β-2-Glycoprotein 1-dependent Macrophage Uptake of Apoptotic Cells

The recognition and removal of apoptotic cells is critical to development, tissue homeostasis, and the resolution of inflammation. Many studies have shown that phagocytosis is regulated by signaling mechanisms that involve distinct ligand-receptor interactions that drive the engulfment of apoptotic cells. Studies from our laboratory have shown that the plasma protein β-2-glycoprotein 1 (β2GP1), a member of the short consensus repeat superfamily, binds phosphatidylserine-containing vesicles and apoptotic cells and promotes their bridging and subsequent engulfment by phagocytes. The phagocyte receptor for the protein/apoptotic cell complex, however, is unknown. Here we report that a member of the low density lipoprotein receptor-related protein family on phagocytes binds and facilitates engulfment of β2GP1-phosphatidylserine and β2GP1-apoptotic cell complexes. Using recombinant β2GP1, we also show that β2GP1-dependent uptake is mediated by bridging of the target cell to the phagocyte through the protein C- and N-terminal domains, respectively.

The phagocytosis and clearance of apoptotic cells occurs through redundant and highly conserved mechanisms that trigger specific signaling pathways (1,2). To accomplish this, both the apoptotic cell and phagocyte have developed a diverse array of distinct ligand-receptor systems that drive the recognition and uptake of dying cells (3). Phosphatidylserine (PS), 2 a plasma membrane inner leaflet phospholipid, becomes externalized at the surface of apoptotic cells and serves as a key ligand that is recognized by macrophages. Although PS-mediated phagocyte uptake can occur through a direct PS/PS receptor-mediated mechanism (4), many lipid binding proteins including protein S (5), complement (6), lactadherin/MFG-E8 (7,8), and ␤2GP1 (9, 10) have been shown to participate as bridging intermediaries by virtue of their ability to bind PS on the apoptotic cell surface and "cross-link" the target cell to the phagocyte.
␤2GP1, also known as apolipoprotein H, is a 50-kDa anionic phospholipid binding plasma glycoprotein that is a member of the short consensus repeat (SCR) superfamily. It is composed of four domains, each of which is cross-linked by two pairs of intradomain disulfide bridges and a PS binding fifth domain (11)(12)(13) at the C terminus with three sets of disulfide bonds (14,15). In addition to binding negatively charged liposomes and PSexpressing apoptotic cells, ␤2GP1 also binds to activated platelets (16,17) and endothelial cells (18) and inhibits ADP-induced platelet aggregation (16,19,20) and lipid-dependent prothrombinase activity (17,21,22). Studies have shown that the binding of ␤2GP1 to PS vesicles and apoptotic cell surfaces results in an enhanced recognition and clearance of these complexes by phagocytes both in vitro (9,23) and in vivo (10,24). These data suggest the involvement of a specific macrophage cell surface receptor that mediates recognition and engulfment of ␤2GP1apoptotic cell complexes. In this manuscript we provide evidence that supports a role for the LRP receptor family in ␤2GP1dependent clearance.

EXPERIMENTAL PROCEDURES
Materials-Phospholipids were purchased from Avanti Polar Lipids (Alabaster, AL). Media, fetal calf serum, 5-chloromethylfluorescein diacetate (CMFDA), and fluorescein isothiocyanate-labeled streptavidin were from Invitrogen. The bacterial expression vector encoding human receptor-associated protein (RAP) was kindly provided by Dr. Joachim Herz (University of Texas Southwestern Medical Center, Dallas, TX). Mouse monoclonal anti-human RAP antibody was obtained from Oxford Biomedical Research (Oxford, MI). Endotoxin levels were determined with the Pyrochrome LAL reagent (Associates of Cape Cod Inc., East Falmouth, MA). Fluoresbrite polychromatic red microspheres was from Polysciences Inc.(Warrington, PA). Phorbol 12-myristate 13-acetate, NHS-LC-Biotin, actinomycin D, and all other reagents unless otherwise stated were from Sigma-Aldrich.
Tissue Culture-Human THP1 monocytic leukemia cells were cultured in RPMI 1640 containing 10% fetal calf serum and 100 units/ml penicillin/streptomycin. Mouse monocytic leukemia cells L5178YS (ATCC) were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. Cells were maintained in a humidified atmosphere of 5% CO 2 at 37°C. Differentiation of THP1 and verification of their macrophage/phagocytic phenotypic was done as described previously (9). L5178YS were triggered into apoptosis with 0.5 g/ml of actinomycin D for 12 h. Apoptosis was verified by the ability of the cells to bind fluorescein isothiocyanate-annexin V.
Purification of ␤2GP1-␤2GP1 was purified from human plasma and antibodies generated as described previously (23,25,26). Expression and purification of recombinant full-length human ␤2GP1 and its domain-deleted mutants were performed using a Pichia pastoris expression system (27). All proteins were tested for purity by SDS-PAGE and Western blotting. Authenticity of the recombinant proteins was determined by N-terminal sequence analysis. Endotoxin was removed from protein preparations by multiple extractions with Triton X-114 (28). The absence of endotoxin was confirmed using the endotoxin assay kit (Associates of Cape Cod, East Falmouth, MA).
Purification, Biotinylation, and Affinity Coupling of RAP-RAP was purified as described previously (29). Biotinylated RAP was prepared by incubating RAP (30 nmol) with 300 nmol of N-hydroxy succinimide ester of LC-Biotin (Pierce) for 2 h at 4°C. The protein was then dialyzed against PBS to remove hydrolyzed/ unreacted succinimide. For affinity coupling, RAP or control human serum albumin (60 nmol) was incubated for 2 h at 4°C with 1 ml of Affi-Gel 10 beads (Bio-Rad). The beads were washed with ice-cold PBS to remove uncoupled proteins and stored as a 50% suspension in PBS. Coupling efficiency was determined by colorimetric protein analysis before and after incubation with activated beads and was found to be Ͼ95%. ␤2GP1-RAP Interactions-Binding of ␤2GP1 to RAP was assessed by Western blotting, ELISA, and RAP affinity chromatography. For Western blotting, purified ␤2GP1 or RAP was separated on SDS-PAGE. The proteins were transferred to polyvinylidene difluoride membranes, blocked with 1% egg albumin, and incubated for 2 h at 20°C with RAP (immobilized ␤2GP1) or ␤2GP1 (immobilized RAP). Protein binding was visualized using anti-RAP-anti-mouse IgG peroxidase or rabbit anti-␤2GP1-anti-rabbit IgG peroxidase. For ELISA, 96-well ELISA plates  were resolved by SDS-PAGE and immobilized on polyvinylidene difluoride membranes. The membranes was blocked with egg albumin and incubated with ␤2GP1 (5 g/ml) followed by rabbit anti-␤2GP1/antirabbit IgG peroxidase (lanes 1 and 2) or RAP (5 g/ml) followed by mouse anti-RAP/anti mouse IgG peroxidase (lanes 3 and 4). B, ELISA plates were coated with ␤2GP1 (E) or anti-RAP IgG (positive control, F) overnight. The wells were blocked with 1% egg albumin and incubated for 1 h at 20°C with serial dilutions of biotinylated RAP. The plates were then washed, and RAP binding was evaluated after incubation with horseradish peroxidase-conjugated streptavidin. C, ␤2GP1 (2 mg/ml, 1 ml) was passed through immobilized human serum albumin (HSA), RAP, or anti-␤2GP1 IgG (positive control). The beads were washed with 3 bed volumes of Tris-buffered saline, and bound proteins were eluted with glycine-HCl (100 mM, pH 2.4). Fractions were analyzed for protein content and for ␤2GP1 by dot-blotting using rabbit anti-human ␤2GP1 IgG. D, fluorescence and bright field photomicrographs of macrophages incubated with Biotin-RAP (5 g/ml) and phycoerythrin-labeled streptavidin.
were coated overnight with ␤2GP1 or anti-RAP IgG (positive control). The wells were then blocked with 1% egg albumin and incubated for 1 h at 20°C with serial dilutions of biotinylated RAP. The plates were then washed and RAP binding evaluated after incubation with horseradish peroxidase-conjugated streptavidin. For RAP affinity chromatography, 2 ml of immobilized RAP was packed into a mini column attached to a UV detector/ recorder. A solution of ␤2GP1 (2 mg) was passed through the column, and the column was washed with 3 bed volumes of Tris-buffered saline. Bound proteins were eluted with 100 mM glycine, pH 2.5. Fractions were assayed for protein and analyzed by dot blotting using anti-␤2GP1. The chromatography profile was compared using immobilized human serum albumin or immobilized anti-␤2GP1 as negative and positive controls, respectively.
RAP-Macrophage Interactions-RAP binding to macrophages was assessed by fluorescence microscopy. Macrophages cultured on glass coverslips were incubated with biotinylated RAP (5 g/ml) for 30 min on ice. RAP binding was assessed by fluorescence microscopy after incubation with phycoerythrinlabeled streptavidin (Invitrogen).
Phagocytosis-For flow cytometry, phorbol 12-myristate 13-acetate-differentiated THP-1 cells (macrophages) in serumfree RPMI 1640 were incubated with fluorescent PS-containing vesicles (0.2 mg/ml) or fluorescent microspheres in the absence or presence of ␤2GP1 or its domain-deleted fragments (2 M). After 1 h at 37°C, the macrophages were washed with PBS to remove unbound vesicles followed by treatment with 0.5% trypsin for 2 min to remove vesicles adhered to the macrophage cell surface. The macrophages were finally washed with PBS and analyzed for vesicle uptake by fluorescence-activated cell sorter. For the RAP inhibition experiments, macrophages were incubated with RAP at the indicated concentrations for 30 min at 37°C before the addition of fluorescent microspheres, PS vesicles, or PS vesicles/␤2GP1. For microscopy, control or apoptotic L5178YS cells in complete medium were labeled with CMFDA for 30 min at 37°C. The cells were washed with PBS to remove excess CMFDA and resuspended in serum-free RPMI 1640 at 1 ϫ 10 6 cells/ml. Macrophages were incubated with fluorescent control or apoptotic L5178YS (apoptotic cells:macrophages ϭ 10:1) for 1 h at 37°C in the absence or presence of plasma or recombinant ␤2GP1 or its domain-deleted mutants (2 M). The macrophages were then washed and fixed with 2% paraformaldehyde before fluorescence microscopy. Photomicrographs were assessed for phagocytosis by merging at least eight independent sets of green fluorescent (target cells) and bright field (macrophages and target cells) images and counting the fraction (%) of phagocytes that contained green fluorescent target cells (intensity of fluorescence was not considered). In some experiments, macrophage/target cell complexes were incubated with fluorescent microspheres to distinguish between macrophages and apoptotic cells before fluorescence photomicrography (data not shown). For the RAP inhibition experiments macrophages were incubated with RAP at the indicated concentrations for 30 min at 37°C before the addition of target cells or target cell/␤2GP1 complexes.
Data Analysis-Data were analyzed using the graphical data analysis software packages Microsoft Excel (Seattle, WA) and SlideWrite Plus (Encinitas, CA). Data presented are the mean Ϯ S.E. of at least three independent experiments.

RAP Inhibits ␤2GP1-dependent Uptake of PS Vesicles by
Macrophages-RAP, a molecular chaperone that regulates folding of LRP during protein synthesis (30), has also been shown to universally inhibit ligand interactions with all LRP family members (31)(32)(33)(34)(35). Hence, it has extensively been used as an antagonist in the study of LRP function (32)(33)(34)(35)(36). Because it has been previously demonstrated that RAP inhibits both binding of ␤2GP1-PS complexes to LRP (37) and LRP-dependent apoptotic cell uptake (34), we determined whether LRP receptors might be involved in ␤2GP1-dependent vesicle uptake. Macrophages were incubated with PS vesicle-␤2GP1 complexes in the presence or absence of RAP. The data presented in Fig. 1 show that RAP inhibited ␤2GP1-dependent PS vesicle uptake. Inhibition was specific for ␤2GP1-dependent uptake since RAP did not affect vesicle uptake in the absence of ␤2GP1 (Fig. 1A). Furthermore, RAP had no effect on phagocytosis of fluorescent microspheres (data not shown).
Three independent assays indicated that the inhibitory effect of RAP on phagocytosis was not due to binding to ␤2GP1. First, ␤2GP1-immobilized on polyvinylidene difluoride membranes was incubated with RAP followed by monoclonal anti-human RAP IgG/anti-mouse IgG peroxidase. Conversely, RAP immobilized on polyvinylidene difluoride membranes was incubated with ␤2GP1 followed by rabbit anti-␤2GP1 IgG/anti-rabbit IgG peroxidase. Both methods failed to demonstrate any affinity  FEBRUARY 15, 2008 • VOLUME 283 • NUMBER 7 JOURNAL OF BIOLOGICAL CHEMISTRY 3763 between these proteins ( Fig. 2A). ELISA binding experiments indicated that ␤2GP1 did not interact with RAP (Fig. 2B). Furthermore, attempts to bind and elute ␤2GP1 from a RAP affinity column also failed to demonstrate any interaction between these proteins (Fig. 2C). Immunofluorescence microscopy, on the other hand, revealed that biotinylated RAP bound directly to human THP-1 macrophages (Fig. 2D). Similar results were obtained with murine J774.1 macrophages (data not shown). Taken together with the known specificity of RAP for the LRP receptor family (30,36), the data presented above suggest that one or more members of this family is involved in ␤2GP1-dependent PS vesicle recognition and clearance by macrophages.

LRP Receptor-mediated Uptake of ␤2GP1-PS Complexes
RAP Inhibits ␤2GP-dependent Macrophage Uptake of Apoptotic Cells-The observations that RAP inhibits PS vesicle uptake in a ␤2GP1-dependent manner raises the possibility that LRP family members also participate in the recognition and phagocytosis of PS-expressing apoptotic cells. To test this, macrophages were incubated with RAP before the addition of apoptotic L5178YS-␤2GP1 complexes (target). As shown in Fig. 3, RAP reduced the extent of phagocytosis to levels comparable with apoptotic cells alone (no ␤2GP1). Similar to the results obtained with PS vesicles, RAP did not inhibit the uptake of apoptotic L5178YS in the absence of ␤2GP1.
Binding ␤2GP1-PS Complexes to Phagocyte LRP Receptors Requires Domain 1-Biophysical studies on interactions of ␤2GP1 with PS-containing membranes (38 -40) indicated that binding of ␤2GP1 to anionic lipid membranes through domain 5 induces conformational alterations to the tertiary structure of the protein that might render domain 1 accessible for the formation of bivalent interactions between apoptotic cells and macrophages. To test this possibility, we generated wild type and domain-deleted mutants of ␤2GP1 (Fig. 4A) and determined their ability to bind PS expressing apoptotic cells by flow cytometry. Fig. 4B shows that similar to the results obtained with PS vesicles (41), recombinant D1-D5 and D2-D5 bound apoptotic cells, whereas the domain 5-deleted mutant protein (D1-D4) did not. The inability of D1-D4 to bind PS is also shown by its failure to block plasma-␤2GP1-dependent clearance of PS vesicles by macrophages (Fig. 4C). Interestingly, whereas the domain 1-deleted mutant protein (D2-D5) bound PS-expressing vesicles (41) and apoptotic cells (Fig. 4B), it inhibited phagocytosis of plasma-␤2GP1-PS complexes (Fig.  4C). As expected, the addition of full-length recombinant ␤2GP1 (which is similar to plasma-␤2GP1) did not block vesicle clearance. Furthermore, because D1-D4 did not inhibit phagocytosis of plasma-␤2GP1-PS complexes, it is proposed that binding of domain 5 to PS induces a conformational change in the protein that exposes an LRP recognition motif on domain 1. Taken together the above studies indicate an important role for domain 1 of ␤2GP1 in apoptotic cell clearance by macrophages.  The macrophages were then washed, incubated with 0.1% trypsin to remove targets adhered to the macrophage surface, and fixed before analysis as described in Fig. 3. AC, apoptotic cells. Data are the mean Ϯ S.E. of three independent experiments. Domain 1 of the ␤2GP1 Is Critical for Apoptotic Cell Recognition by Phagocytes-To further establish the potential importance of domain 1 in apoptotic cell recognition, we tested the ability of the recombinant proteins to promote phagocytosis of apoptotic cells. Fig. 5 shows that whereas (wild type) D1-D5 was capable of promoting phagocytosis of apoptotic cells, deletion of domain 1 (D2-D5) or domain 5 (D1-D4) resulted in a reduced ability of these recombinants to promote phagocytosis. Taken together with the data in Fig. 4, the above experiments indicate the requirement of both domains D1 and D5 for ␤2GP1dependent phagocytosis of apoptotic cells.

DISCUSSION
The efficient recognition and removal of apoptotic cells is critical for the maintenance of homeostasis, resolution of inflammation, wound healing, tissue remodeling, and embryogenesis (8,42,43). The importance of this process is underscored by observations that phagocytes have established a repertoire of redundant receptor/ligand systems and mechanisms that identify and regulate the clearance of dying cells (3,44). Previous results from our laboratory have indicated that ␤2GP1 binds PS externalized on the surface of apoptotic cells and promotes their recognition and engulfment by macrophages. The phagocyte receptor and the ␤2GP1 domains critical to this process, however, have not been identified.
Members of the LRP gene family play a prominent role in cell physiology. This family of cell surface receptors is composed of several members that recognize structurally diverse extracellular ligands and internalize them for degradation by lysosomes (45). As demonstrated earlier (6,35,37,46), two members of the LRP family, LRP-1 (CD91) and LRP-2 (megalin), are particularly important in the clearance of apoptotic debris. Moreover, LRP-2, a receptor lining the absorptive epithelia of the kidney, has been shown to bind ␤2GP1 in vitro, and this affinity is augmented in the presence of PS (37). To determine whether LRP might also play a role in the ␤2GP1-dependent recognition and clearance of apoptotic cells by macrophages, we tested the ability of RAP, an universal antagonist of LRP-ligand interactions (31)(32)(33)(34)(35), to block phagocytosis of PS vesicles and apoptotic cells by macrophages. The inclusion of RAP effectively blocked ␤2GP1-dependent, but not ␤2GP1-independent clearance of PS vesicles and apoptotic cells. Although these data suggest a role for the LRP family in the clearance of ␤2GP1-apoptotic cell complexes, the identity of the receptor subtype is not known. Because prior binding of ␤2GP1 to PS is critical for the recognition of these complexes by macrophages (9), LRP-2 is an unlikely candidate since it has been shown to mediate endocytosis of the lipid-free form of ␤2GP1 (37). On the other hand, apolipoprotein E receptor 2Ј, also a member of the LRP family, has been shown to bind ␤2GP1-phospholipid complexes but not the lipid-free protein (47), making it a candidate that warrants further investigation.
It has been previously shown that binding of ␤2GP1 to anionic lipid surfaces leads to major alterations to the proteins conformation (38 -40). To determine whether such ligand (PS)-induced conformational changes might be critical for macrophage recognition, we generated recombinant human ␤2GP1 deleted at domains 1 or 5. Similar to experiments with PS-containing vesicles (41), our data shows that domain 5 is indeed critical for apoptotic cell binding and subsequent phagocytosis, whereas domain 1 is crucial for macrophage recognition. This is concluded from results which showed that the domain 5-deleted protein (D1-D4) did not inhibit phagocytosis of PS/plasma-␤2GP1 complexes (Fig. 4C) or enhance phagocytosis of apoptotic cells (Fig. 5). Moreover, the domain 1-deleted protein (D2-D5) did not enhance phagocytosis (Fig. 5) despite the fact that it binds apoptotic cells (Fig. 4B) FIGURE 6. Proposed interactions between PS, ␤2GP1 domains, and macrophage LRP receptors. Binding of wild type ␤2GP1 to PS on cell surfaces via D5 results in a key conformational alteration to the protein that exposes a cryptic LRP recognition motif on D1. Although the domain 1-deleted mutant, D2-D5, binds PS similar to the wild type, it is unable to promote apoptotic cell recognition by LRP due to absence of D1. The domain 5-deleted protein D1-D4 on the other hand does not promote apoptotic cell clearance because of its inability to bind to PS and, hence, elicit the crucial conformational change at D1. that exposes a critical motif on domain 1. This facilitates its interaction with the macrophage, thereby bridging ␤2GP1-apoptotic cell complexes to the phagocyte (Fig. 6). The involvement of a conformational alteration on domain 1 subsequent to PS binding is underscored by the observation that D1-D4, which cannot bind PS (Ref. 41 and Fig. 4B), does not block phagocytosis of plasma-␤2GP1-PS complexes (Fig. 4C).
In summary, we have identified a two-step mechanism that is likely responsible for the recognition of ␤2GP1-apoptotic cell complexes by macrophages. First, binding of ␤2GP1 to PS on apoptotic cell surfaces occurs through the proteins fifth domain, which results in a conformational alteration that exposes a cryptic epitope on domain 1. Second, the newly exposed epitope serves as a ligand for LRP receptors on the phagocyte surface. Given the role LRP family members and ␤2GP1 in autoimmune disease and atherosclerosis (48 -51), the data presented here set the stage for identifyingtheLRPreceptorsubtypethatiscriticalfor ␤2GP1dependent apoptotic cell recognition and for a better understanding of the relationship between LRP, autoimmune diseases, and impaired clearance of apoptotic debris.