Neutrophils undergo constitutive death by apoptosis, leading to safe nonphlogistic phagocytosis and clearance by macrophages. Recent work has shown that before secondary necrosis, neutrophils exhibiting classical features of apoptosis can progress to a morphologically defined late apoptotic state. However, whether such neutrophils could be safely cleared was unknown. We now report that human late apoptotic neutrophils could be purified from cultured neutrophil populations undergoing constitutive death and were subsequently ingested by human monocyte-derived macrophages by serum-independent mechanisms that did not trigger the release of IL-8 or TNF-α. Such ingestion was specifically inhibited by Abs to thrombospondin-1 and the αvβ3 vitronectin receptor. Murine bone marrow-derived macrophage phagocytosis of late and early apoptotic neutrophils occurred by similar mechanisms, proceeding with the same efficiency as that observed for wild-type controls when macrophages from αm−/− or β2−/− mice were used. We conclude that specific nonphlogistic, β2 integrin-independent mechanisms involving thrombospondin-1 and αvβ3 allow macrophages to ingest late apoptotic neutrophils without eliciting inflammatory cytokine secretion.
Neutrophils and their toxic contents are vital for host defense, but may also mediate undesirable tissue injury in a wide range of inflammatory diseases (1, 2). However, a growing body of evidence indicates that neutrophils can be safely eliminated from inflamed tissues, promoting resolution of the inflammatory response, by constitutively undergoing apoptosis (3, 4, 5, 6, 7, 8, 9). This is a programmed form of cell death that leads to swift phagocyte recognition, uptake, and degradation of intact senescent neutrophils, preventing leakage of noxious contents from the dying cells and failing to elicit proinflammatory responses from the ingesting phagocyte (10, 11, 12, 13). Consequently, safe phagocytic clearance of neutrophils undergoing constitutive apoptosis is viewed as a key control point in the inflammatory response.
However, the molecular mechanisms mediating safe phagocytic clearance of apoptotic cells remain poorly understood; an increasing number of phagocyte receptors (see Refs. 14, 15, 16 for reviews) have been implicated in vitro. This complexity may reflect the fact that studies have frequently involved administration to phagocytes of “meals” consisting of heterogeneous populations containing cells at various stages of the death program, including secondary necrosis (17). By contrast, human neutrophils undergoing constitutive apoptosis during overnight culture contain a mixture of histologically normal neutrophils (not ingested by phagocytes) and intact cells that exclude trypan blue and propidium iodide (PI)3 and exhibit classical morphologic features of early apoptosis (3). These include nuclear coalescence and chromatin condensation accompanied by well-defined surface changes such as the capacity to bind annexin V (18, 19, 20). By separating intact apoptotic neutrophils from histologically normal neutrophils within the same population of aging cells, we confirmed that phagocytes recognized only the apoptotic neutrophils (3). Nevertheless, even this apparently “clean” model system has proved more complex than was first thought. Beyond 18 h in culture, a steadily increasing proportion of senescent neutrophils exhibit a characteristic late apoptotic morphology in which nuclear degradation or so-called evanescence is accompanied by electron microscopic evidence of limited granule fusion with the plasma membrane (21, 22). To avoid confusion, we propose that neutrophils with classical features of apoptosis (3) should be regarded as early apoptotic cells.
We have become interested in whether late apoptotic neutrophils are recognized by macrophages (Mφs) and, if so, the molecular mechanisms and consequences of this event. A number of different types of phagocytes can deploy the phagocyte surface αvβ3 vitronectin receptor integrin to present bridging thrombospondin 1 (TSP1) to apoptotic cells and promote phagocytosis without inciting proinflammatory secretory responses (10, 11, 23, 24). This mechanism was a strong candidate for Mφ recognition of late apoptotic neutrophils, because 1) these cells exhibit limited fusion with the plasma membrane of granules containing proteins capable of binding TSP1 (21, 22, 25); 2) a very recent report indicates that late apoptotic neutrophils bind TSP1 with such efficiency that this can be demonstrated with soluble biotinylated TSP1 (19); and 3) myeloid dendritic cells bind late apoptotic cells via an αv-mediated mechanism (26). However, equally strong candidate phagocyte receptors were those of the β2 integrin family that bind opsonic complement fragments, type 3 complement receptor (CR3; αmβ2 or CD11b/CD18) and CR4 (αxβ2 or CD11c/CD18). Not only does ligation of these receptors fail to stimulate the release of inflammatory mediators from Mφs (27, 28), but there is also evidence that they can bind, via opsonic complement fragments, populations of dying cells (17, 29). Furthermore, β2 integrins can also bind denatured proteins (30), which might be exposed by late apoptotic cells.
Therefore, in this study we set out to determine whether it was possible to purify late apoptotic neutrophils from cultured neutrophil populations undergoing constitutive death, whether such cells were nonphlogistically ingested by Mφs, and whether either Mφ TSP1/αvβ3 or β2 integrins mediated phagocytosis of late apoptotic neutrophils.
Materials and Methods
Tissue culture materials
All chemicals were purchased from Sigma (St. Louis, MO) unless otherwise indicated. Culture media (HBSS, IMEM, and DMEM) and supplements (100 U/ml penicillin, 100 μg/ml streptomycin, 2 mM glutamine, and 10% FCS) were obtained from Life Technologies (Grand Island, NY). Sterile tissue culture plasticware was purchased from Falcon Plastics (Cockeysville, MD).
Abs and peptides
mAb to the human vitronectin receptor integrin (31) was the αVβ3-specific mAb 23C6 (IgG1; provided by Prof. M. Horton, St. Bartholomew’s Medical School, London, U.K.). V. Dixit (University of Michigan, Ann Arbor, MI) provided mTSP1, a rabbit affinity-purified polyclonal Ab specific for mouse TSP1 (32), and A6.1, an IgG1 murine mAb for human TSP1 (33). H9.2B8, a hamster mAb specific for the mouse integrin αv chain (34) was obtained from PharMingen (San Diego, CA). The control mAb was mAb OX-7 (from Serotec, Banbury, Oxon, U.K.), an IgG1 that recognized Thy 1.1. The tetrapeptides Arg-Gly-Asp-Ser (RGDS) and Arg-Gly-Glu-Ser (RGES) were obtained from Sigma.
Neutrophils were isolated from fresh citrated normal human blood by dextran sedimentation and plasma-Percoll discontinuous density gradient centrifugation and were “aged” in tissue culture in IMEM with 10% autologous platelet-rich plasma-derived serum (PRPDS) so as to undergo apoptosis, exactly as previously described (3). Late apoptotic neutrophils were purified from 22-h aged polymorphonuclear neutrophil granulocyte (PMN) by discontinuous plasma-Percoll density gradient centrifugation; as described later in this report, >95% pure late apoptotic cells were isolated at the interface between platelet-poor plasma (PPP) and 31% Percoll in PPP, whereas a mixed population of late apoptotic and early apoptotic cells was found at the interface of 31 and 42% Percoll in PPP; the latter was not used in phagocytosis assays. A mixed population of early apoptotic and nonapoptotic neutrophils free of late apoptotic cells was retrieved at the 42–51% Percoll in PPP interface; this last population was used as early apoptotic neutrophils in studies of recognition by Mφ (see below), as all our previous studies have used similar mixed populations that routinely arise after overnight aging in culture, because such cells are >99% viable by trypan blue exclusion (3, 6, 23, 24). Gradients were centrifuged at 610 × g for 24 min at 4°C. Human monocytes (>90% pure) were prepared by counterflow centrifugation as described previously (35) and cultured for 5 days in IMEM with 10% autologous prpds to mature into Mφ as previously described (35).
Isolation and culture of mouse Mφ
Bone marrow was harvested from BALB/c mice and plated in 96-well plates in DMEM containing 10% FCS and 10% L929 cell-conditioned medium as a source of M-CSF. Bone marrow Mφ were used in phagocytosis assays after 7 days of growth in culture as previously described (34).
In some experiments Mφ were prepared as described above from gene-targeted mice and wild-type strain controls as follows: αm (CD11b)−/− BALB/c mice and BALB/c controls were used as previously described (36); β2 (CD18) null mice were crossed from 129sv/eg onto the C57BL/6J background, and F2 CD18−/− animals were used as founders for the CD18−/− mice used in the current experiments, with control mice bred from 129sv/eg and C57BL/6J in parallel (37).
Assessment of PMN morphology and flow cytometric analysis
Early apoptotic and late apoptotic PMN were assessed by microscopic examination of cytocentrifuge preparations fixed in methanol and stained with May-Giemsa or by transmission electron microscopy as previously described (3). Binding of FITC-conjugated annexin V in ice-cold PBS containing calcium and magnesium as previously described (18, 20) was used to assess exposure of phosphatidylserine exposure. PI was used to assess plasma membrane permeability. Cells were exposed to PI at 1 μg/ml for 120 s immediately before flow cytometric analysis; a positive control was provided by brief heating of 22-h aged neutrophil populations to 100°C. Labeled cells were applied to a Becton Dickinson FACScan flow cytometer (Mountain View, CA) that automatically and simultaneously measured the fluorescence of individual cells identified by their size-dependent light-scattering properties.
A coded, observer-blind, microscopically quantified phagocytic assay of Mφ ingestion of apoptotic PMNs, which has been extensively described, illustrated, and validated (3, 23, 24, 38), was used in these studies. Apoptotic and late apoptotic PMNs prepared as described above from a single population of 22-h aged PMNs were washed once in HBSS and suspended in IMEM, and 0.5 × 106 PMN in 50 μl of medium were added to each washed well of Mφ cultured in 96-well plates. After interaction for 30 min at 37°C in 5% CO2, the wells were washed in cold (4°C) 0.9% saline to remove noningested PMNs, and then the Mφ monolayer was fixed in 2% glutaraldehyde in saline for 2 min and stained for myeloperoxidase (MPO), and the proportion of Mφ-ingesting PMNs was counted by inverted light microscopy, exactly as previously described (3, 23, 24, 38). Because of a tendency of nonapoptotic neutrophils to adhere to mouse bone marrow Mφ after 30-min interaction, mouse Mφ were then trypsinized, and a separate cytocentrifuge preparation was prepared for each well as previously described (11, 35). These were fixed with 2% glutaraldehyde, stained for MPO, and finally counterstained with Hemalum (BDH, Poole, U.K.). The proportion of mouse Mφ-containing, brown-staining, MPO-positive PMNs was then counted.
Effects of serum
To determine whether there were effects of serum on the phagocytosis of early apoptotic and late apoptotic PMN, such PMNs were interacted with human monocyte-derived Mφ in the presence of 15% normal nonheated autologous human PRPDS or serum prepared in glass and 15% normal heated (56°C for 30 min) autologous human PRPDS or serum from glass.
Effects of Abs
These were determined as previously described (23, 24, 34, 38). Mφ in 96-well plates were washed, and 50 μl of Ab at the desired concentration in IMEM was added to each well. The plates were incubated for 15 min at 4°C, followed by addition of 0.5 × 106 PMN in 10 μl of IMEM at 37°C, and then interacted for 30 min under standard conditions.
Effects of soluble inhibitors
Various inhibitors were included in the interaction medium. The tetrapeptide RGDS and the control peptide RGES were made up in IMEM before being added to the interaction medium to achieve the desired concentration (1 mM).
Assessment of Mφ response to ingestion of late apoptotic cells
Human Mφ were washed with HBSS, then incubated with apoptotic PMN, late apoptotic PMN, and control particles in IMEM for 30 min. The noningested apoptotic PMN and control particles were washed away, and IMEM was added for 16 h. Supernatants were collected at this time point because cytokine secretion stimulated by opsonized zymosan was maximal (13). Supernatants were centrifuged at 6000 rpm for 3 min to remove particulate debris. Cytokine concentrations in the culture supernatants were assayed by ELISA for TNF-α and IL-8 using specific assays as previously described (11).
Purification of late apoptotic neutrophils
In keeping with the studies by Hébert et al. (21), we observed increasing proportions of apparently anucleate late apoptotic neutrophils after 18 h of culture of normal human peripheral blood neutrophils (Figs. 1 and 2). We reasoned that it might be possible to purify late apoptotic neutrophils by density centrifugation on discontinuous Percoll-plasma gradients. From neutrophil populations aged in vitro for 22 h, we were able to obtain a fraction from the 0/31% Percoll interface containing >95% pure late apoptotic cells (Fig. 3,A) with typical nuclear evanescence (21) and other characteristic features confirmed by electron microscopy (Fig. 3,B) and a distinct fraction of aged neutrophils from the 42/51% Percoll interface displaying a mix of morphologically normal and early apoptotic neutrophils (Fig. 3,C) closely similar to the overnight-aged neutrophil populations used in our previous studies of phagocyte recognition of early apoptotic neutrophils (23, 24, 34, 38). Typically around 50% of cells in this fraction displayed early apoptotic morphology (Fig. 3,C; see also Fig. 4 B).
When the phenotype of purified late apoptotic neutrophils from the 0/31% Percoll interface was further examined by flow cytometry, nonfixed cells stained with FITC-labeled annexin V, indicating exposure of phosphatidylserine (Fig. 4,A) to a degree similar to that exhibited by the 50% of early apoptotic cells in the fraction from the 42/51% Percoll interface (Fig. 4,B, lower right quadrant). The late apoptotic neutrophil population exhibited slightly greater permeability to PI than early apoptotic cells (compare Fig. 4, A and B), although much less than that exhibited by heat-permeabilized 22-h aged neutrophils (Fig. 4 C). These data indicated that late apoptotic neutrophils exhibited some increase in plasma membrane permeability despite being morphologically intact. However, the late apoptotic neutrophil fraction isolated from 22-h aged neutrophil populations did not begin to undergo secondary necrosis, as evidenced by detectable spontaneous release of the granule constituent MPO, until recultured under standard conditions for at least 4 h (data not shown).
Human monocyte-derived Mφ phagocytosis of late apoptotic cells is nonphlogistic
Previously we (10, 11, 39) and others (13) reported that uptake of apoptotic cells did not trigger release of proinflammatory cytokines from Mφs and other phagocytes. However, Mφ did release such cytokines when ingesting debris from granulocytes that had undergone secondary necrosis after constitutive apoptosis (39). Consequently, it was important to determine the response made by Mφ taking up late apoptotic neutrophils. Following a 30-min interaction with late apoptotic neutrophils as described in Materials and Methods, 40.2 ± 1.8% (mean ± SE; n = 6) of human monocyte-derived Mφ ingested late apoptotic neutrophils. However, by contrast with uptake of opsonized zymosan, there was no release of the proinflammatory cytokines, IL-8 or TNF-α, indicating that uptake of late apoptotic neutrophils was also nonphlogistic (Table I).
|Stimulus .||TNF-α (ng/ml) .||IL-8 (ng/ml) .|
|Control||2.60 ± 0.36||8.33 ± 1.35|
|Opsonised zymosan||589.29 ± 1.32*||62.75 ± 9.42*|
|Early apoptotic PMN||2.81 ± 1.49||4.81 ± 0.99|
|Late apoptotic PMN||1.61 ± 0.61||5.90 ± 1.45|
|Stimulus .||TNF-α (ng/ml) .||IL-8 (ng/ml) .|
|Control||2.60 ± 0.36||8.33 ± 1.35|
|Opsonised zymosan||589.29 ± 1.32*||62.75 ± 9.42*|
|Early apoptotic PMN||2.81 ± 1.49||4.81 ± 0.99|
|Late apoptotic PMN||1.61 ± 0.61||5.90 ± 1.45|
Note that phagocytosis of neither early apoptotic PMN by 38.3 ± 3.0% of Mφ nor phagocytosis of late apoptotic PMN by 48.4 ± 1.9% of Mφ stimulated cytokine release, by contrast with positive control stimulus of opsonised zymosan. Date are mean ± SE, n = 4; the only significant differences from control observed were in cytokine release after opsonized zymosan (*, p < 0.001).
Human Mφ phagocytosis of late apoptotic cells is not modulated by serum, but is inhibited by Abs to TSP1 and αvβ3
Although our previous studies have routinely used human neutrophils and Mφ cultured in autologous serum obtained by recalcifying platelet-rich plasma (PRPDS), we have routinely washed both cell types before interaction in the absence of added serum (3, 23, 24, 35, 38). A recent report, using an assay of aged neutrophil interaction with human Mφ in which much of the interaction signal appeared to be tethering of apoptotic cells rather than phagocytosis, suggested that interaction could be markedly enhanced by the presence of up to 15% serum as a source of complement (29). Nevertheless, the inclusion of 15% autologous serum prepared in glass did not enhance human monocyte-derived phagocytosis of either late apoptotic or early apoptotic neutrophil fractions in our assay (Fig. 5). Furthermore, the use of 15% PRPDS also failed to enhance phagocytosis of each cell type (data not shown). By contrast, mAbs (but control mAb) to TSP1 and αvβ3 significantly inhibited human monocyte-derived Mφ phagocytosis of both late and early apoptotic neutrophil fractions (Fig. 6), but not uptake of opsonized RBC used as controls (data not shown for clarity).
Murine bone marrow-derived Mφ phagocytosis of late apoptotic neutrophils is also inhibited by Abs to TSP1 and αv
Our previous work suggested that murine bone marrow-derived Mφ use mechanisms similar to those of human monocyte-derived Mφ in uptake of human early apoptotic neutrophils; inhibition by RGDS peptide and anti-murine αv mAb is demonstrable (34). Not only did these reagents inhibit uptake of late apoptotic neutrophils by murine bone marrow-derived Mφ (but not control RGES peptide or control mAb; Fig. 7), but inhibition was observed with an affinity-purified rabbit Ab to TSP1 (but not by control rabbit IgG). No reagent used inhibited uptake of opsonized erythrocytes (data not shown for clarity).
β2 integrins are not necessary for efficient Mφ uptake of early and late apoptotic neutrophils
Because ligation of Mφ CR3 and CR3-mediated uptake of particles have been reported not to activate proinflammatory responses from Mφ (27, 28), we considered it important to examine the role of α2 integrins in Mφ uptake of early and late apoptotic neutrophils, particularly the αmβ2/CD11b CD18 CR3 integrin. Consequently, we sought definitive evidence of a requirement for αmβ2 or other β2 integrins by studying Mφ from gene-targeted mice. In experiments using bone marrow-derived Mφ from αm−/− mice and wild-type controls, no difference was observed in the proportion of Mφ ingesting late apoptotic neutrophils; these were highly efficiently ingested by 71.8 ± 5.3% of αm−/− Mφ and 66.7 ± 5.9% of wild-type Mφ (mean ± SE; n = 3; i.e., three mice in eachgroup; experiments were also performed in triplicate in Mφ from each mouse). These findings were confirmed in a more extensive series of experiments using bone marrow-derived Mφ from β2−/− and wild-type control mice; there was no difference in the proportion of Mφ ingesting either apoptotic or late apoptotic neutrophils (Fig. 8). Indeed, although recognition of late apoptotic neutrophils was generally slightly greater in absolute degree than uptake of early apoptotic cells in each set of experiments, perhaps reflecting unavoidable dilution of early apoptotic cells by copurified nonapoptotic neutrophils, the mechanism of recognition by bone marrow-derived Mφ from wild-type or β2−/− mice appeared very similar despite variation in the baseline degree of phagocytosis between sets of experiments consistent with previous experience (11, 34). Thus, specific inhibition of bone marrow-derived Mφ uptake of late apoptotic neutrophils was observed with RGDS peptide at 1 mM, but not control RGES peptide, and was seen also with Ab to murine vitronectin receptor (αv), but not control Ab (Fig. 8).
Although previous work (21) has demonstrated that neutrophils undergoing constitutive death by apoptosis can pass from the early apoptotic state (as defined in our earlier studies (3)) to a late apoptotic state for some hours before finally undergoing secondary necrosis, it has been unknown whether late apoptotic neutrophils can be safely cleared by Mφs. In this report we have demonstrated that fractions of morphologically defined late apoptotic neutrophils can be purified from populations of senescent human neutrophils. Furthermore, in keeping with previous data obtained with populations of ‘early’ apoptotic neutrophils (10, 11), late apoptotic neutrophils were efficiently ingested by human monocyte-derived Mφ without triggering release of the pro-inflammatory mediators IL-8 and TNF-α. By contrast with a previous report (29), the proportion of human monocyte-derived Mφ ingesting early apoptotic neutrophils was not increased by the presence of serum and no increase in uptake of late apoptotic neutrophils was observed. These data argued against a role for Mφ β2 integrins in uptake of dying neutrophils. However, the uptake of late apoptotic neutrophils by both human monocyte-derived Mφ and murine bone marrow-derived Mφ was inhibited by Abs to TSP1 and the vitronectin receptor, as was ingestion of early apoptotic neutrophils. Furthermore, strong evidence against a requirement for β2 integrins in efficient Mφ phagocytosis of either early or late apoptotic neutrophils was provided by the failure of Mφs from αm−/− and β2−/− knockout mice to exhibit any defect in phagocytosis of either target.
Our first major conclusion from these data is that progression beyond the (early) apoptotic state originally defined in our studies of the constitutive death of cultured neutrophils (3) does not deny senescent neutrophils the opportunity for safe clearance while in the late apoptotic state before potentially deleterious secondary necrosis. Although our assessment of the proinflammatory secretory response from Mφs taking up late apoptotic neutrophils was limited to assay of Mφ release of the key chemokine IL-8 and the master inflammatory mediator TNF-α, previous studies (11, 12, 13) indicate that other classes of inflammatory mediator are unlikely to be released by Mφ under these circumstances. The current data re-emphasize the tissue-protective potential of neutrophil clearance by apoptosis, demonstrating that even cells reaching what might be termed the “last chapter” of constitutive death can be cleared safely.
However, our data also indicate that it may prove difficult to selectively define the potential for tissue protection afforded by specific clearance of late apoptotic neutrophils in vivo. This is because Mφ phagocytosis of both early and late apoptotic neutrophils appears likely to involve similar mechanisms in which TSP1 and the vitronectin receptor play a major role. Thus, it may prove difficult to inhibit uptake of early apoptotic cells selectively, because this would be necessary for a formal assessment of the capacity of Mφ phagocytosis of late apoptotic neutrophils to serve as a last line of tissue defense in inflammation. Nevertheless, despite a growing understanding of molecular mechanisms mediating phagocytic clearance, we still know very little of those mechanisms that predominate in the clearance of apoptotic cells from various sites in the living mammal. Future work should take into account the possibility that clearance mechanisms dedicated to removal of late apoptotic cells could exist. Some support for this possibility can be drawn from the observation by Rubartelli et al. that myeloid dendritic cells may use αv integrins in selective ingestion of late apoptotic cells (26).
The second major finding of our study also emphasizes that much remains to be learned about mechanisms mediating Mφ ingestion of cells dying by apoptosis. The data clearly demonstrate that expression of β2 integrins is not necessary for efficient phagocytosis of either early or late human apoptotic neutrophils by murine bone marrow-derived Mφs (which, nevertheless, appear to use TSP1/vitronectin receptor-mediated mechanisms similar to those exhibited by human monocyte-derived Mφ). The current data are in keeping with our earlier work (3, 38) in an assay system in which neutrophils underwent apoptosis in the presence of autologous PRPDS, a source of complement that, according to a recent report, apoptotic cells may activate so that they become coated with opsonic complement fragments (29). Nevertheless, when such cells were interacted in the absence of added serum with human monocyte-derived Mφ, no defect in phagocytosis of apoptotic neutrophils was observed despite functionally validated Ab blockade of Mφ, CR1, CR3, and CR4 receptors (38), nor was any obvious defect in phagocytosis exhibited by monocyte-derived Mφ prepared from a patient with severe congenital β2 deficiency (40). However, our findings must be set against 1) growing evidence that the first component of complement, C1q, could bridge apoptotic cells to phagocytes in a manner similar to that proposed for TSP1 (41, 42), and 2) compelling data suggesting that Mφ receptors for opsonic complement fragments could be important in amplifying efficient phagocytosis of dying cells (17, 29). Nevertheless, it is notable that these latter reports have either used mixed populations of dying cells, including cells in secondary necrosis (17), or have used assays of interaction with Mφ that include a large tethering element (29) rather than our own extensively validated assay of completed phagocytosis. Further studies will be needed to resolve the importance of complement components in removal of neutrophils at various stages of the apoptotic death program.
To conclude, our studies demonstrate that late apoptotic neutrophils can be ingested by Mφs via specific mechanisms uncoupled from secretory proinflammatory responses. Such mechanisms can operate efficiently in the absence of Mφ β2 integrins, however, emphasizing the need for further characterization of the role of complement components in the safe clearance of cells dying by apoptosis.
The following are thanked for helpful discussions and/or the provision of reagents Dr. Eric Brown (Washington University, St. Louis, MO), Drs. Vishva Dixit and Karen O’Rourke (University of Michigan, Ann Arbor, MI), Prof. Mike Horton (University College London Medical School, London, U.K.), and Dr. Alex Law (Medical Research Council Immunochemistry Unit, Oxford, U.K.). Carolyn Gilchrist provided expert secretarial assistance.
This work was supported by the Wellcome Trust (Grant 047273). Y.R. was a Dorothy Hodgkin Fellow of the Royal Society, and F.L. is supported by the Howard Hughes Medical Institute and the Arthritis Foundation.
Abbreviations used in this paper: PI, propidium iodide; TSP1, thrombospondin 1; CR3/4, complement receptor type 3/type 4; Mφ, macrophage; MPO, myeloperoxidase; PMN, polymorphonuclear neutrophil granulocyte; PPP, platelet-poor plasma; PRPDS, platelet-rich plasma-derived serum; RGDS, Arg-Gly-Asp-Ser; RGES, Arg-Gly-Glu-Ser.