The C4b-binding Protein-Protein S Complex Inhibits the Phagocytosis of Apoptotic Cells*

The phagocytosis of apoptotic cells is a complex process involving numerous interactions between the target cell and the macrophage. We have examined a role of the major soluble inhibitor of the classic and lectin complement pathways, C4b-binding protein (C4BP), in the clearance of apoptotic cells. The major form of C4BP present in blood is composed of seven (cid:1) -chains and one (cid:2) -chain, which binds protein S (PS). Approximately 70% of all PS in human plasma is trapped in such a complex and is able to localize C4BP to the surface of apoptotic cells due to the high affinity to phosphatidylserine. Free PS has recently been shown to enhance phagocytosis of apoptotic cells by macrophages. We observed a stimulatory effect of free PS on the engulfment of apoptotic cells (BL-41 and Jurkat) by primary human macrophages or THP-1 cells and a decrease of activity in serum depleted of PS in agreement with previous results. However, we also show that the process is strongly inhibited in the presence of the C4BP-PS complex. Addition of the C4BP-PS complex to serum deficient in both molecules abolished the enhancing effect of serum on phagocytosis. The effect of both free PS and the C4BP-PS complex could be inhibited with monoclonal antibody directed against the Gla domain of PS. Although the presence of the C4BP-PS complex on

The process of apoptosis marks superfluous cells with signals that direct recognition, engulfment, and degradation by phagocytes in a very complex process involving a number of molecules present on the apoptotic cell and the macrophage as well as molecules recruited from serum. The macrophages silently engulf the apoptotic cells and prevent leakage of contents from the dying cells that would otherwise lead to inflammation. Cells undergoing apoptosis expose many marker molecules that are recognized by the macrophages such as phosphatidyl-serine (normally restricted to the inner leaflet of the cell membrane), differentially glycosylated molecules, pathogen-like apoptotic cell-associated molecular patterns, thrombospondin binding sites, oxidized low density lipoprotein-like sites, and intercellular adhesion molecule 3. Some of these molecules are recognized by complement component C1q (1) and mannose binding lectin (MBL) 1 (2) leading to deposition of complement factors C3b and iC3b (3). Receptors on the phagocyte include scavenger receptors, CD14 (4), ␤ 2 integrins (3), complement receptor 1 (CR1), ␤3/␤1/␤5 integrins, and still poorly characterized receptors for C1q and phosphatidylserine. Furthermore, a number of soluble molecules bridging the apoptotic target and phagocytes are known such as thrombospondin, C1q, MBL, C3b/iC3b, ␤ 2 -glycoprotein I, and the recently identified vitamin K-dependent anticoagulant protein S (PS) (5). PS is a 75-kDa blood glycoprotein that acts as a non-enzymatic cofactor to activated protein C in the degradation of activated coagulation factors Va and VIIIa (6,7). PS is composed of an N-terminal Gla domain followed by a thrombin-sensitive region, four epidermal growth factor (EGF)-like domains, and a sex hormone binding globulin (SHBG)-like region (schematically shown in Fig. 5). The SHBG-like region consists of two laminin G-like domains (LG domains) both of which are involved in a high affinity binding to the ␤-chain of an abundant complement inhibitor C4b-binding protein (C4BP) (8). The Gla domain, binds with high affinity (K D 100 nM) to negatively charged phosphatidylserine, which is exposed in the early stages of apoptosis. We have shown previously that PS localizes C4BP to the surface of apoptotic cells (9). The most common form of C4BP consists of seven identical ␣-chains and one ␤-chain (schematically shown in Fig. 5), consisting of eight and three complement control protein (CCP) domains, respectively, but forms with six ␣-chains and one ␤-chain (␣6␤1), and seven ␣-chains only (␣7␤0) are also present (10). C4BP inhibits the classic and lectin pathway of complement by binding activated complement factor C4b and thereby inhibiting the formation of the crucial complement enzymatic complex C3-convertase and accelerating its natural decay. C4BP also acts as a cofactor to serine protease factor I in the degradation and inactivation of C4b (11) and C3b (12). C4BP is an acute phase protein (13), but only the ␣7␤0 form increases in expression during inflammation (14), ensuring that PS (the ␤-chain ligand) will not be depleted from plasma, which could otherwise lead to severe thrombotic disorders. 1 The abbreviations used are: MBL, mannose binding lectin; CR, complement receptor; C4BP, C4b-binding protein; CCP, complement control protein; HBSS, Hanks' balanced salt solution; PE, phycoerythrin; PS, protein S; EGF, epidermal growth factor; SHBG, sex hormone binding globulin; PBS, phosphate-buffered saline; CFSE, 5-(and 6-)carboxyfluorescein diacetate succinimidyl ester; mAb, monoclonal antibody; ELISA, enzyme-linked immunosorbent assay; SLE, systemic lupus erythematosus.
It has been shown that purified PS was equivalent to serum in its ability to stimulate macrophage phagocytosis of apoptotic lymphoma cells (5). Immunodepletion of PS with polyclonal antibody eliminated the prophagocytic activity of serum. However, this previous study did not address a role of the C4BP-PS complex. Only 30% of all PS is free in plasma (7.5 g/ml), the other 70% is in complex with C4BP. Binding to C4BP heavily modulates the function of PS. PS fully looses its anticoagulant property when bound to C4BP, and once in complex it may also loose the possibility to interact with its putative receptor on the phagocyte as PS is known to interact with tyrosine kinase receptor Sky via the SHBG domain (15). In the present study we have analyzed the role of free PS versus the C4BP-PS complex in the phagocytosis of apoptotic cells. We reproduce the observation that free PS stimulates phagocytosis, but strikingly we show that PS in complex with C4BP, by far the most common form of PS in serum, strongly inhibits phagocytic uptake of apoptotic cells. These data may indicate that PS will target the C4BP-PS complex to the apoptotic cell surface where it functions as a complement regulator to prevent secondary necrosis following complement attack.

MATERIALS AND METHODS
Cells-Human BL-41 cells (B cells, Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH, Germany) were cultured in RPMI supplemented with 10% heat-inactivated fetal calf serum, 3.4 mM Lglutamine, 100 units/ml penicillin and streptomycin, and 50 M 2-mercaptoethanol. Jurkat cells (T cells, ATCC) were cultured in the same medium without 2-mercaptoethanol.
Human monocytes were prepared from buffy coat by a standard method. Briefly, the fresh buffy coat (Lund University Hospital) was mixed with equal amounts of phosphate-buffered saline (PBS), and dextran was added to 0.6% (v/v). The mixture was left at room temperature for 1 h, and then the upper phase was withdrawn and centrifuged for 10 min at 1000 rpm. After two washes of the cells in PBS, Ficoll was layered under the cell suspension, and the gradient was centrifuged for 35 min at 1460 rpm. The white blood cells were transferred to a new tube and centrifuged again at 1460 rpm for 35 min. The pellet of white blood cells was suspended in PBS, and the cell concentration was determined. The cells were centrifuged (1000 rpm, 10 min) and resuspended at 2⅐10 6 cells/ml. Two million cells were seeded in each well of a 24-well plate in RPMI supplemented with transferrin (Sigma, 25 g/ ml), gentamycin (Invitrogen, 50 g/ml), L-glutamine (Invitrogen, 2 mM), and granulocyte macrophage-colony stimulating factor (R & D Systems, 5 ng/ml). The cells were allowed to differentiate for 7 days before the phagocytosis assay was performed.
Apoptosis-Human BL-41 cells (0.5⅐10 6 cells/ml) were treated with 200 g/ml etoposide (Sigma) in growth medium for 3 h at 37°C. To monitor the amount of apoptotic and necrotic cells the cells were labeled with annexin-V-phycoerythrin (PE, Molecular Probes) and Viaprobe (7-aminoactinomycin D, Molecular Probes) and analyzed by flow cytometry. Around 50% of the cells was positive for annexin-V after treatment of etoposide for 3 h, 40% was double negative, and 10% was double positive for annexin-V and Viaprobe. Human Jurkat cells (1⅐10 6 cells/ ml) were treated with 0.5 M staurosporine (Sigma) for 3 h at 37°C, and the relative amount of annexin-V-and Viaprobe-positive cells was in the same range as for the BL-41 cells.
Phagocytosis Assay-After induction of apoptosis of the BL-41 cells or Jurkat cells, they were washed two times in PBS and then labeled with 5-(and 6-)carboxyfluorescein diacetate succinimidyl ester (CFSE, 9 ng/ml) for 20 min in 37°C at a cell concentration of 1⅐10 6 cells/ml. Both labeled apoptotic cells and macrophages were washed three times with Hanks' balanced salt solution (HBSS). The labeled apoptotic cells were incubated with the macrophages at a ratio of 2:1 in HBSS, supplemented with 0.75 mM CaCl 2 (final concentration: 2.5 mM), for 1 h at 37°C. The wells were then washed two times with HBSS, to get rid of most non-phagocytosed apoptotic cells, before the macrophages were labeled with anti-CD11b and anti-CD14, conjugated with PE, at room temperature for 45 min. The cells were detached from the plate with 1% lidocaine (Sigma) in FACS buffer (PBS, 30 mM NaN 3 , 1% bovine serum albumin) and analyzed by flow cytometry.
Proteins and Sera-The C4BP-PS complex was purified from human plasma as described before (16). Human recombinant PS was expressed in human embryonic kidney (HEK) 293 cells (ATCC number 1573-CRL) and purified essentially as described before (17,18). C4BP with ␤-chain but without PS was prepared by incubation of C4BP-PS in 4 M guanidium chloride, subsequent gel filtration, and dialysis. Human serum deficient in the C4BP-PS complex and PS (deficient in both the C4BP-PS complex and free PS) was prepared by passing fresh serum consecutively through two HiTrap columns (Amersham Biosciences), one coupled with MK 104, a mouse mAb directed against CCP1 of the ␣-chain of C4BP, and the other with MK 21, a mouse mAb directed against the Gla domain of PS. The flow-through was collected, and the depleted serum was stored in aliquots at Ϫ70°C.
C1q was purified by ion exchange on Biorex 70 followed by gel filtration as described before (19), and the remaining serum was used as serum deficient in C1q.
ELISA for Determination of C4BP and PS Concentrations-The concentration of C4BP and PS in normal human serum and serum deficient in both proteins was determined by ELISA. Plates were coated with antibody, rabbit PK 9008 (for detection of C4BP) and PK 7909 (for detection of PS), overnight at 4°C at 10 g/ml in 75 mM sodium carbonate, pH 9.6. The plates were quenched with 50 mM Tris-HCl, 0.15 M NaCl, 0.1% Tween, pH 7.5 (washing buffer) supplemented with 3% fish gelatin for 1 h. In the detection of PS, 2 mM CaCl 2 was used in all steps. After quenching, the plates were washed three times with washing buffer, and then the serum was added for 4 h. After another three washes the MK 104 (for detection of C4BP, 10 g/ml) and MK 54 (for detection of PS, 10 g/ml) were added and incubated for 1 h. Goat anti-mouse antibody, conjugated with horseradish peroxidase (Dako), was added after another three washes and incubated for 1 h. After the last three washes the plates were developed using 1,2-phenylenediamine dihydrochloride as substrate (Dako), and the absorbance at 490 nm was measured spectrophotometrically.
Hemolytic Activity of Human Serum-Hemolytic assays of normal human serum and sera deficient in C1q, C4BP, and PS were performed. Sheep erythrocytes were washed three times with DGVB 2ϩ (2.5 mM veronal buffer, pH 7.3, containing 70 mM NaCl, 140 mM glucose, 0.1% gelatin, 1 mM MgCl 2 , and 0.15 mM CaCl 2 ). The cells were incubated with a complement-fixing antibody (Amboceptor diluted 1:3000, Boehringverke) at a concentration of 0.5⅐10 9 cells/ml for 20 min at 37°C. After two washes with DGVB 2ϩ the cells were incubated for 1 h at 37°C with various concentrations of serum diluted in DGVB 2ϩ . The samples were centrifuged and the amount of lysed erythrocytes was determined by spectophotometric measurement of the amount of released hemoglobin at 405 nm.
Preparation of Fab Fragments of Monoclonal Antibodies-Fab fragments of mouse mAb MK 21, directed against the Gla domain, MK 54, directed against EGF domain 1, MK 34, directed against the SHBG-like domain of PS, and MK 104, directed against CCP1 of the ␣-chain of C4BP, were prepared by papain digestion using the ImmunoPure Fab preparation kit (Pierce) according to the manufacturer's instructions.

RESULTS AND DISCUSSION
The disposal of unwanted cells via apoptosis is a tightly controlled process to prevent inflammation and autoimmunity. A large number of factors are involved both on the apoptotic cell and the phagocyte as well as bridging molecules derived from serum. Among these proteins that bind to apoptotic cells there are several complement components (20). Binding of the early components of the classic pathway is thought to be very important in the clearance, since deficiencies in these components although rare are the strongest risk factors for the development of systemic lupus erythematosus (SLE). SLE is characterized by the presence of autoantibodies against cell components present on apoptotic cells that normally are not exposed for prolonged time to the immune system (21). The potential role of C1q in clearance of apoptotic cells was supported by the observation that C1q and MBL stimulate uptake of apoptotic cells by macrophages (22). In addition, the collagenous tail of C1q and MBL binds to calreticulin on the macrophage, which will then give a signal via surface molecule CD91 for ingestion of the apoptotic cell by macropinocytosis (22). Furthermore, both C1q-deficient humans and C1q-deficient mice develop SLE-like disease (23). CR1 and CR3 are also involved as they recognize C3b/iC3b deposited on the surface of apoptotic cells. The complexity of phagocytosis of apoptotic cells may reflect redundancy necessary for health, but it may also be simplified in particular tissues and circumstances in vivo.
Recently, PS was proposed to act as the major phagocytic factor in the uptake of apoptotic cells present in serum (5). This study, however, did not consider that 70% of all expressed PS in man is irreversibly bound to C4BP, and this study did not address the effect of the complex. Activity of PS was first identified in bovine serum that does not contain the C4BP-PS complex. The reason why these two proteins form a high affinity complex in human blood has long been an unsolved question. PS bound to C4BP cannot exert its anticoagulant function in the protein C pathway of inhibition of coagulation. Since the SHBG-like domain of PS binds to the ␤-chain of C4BP, this only leaves the Gla domain free to exert any additional function. Gla domains are present in a number of proteins, mostly coagulation factors, and can bind to negatively charged phospholipids such as phosphatidylserine in the presence of calcium ions (24,25). Previously we showed that the Gla domain of PS localizes both free PS and C4BP-PS to the surface of apoptotic cells (9,26).
In the present study we have investigated the role of free PS versus the C4BP-PS complex in the phagocytosis of apoptotic cells. The results presented were obtained in a phagocytosis assay in which BL-41 cells were made apoptotic with etoposide (50% of the whole population) and presented to primary human macrophages for engulfment. Furthermore, similar results were obtained using both Jurkat T cells and BL-41 cells and both the monocytic cell line THP-1 and primary macrophages. First we confirmed that in our assay human serum stimulated phagocytosis of apoptotic cells. When live and apoptotic BL-41 cells were added to macrophages in the presence or absence of 10% human serum, we found that the phagocytosis of apoptotic cells increased significantly when serum was added (Fig. 1) and that the effect was dose-dependent (not shown). To allow comparison of data obtained in different experiments, the amount of engulfed live cells was set to zero for each experiment. The phagocytic index for live cells increased in the presence of 10% human serum but to a much lower extent than for the apoptotic cells (Fig. 1).
To study the role of C4BP, PS, C4BP-PS, and C1q we have purified these proteins and prepared corresponding deficient sera. C4BP, C4BP-PS, and C1q were purified from human plasma, whereas PS was expressed recombinantly. The proteins were more than 95% pure as shown in Fig. 2D (10% SDS-PAGE under reducing conditions followed by a Coomassie staining). The C4BP-PS complex appears as a 70-kDa band (␣-chains that are predominant polypeptide in the complex). PS appears as a 75-kDa band and C1q as a 30-kDa band (Fig. 2D). Serum depleted from both PS and C4BP (PS/C4BP-deficient serum) was prepared by passing freshly prepared normal human serum through two affinity columns coupled with monoclonal antibodies against the two proteins. Using ELISA we show this preparation to be completely depleted of both C4BP and PS (Fig. 2, A and B). Another batch of human serum was depleted from C1q by ion exchange chromatography. A hemolytic assay measuring complement activation (classic pathway) on the surface of sheep erythrocytes was used to ascertain that normal human serum and PS/C4BP-deficient serum were highly active, while C1q-deficient serum lost its ability to be activated via the classic pathway (Fig. 2C).
In the present study we confirm that free PS is one of several serum factors required for phagocytosis of apoptotic cells (Fig.  3A). Purified PS increased phagocytosis in a dose-dependent manner, while PS/C4BP-deficient serum had a lower but not entirely abolished stimulatory effect on phagocytosis (Fig. 3A). This could be due to the fact that intact C1q was still present in our serum as shown by the hemolytic assay. Great care has to be exercised while depleting sera with antibodies, since C1q can easily be co-depleted when using for example a polyclonal antibody that will form immune complexes with its target. We have used monoclonal antibodies to prepare deficient sera, and we have shown that C1q was intact in our serum lacking C4BP and PS. Interestingly, we have observed that C1q-deficient serum entirely lost its ability to stimulate phagocytosis (Fig.  3B). Furthermore, purified C1q showed an increase of the phagocytic index when present in the assay (Fig. 3B).
Interestingly, we have found that addition of the C4BP-PS complex in the assay resulted in a drastically reduced amount of phagocytosed apoptotic cells (Fig. 3A). To further confirm this effect, the C4BP-PS complex or free PS was added to PS/C4BP-deficient serum. PS added to deficient serum yielded a phagocytic index similar to intact serum, while addition of the complex again showed strong inhibitory effect (Fig. 3A). When C4BP containing ␤-chain but no PS was added together with 10% normal serum, the phagocytic effect of serum decreased (Fig. 3A), which can be explained by the formation of the complex between C4BP and the free PS in the serum. The inhibitory effect of the complex is not surprising as it has been shown recently that phagocytosis of Streptococcus pyogenes is inhibited by the binding of C4BP to the M proteins on the surface of the bacteria (27). C4BP bound to the bacteria is able to strongly inhibit the activation of complement at least via the classic and lectin pathways, and the bacteria can therefore avoid elimination and survive in its host. Apart from regulating complement activation on the surface of the bacteria, it now appears that the binding of C4BP also inhibits the phagocytosis by macrophages. The effect of C4BP observed in the present study could be due to inhibition of C3b deposition on the surface of apoptotic cell. Also the size of the C4BP-PS complex could sterically block interactions between other phagocytic stimulators and their receptors.
To further characterize mechanisms by which PS and C4BP-PS exert their functions we have prepared a set of Fab fragments, to avoid activation of C1q and phagocytosis via Fc receptor. Fab fragments of several monoclonal antibodies directed against various parts of PS were used. The Fab fragments were added in the assay together with either the C4BP-PS complex or PS alone. MK 21, directed against the Gla domain of PS, decreased the phagocytosis degree seen for free PS (Fig. 4). The most probable explanation for this observation is that MK 21 blocks binding of PS to phosphatidylserine. MK 34, directed against the SHBG-like domain on PS, did not significantly decrease the phagocytic index, although there was a tendency toward inhibition of the PS effect. MK 34 abrogates the binding between PS and C4BP only partially (28), indicating that PS could possibly still interact with a receptor on the phagocyte even if MK 34 is bound to it. PS is known to interact with members of the Axl/Sky/Mer receptor family, and they may be the target on the macrophage (29). When adding the Fab fragments from MK 54, directed against EGF1, together with free PS there was no effect on the phagocytosis (Fig. 4). The inhibitory effect of the C4BP-PS complex was abolished when the Fab fragments from MK 21 were added (Fig. 4), while MK 34 could not block the activity of the complex (Fig. 4). Apparently, to exert its inhibitory function C4BP must be localized to the surface of apoptotic cell via the Gla domain of PS. MK 104, directed against CCP1 on the ␣-chain of C4BP, did interfere with the inhibitory effect of the complex to some extent but could not entirely block it (Fig. 4).
Taken together, free PS stimulates phagocytosis of apoptotic cells, while the same protein in complex with C4BP has an inhibitory effect on this process (see proposed mechanisms in Fig. 5). It appears that PS and C4BP are a part of an intriguing network of signals that stimulate and inhibit the process of phagocytosis of apoptotic cells to fine tune it to particular situations in different in vivo conditions. The current data indicate that the degree by which PS contributes to phagocytosis is mainly determined by the C4BP-PS complex. Although the presence of the C4BP-PS on apoptotic cells may lead to decreased phagocytosis, it may, importantly, prevent secondary necrosis because it inhibits further complement attack.