Heparin Influence on the Complex of Serum Amyloid P Component and Complement C4b-binding Protein*

Serum amyloid P component (SAP) forms a calcium-dependent complex with C4b-binding protein (C4BP) in human serum. This study demonstrated that heparin interacted with SAP in a calcium-dependent manner and prevented formation of the SAP*C4BP complex. Furthermore, the SAP-heparin interaction interfered with SAP binding to membranes. Therefore, all three of these interactions involved similar sites on SAP, or each interaction sterically obstructed the other binding sites. In addition to heparin, SAP bound to heparan sulfate and chondroitin sulfate. In each case, a distinct multimeric species was generated. Gel filtration and sucrose density gradient ultracentrifugation suggested that heparin and heparan sulfate produced a dimer of SAP. The dimer appeared to be the most stable struc- ture since it was not dissociated by excess heparin. While low molecular weight heparin interacted with SAP and inhibited SAP association with membranes, the SAP dimer was not detected in sucrose density gradient ultracentrifugation studies. Polybrene prevented the interaction between SAP and heparin in both a purified system and in human serum that was enriched in SAP and heparin. In contrast, Polybrene did not seem to alter the SAPoC4BP complex. While the function of the SAPaC4BP complex is unknown, it may be important for regulation of complement and/or transport of SAP to sites in the body. Dissociation of the SAPoC4BP complex by added to a solution (1.6 ml) of phospholipid vesicles (25 pg, phosphatidylcholine: phosphatidylserine, 75:25) and SAP (7 pg) in the absence of heparin (0), in the presence of heparin (m, 1.19 pg), and in the presence of various amounts of low molecular weight heparin (0, 0.6 pg; 0, 3 pg). Protein-membrane binding was determined from the light scattering intensities. Equation 1 was used to determine M2/Ml, the molecular weight ratio, of the protein-phospholipid complex to that of the phospholipid.

Serum amyloid P component (SAP) forms a calciumdependent complex with C4b-binding protein (C4BP) in human serum. This study demonstrated that heparin interacted with SAP in a calcium-dependent manner and prevented formation of the SAP*C4BP complex. Furthermore, the SAP-heparin interaction interfered with SAP binding to membranes. Therefore, all three of these interactions involved similar sites on SAP, or each interaction sterically obstructed the other binding sites. In addition to heparin, SAP bound to heparan sulfate and chondroitin sulfate. In each case, a distinct multimeric species was generated. Gel filtration and sucrose density gradient ultracentrifugation suggested that heparin and heparan sulfate produced a dimer of SAP. The dimer appeared to be the most stable structure since it was not dissociated by excess heparin. While low molecular weight heparin interacted with SAP and inhibited SAP association with membranes, the SAP dimer was not detected in sucrose density gradient ultracentrifugation studies. Polybrene prevented the interaction between SAP and heparin in both a purified system and in human serum that was enriched in SAP and heparin. In contrast, Polybrene did not seem to alter the SAPoC4BP complex. While the function of the SAPaC4BP complex is unknown, it may be important for regulation of complement and/or transport of SAP to sites in the body. Dissociation of the SAPoC4BP complex by sulfated polysaccharides such as heparin may be a physiological response that could be important during tissue damage or complement activation.
C4BP' functions as an inhibitory protein of the complement cascade. It is a large protein which consists of seven long, tentacle-like a-subunits and a shorter tentacle referred to as the P-subunit. The large subunits function by binding C4b  and promoting its activation by factor I (Scharfstein et al., 1978;Nagasawa and Stroud, 1977). C4BP binds about 60% of the vitamin K-dependent protein S in serum (Dahlback and Stenflo, 1981;Dahlback, 1983) through its small subunit Dahlback, 1988, 1990). The protein S-C4BP interaction is calciumdependent and of high affinity (Schwalbe et al., 1990a;Dahlback et al., 1990). The role of free protein S is to inhibit blood * This work was supported by Grants HL15728 (to G. L. N.) from the National Institutes of Health and B91-13X-07143-07B (to B. D.) from the Swedish Medical Research Council. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The abbreviations used are: C4BP, C4b-binding protein; SAP, serum amyloid P component. clot formation by serving as a cofactor for activated protein C (Walker, 1980;Walker et al., 1987). Protein S that is complexed to C4BP cannot function as an anticoagulant (Comp et al., 1984;Bertina et al., 1985;Dahlback, 1986). The role of protein S in the protein S-C4BP complex may be to localize C4BP at certain membrane sites, thereby protecting certain tissues or cells from complement activity. The protein S-C4BP complex interacts with specific membranes through the protein S component (Schwalbe et al., 1990a). Recently, we have reported that C4BP also forms a calcium-dependent complex with serum amyloid P component (SAP) in serum (Schwalbe et al., 1990b).
SAP is a member of the pentraxin family of proteins (for reviews see: Pepys and Baltz, 1983;Skinner and Cohen, 1988). These proteins are characterized by five identical subunits which are noncovalently associated in a disc-like structure. SAP is a decamer with two pentameric discs noncovalently bound face to face (Painter et al., 1982;Skinner and Cohen, 1988). SAP is a normal constituent of serum and is a normal tissue protein (Dyck et al., 1980;Breathnach et al., 1981). While the biological function of SAP is not known, the fact that no deficiency or alteration in the amino acid sequence has been reported suggests an important function (Pepys and Baltz, 1983). It was suggested that SAP may act as an important regulatory protein in the complement and/or coagulation systems (Schwalbe et al., 1990b).
SAP has also been reported to interact with immobilized single-stranded and double-stranded DNA in a calcium-dependent manner (Pepys and Butler, 1987). Furthermore, SAP was able to displace H-1 and interact with chromatin (Butler et al., 1990). It was proposed that SAP may solubilize extracellular DNA and help clear it from the circulatory system. In addition, SAP undergoes self-aggregation in the presence of calcium (Baltz et al., 1982;Schwalbe et al., 199Ob). Heparin and other glycosaminoglycans have been shown to inhibit this polymerization (Hamazaki, 1987(Hamazaki, , 1989. This latter property raised the question of whether heparin might also influence the SAP-C4BP interaction, thereby altering the distribution of SAP and C4BP in serum. Heparin is a normal constituent of mast cells and is released by these cells when their Fc, receptors are cross-linked (Kimball, 1986). Heparin is also used as a clinical anticoagulant (for review see Becker and Alpert, 1990). It functions by enhancing the activity of antithrombin 111, which inactivates thrombin and other coagulation proteases (Mourey et al., 1990).
This study was initiated to examine the potential influence of heparin on the C4BP.SAP complex. We report that heparin did interact with SAP in the presence of calcium and did prevent formation of the SAP. C4BP complex. Thus, heparin released from mast cells may function as a regulator of the SAP. C4BP complex in serum.

MATERIALS AND METHODS
Human C4BP was purified as previously described (Dahlback, 1983) and was quantitated by its extinction coefficient of 14.1 (Perkins et al., 1986). Human SAP used in this investigation was either prepared by published methods (Thompson and Enfield, 1978) or was purchased from Sigma. Gel electrophoresis under denaturing conditions showed essentially homogeneous protein preparations (Laemmli, 1970). In addition, there appeared to be little difference in the properties of the two preparations. The glycosaminoglycans used in this study were purchased from Sigma. Chondroitin sulfate and heparan sulfate were from a bovine source, and heparin (186.8 units/ mg) was from a porcine source. Low molecular weight heparin was reported to have a molecular weight of 4,000-6,000. Polybrene was also purchased from Sigma.
Proteins were radiolabeled by reductive methylation using radioactive formaldehyde as described by Jentoft and Dearborn (1983). Briefly, the procedure consisted of adding 20 mM NaCNBH3 (Sigma) followed by radioactive formaldehyde (50 pCi/mg of protein), to protein (0.5 to 0.99 mg/ml) in 50 mM phosphate buffer at pH 7.0 containing 0.1 M NaCl. The reaction mixture was incubated for 4 h at room temperature and was then dialyzed at 4 "C against 0.05 M Tris buffer, pH 7.5, containing 0. Sucrose density gradient ultracentrifugation was carried out using isokinetic gradients as previously described (McCarty et al., 1974). The centrifuge tubes (14-X 95-mm polyallomer, Beckman) contained a 10-ml sucrose gradient. Samples (0.3 or 0.4 ml) were applied to the top of the centrifuge tube, and they were centrifuged at 40,000 rpm at 5 "C for 22 h in a Beckman model SW 40 rotor and Beckman model L5-50 preparative ultracentrifuge.
The gradients were fractionated by pumping a solution of 31% sucrose into the bottom of the tube so that fractions were collected starting from the top of the tube. Fractions were collected by drop counting, and fraction volume was determined. Ultracentrifugation was carried out in a 50 mM Tris buffer, pH 7.5, that contained 0.1 M NaC1, 0.1% bovine serum albumin, plus either 1 mM calcium or 1 mM EDTA as indicated (Schwalbe et al., 1990b). Some samples also contained serum. In this case, 0.05 ml of serum was diluted to 0.3 ml, or 0.1 ml of serum was diluted to 0.4 ml with buffer. When serum was used, the purified radiolabeled proteins were added to the samples and they were incubated for 30 min at 37 "C prior to centrifugation. Buffer that did not contain bovine serum albumin was used for the standards that were to be detected by absorbance at 280 nm. Sedimentation positions for the samples were highly reproducible with a variation of 50.3 ml for the sedimentation position for multiple runs of the similar samples.
Each of the radiolabeled protein samples was sedimented in a sucrose density gradient to show protein and radiochemical purity. The sedimentation patterns were analyzed for absorbance at 280 nm and for radioactivity by scintillation counting. A Beckman SL 5000 TD scintillation counter was used. Both detection methods gave one symmetrical peak at the anticipated sedimentation position (Schwalbe et al., 1990b). The data are reported as counts per min or as a percentage of the total counts recovered. Bovine brain phosphatidylserine and phosphatidylcholine were obtained from Sigma. The supplier reported the phospholipids to be at least 98% homogenous. Small unilamellar vesicles consisted of 25% phosphatidylserine and 75% phosphatidylcholine and were produced by sonication as described by Nelsestuen and Lim (1977). Briefly, phospholipids were mixed in an organic solvent and dried under a stream of nitrogen. They were stored under high vacuum for 4 h and were then dispersed in buffer. They were cooled in an ice bath and subjected to 9 X 30 s bursts of sonication by a Heat Systems model W185 sonifier. The preparation was then chromatographed on Sepharose 4B (Huang, 1969), and the fractions containing small unilamellar vesicles were pooled. Concentration of phospholipid was determined from organic phosphate measurements (Chen et al., 1956) using a phosphorus to phospholipid ratio of 25.
Protein-phospholipid vesicle interactions were detected by light scattering intensity measurements at 90' as previously outlined (Nelsestuen and Lim, 1977). The measurements were made in a Perkin-Elmer-Hitachi model MPF 44A fluorescence spectrophotometer. All measurements were conducted in 0.05 mM Tris buffer con-taining 0.1 M NaCl and calcium as needed. The light scattering intensity measurements (wavelength = 320 nm) were used to obtain weight-average molecular weight ratios of protein-vesicle complexes (M2) to vesicles alone (M,) from Equation 1 (Nelsestuen and Lim, 1977) 12/I1 = ~M2/Mll'~~an~/ac2~/~~n,/~cl~12 (1) In this relationship, I is the light scattering intensity of the sample, M is the weight-averaged molecular weight, and anlac is the refractive index increment (0.19 for protein and 0.17 for phospholipid). The subscript 2 denotes the protein-vesicle complex, and subscript 1 designates vesicles alone. The light scattering intensities returned to their original values upon addition of EDTA, indicating that the protein-membrane interaction was calcium-dependent. In all cases, light scattering from free protein was subtracted as a background.

Binding of SAP to C4BP, Phospholipid Vesicles, and Various
Glycosaminoglycans-Earlier studies reported that C4BP, phospholipid vesicles, heparin, heparan sulfate, and chondrotin sulfate all prevented calcium-induced self-association of SAP (Schwalbe et al., 1990b;Hamazaki, 1987Hamazaki, , 1989. In the case of the glycosaminoglycans, the nature of this inhibition and the physical nature of the products formed were not determined. Fig. 1 shows the sedimentation positions of radiolabeled SAP in a sucrose density gradient. The sedimentation coefficients were determined by comparison to protein standards (Schwalbe et al.,199Ob). The sedimentation coefficients for SAP and for the SAP-C4BP complex were 7.5 and 11, respectively. The results also show the sedimentation position of radiolabeled SAP that was mixed with heparin.
This mixture gave a single sedimentation peak and, therefore, a discrete complex. The sedimentation position of the SAPheparin complex was identical with that of the SAP.C4BP complex.
The large heparin-induced shift in the sedimentation position of SAP could most easily be explained by formation of a dimer of the normal SAP species. Dimerization of spherical particles should cause a 1.6-fold increase in the sedimentation coefficient (van Holde, 1985) to an expected value of 12 for SAP. The observed sedimentation value of about 11 was slightly smaller but well within a reasonable range for dimerization of a nonspherical species. Thus, the results suggested that interaction of SAP with heparin generated a protein dimer. Furthermore, concentrations of heparin that corresponded to 115 times the amount needed to generate the maximum yield of SAP dimer (see below) did not shift or diminish the yield of this peak (data not shown). This sug-Volume  The sedimentation positions of radiolabeled SAP in the presence of various glycosaminoglycans are shown in Fig. 2. In every case, the carbohydrates interacted with SAP and appeared to form a single multimeric form of SAP. Similar results were obtained with heparin or heparan sulfate (Fig. 2,  A and B ) . The complex formed with chondroitin sulfate had a somewhat higher sedimentation coefficient (Fig. 2C). Other experiments indicated that chondroitin sulfate was also different in that a large amount was needed to generate the complex (data not shown). This suggested that chondroitin sulfate had a lower affinity for SAP than did heparin or heparan sulfate. Low molecular weight heparin did not produce the S = 11 complex (data not shown) and did not appear to be competent in forming the discrete SAP complex that is characterized by this sedimentation value.
Gel filtration chromatography showed the elution positions of radiolabeled SAP by itself, in the presence of heparin, or in the presence of C4BP (Fig. 3). A problem with gel filtration of SAP was its apparent tendency to interact with the column matrix and generate a trailing edge of protein in the elution profile, despite the presence of bovine serum albumin. Nevertheless, the results showed that the primary heparin-SAP complex eluted at a position approximately midway between free SAP and the SAP. C4BP complex (Fig. 3). Thus, despite similar sedimentation coefficients, the SAP. C4BP complex had a substantially larger hydrodynamic radius than the SAPheparin complex. This indicated that the SAP. C4BP complex, which has a larger molecular weight, also has a much larger frictional coefficient than the SAP-heparin complex. This would be expected for SAP.C4BP (Mr = 235,000 plus 570,000) uersus SAPz (Mr = 2 X 235,000). Heparin appeared to make a minor contribution to the mass of the SAP-heparin complex (see below). SAP binds to negatively charged phospholipid vesicles in a  calcium-dependent manner. Experiments were conducted to determine if this interaction competed with heparin binding. This approach was also used to determine if low molecular weight heparin interacted with SAP. Calcium addition to a solution of SAP plus phospholipid vesicles resulted in an increase in the light scattering signal indicating that SAP bound to the phospholipid vesicles (Fig. 4). In contrast, when heparin was present, there was little calcium-dependent increase in the light scattering signal, indicating that heparin prevented SAP-phospholipid binding. Low molecular weight heparin also inhibited SAP-membrane binding. However, the amount of low molecular weight heparin needed to abolish SAP association with membranes was severalfold higher than the amount of heparin needed. This suggested that the binding site for phospholipid vesicles was related to, or blocked by, the heparin-SAP or low molecular weight heparin-SAP interaction.
Heparin Dissociation of the SAP. C4BP Complex-Both C4BP and heparin form calcium-dependent complexes with SAP. Additional studies were carried out to determine if heparin was able to effectively compete with C4BP for interaction with SAP. [3H]C4BP (7.9 nM) and a slight excess of [14C]SAP were sedimented in sucrose density gradients (Fig.  5). The sedimentation patterns of this mixture in the absence All protein samples were diluted to 0.3 ml with buffer containing 1 mM calcium and 0.1% bovine serum albumin. of heparin showed that both proteins sedimented primarily at the position of the SAP.C4BP complex (Fig. 5A). The SAP. C4BP interaction is of high affinity (Schwalbe et al., 1990b), and SAP which sedimented at the position of the SAP monomer represented excess material (Fig. 5A). When heparin was included in the sample (Fig. 5B), all the SAP sedimented at the position of the heparin-SAP complex. While this was indistinguishable from the position of the SAP. C4BP complex, it was possible to show that C4BP was no longer complexed with SAP. That is, the sedimentation of C4BP was slower when heparin was present (Fig. 5B). Thus, heparin prevented C4BP from interacting with SAP. Panel C shows difference plots for the sedimentation of C4BP when complexed with SAP (Fig. 5A) uersus when these proteins were sedimented in the presence of excess heparin (Fig. 5B). The percent of radioactivity recovered in each fraction was determined, and the difference of values obtained in the presence of heparin minus those obtained in the absence of heparin is shown. The results showed that heparin caused C4BP to sediment at a lower value as expected if it became dissociated from SAP. The amounts of heparin added in the two experiments shown were 6.2 pg/ml and 62 pg/ml (Fig.  5B). Since there was no difference in the results, it appeared that the lower amount of heparin was sufficient to displace all the C4BP from the SAP. C4BP complex.
The amount of heparin required to bind 13 nM SAP was determined from the recovery of radioactivity sedimenting at the position of the SAP dimer (Fig. 6). Approximately 0.08 pg of heparin gave maximum recovery of SAP in this complex. If a 1:l ratio of SAP to heparin in the complex is assumed, (0.96 pg, 10,900 cpmlpg) plus various amounts of heparin were mixed in a 0.3-ml sample and sedimented in a sucrose density gradient as described in Fig. 1. The amount of I 4 C which sedimented at the position of the SAP-heparin complex (approximately 5.5 to 7.5 ml, Fig. 2 A ) was determined and is plotted as a function of heparin added to the sample. The buffers used contained 1 mM calcium and 0.1% bovine serum albumin. All samples of proteins were loaded onto the centrifuge tube in a 0.3-ml final volume.
this amount of heparin would correspond to an average molecular weight of 20,000. This value was close to the midpoint for the molecular weight of commercial heparin samples (Teng and Teller, 1981). Thus, the mass of heparin in the S = 11 complex was only about one-tenth the mass of SAP and probably corresponds to 1 or 2 molecules of heparin per SAP dimer. Demonstration of Heparin Dissociation of the SAP'C4BP Complex in Serum-A mixture of serum plus an amount of radiolabeled SAP that was approximately equal to the reported level of SAP in that of serum was sedimented in a sucrose density gradient. This mixture generated a broad peak of [ 14C]SAP indicative of a two-component mixture consisting of approximately a 1:l ratio of free and C4BP-complexed SAP (Fig. 7A). This was expected from previous results (Schwalbe et al., 1990b). When heparin was also included in this mixture, all of the SAP sedimented in the S = 11 position and would consist of C4BP. SAP complex and/or the heparin-SAP complex (Fig. 7A). The influence of heparin on the sedimentation position of [3H]C4BP illustrated a shift that was consistent with heparin-induced dissociation of the SAP. C4BP complex (Fig. 7B). This shift was more easily seen in the difference plots shown in Fig. 7C, where C4BP was decreased at higher S values and increased at lower S values, as expected for dissociation from SAP. These data indicated that heparin was able to prevent SAP from interacting with C4BP in serum. The degree of dissociation of the SAP. C4BP complex was dependent on the amount of heparin added to the serum (Fig.  7C). The amount of heparin needed to initiate SAP.C4BP dissociation was low and corresponded to approximately therapeutic levels of heparin (0.4-1.0 unit/ml or 2.5-6.25 pg/ml (Holm et al., 1986;Boneu et al., 1989), Fig. 7C). These results suggested that physiological levels of heparin may be capable of altering the amount of the C4BP. SAP complex in serum and of generating a physiological effect that this might produce.
The effect of Polybrene on the SAP-heparin complex was monitored. Polybrene is a cationic polymer that binds anionic polymers such as heparin. Once again, sufficient radiolabeled SAP was added to give approximately a 1:l ratio of free SAP and SAP.C4BP. The sedimentation showed a broad double peak which was essentially unaffected by polybrene (Fig. 8 A ) . When heparin was included in the mixture, the [I4C]SAP peak became very narrow and sedimented at the position of The percent of recovered ['HH]C4BP was determined for each fraction, and the corresponding value obtained for a sample that did not contain heparin was subtracted. This procedure was similar to that shown in Fig. 5C. Buffers contained 1 mM calcium and 0.1% bovine serum albumin in all the experiments. the SAP-heparin complex (Fig. 8B, minus Polybrene). In addition, when Polybrene was also added in the mixture, it appeared that the sedimentation pattern was similar to the broad sedimentation peak of free SAP plus the SAPeC4BP complex. Thus, Polybrene prevented formation of the heparin-SAP complex but appeared to have little influence on the SAP. C4BP complex.

DISCUSSION
The purpose of this study was to determine how heparin or other glycosaminoglycans might influence the SAP. C4BP interaction in serum (Schwalbe et al.,199Ob). Heparin has been shown to interact with SAP as evidenced by its ability t o prevent SAP polymerization (Hamazaki, 1987(Hamazaki, , 1989. If heparin also influenced SAP. C4BP interaction, it may have a physiological role of altering the distribution of SAP and C4BP. The results of this study demonstrated a competition between heparin and C4BP for binding to SAP. The heparin-SAP complex was a discrete entity, probably corresponding to a complex that included one or two heparin molecules plus two SAP decamers. This (SAP)* species appeared to be the most stable structure since excess heparin did not dissociate this complex. The levels of heparin required for these interactions may be physiologically relevant.
While the role of heparin-dependent dissociation of the SAP.C4BP complex is not known, this effect has intriguing possibilities that may form the basis for future research. For (1.92 pg, 10,900 cpm/pg) added to 0.05 ml of human serum and diluted to a total sample volume of 0.3 ml. The two sedimentation patterns are for plus (0) and minus (0) Polybrene (50 pg) in the sample. The experiments in panel E are the same as in panel A except that 10 pg of heparin was added to both samples. Again, samples with (0) and without (0) Polybrene were run. All the experiments contained 1 mM calcium and 0.1% bovine serum albumin. example, during tissue damage or complement activation, mast cells may be triggered to release heparin and other components (Kimball, 1986). Along with other effects, this may result in dissociation of the SAP.C4BP complex and formation of the SAP-heparin complex. Dissociation of SAP may influence, either positively or negatively, the action of C4BP, thereby influencing complement fixation. Alternatively, the SAP-heparin complex may trigger another reaction or serve as a regulatory signal. For instance, SAP interaction with heparin might help modulate the anticoagulant role of heparin. Otherwise, heparin released from mast cells might also displace SAP from other sites in the body, thereby producing alterations in tissue properties. Thus, while precise roles of these several interactions are not known, their existence suggests numerous possibilities which will require future examination.
In addition, the acute phase response is characterized by a modification of the concentration of several plasma proteins (Kushner, 1982;Pepys and Baltz, 1983). This response is stimulated by infection, tissue damage, and cell death. The serum levels of C4BP (Saeki et al., 1989) are reported to increase during the acute phase response. For this reason, the serum levels of free C4BP or C4BP complexed to protein S (Dahlback and Stenflo, 1981) might be elevated in the acute phase. This may constitute an important relationship as well. Thus, extensive future studies will be needed to examine these many possible roles for the influence of heparin and other factors on SAP and its many interactions.