Elena Csernok PhD, Oskar-Alexander-Str. 26, 24576 Bad Bramstedt, Germany.
Since proteinase 3 (PR3)-ANCA interact with PR3 on the surface of apoptotic polymorphonuclear neutrophils (PMN) and ingestion of apoptotic PMN is known to modulate macrophage inflammatory reactions, we raised the question whether PR3-ANCA-opsonized apoptotic PMN influence the uptake by macrophages and their state of activation. We therefor analysed the effects of PR3-ANCA-opsonized apoptotic PMN on the uptake process by enzymatic assay. We further investigated the production of TNF-α, IL-10, IL-12 and the secretion of lipid inflammatory mediators (TxB2, leukotriene B4 (LTB4) and prostaglandin E2 (PGE2)) by human monocyte-derived macrophages using FACS and ELISA methods. We show that PMN-opsonization by PR3-ANCA substantially enhances phagocytosis by macrophages and thereby triggers the production of TNF-α and TxB2. These in vitro findings indicate that PR3-ANCA opsonization of apoptotic PMN might be an important mechanism in the pathogenesis of Wegener's granulomatosis (WG), prompting macrophages to produce proinflammatory mediators. These mediators, mainly TNF-α, might prime further PMN leading to perpetuation of the known priming-dependent mechanisms of ANCA action.
Proteinase 3 (PR3)-ANCA are the serological hallmark of Wegener's granulomatosis (WG) [1–3]. Whereas the diagnostic relevance of ANCA is widely accepted, their pathogenic role remains controversial . ANCA may gain access to the intracellular PR3 via the translocation process to the cell surface during activation of polymorphonuclear neutrophils (PMN), e.g. due to TNF-α, or during programmed cell death . Apoptotic PMN are removed from sites of inflammation and the circulation by macrophages, a process leading to active suppression of inflammation . In contrast PMN opsonized, for example, with mouse anti-CD45/rabbit anti-mouse IgG after phagocytosis have been shown to lead to release of TNF-α and other mediators , i.e. to a process enhancing inflammation.
TNF-α is elevated in plasma of patients with active WG and its expression is enhanced in cells in damaged tissue in glomerulonephritis, as well as in peripheral blood mononuclear cells (PBMC) [8–11], indicating a key role for this cytokine in WG. However, so far it is enigmatic how mononuclear cells might be stimulated to enhance TNF-α transcription.
Since ANCA opsonize apoptotic PMN, as demonstrated by Gilligan et al. , and the clearance of apoptotic material is achieved by scavenger macrophages, we investigated the effects of PR3-ANCA opsonization of apoptotic PMN on the clearance by macrophages and their resulting state of activation, including TNF-α production, in order to delineate how this process may lead to enhanced inflammation.
We report that PR3-ANCA opsonization accelerates ingestion by macrophages and leads to the production of TxB2 and TNF-α. This mechanism might represent a link between ANCA and TNF-α induction and therefore most likely is important for the inflammation occurring in vasculitis.
Materials and methods
Sera from 10 patients (PR3-ANCA+, titre 1:1024 to 1:4096) with active untreated WG were chromatographed on protein G-Sepharose (Pharmacia, Freiburg, Germany). The protein concentration in the eluted fractions was evaluated using both chemical (BCA protein assay reagent; Pierce Europe, Heidelberg, Germany) and immunological assays. The IgG concentration was always > 90%. Affinity-purified PR3-ANCA containing IgG fractions were endotoxin-free, as assayed by the kinetic-OLC Limulus amoebocyte cell lysate test (Boehringer Ingelheim, Heidelberg, Germany).
Irrelevant IgG was prepared identically using sera from 10 healthy volunteers (ANCA−).
Human macrophages were prepared from PBMC, as previously described by Haslett et al. . Briefly, PBMC were plated at 4 × 106/ml in 24-well plates for 60 min in RPMI, after which non-adherent cells were washed out with PBS. The macrophages were cultured for 3 days in RPMI medium containing 10% pooled human serum (Sigma, Deisenhofen, Germany). The medium was changed at 3 days. For all experiments, macrophages had been cultured for 8 days before use, yielding 1–2 million macrophages per well.
Induction of apoptosis
Human PMN were used as a source of apoptotic cells. PMN were isolated from the same healthy volunteers as the macrophages used in each experiment as described earlier . Briefly EDTA-anti-coagulated blood was centrifuged in a Ficoll–Paque gradient (Pharmacia, Uppsala, Sweden), erythrocytes were sedimented with polyvinyl alcohol (Merck, Hohenbrunn, Germany) and residual erythrocytes were removed by hypotonic lysis. After washing and resuspension purity was > 98% (Pappaenheim staining) and viability was > 90% (trypan blue exclusion). Apoptosis was induced by cycloheximide (10 μg/ml, 18 h) or by 10 min of UV irradiation at 254 nm and subsequent culture in RPMI plus 10% heat-inactivated fetal calf serum (FCS; Life Technologies, Eggenstein, Germany) at 37°C, 5% CO2 for 3–4 h. At that time, apoptosis (assessed morphologically by light microscopy of stained cytocentrifuged cells and by flow cytometry using AnnexinV–FITC and ToPro3 staining) was between 40% and 60%, whereas necrosis was 1–2%.
Binding of ANCA to apoptotic cells
Binding of affinity-purified ANCA containing IgG to viable/apoptotic PMN was assessed using 10 μg/ml of IgG purified by protein G affinity chromatography (range of concentrations tested 0·1–20 μg/ml). For each experiment, the binding of equivalent amount of purified pooled human IgG to viable/apoptotic cells was tested in parallel. After a 30-min incubation at 4°C, cells were washed twice in PBS containing 1% human serum. As described by Gilligan et al. , binding of ANCA to apoptotic neutrophils was confirmed by indirect immunofluorescence (IIF). As an additional control ANCA-containing IgG was preincubated with purified PR3 (100 μg/ml) for 30 min and also used for IIF.
Phagocytosis of apoptotic cells
Apoptotic PMN preincubated with ANCA, irrelevant IgG or medium alone were added at day 8 to human monocyte-derived macrophages at 5 × 106 cells/well of a 24-well plate in RPMI medium and incubated for 1 h at 37°C/5% CO2. The monolayer was then washed vigorously with ice-cold PBS to remove bound but uningested PMN as was proven by light microscopy. Cells were then lysed by sonication and myeloperoxidase (MPO) as a marker of ingested PMN was detected by K-blue enzymatic assay as described earlier . The macrophages themselves were routinely negative for MPO. In earlier experiments the expected strict correlation between the amount of ingested cells as determined by light microscopy and MPO activity was confirmed (data not shown).
Collection of supernatants
Before use, macrophage monolayers were washed with PBS, then RPMI medium without human serum was added. Apoptotic cells (5 × 106/well) were added for 1 h (yielding a ratio of 3:1 apoptotic PMN:macrophages) and removed by washing with PBS. Macrophages were further cultured in RPMI and supernatants were collected 18 h later. Supernatants were stored in aliquots at −70°C. Cells were collected, washed with PBS and analysed for the presence of intracytoplasmic cytokines by flow cytometry.
Analysis of cytokine: intracytoplasmic cytokine detection by flow cytometry
To increase the quantity of cytokine available for detection within each cell, monensin (2 μmol/l) was added to block the export of newly synthesized cytokine from the Golgi apparatus. Cultured cells were mechanically detached, pooled and washed twice in staining buffer (PBS without Ca2+ and Mg2+, 0·1% bovine serum albumin (BSA), 0·1% sodium azide, pH 7·4), then fixed in fixation buffer (4% formaldehyde in PBS). The cells were washed twice and resuspended in 100 μl permeabilization buffer (PBS without Ca2+ and Mg2+, 0·1% BSA, 0·1% sodium azide, 0·1% saponin, pH 7·4) containing a previously determined optimal concentration (0·25–0·5 μg of fluorochrome-conjugated anti-cytokine antibodies (all obtained from PharMingen, Hamburg, Germany)) for intracellular antigens (IL-10, IL-12, TNF-α) or appropriate isotype controls in 100 μl buffer. Incubation was performed at 4°C for 30 min in the dark. After incubation cells were washed twice in permeabilization buffer. Thoroughly resuspended cells in buffer (1% formaldehyde in PBS) were further analysed by flow cytometry (FACSCalibur; Becton Dickinson, Heidelberg, Germany) using the CellQuest software.
Analysis of cytokine by ELISA
Cytokine concentrations (IL-10 and IL-12) in culture supernatants were determined by ELISA with specific kits (PharMingen). The threshold of the assays for both cytokines was 13 pg/ml.
Analysis of eicosanoids
Prostaglandin E2 (PGE2), leukotriene B4 (LTB4) and TxB2 were quantified by ELISA, using commercial immunoassays (R&D Systems, Hamburg, Germany). Assays were performed according to the instructions provided. The sensitivity of the assays ranged from 8 to 36 pg/ml.
Wilcoxon matched pairs test was used to compare data from the different treatment groups. P < 0·05 was considered statistically significant.
ANCA binding to the surface of apoptotic PMN
As demonstrated earlier by Gilligan et al. , apoptotic PMN incubated with PR3-ANCA-containing IgG fractions showed a surface staining in IIF, in contrast to viable cells, demonstrating that PR3-ANCA binds to the surface of apoptotic but not viable PMN. Only minimal surface staining was seen with irrelevant IgG or PR3-ANCA-containing IgG fractions which were preincubated with PR3. Under the chosen incubation conditions and concentrations > 90% of apoptotic PMN stained positive in indirect immunofluorescence technique (IFT) using PR3-ANCA-containing IgG fractions (data not shown).
PR3-ANCA-opsonized apoptotic PMN are ingested in higher amounts than control cells
As depicted in Fig. 1, MPO activity in macrophage lysates was about twice as high in the ANCA group compared with irrelevant IgG or pure PMN (17·5 U/ml versus 10 U/ml versus 9·2 U/ml). This difference was statistically significant (P = 0·027, n = 6). These data indicate much higher phagocytosis of opsonized versus non-opsonized cell material.
IL-10, IL-12 and TNF-α production by macrophages
In Fig. 2 a scatter plot of the mean fluorescence channels for intracytoplasmic staining of macrophages for IL-10, IL-12 and TNF-α is given (n = 6). There were no significant differences between IgG-treated cells and those cells incubated with pure apoptotic PMN in the production of all three cytokines, nor between ANCA-opsonized cells versus IgG-treated neutrophils for IL-10 and IL-12. The most striking difference was seen for TNF-α, which was significantly up-regulated in macrophages incubated with ANCA-opsonized PMN compared with irrelevant IgG-treated or pure apoptotic PMN (P < 0·05). Macrophages not challenged with apoptotic PMN were routinely negative for all three cytokines, as were macrophages incubated with PR3-ANCA alone.
Release of IL-10, IL-12, LTB4, TxB2 and PGE2
While the IL-10 concentration was below the detection threshold in all macrophage culture supernatants (n = 6), there were detectable amounts of IL-12 in all supernatants but without significant differences between the groups (pure PMN 22·5 ± 11·3 pg/ml; PMN + ANCA 24·3 ± 7·7 pg/ml; PMN + IgG 27·7 ± 20·3 pg/ml). In contrast, there was a significant, about three-fold, rise in TxB2 concentration after incubation with ANCA-opsonized apoptotic cells (Fig. 3). There also was a slight but significant higher TxB2 production in macrophages incubated with pure apoptotic cells when compared with unchallenged macrophages (mean 581 ± 244 versus 294 ± 175 pg/ml, P = 0·02, n = 6).
PGE2 was detectable in appreciable amounts in three out of six culture supernatants from ANCA-opsonized cells challenged macrophages, but in none of the other supernatants. This difference (with ANCA 80 ± 118 pg/ml versus 0 pg/ml in all other supernatants) was not significant.
Macrophages produced detectable amounts oft LTB4 after ingestion of apoptotic PMN with or without treatment. ANCA and IgG opsonization of PMN led to higher LTB4 release (16·9 ± 7·9 pg/ml and 21·1 ± 10·6 pg/ml) than phagocytosis of untreated apoptotic PMN (14·7 ± 7·3 pg/ml), a difference significant only for the IgG group (P = 0·018).
ANCA can bind to the surface of apoptotic PMN . Whether this process might contribute to inflammation was not clear. The data presented here might suggest that phagocytosis of PR3-ANCA-opsonized PMN by macrophages leads to the release of proinflammatory mediators, especially TNF-α, which then might contribute to priming and recruitment of further PMN leading to a cascade in which opsonized primed PMN not only release reactive oxygen species and enzymes [14,15] but also might be phagocytosed by further scavenger macrophages leading to a self-perpetuating inflammation. Since TNF-α is a bifunctional regulator of the inflammatory response, in this scenario low and moderate concentrations might prime PMN, enhance leucocyte recruitment to the inflammatory site, induce expression of granulocyte-macrophage colony-stimulating factor (GM-CSF) and prolong PMN survival, whereas high concentrations might promote further cellular apoptosis . This scenario might explain a major pathway of TNF-α induction in WG, where TNF-α has been demonstrated to be highly expressed in cells infiltrating the tissue, e.g. during active glomerulonephritis .
Clearance of apoptotic material by macrophages is a critical process in the resolution of inflammation. Several recognition mechanisms for non-opsonized apoptotic cells have been described, involving, for example, the CD36/votronectin receptor binding to thrombospondin  and phosphatidylserine-specific receptors . Ingestion via those mechanisms does not lead to inflammation, and Fadok et al. provided evidence that via autocrine/paracrine mechanisms cytokine production is actively inhibited . In contrast, phagocytosis of opsonized particles via Fc receptors induces strong proinflammatory reactions with release of IL-1β, IL-8, IL-10, IL-12, GM-CSF and TNF-α as well as eicosanoids .
Macrophages which have ingested apoptotic Jurkat cells opsonized with anti-phospholipid antibodies possess a higher TNF-α secretion. This observation demonstrates that opsonization of apoptotic cells by autoantibodies and consecutive phagocytosis leading to proinflammatory signals is not restricted to ANCA but might be of importance in other autoimmune diseases, such as the anti-phospholipid syndrome . In systemic lupus erythematosus (SLE) autoantigens are surface-expressed, e.g. on keratinocytes , possibly leading to inflammation via the same mechanisms.
In contrast to other works  using artificial particles such as opsonized zymosan or apoptotic cells opsonized with mouse anti-CD45 and rabbit anti-mouse as secondary antibodies, IL-10 production was not stimulated in our experiments. These differences might be explained by the more physiologic human antibody in contrast to mouse and rabbit antibodies or zymosan used in other works, probably leading to activation of yet unknown different intracellular signalling pathways. The same was found for the eicosanoid pathway: using artificially opsonized material both the cyclooxygenase and the lipoxygenase pathway are activated leading to thromboxane, prostaglandin and leukotriene release . In contrast, with PR3-ANCA we found a significant increase only in thromboxane production, again indicating a different regulation. Beside the possibly of a more physiologic situation in our experiments using human antibodies, another explanation might be a different time kinetics in the mentioned experiments. Another methodological explanation for the differences observed between our data and those of Fadok et al. might be that in their work different numbers of PMN were added to the macrophages, leading to uptake of equal amounts of opsonized and unopsonized cells, whereas in our study macrophages were challenged with a constant PMN number leading to a higher uptake of ANCA-opsonized PMN when compared with non-opsonized PMN. However, since the differences in cell uptake were much lower than those in mediator production, this could not be the only mechanism responsible. Furthermore, assuming that the macrophage reaction only depends on the amount of ingested cells, one would expect equal differences for all mediators, in contrast to our observations.
However, the likely importance for the pathogenesis of WG and possibly other ANCA-associated vasculitis results from the fact that PMN are the most abundant cells not only in circulation but also at sites of inflammation in the special microenvironment of necrotizing vasculitis and granuloma. Furthermore, since the turnover of PMN is much higher than that of lymphocytes, for example, much more apoptotic material is available.
Macrophages are present in lesional tissue, e.g. the typical WG granuloma, as well as in the lung. Furthermore, human mesangial cells, which in many ways behave like macrophages, are also known to phagocytose non-opsonized apoptotic PMN , an important mechanism for clearing renal inflammation not leading to release of proinflammatory mediators. Since they are also able to express Fc receptors , cells capable of performing the mechanism described in this study are present in the most affected localization in WG.
Suggesting that proinflammatory signals play a significant role in the pathogenesis of ANCA-associated vasculitis, results of this study may support new therapeutic anti-TNF-α approaches in WG.
The authors are grateful to Monika Backes and Barbara Hertel for excellent technical assistance.