The Interaction of a Ca2+-dependent Monoclonal Antibody with the Protein C Activation Peptide Region EVIDENCE FOR OBLIGATORY Ca2+ BINDING TO BOTH ANTIGEN AND ANTIBODY*

Protein C undergoes Ca2+-induced conformational changes required for activation by the thrombin-thrombomodulin complex. A Ca2+-dependent mono- clonal antibody (HPC4) that blocks protein C activation was used to study conformational changes near the activation site in protein C. The half-maximal Caz+ dependence was similar for protein C and y-carboxy- glutamic acid-domainless protein C for binding to HPC4 (205 f 23 and 110 f 29 p~ Ca2+, respectively), activation rates (214 f 22 and 210 f 37 p ~ ) , and intrinsic fluorescence of y-carboxyglutamic acid-do-mainless protein c (176 f 34 pM). Protein c heavy chain binding to HPC4 was half-maximal at 36 p~ Ca2+, although neither the heavy chain nor HPC4 sep-arately bound Ca" with high affinity. The epitope was lost when the activation peptide was released. A synthetic peptide, P(6-17), which spans the activation site, exhibited Ca2+-dependent binding to HPC4 (half-maximal binding = 6 p~ Ca2+). Thus, each decrease in antigen structure resulted in a reduced Ca2+ requirement for binding to HPC4. Tb3+ and Ca2+ binding stud- ies demonstrated a Ca2+-binding site in HPC4 required for high affinity antigen binding. These studies provide the first direct evidence for a Ca2+-induced conformational

The Interaction of a Ca2+-dependent Monoclonal Antibody with the Protein C Activation Peptide Region EVIDENCE FOR OBLIGATORY Ca2+ BINDING TO BOTH ANTIGEN AND ANTIBODY* (Received for publication, July 24,1987) Deborah J. StearnsS, Shinichiro Kurosawa, Peter J. Simss, Naomi L. Esmon Protein C undergoes Ca2+-induced conformational changes required for activation by the thrombinthrombomodulin complex. A Ca2+-dependent monoclonal antibody (HPC4) that blocks protein C activation was used to study conformational changes near the activation site in protein C. The half-maximal Caz+ dependence was similar for protein C and y-carboxyglutamic acid-domainless protein C for binding to HPC4 (205 f 23 and 110 f 29 p~ Ca2+, respectively), activation rates (214 f 22 and 210 f 37 p~) , and intrinsic fluorescence of y-carboxyglutamic acid-domainless protein c (176 f 34 pM). Protein c heavy chain binding to HPC4 was half-maximal at 36 p~ Ca2+, although neither the heavy chain nor HPC4 separately bound Ca" with high affinity. The epitope was lost when the activation peptide was released. A synthetic peptide, P(6-17), which spans the activation site, exhibited Ca2+-dependent binding to HPC4 (halfmaximal binding = 6 p~ Ca2+). Thus, each decrease in antigen structure resulted in a reduced Ca2+ requirement for binding to HPC4. Tb3+ and Ca2+ binding studies demonstrated a Ca2+-binding site in HPC4 required for high affinity antigen binding. These studies provide the first direct evidence for a Ca2+-induced conformational change in the activation region of a vitamin Kdependent zymogen. Furthermore, Ca2+ binding to HPC4 is required for antigen binding. The multiple roles of Ca2+ described may be useful in interpretation of other metal-dependent antibodylantigen interactions.
Protein C, a member of the vitamin K-dependent plasma zymogens, is activated in a Ca2+-dependent reaction by the thrombin-thrombomodulin complex to form an anticoagulant enzyme (1,2). As with other proteins of this class, calcium plays a central role in the activation and function of protein C. Potential structures involved in Ca2+-dependent reactions of protein C have been identified. These include two of the post-translational modifications: the y-carboxyglutamic acid residues (Gla') located in the amino-terminal portion of the search Grants R01 HL-29807, R01 HL-30340 (to C. T. E.), and R01 * This work was supported by National Institutes of Health Re-HL-36061 (to P. J. s.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "acluertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ Recipient of a research fellowship award of the Oklahoma Affiliate of the American Heart Association.
5 Established Investigator of the American Heart Association.
The heavy chain of protein C contains the functional serine protease domain of the molecule and is linked to the Ca2+binding light chain by a single disulfide bond. Although activation of protein C is Ca2+-dependent, the site of cleavage (Arg"-Leu13) is located on the heavy chain, far removed from known Ca2+-binding sites.
Calcium has novel influences on protein C activation both on the membrane surface and in solution. Whereas Ca2+ is required for activation by the thrombin-thrombomodulin complex, it inhibits activation by free thrombin (11). The Ca2+ concentration dependence of these two processes is indistinguishable and correlates with changes in protein C conformation monitored by intrinsic protein fluorescence (12). At least two possible mechanisms could be responsible for these observations. One is that Ca2+ alters protein C conformation at the cleavage site for the activator. An alternative is that Ca2+-induced conformational changes occur distal to the cleavage site and are important for secondary binding interactions with the thrombin-thrombomodulin activation complex.
As one means of monitoring conformational changes in such complex systems, conformationally sensitive metal-dependent antibodies have often been employed. Distinct metalinduced structural transitions relating to function have been described for both prothrombin and Factor IX by the selective application of metal-dependent antibodies (13,14). Ca2+-dependent monoclonals to human protein C (HuPC) have been described which bind to epitopes on the light chain either in the Gla domain (15) or the epidermal growth factor homology domain (10,16).
Using an analogous approach, we have isolated a Ca2+dependent monoclonal antibody to human protein C, termed HPC4, that blocks protein C activation by thrombin-thrombomodulin both in uiuo and in vitro (17). In the presence of Ca2+, this antibody binds to protein C, but not activated protein C, which permitted development of a functional assay for HuPC (18).
The studies presented here have investigated the mechanism of this Ca2+-dependent antibody/antigen interaction. One of the potential complications we considered, although seemingly unlikely, was that Ca2+ binding to the antibody was an obligatory event in high affinity antigen binding. The evidence presented demonstrates multiple roles for Ca2+ in this antibody/antigen interaction.
One is Ca2+-induced expression of the protein C epitope which overlaps the acti-A ea2+-binding Antibody Binds the Protein C Activation Site 827 vation cleavage site recognized by thrombin-thrombomodulin; the other is that occupancy of a Ca2+-binding site in the antibody is necessary for high affinity antigen binding.
Solutions-For experiments requiring metal-free conditions, buffers were dialyzed extensively in polypropylene containers with Chelex 100 in the dialysate to remove divalent cations. Protein solutions contained 1-2 mM EDTA and were dialyzed at 4 "C against the appropriate buffer containing Chelex.
Extinction Coefficients and Mokculur Weights for Proteins-The following values were used for calculating protein concentration: thrombin, E:' c, = 21, 37,000 (19); thrombomodulin, E ! : , 150,000 (22). The extinction coefficient for reduced carboxymethylated (RCM) heavy chain was estimated to be 19.7 based on a dye binding assay for protein content (Bio-Rad) using human protein C as the standard. The molecular weight used for the RCM heavy chain was 40,000 (3). Preparation of Synthetic Peptides-Solid-phase synthesis of peptides was performed with an Applied Biosystems 430A peptide synthesizer using the t-butoxycarbonyl chemistry (23). Peptides were cleaved by treatment with anhydrous hydrogen fluoride. Purity of the peptides as assessed by reverse-phase high pressure liquid chromatography was >go%. Molecular weight of the peptides was estimated by summation of the individual anhydrous amino acid molecular weights with correction for peptide bond formation. Peptide concentrations were estimated by reference to the absorbance at 220 nm of 1 mM peptide solutions in purified water.
Data AnaLysis-Values for half-maximal metal ion concentrations were obtained by nonlinear least-squares regression analysis of the data.
Monoclonal Antibody Production-BALB/c mice were injected peritoneally with 50-100 pg of HuPC in complete Freund's adjuvant. The HuPC immunization was repeated after 3 weeks (emulsified in incomplete Freund's adjuvant) and 6 weeks (in TBS). Four days later, spleen cells were fused with the mouse myeloma cell line PX63AG8-653 according to the method of Kohler and Milstein (24) using 35% polyethylene glycol 1450.
Supernatants from fused cells were screened for antibody production 4 weeks later by solid-phase enzyme-linked immunoadsorbent assay in the presence and absence of 5 mM Ca".
Positive clones of interest were recloned at least two times by limiting dilution onto murine peritoneal lavage feeder cells.
For production of ascites fluid, BALB/c mice were initially primed with pristane and 14 days later injected peritoneally with 0.1 ml of 10 mg/ml cyclophosphamide. Twenty-four hours later, 3-6 X IO6 cells were injected intraperitoneally. After 7-10 days, ascites fluid was collected. The monoclonal antibody, HPC4, was purified from ascites fluid by NH4S0, fractionation, followed by chromatography on QAE-Sephadex Q-50 (17) or by HuPC-Affi-Gel affinity chromatography.
Calcium Dependence of HuPC, HuGDPC, and RCM Heavy Chain Binding to HPC4-HPC4 coupled to Immunobeads was incubated overnight at 4 "C in TBS, 0.1% gelatin, 1 mM EDTA, pH 7.5, and then washed extensively with TBS, 0.1% gelatin, pH 7.5 (Chelextreated). Radiolabeled proteins were added to 100 p1 of HPC4-coupled beads with increasing concentrations of Caz+ in TBS, 0.1% gelatin, pH 7.5. Total volume was 200 pl. The solutions were incubated with mixing (2 h, 25 "C) and washed with gelatin buffer containing the appropriate amount of Ca2+, and the beads were counted in an NE 1600 y counter (Nuclear Enterprises, Ltd.). Control samples included solutions with no added Ca2+ and 1 mM EDTA, respectively. Final antigen concentrations ranged from 0.04 to 0.1 pM in several experiments, which were sufficiently low to ensure an excess of antibodybinding sites. Base-line counts/minute determined in the presence of 1 m M EDTA (5-15% of total counts added) were subtracted from the total counts/minute bound. Maximal binding of the '261-labeled RCM heavy chain was 80-90% of the total amount added and 60-70% of the added '251-labeled HuPC or Gla-domainless protein C.
Fluorescence Studies-All fluorescence spectra were obtained on an SLM-8000 fluorescence spectrophotometer (SLM-Aminco, Urbana, IL) equipped for stirring and temperature control. The temperature of the cuvette was maintained at 23 'C, and shutters were kept closed except during scans to minimize photodegradation of the sample. Correction for wavelength response of the photomultiplier tube was performed using correction factors supplied by the manufacturer. Integration of all emission spectra was performed in wave numbers.
Tryptophan was excited at 285 nm (4-nm slit widths), and emission was recorded at 2-nm intervals (4-nm slit widths). Emission peak intensity was quantified by integration of the fluorescence from 305 to 400 nm. In initial experiments, background signal due to titration of the buffer solvent was negligible relative to the sample intensity ( 4 % ) . Therefore, background correction of the tryptophan emission spectra was not done.
Samples for the terbium titration studies were excited at 285 nm (2-nm slit widths) using an SB 300 UV band-pass filter (Oriel) in the excitation path. Tb3+ ion emission intensity was recorded at 2-nm intervals and integrated from 538 to 552 nm. Contributions from light scattering were measured as the harmonic of the excitation wavelength recorded at 570-580 nm. For each T b 3 + concentration, the emission intensity of the solvent blank titrated in parallel with the sample was subtracted from that of the protein solution to give only the protein-dependent T b 3 + fluorescence.
Titrations for the Fluorescence Studies-For the intrinsic fluorescence studies, HPC4 antibody (1 p~) in TBS, pH 7.5 (Chelextreated), in the presence or absence of 2 p~ P(6-17) was titrated with CaC12 or MgClz diluted in the same buffer. HPC4 titrated with the synthetic peptide P(1-121, P(6-171, or P(15-27) was at a final concentration of 5 pM, and the solution contained either 1 mM EDTA or 1 mM CaClZ before addition of peptides. The emission intensity (305-400 nm) was recorded 5 min after each addition of titrant. Under all experimental conditions, sample dilution due to addition of titrant contributed <4% of the observed change in the signal.
For the terbium binding studies, affinity-purified HPC4 was dialyzed against 0.1 M MES, pH 6.0, containing Chelex. The antibody was diluted to 1 FM final concentration in 0.1 M MES, 0.1% gelatin, pH 6.0 (Chelex-treated), in the presence or absence of 4 p~ P(6-17). The solutions were titrated with T b 3 + at the indicated concentrations. In this buffer system, scatter from the protein sample did not increase significantly above that of the buffer blank, in contrast to large increases in scatter (protein precipitation) observed in the absence of gelatin or at higher pH.
Equilibrium Dialysis-Calcium binding to RCM heavy chain, HPC4 antibody, or a mixture of the two proteins was determined by equilibrium dialysis using ''CaClz as described elsewhere (25). The dialysis experiments were performed at room temperature in 0.25-ml wells for 24 h, a time sufficient for the system to reach equilibrium.

RESULTS
Interaction of HPC4 Antibody with HuPC-One of the monoclonal antibodies, HPC4, demonstrated Ca"-dependent binding to solid-phase HuPC using standard enzyme-linked immunosorbent assay screening procedures. HPC4 did not bind to bovine protein C either in the presence or absence of Ca2+ (data not shown). T o determine which portion of HuPC contained the antibody-binding epitope, HuPC was reduced, carboxymethylated, and chromatographed on an HPC4-Affi-Gel 10 affinity column. One peak was observed in the breakthrough fractions, and a single peak was eluted with 2 mM EDTA (Fig. L4). Contrary to expectations, the bound protein corresponded to the RCM heavy chain (Fig. 1 B ) which exhibited the characteristic doublet (3).
The role of the activation peptide region was examined by affinity chromatography on an HPC4-Affi-Gel 10 column. As expected, HuPC bound to the antibody column in a Ca2+dependent manner (Fig. 2 A ) . Following activation, the protein C no longer bound (Fig. 2 B ) . The RCM heavy chain also bound in the presence of Ca'+ and was eluted with EDTA ( Fig. 2C). When the eluate was treated with thrombin, the heavy chain no longer bound to HPC4 (Fig. 2 0 ) . By sodium dodecyl sulfate gel analysis, the thrombin-treated RCM heavy chain was indistinguishable from activated protein C heavy chain (data not shown). These results suggested that the Ca2+dependent HuPC epitope was in or near the activation peptide region of the heavy chain. This is supported by the previous observation that HPC4 inhibits HuPC activation by thrombin-thrombomodulin in the presence of Caz+ (17).
Ca'+ Dependence of Antigen Binding to HPC4 Antibody-The Ca2+ dependence of the HPC4lantigen binding was examined by incubating radiolabeled antigens with immobilized HPC4 (Fig. 3). The Ca'+ concentration resulting in halfmaximal binding to HPC4 was 36 f 5 PM for the RCM heavy chain, 110 f 29 PM for HuGDPC, and 205 f 23 PM for HuPC. epitope, we would have expected half-maximal binding to the antibody to occur at a Ca'+ concentration similar to the high affinity site (Kd = 60 p~) in bovine GDPC (12). The observation that HuPC displayed half-maximal binding to HPC4 at 205 PM Ca'+ seemed inconsistent with this hypothesis. Therefore, we examined the influence of Ca'+ on the initial rate of HuPC and HuGDPC activation and on the conformation of these proteins. Analysis of the Ca2+ dependence of the activation rates by thrombin-thrombomodulin (Fig. 4A) revealed half-maximal activation at 214 f 22 pM Ca2+ for HuPC and at 210 f 37 p~ Ca'+ for HuGDPC. Fig. 4B depicts Ca'+-induced quenching of the intrinsic fluorescence of the proteins. For HuGDPC, the half-maximal change in tryptophan fluorescence occurred at 176 f 34 pM Ca2+, in reasonable agreement with that found for activation (210 p~) , but somewhat higher than that observed for binding to HPC4 (110 p~) .
HuPC showed a maximal decrease in tryptophan fluorescence of 14% with the half-maximal change occurring at 390 f 77 p~ Ca'+, which is higher than that required for activation (214 p~) or binding to HPC4 antibody (205 p~) . ea2+ Binding to Components of the Antigen-HPC4 Complex-When equilibrium dialysis experiments with 45Ca2+ were performed with HPC4 (50 p~) or RCM heavy chain (35 PM), no high affinity Ca"-binding site was detected in either protein (at 0.8 mM Ca", 0.36 site in RCM heavy chain and 0.32 site in HPC4). However, when dialyzed together (15 p~ RCM heavy chain, 30 p~ HPC4), the results indicated between 2 and 3 mol of Ca'+ bound per mol of complex at 2 mM Ca2+, assuming a 2:l stoichiometry of RCM heavy chain to HPC4 in the complex (data not shown; see below). This suggested that the Ca'+ dependence of complex formation was mediated not only by the antigen conformation, but also by contributions directly from the antibody. Due to the complexity of this system and the expense and scarcity of the RCM heavy chain, further analysis of this system was not attempted.
Identification of the Protein C Epitope-Since the activation peptide region was required for antibody binding, three synthetic peptides were prepared which span this region of the heavy chain (Fig. 5). These peptides were then assayed for  metal (F,,).  their effect on '251-labeled HPC4 binding to solid-phase HuPC (Fig. 6). P(6-17) inhibited the binding of HPC4 to protein C with half-maximal inhibition occurring at ~0 . 5 pM peptide. The activation peptide, P(1-12), did not inhibit; and P(15-27) inhibited very little (30%) and only at the highest concentration (1 mM peptide).

-
The interaction of HPC4 with the synthetic peptides was also studied by monitoring intrinsic protein fluorescence in the presence and absence of Caz+ (27,28). This approach was used because any changes observed would be directly attributable to changes in the antibody resulting from peptide binding since the peptide that binds does not contain aromatic amino acids. The intrinsic fluorescence of HPC4 increased when titrated with P(6-17) in the presence of 1 mM Ca2+, reaching a maximum by the expected 2:l ratio of peptide to antibody (Fig. 7). In 1 mM EDTA, the fluorescence also increased, but this required higher peptide concentrations. The other two peptides did not significantly change the HPC4 fluorescence in the presence of Ca2+. The combined data of Figs. 6 and 7 demonstrate that P(6-17) contains the Ca2+dependent epitope.
Role of ea2+ Binding to HPC4 in Complex Formation-The Ca'+ dependence of HPC4/P(6-17) binding was also studied using fluorescence methods. The intrinsic fluorescence of HPC4 in the presence of P(6-17) increased when titrated with Ca'+ with the half-maximal change occurring at 6.5 1.2 p~ CaZ+ (Fig. 8). Mg2' had no effect on HPC4 intrinsic fluores-  (1 p~) was titrated with metal ions in the presence or absence of 2 p~ P(6-17). Tryptophan emission changes were monitored as described under "Experimental Procedures." Fo represents the HPC4 emission peak area (&peptide) in the absence of added metal. 0 and 0, Ca2+ titration of HPC4 or HPC4 + peptide, respectively, W and 0, Mg2+ titration of HPC4 or HPC4 + peptide, respectively. cence in the presence or absence of the peptide. Ca2+ titration of the antibody in the absence of peptide showed a small (5%) quenching of the intrinsic fluorescence, suggesting a possible Ca2+ interaction with the antibody.
T b 3 + is frequently used to investigate Ca2+-binding sites in proteins because its ionic radius and coordination properties are similar to those of Ca2+ (29, 30). It is a sensitive probe since Tb3+ fluorescence emission becomes greatly enhanced when the ion binds to a site on a protein close enough to a tryptophan or tyrosine residue(s) to allow efficient singletsinglet energy transfer. P(6-17) binding to HPC4 provided a unique system to monitor Tb3+ binding since only the antibody contained donor aromatic amino acids for enhanced Tb3+ fluorescence.
Titration of HPC4 with Tb3+ resulted in increased Tb3+ fluorescence that was half-maximal at a free Tb3+ concentra- Terbium binding by HPC4 in the presence and absence of P(6-17). HPC4 (1 PM) was titrated with T b 3 + in the presence (0) or absence (0) of 4 p~ P(6-17). The buffer used was 0.1 M MES, 0.1% gelatin, 0.02% NaN3, pH 6.0. HPC4 tryptophan was excited at 285 nm, and Tb3+ emission intensity was integrated from 528 to 552 nm as described under "Experimental Procedures." At each T b 3 + concentration, fluorescence due to buffer background was subtracted from the total emission peak area to give only proteindependent T b 3 + fluorescence (Fm). Half-maximal T b 3 + concentrations were determined after correction for bound T b 3 + assuming one T b 3 + lon/antibody-antigen ' complex. tion of 34 k 11 PM (Fig. 9). Tb3+ titration of HPC4 in the presence of P(6-17) also resulted in enhanced Tb3+ fluorescence, but the half-maximal free Tb3+ concentration decreased to 2 +-0.9 WM. Control experiments in which the peptide was titrated under the same conditions showed no enhancement of T b 3 + fluorescence (equivalent to buffer), which eliminated the possibility that the peptide alone was responsible for the changes in fluorescence. The 17-fold increase in affinity for T b 3 + by the peptide-HPC4 complex directly demonstrates the A Ca2+-binding Antibody Binds the Protein C Activation Site 83 1 transformation of a low affinity metal ion-binding site in HPC4 to a high affinity site.
To determine if Ca2+ binding to the antibody-peptide complex had similar requirements, Ca2+ binding studies were done using the Hummel-Dreyer gel filtration technique (31). P(6-17), HPC4, and mixtures of the two were applied to the column, and their ability to bind Ca2+ was determined. P(6-17) did not bind Ca2+ (Fig. lOA), as evidenced by the lack of either a 45Ca2+ peak eluting with the peptide or a trough at the inclusion volume (22-23 ml). HPC4 alone also did not show significant Ca2+ binding (Fig. 1OB). There was some indication of a 45Ca2+ peak coincident with the Azso peak (0.3 mol of Ca2+/mo1 of HPC4), but no discernible trough. When the experiment was performed with HPC4 and a 3.5-fold molar excess of peptide (Fig. lOC), Ca2+ binding increased to 1.76 mol of Ca2+/mo1 of HPC4. The Ca2+ binding was constant across the Azao peak (1.72-1.77 mol of Ca2+/mo1 of HPC4), and analysis of the trough data indicated an average of 1.65 mol of Ca2+/mo1 of HPC4. Mg2+ had no effect on Ca2+ binding by the peptide-HPC4 complex (1.41 and 1.52 mol of Ca2+/mo1 of HPC4 in 0.02 mM ca2+ or 0.02 mM Ca2+, 1 mM M P , respectively). This was consistent with the inability of M$+ to affect the intrinsic fluorescence of HPC4 f peptide.

DISCUSSION
During the course of these studies, the evolving characteristics of the Ca2+-dependent HPC4/antigen interaction posed some interesting problems. The observation that the Ca2+dependent HuPC epitope was independent of the light chain Gla domain was compatible with known Ca2+ binding characteristics of bovine protein C and GDPC (11,12). The apparent Kd of this site on the bovine molecule is =60 p~, well below the observed Ca2+ dependence for HuPC/HPC4 binding (205 p~ Ca2+). Analysis of the Ca2+ dependence of activation and conformational changes in human GDPC revealed that this site has a lower affinity (~2 0 0 p~ Ca2+) than the bovine protein. The observation that HuGDPC required less Ca2+ for antibody binding (110 p~) than for activation (210 p~) may be related to the fact that the antibody binds the Ca2+-stabilized form of GDPC with high affinity, thereby favoring the Ca2+-complexed conformer of GDPC. The resulting shift in equilibrium could reduce the Ca2+ concentration dependence for this process. Overall, these results show that Ca2+ binding to the Gla-independent site, probably in the light chain epidermal growth factor homology domain (lo), alters the conformation of the active peptide region of the heavy chain and suggests that this change in conformation is critical for substrate presentation to the thrombin-thrombomodulin complex. In addition, these studies provide the first direct evidence for a Ca2+-induced conformational change in the activation site of a vitamin K-dependent zymogen.
What remained unresolved was the observed Ca2+ dependence of RCM heavy chain binding to HPC4. The half-maximal binding decreased to 36 p~ Ca2+ for a protein with less secondary structure than the native molecule. Furthermore, neither the RCM heavy chain nor the HPC4 antibody had the ability to bind Ca2+ with the expected high affinity. Our working model for the system was the classical one, wherein Ca2+ binding to the antigen results in a conformation change exposing the epitope on the antigen. However, none of the data for the RCM heavy chain/HPC4 interaction could be explained by this mechanism. In addition, the observation that Ca2+ binding increased when both the RCM heavy chain and HPC4 were present together suggested that the role of Ca2+ in this system was more complicated than anticipated.
The availability of the synthetic peptide P(6-17) permitted studies to resolve the paradoxical results obtained with the RCM heavy chain. This peptide, which spans the activation site on protein C, contains the epitope recognized by HPC4. Like the native molecules, the HPC4/P(6-17) interaction was Ca2+-dependent. Furthermore, the peptide alone did not bind Ca2+, but in the presence of the antibody, Ca2+ binding increased to m 2 mol of Ca2+/mo1 of HPC4. These observations, in combination with the ability of the peptide to interact weakly with HPC4 in the presence of EDTA, led to the possibility that the antibody itself contained a Ca2+-binding site required for high affinity binding to the antigen. This was directly demonstrated by Tb3+ binding experiments. The antibody alone bound Tb3+, with half-maximal binding occurring at 34 p~ Tb3+. In the presence of P(6-17), the apparent affinity for Tb3+ increased 17-fold. In this system, the only possible donor for energy transfer to the Tb3+ ion was the antibody. The simplest interpretation of these results is that binding the antigen transforms the metal-binding site in the antibody from low to high affinity. Further studies are necessary to establish exactly where the ion binds in HPC4 and the mechanism of antigen-induced changes in metal ion affinity. It is possible that the site is directly in the HPC4 combining region of the variable domains or, alternatively, that metal binding to HPC4 affects antigen binding by allosteric mechanisms. By comparing the Ca2+ dependence of HuPC binding with that of the peptide, at least two roles have emerged for Ca2+ in mediating antigen binding to HPC4. The first incorporates the classical mechanism where Ca2+ binds to HuPC or Hu-GDPC and stabilizes a particular conformation. This results in a conformation recognized by both the thrombin-thrombomodulin complex and HPC4. The second role for Ca2+ is stabilization of the antigen-HPC4 complex. This is mediated by a metal ion-binding site in the antibody. The two Ca2+dependent events are clearly related, but can be discriminated. As the antigen structure decreases, the Ca2+ requirement for antigen binding decreases, presumably because the epitope becomes less conformationally constrained. These studies provide the first evidence to our knowledge of a Ca2+ requirement for antibody/antigen binding where a critical role for Ca2+ interaction with the antibody is implicated.
Like protein C, the other vitamin K-dependent proteins require metal ions for optimal expression of function. Recently, several laboratories (13-16) have described the role of cations in structure/function relationships of prothrombin, Factor IX, and protein C by using various metal-dependent antibodies. Whereas these immunological approaches are ideally suited to the study of ion specificity in these complex systems, interpretation of metal involvement in the antigen/ antibody interactions may be more complex than has previously been appreciated. This study demonstrates that the role of the metal ion is not necessarily limited to expression of an epitope or stabilization of a particular protein conformation. Furthermore, the inability of M%+ to support HPC4-P(6-17) complex suggests that metal ion specificity may be related to requirements of the antibody. Whether HPC4 constitutes a unique antibody or represents a class of antibodies remains to be determined.

A ea2+-binding Antibody
Binds the Protein C Activation Site