

Statistical analysis was performed using GraphPad Prism 8 software (GraphPad Software, San Diego, CA). Statistical differences were determined using a one-way ANOVA followed by a Dunnett’s test for multiple comparisons (more than two groups) or an unpaired t test, nonparametric Mann–Whitney U test, or two-way ANOVA followed by a Bonferroni test for multiple comparisons (in vivo data). A p value <0.0332 was considered statistically significant.

This study involves human participants and was approved by the Ethik-Kommission der Ärztekammer Westfalen-Lippe und der Westfälischen Wilhelms-Universität Münster (no. 2015-439-f-S and 2016-243-f-S). The involved animal studies were approved by the State Office for Nature, Environment, and Consumer Protection (North Rhine–Westphalia, Germany, no. 84-02042016-A431).

Our immunofluorescence studies demonstrated perivascular IgA-IC deposition always in lesional skin, but also in clinically uninvolved skin, in 33 of 39 IgAV patients (84.6%) (Fig. 1A, 1B), whereas psoriasis patients and CTRLs displayed no IgA-IC deposition (Supplemental Fig. 1A, lower row). IgA deposition was accompanied by perivascular C3 deposition in lesional skin in 9 of 28 (32.1%) patients with IgAV (Fig. 1B). In addition, four patients who were in remission following a cutaneous IgAV flare also demonstrated IgA-positive (and in two patients C3-positive) vessels in nonlesional skin. Notably, we also observed, exclusively in IgAV, that IgA-ICs colocalized with some intravascular PMNs in lesional skin (Fig. 1B).

When attempting to detect PMN NETosis in situ, we detected a marked presence of elastase, MPO, and H3Cit in early vasculitic lesions (clinically characterized by partially blanchable purpuric macules) in 92% of our IgAV patients (Fig. 1C, lesional skin), which clearly colocalized with DNA on extracellular filamentous structures; the latter were partially anchored to walls of dilated blood vessels, some of which showed early fibrinoid necrosis (Fig. 1B). In contrast, these markers were almost absent in unaffected skin in IgAV (Fig. 1C, nonlesional skin, 92% of our patients) and were completely absent in skin of CTRLs (Supplemental Fig. 1A, upper and middle rows), although weakly present in very few areas of psoriasis, but without relationship to blood vessels (Fig. 1C, psoriasis skin). In all cases of IgAV, occurrence of NETs was associated with the presence of PMNs in lesional skin (Fig. 1D).

Our biopsy series confirmed that perivascular IgA-IC deposition alone was insufficient to elicit vasculitis. Because IgA levels are often raised in the acute phase of IgAV and because IgA was bound to some intravascular PMNs in situ, we wondered whether IgA-IC binding to circulating PMNs would be a feature in IgAV and whether this binding was related to the serum concentration of IgA-ICs.

Indeed, we found significantly higher serum concentrations of IgA-ICs comprising IgA/IgG in IgAV (Fig. 2A, *p = 0.04) and significantly higher levels of total serum IgA (Fig. 2B) compared with CTRLs (p = 0.32) and patients with psoriasis (p = 0.05), respectively.

Binding of IgA to the cell surface of several peripheral PMNs was detected in 72.7% (8 of 11) of cytospin preparations from IgAV patients, but only during acute flares of IgAV (Fig. 2C, 2D). In the cytospin preparations of CTRLs (n = 8) no PMNs or a very low percentage of PMNs showed distinct positivity for IgA.

Incubation of CTRL PMNs with heat-aggregated or polymeric IgA resulted in a significantly enhanced release of NET DNA compared with incubation with non–heat-aggregated or monomeric IgA (Fig. 3A, 3B, *p ≤ 0.05). IgA isolated from the serum of IgAV patients instigated a significantly greater release of DNA from PMNs compared with IgA from CTRLs (Fig. 3C, ***p = 0.0008; Supplemental Fig. 1B, *p = 0.02; Supplemental Fig. 1C, *p = 0.03). This NET release could be abolished by blocking FcαR (CD89) with neutralizing Abs, verifying that CD89-mediated binding of IgA to PMNs was the specific stimulus for NETosis (Fig. 3D, **p = 0.001). PMNs from CTRLs exposed and then re-exposed to total IgA displayed increased NETosis in comparison with PMNs not preincubated with total IgA (Fig. 3E, *p = 0.03). In patients with IgAV the overall susceptibility of PMNs for NETosis was increased in comparison with psoriasis patients and CTRLs, as incubation with PMA, a potent general stimulus of NETosis, resulted in significantly higher NET release than that seen in CTRLs (Supplemental Fig. 1D, *p = 0.02).

Notably, PMNs from some IgAV patients exhibited a degree of spontaneous NET formation (Fig. 3F, *p ≤ 0.05), a phenomenon not observed in CTRLs and psoriasis patients.

Because complement activation has been linked to the pathogenesis of vessel damage in IgA nephropathy or antineutrophil cytoplasmic Ab (ANCA)–associated vasculitis and to the activation of PMNs, we determined whether the presence of complement proteins augments IgA- and IgA-IC–induced PMN NETosis. As shown in (Fig. 3G, addition of complement-containing serum yielded no significant difference in NET release compared with heat-inactivated serum, which contains no functional complement proteins (Fig. 3G, p > 0.99).

There were no signs of systemic complement activation in the serum of our IgAV patients as measured by serum C3 and C4 levels and IgA/C3 and IgA/C4 ratios (Supplemental Fig. 1E, 1F). Only one patient had moderately low C3 levels, which were accompanied by normal C4 levels, and one had moderately low C4 levels but normal C3 levels.

Because PMNs from IgAV patients, but to a lesser extent also those from psoriasis patients, exhibited higher NETosis compared with CTRLs, we wondered whether PMNs were additionally being primed by specific cytokines in the circulation. We found that of the several measured cytokines, only IL-6 and IL-18 were significantly elevated in the serum in IgAV (Fig. 3H, ***p ≤ 0.001; (Fig. 3I, *p ≤ 0.05). When we preincubated CTRL PMNs with IL-6 and IL-18 at concentrations similar to those seen in IgAV patients, the PMNs exhibited significantly increased NET release in response to heat-aggregated IgA after IL-6 (Fig. 3J, *p ≤ 0.05), but not after IL-18 (Fig. 3K) exposure.

To better elaborate the consequences of exposure of circulating PMNs to IgA-ICs in IgAV, we studied the interaction of PMNs with heat-aggregated IgA (IgAV, n = 6; CTRLs, n = 6) in a flow system incorporating activated ECs under a shear stress of 0.8 dyne/cm², resembling the vascular microenvironment seen in postcapillary venules in IgAV (22). EC activation, with upregulation of adhesion molecules, is a recognized requirement for development of vasculitis in vivo following perivascular deposition of ICs (4, 5, 23).

With no prestimulation of ECs, rolling of PMNs was only occasionally observed. After stimulation with TNF-α, PMN rolling was enhanced. Addition of IgA induced longer rolling times, associated with enhanced adhesion of PMNs to the EC layer, in each of the experiments (Fig. 4A, *p = 0.01, **p = 0.001, ***p ≤ 0.001). Prior to addition of IgA we observed rolling and persistent adhesion of some PMNs (5–10 PMNs per microscope field of view) to TNF-stimulated ECs, but 15 min after IgA was added their number increased (10–20 PMNs per field of view), predominantly due to prolonged adherence of IgA-IC–binding PMNs to ECs (Fig. 4B) at the expense of a shortened rolling time (1–4 s versus 3–6 s). In contrast, most PMNs without IgA-ICs remained in the circulation After another 30 min, 64.4% (counted per microscope field of view) of the adherent PMNs released single NET strains under shear stress conditions (Fig. 4C, white arrow). Nonadherent PMNs could not be observed to release NET at any time during the experiment (immunofluorescence staining during the flow experiment). Surface IgA-positive PMNs from IgAV patients adhered faster and in higher numbers to ECs than did surface IgA-negative CTRL PMNs (Fig. 4D, 4E), without the need for further exposure to IgA-ICs, and these surface IgA-positive PMNs from IgAV patients also released NETs after 20 min without addition of IgA-ICs to the flow system (Fig. 4F).

The release of NETs resulted in a high rate of endothelial cell death, which was higher with NETs from IgAV patients (Fig. 5A, **p = 0.0038, ***p = 0.0003) than from CTRL patients. The composition of NETs released from IgAV PMNs revealed significantly higher amounts of elastase and MPO compared with NETs from CTRL and psoriasis PMNs (Fig. 5B, *p = 0.012, **p = 0.009; (Fig. 5C, p = 0.005).

Under flow conditions PMNs incubated with IgA started to release increased amounts of NETs after 30 min, and by 24 h the EC layer showed marked damage at sites of NET release in all four independent experiments (Fig. 5D). Treatment with DNase during the time of NET release always abolished these damaging effects (Fig. 5E).

Mice lack the Fcα receptor, and there is no robust murine model of IgAV. PMN NETosis is, however, also induced by IgG binding, and so to demonstrate that IC-mediated NETosis by PMNs is the cause of vessel damage in IC vasculitis in vivo, we studied the IgG-related vasculitis in the reversed passive Arthus reaction.

Two hours after elicitation of the Arthus reaction, ears of PBS-treated mice (n = 14) showed initial characteristic petechiae and hemorrhagic macules, whereas mice injected with DNase (n = 10) developed significantly fewer and smaller lesions (Fig. 6A, ***p ≤ 0.0001).

Injection of a PAD I inhibitor (n = 20), to inhibit citrullination of histone H3 and condensation of histones in PMNs, thus preventing release of NETs, also led to significantly decreased disease scores in comparison with the PBS-treated group (n = 20) at 4 h (Fig. 6B, *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001).

MPO knockout mice (n = 13) also developed significantly less petechiae and hemorrhagic areas than the wild-type mice (n = 13) at 5 and 10 h post elicitation (Fig. 6C, *p ≤ 0.05, **p ≤ 0.01), indicating the relevance of MPO. To verify the release of NETs, we performed immunofluorescence staining with H3Cit and MPO (Fig. 6D, n = 3 per experiment).

In contrast to the common belief that the deposition of IgA-ICs along vascular walls is the one major factor igniting IC vasculitis (3, 7, 11, 13), we have demonstrated that during episodes of active IgAV, circulating PMNs additionally need to be primed by the elevated levels of circulating IgA-ICs and that this priming greatly amplifies the ensuing processes critical for local vessel destruction: an enhanced capacity of PMNs to both adhere to the postcapillary venule wall and to release NETs in proximity to ECs. These processes are specific to IC vasculitides.

We also demonstrated that this priming of PMNs only occurs following exposure to large polymeric IgA molecules present in the serum of IgAV patients and in vitro to heat-aggregated IgA, but not to the far more abundant monomeric IgA. The fact that we were able to abolish these effects by a specific mAb to FcαRI (CD89) further suggests that these effects are mediated through cross-linking of FcαRI (CD89) by polymeric IgA or IgA-ICs.

Priming and subsequent PMN activation did not require the presence of an intact complement system, as we found no significant differences in our experiments when functional complement proteins were included in our in vitro model.

This is remarkable because complement is relevant in IgA nephropathy (24) and ANCA vasculitis (25), and because deposits of C3 and other complement components are seen deposited around some cutaneous vessels in IgAV (reviewed in Ref. 26).

Complement activation may be less essential for IgAV in the skin than in other organs, as 1) we found no signs of complement activation in serum of our patients, 2) complement is not required for the cutaneous murine Arthus reaction (5, 27), and 3) C3 was less regularly detected around vessels in skin-limited IgAV (∼30%) than in kidney biopsies in IgA nephropathy (>90%) (28; J. Barratt, unpublished data). Complement activation may be essential for damage of the kidney or other affected organs in systemic IgAV other than in skin-limited IgAV (26). The tissue-specific microenvironment, combined with genetic differences in complement genes (24), may offer one clue as to why in the spectrum of IgA-IC diseases we see either isolated IgA nephropathy, skin-limited IgAV (our cohort), or both.

It is accepted that polymeric IgA and denatured IgA, but not monomeric IgA, are capable of binding the lectin pathway carbohydrate recognition molecule, MBL (29; reviewed in Ref. 24), and IgA aggregates have been reported in vitro to activate alternative pathway of complement (30). Our data clearly show that complement is not, however, relevant to IgA-primed NETosis by PMNs, both by native IgA isolated from patients and by heat-aggregated IgA.

In addition, PMNs from patients with IgAV exhibited spontaneous NETosis in static in vitro assays, and these PMNs responded to PMA stimulation with a faster release and greater amount of NETs than PMNs from CTRLs or psoriasis patients. Although PMNs in psoriasis displayed a slightly increased release of NETs compared with CTRLs, there was no spontaneous NET release. The slightly raised NETosis is likely due to prestimulation of PMNS by IL-6, which is elevated in both IgAV and psoriasis (31) and which primed PMNs in vitro to significantly increase NETosis compared with IgA alone. Although IL-18 has been shown previously to induce NETosis (32), we detected no augmenting or priming effect in our experiments. The molecular basis for a potential modulatory effect of IL-18 on NETosis has not been elucidated (32, 33). IL-6 has been shown to induce NETosis (34), whereas in our study it stood out more as a primer for NETosis by PMNs. Such a priming effect could have been suspected from the observation that increased IL-6 levels were associated with increased NETosis rates in, for example, COVID-19, and that IL-6 expression coincides with increased expression of genes functional in NETosis (35).

The lower threshold to undergo NETosis in the absence of an external stimulus and the increased release of NETs after PMA stimulation have only been described in a very limited number of other settings, that is, one study each in ANCA-associated vasculitis (36–38) and systemic lupus erythematosus (39), and more recently in the context of SARS-CoV-2 infection (40). Thus, although a moderate preactivation of circulating PMNs by cytokines is a conceivable scenario in some systemic inflammatory diseases, our data show that in IgAV it is exposure to IgA-ICs that specifically drives the extraordinary PMNs priming for vessel damage. Supporting this, when PMNs from CTRLs were exposed to complexed (but not monomeric) IgA, they displayed increased NETosis, adopting an “IgAV-like” phenotype. Although these in vitro and in vivo IgA-IC–specific effects on PMN function and capacity for NETosis appear to be specific for IgAV, it is conceivable other ICs exert similar effects through cross-linking of PMN Fcγ receptors.

The site of vessel damage in IC vasculitis commonly corresponds to the site of leukocyte transmigration (5, 6, 23), where damaging events start at the luminal aspect of the vessel (23) and thus at a location where cytotoxic reagents under physiological circumstance would rapidly become cleared by the bloodstream. To investigate whether IgA-IC–induced PMNs could cause endothelial injury we mirrored the in vivo blood flow conditions in postcapillary venules (22) and confirmed that binding of IgA-ICs to circulating PMNs promotes and augments PMN adherence to ECs. Adherence of PMNs was a prerequisite for generation of a maximal (or spontaneous) NETosis response. Binding of IgA-ICs to circulating PMNs without adherence to ECs was not sufficient to induce meaningful PMN-NETosis, explaining why NETosis is not observed in the circulation in vivo in IgAV or other systemic diseases. In IgAV for marked NETosis to occur, PMNs require both exposure to IgA-ICs in the circulation and adherence to ECs, as has similarly been observed for the oxidative burst and degranulation (5, 11). Data from previous reports of NET-induced damage of ECs (41–45) relate mostly to static in vitro models and do not take into particular consideration that cytotoxic agents would be expected to flow away. In our perfusion system, instead of flowing away, NETs colocalized spatially and temporally with sites of loss of integrity in the EC layer. This effect was abrogated, in both our perfusion system and in a separate study (46), with DNase treatment. The spatial and temporal colocalization of strands of NETosis with ECs reflects the processes we visualized in vivo in lesional skin of IgAV patients where NET proteins were located on the luminal side of postcapillary venules and associated with damaged blood vessels, especially in incipient lesions of IgAV, as has been shown in mouse models of IgG-IC vasculitis (10, 47).

In psoriasis, we observed only weak NETosis, and it was localized in the epidermis rather than with vessels (48). Spatial localization of NETosis in vivo in disease requires careful evaluation and use of control tissue. Demonstration of filamentous NETs has technical limitations in tissues with high cellular density and has only been successful in a few studies where its association with destruction of vessels has but convincingly been shown in glomerular capillaries in lupus nephritis and ANCA-associated vasculitis (15, 39, 49). In this study, we were able to directly demonstrate involvement of NETosis in the skin in IgAV.

By neutralizing components of NETosis in vivo using DNAse treatment or a PAD inhibitor (50) we showed that NETs are critical contributors to vessel damage in IC vasculitis. The observation that the Arthus reaction was only delayed, but not abolished, in MPO-deficient mice supported the contribution of MPO, but also showed that more components than MPO are involved in the process of vessel damage. In this way, we confirmed that NETs are significant factors in mediating vessel damage in IC-mediated vasculitis in vivo.

We suggest the following paradigm as an explanation for the development of cutaneous, and possibly systemic, IgAV lesions: 1) intermittently raised levels of circulating IgA-ICs (possibly due to concurrent infection or drug intake) lead to 2) binding of IgA-ICs to PMNs in the circulation, which results in 3) PMN prestimulation, lowering the threshold for NETosis, but not eliciting NETosis yet without adhesion of PMNs; 4) additional, albeit minor, PMN prestimulation also occurs through exposure to cytokines (e.g., IL-6), 5) perivascular deposition of IgA-ICs, and 6) activation of ECs, with expression of adhesion molecules that results in 7) IgA-IC bearing PMNs firmly adhering to ECs, resulting in 8) complete PMN activation and marked release of NETs, which 9) anchor to the luminal side of the EC layer without being cleared by the bloodstream, and 10) cause destruction of the postcapillary venule walls. Without priming of PMNs by transiently elevated circulating IgA-ICs, already deposited perivascular IgA will not lead to vasculitic inflammation (as observed after flares of IgAV). In contrast to IgA nephropathy, complement appears to play a minor role in the pathomechanism of cutaneous vessel damage.

Our new concept for the evolution of vessel wall injury in IgA vasculitis would explain why IgA-ICs are found around blood vessels (and likely also in kidneys and other organs) without causing tissue damage. As such, a potential therapeutic goal in IC-mediated vasculitides and IC-mediated diseases would be to prevent PMN priming and thereby the ensuing PMN-mediated vessel damage through NETosis.

We are grateful to the Study Group of Immunology (University of Muenster, Muenster, Germany) for advice, technical assistance, and expert help with experiments. We thank J.S. Bhachu (Department of Cardiovascular Sciences, University of Leicester, Leicester, U.K.) for the help to isolate IgA from serum samples of patients. We are thankful for all our patients and healthy subjects taking part in this study.

The authors have no financial conflicts of interest.