Molecular Mechanism for Familial Protein C Deficiency and Thrombosis in Protein CVermont ( G1u20 + Ala and Val34 + Met)*

The role of two protein C y-carboxyglutamic acid domain mutations in familial thrombosis, protein CVemont (Bovill, E. G., Tomczak, J. A, Grant, B., Bhushan, F., Pillemer, E., Rainville, I. R., and Long, G. L. (1992) BZood 79, 1456-1465), was investigated. Two single mutations (G1uZ0 +Ala and VaP4 + Met) and the naturally occurring double mutation were created by site-directed mutagen- esis and were expressed in human kidney 293 cells. Purified recombinant protein C with the mutation gluta- mate to alanine at position 20 is defective in the assays of activated partial thromboplastin time, factor Va inac- tivation, and fibrinolysis. Mutation from valine to me-thionine at position 34 has only a minor effect. Activation of G1uZo mutants by thrombin-thrombomodulin was not enhanced by phospholipid vesicles and showed a different calcium dependence compared with the wild type, suggesting that GlaZ0 is important in the interac- tion of the protein C Gla domain with a phospholipid-mediated site on the thrombomodulin molecule. G1uZ0- substituted protein C is not inhibited by calcium ion in its interaction with the calcium-dependent monoclonal antibody H-11, suggesting that this

The role of two protein C y-carboxyglutamic acid domain mutations in familial thrombosis, protein CVemont (Bovill, E. G., Tomczak, J. A, Grant, B., Bhushan, F., Pillemer, E., Rainville, I. R., and Long, G. L. (1992) BZood 79, 1456-1465), was investigated. Two single mutations (G1uZ0 +Ala and VaP4 + Met) and the naturally occurring double mutation were created by site-directed mutagenesis and were expressed in human kidney 293 cells. Purified recombinant protein C with the mutation glutamate to alanine at position 20 is defective in the assays of activated partial thromboplastin time, factor Va inactivation, and fibrinolysis. Mutation from valine to methionine at position 34 has only a minor effect. Activation of G1uZo mutants by thrombin-thrombomodulin was not enhanced by phospholipid vesicles and showed a different calcium dependence compared with the wild type, suggesting that GlaZ0 is important in the interaction of the protein C Gla domain with a phospholipidmediated site on the thrombomodulin molecule. G1uZ0substituted protein C is not inhibited by calcium ion in its interaction with the calcium-dependent monoclonal antibody H-11, suggesting that this mutation has lost the calcium-induced, lipid-independent conformational transition of the protein C Gla domain. These data indicate that the loss of GlaZ0 causes the major familial dysfunction of protein C to associate with phospholipid as well as to undergo Ca2+-dependent, lipid-independent conformational changes and are consistent with the importance of GlaZ0 in both external and internal Ca2+ binding based upon the x-ray-derived structure of the homologous Gla domain in bovine prothrombin.
Protein C (PC)' is a vitamin K-dependent plasma protein consisting of a y-carboxyglutamic acid (Gla) domain, two epidermal growth factor domains, and a classical trypsin-like serine protease domain (1). PC can be converted by thrombin complexed with thrombomodulin (TM) on the surface of endogram Project Grant Pol-46703 from the National Institutes of Health * This work was supported by Facility Grant C06-HL39475 and Proand American Heart Grant-in-aid 92011860. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisernent" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. thelia1 cells (2) to activated PC (AF'C), which functions as an anticoagulant by proteolytically cleaving coagulant factors Va (3, 4) a n d VIIIa (5, 6). The first congenital PC deficiency was reported by Griffin et al. (7), and following reports (8,9) led to the association of PC deficiency with thrombotic disease. Newborns with PC homozygous deficiency exhibit severe, lifethreatening purpura fulminans (lo), whereas heterozygotes from clinically affected families experience mainly thromboembolic disease in a much smaller percentage of biochemically affected individuals (11,121. PC deficiency is classified into two categories: type I deficiency, in which both circulating PC antigenic and functional levels in heterozygotes are -50% of the normal range; and type I1 deficiency, in which individuals have normal antigenic levels, but reduced functional activity. Bovill et al. (13) investigated type I1 PC deficiency in a family (protein CVemont) with manifestations of both arterial and venous thrombosis. Direct DNA sequencing of PC exon 2 from genomic DNA of deficient individuals showed two nucleotide substitutions resulting in two point mutations in the Gla domain, G1uZo -j Ala and Val34 + Met (13).
The Gla domain of PC contains 9 glutamic acid residues (14) that are post-translationally modified to y-carboxyglutamic acid residues by vitamin K-dependent carboxylase (15). These y-carboxyglutamic acid residues (positions 6, 7, 14, 16, 19, 20, 25, 26, and 29) are necessary for Ca2+-dependent lipid membrane association (16,17) as well as a lipid-independent structural transition from a random to an ordered structure upon binding of divalent metal cations (18). The influence of individual amino acids of the PC Gla domain was extensively studied by Zhang and Castellino (19)(20)(21). They found that Gla residues at positions 7, 16, 20, and 26 in the Gla domain of PC are essential for the anticoagulant activity ofAPC (19-21). To further understand the functional role of Gla domain residues and to understand the basis of PC dysfunction in familial (protein CVemont) deficiency and thrombosis, we have studied the effects of the double mutation occurring in this affected kindred as well as each of the single mutations.
Construction of Wild-type P C Expression Vector and Site-directed Mutagenesis-Full-length human PC cDNA, a product of two PC cDNA fragments obtained from primer-specific PCR amplification and cDNN h g t l l library clones, respectively, was cloned into the EcoRI site of a SV40/adeno/pBR322-derived expression vector, pDX (24). Site-directed mutagenesis was performed by the PCR-mediated method developed in this laboratory (25) using a linearized (Hind111 digestion) wild-type PC expression vector as template and is shown schematically in Fig. 1. The external PC primers used were as follows: B, 5'-ggggtactagtaacccgggc-CAATTGAGGAGGCTCACCTC-3'; and C, 5"ggttgaattcccgGGCGA-ACTTGCAGTATCT-3' (lower-case letters represent the "Zeno" sequence and restriction enzyme digestion sites for EcoRI and SmaI, respectively), which correspond to positions 497-516 and 44-61 of the protein C cDNA, respectively (14). The sequences of mutagenic primers were 5'-TGCATAGAGGcGATCTGTGAC-3' for G1uZn + Ala and 5"TTTC-CAAAATaTGGATGACAC-3' for VaP4 + Met (lower-case letters represent the desired mutations). The second-round PCR product was digested with SmaI and SalI, and the resulting 313-base pair fragment was resolved on a 1% agarose gel, isolated by electroelution, and ligated into the wild-type PC expression vector from which the corresponding fragment had been removed. The double mutant was generated by using the single mutant PC cDNA with G1uZn + Ala as template and the other mutagenic primer to generate the second mutation. The region of the PCR-mutated fragment in the final PC expression vectors was sequenced (Applied Biosystems Model 373A DNA sequencer) to confirm the desired mutations and to exclude inadvertent mutations introduced by PCR. The sequencing reactions were performed with the TaqDye DeoxyTM terminator cycle sequencing kit (Applied Biosystems Inc.) according to the manufacturer's instruction. ltssue Culture and Dansfection-Human kidney 293 cells were cultured in Dulbecco's modified Eagle's medium, high glucose (Life Technologies, Inc. 11995-016), supplemented with 10% fetal bovine serum and 10 pg of vitamin K,/ml. Forty pg of PC expression vector and 2 pg of neomycin resistance expression vector were cotransfected into human kidney 293 cells (70% confluent in a 10-cm plate) by the calcium phosphate method (26). Resistant clones were selected by growing the cells for 3 weeks in Dulbecco's modified Eagle's medium with 0.5 mdml G418 (Geneticine), and then individual clones were isolated with a cloning ring (27), grown up, and screened for PC level by immunological assay (28). High level producers were propagated to confluence in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 10 pg of vitamin K,/ml, and 0.1 mg/ml G418. Then the plate was washed twice with Tris-buffered saline and added Dulbecco's modified Eagle's medium supplemented with 10 pg/ml vitamin K,. The serum-free conditioned medium was collected after 48 h to produce recombinant protein for further biochemical characterization.
Purification of Recombinant PCs-The purification of recombinant PC was based on the method described by Yan et u1. (29). Conditioned serum-free medium from stably transfected cells was diluted with a n equal volume of 20 m~ Tris (pH 7.4), 150 mM NaCl buffer and adjusted to 5 mM EDTA and 5 mM benzamidine. The diluted conditioned medium was applied to a pre-equilibrated Q-Sepharose fast flow anion-exchange column (Pharmacia Biotech Inc.) and washed with buffer containing 5 then washed with 20 mM Tris (pH 7.4),150 mM NaCl without EDTA and mM EDTA until the media pigment was off the column. The column was eluted with a 0-30 mM CaCl, gradient in the above buffer. Fractions were collected, and PC was detected by SDS-PAGE followed by silver staining (30). Fractions containing PC were combined; diluted with 3 volumes of 20 mM Tris (pH 7.4), 150 mM NaC1, and 5 mM EDTA; and applied to a second Q-Sepharose fast flow column. The column was washed as described above; a linear gradient of 150-500 mM NaCl in 20 mM Tris (pH 7.4) was applied; and fractions were collected. At this point in the purification, wild-type recombinant PC (rPC) was homogeneous as determined by SDS-PAGE. In the case of the mutants, any impurity seen after the second ion-exchange column chromatography step was removed by monoclonal antibody HPC-2 affinity chromatography under conditions described elsewhere (19). rPCs were dialyzed in 50 mM NH4HC0, followed by water with chelating resin (Sigma, C-7901) and concentrated with a Centricon concentrator (Amicon, Inc.). SDS-PAGE and Western blotting were performed by standard methods as described elsewhere (31, 32). The concentrations of rPC were determined by enzyme-linked immunosorbent assay (28) using human plasma PC as a standard.
Chemical Analysis-Gla analysis was performed as described by Kuwada and Katayama (33). The numbers of aspartate and asparagine residues in PC were used as internal standards for the determination of Gla number. The PC Gla peptide (ANSFLyyLRHSS) was employed as a reference standard. P-Hydroxyaspartic acid content of PC was deter-mined as described by Przysiecki et al. (34). NH,-terminal sequences of PC were determined with a n Applied Biosystems Model 470A protein sequencer.
Activation of P C to APC and Amidolytic Assay-Activation of PC with the snake venom activator Protac was performed as described (35) with 1-5 pg of PC and 0.2 unit of Protac incubated at 37 "C for 1 h and then a t 4 "C for 16 h in 1 ml of 50 mM Tris (pH 7.41, 100 mM NaCl to achieve full activation. Activation of PC by thrombin was performed as described elsewhere (36) by incubating PC with a-thrombin (l:lO, w /~) at 37 "C for 2 h in 20 mM HEPES (pH 7.4),150 mM NaCl, and 5 mM EDTA. The mixtures were diluted with an equal volume of water and passed over a sulfopropyl-Sephadex column to remove the thrombin. APC was recovered from flow-through fractions by eluting the column with 10 mM HEPES (pH 7.4), 75 mM NaC1. The achievement of PC full activation was confirmed by the mobility difference between PC and APC on nonreduced SDS-PAGE visualized by Western blotting. The amidolytic activity ofAPC was measured by the hydrolysis of the synthetic substrate PCa as a function of time. Three-hundred pl of synthetic substrate PCa (4 mM in H,O) was mixed with 800 pl of 1.0 M NaC1, 400 pl of 1.0 M Tris (pH 7.4) and used as a stock to measure the APC chromogenic activity. The ability of APC to hydrolyze synthetic substrate was determined by adding 190 p1 of various concentrations of the synthetic substrate PCa (0-0.8 mM) to 50 pl of APC (0.2 pdrnl) and monitoring the para-nitroaniline release by 405 nm absorbance increase using a microplate reader (Molecular Devices) at room temperature.
Activated Partial Thromboplastin Time ( A P T T ) Assay-Quantitative determinations of APC were based on the prolongation of APTT. Staclot protein C clotting assay kits (Diagnostic Stago) were used for APTT assay of APC. One-hundred p1 of APC, 100 pl of protein C-depleted citrated human plasma, and 100 pl of thromboplastin were combined and incubated at 37 "C. Clotting was initiated by adding 100 pl of 0.025 M CaCI,. The assays were performed on a Diagnostica Stago ST4 fibrometer. The conditions were calibrated by a Thrombo calibrator as recommended by the manufacturer.
HEPESPryrode's buffer (pH 7.4) with 5 mM CaCI,, 20 p~ PCPs, and 100 Factor Vu Inuctiuation-Inactivation of factor Va was performed in n~ human factor Va preincubated at 37 "C (36). Inactivation was initiated by the addition of APC to 0.5 nM, and 10 pl of the mixture was aliquoted a t different time intervals thereafter to assay the remaining cofactor activity of factor Va. The factor Va activity was measured as the cofactor required for thrombin formation in the prothrombinase assay (37). The prothrombinase assay mixture consisted of 2 rnl HEPES/ Tyrode's buffer, 1.39 PM prothrombin, 5 nM factor Xa, 20 p~ PCPs, 3 p~ dansylarginine-N-(3-ethyl-l,5-pentanediyl)amide, and 5 mM CaCI,. The generation of fluorescence from binding of dansylarginine-N-(3-ethyll$pentanediyl)amide to the active site of thrombin was measured in a Perkin-Elmer LS-3B fluorescence spectrometer with an excitation wavelength of 335 nm and an emission wavelength of 565 nm. The initial rate of thrombin formation was proportional to the remaining cofactor activity of factor Va under these conditions. Fibrinolysis-tPA-induced fibrinolysis was performed as described by Bajzar et al. (38) with thrombin-activated, plasma-derived PC or rPC. The reaction was performed in a 96-well plate. Human plasma was dialyzed against 20 mM HEPES (pH 7.41, 150 mM NaCl and then diluted one-third with the same buffer. Aliquots of plasma were added to each well of the plate containing CaCl,, human melanoma tPA, and thrombin at final concentrations of 10 mM, 0.6 nM, and 6 nM, respectively, in a final volume of 250 pl in the absence and presence of 20 nM APC. The profile of clotting initiated by thrombin and subsequent fibrinolysis induced by tPA was measured by monitoring the turbidity at 405 nm at 3-min intervals in a microplate reader. The lysis time was judged as the time required for turbidity to decrease to 50% of plateau. Calcium Dependence of PC Actiuation-Activation of PC by 20 nM thrombin or by 2 n~ thrombin, 20 nM thrombomodulin (preincubated for 5 min at 37 "C) was performed by incubation of PC (0.5 PM) with thrombin or the thrombin-thrombomodulin complex (plus or minus 100 p~ PCPs) in 20 mM Tris (pH 7.41, 150 mM NaCl, and 0.1% gelatin in the presence of varying concentrations of CaCl, a t 37 "C. At different time intervals after the start of activation, 30-pl aliquots were stopped by the addition of 10 pl of 0.5 p~ antithrombin I11 and 150 unitdm1 heparin. APC was quantified by adding 100 p1 of chromogenic substrate PCa (0.8 mM) and recording the increase in absorbance a t 405 nm as a function of time. The activity of APC was defined as the change in absorbance/ minute, and the initial rate of APC production was obtained by determining the slope of change in absorbance/minute uersus time of activation.
Epitope Mapping of PC-Binding of PC to monoclonal antibody H-11 (18) was performed in both solid and solution phase. All solutions were Molecular Basis of Protein CW.m,,,n, Hind111 I ,""""""""""  For solid-phase hinding, a 96-well plate (Corning) was coated with 100 n g of plasma-drrived PC or rPC overnight at 4 "C and hlocked with 5% RSAin TRS (pH 7.4) for 2 h a t room temperature. After three washes with TRS, the plate was incuhated overnight at 4 "C with 100 pI of H-11 (10 pg/ml) in 3'% RSA/rRS with varying concentrations of CaCI,. H-11 bound to thc plate was detected by incuhation with 100 p1 of a 1:10,000 diluted anti-mouse I&-prroxidase conjugate (Sigma, A4419) in 1% RSAfl'RS at room temprrature for 3 h, followed by color development to PC was expressrd as the percrntagr ofahsorhancr compared with the ahsorhance without CaCI,.

RESULTS
PCR-mediated site-directed mutagenesis was used to generate two single mutants (Glu'" -* Ala and Val"' * Met) and the double mutant (Glu'" --Ala/ValRI * Met) of protein C (Fig. 1 ). Wild-type PC cDNA and corresponding segments with the above mutations were ligated into the SV40/ndeno-based mammalian expression vector pDX and used to derive stable cell cultures to express rPC. After the selection and screening, the expression levels of stable transfectants used for the production of recombinant protein were 2.8, 2.1, 4.0, and 1.6 pglrnl, for wild-type rPC and GIuZ'' .--. Ala, Val34 * Met. and Glu"' -AlaNal~"" -> Met mutant rPCs, respectively.
All forms of PC were purified to homogeneity based upon SDS-PAGE (Fig. 2). Chemical analysis of Gla and 8-hydroxyaspartate indicated that each construct was properly modified (Table I). NH,-terminal analysis of all forms of PC (first 10 amino acid residues) revealed the appropriate sequences of light and heavy chains for each cycle. indicating that the recombinant proteins were appropriately processed into twochain forms and that the propeptide hnd been properly removed (data not presented).
Both Protac-and thrombin-activated PCs were assessed by chromogenic substrate PCa hydrolysis as a function of time. The fully activated PC achieved similar specific activities with the chromogenic substrate for all of the plasma-derived and recombinant PCs. The turnover numbers of the synthetic substrate PCa were 153 * 5 and 137 * 6 ArnOD/midpg of APC for plasma-derived PC and four rPCs under our conditions, respectively. Based upon equal chromogenic substrate activity of each construct, APCs were adjusted to the same level of serine protease activity for use in the functional assays described below.
Protac-activated PCs with equal chromogenic activities were used for APTT and factor Va inactivation assays. The prolongation of clotting time for each APC species is shown in Fig. 3.  The wild-type rAPC prolonged clotting time was identical to that for plasma-derived APC, and the clotting time obtained with Val34 + Met mutant rAPC was slightly shorter than that obtained with wild-type rAPC. However, clotting times for G1uZ0 + Ala mutant APC and double mutant APC were the same as that for the buffer control alone. These results implicate the G1u2' mutation as the cause of PC dysfunction. Similar results were obtained for the APC inactivation of factor Va, as shown in Fig. 4. In the presence of 20 PM PCPs, factor Va retained 20% of the initial cofactor activity after 15 min of incubation with wild-type rAPC or plasma-derived APC, and Val34 + Met mutant rAPC inactivated factor Va identically to wild-type rAPC. In contrast, factor Va still retained 80% of the initial cofactor activity after 15 min of incubation with G1uZo + Ala mutant rAPC or double mutant rAPC (Fig. 4).
In addition to its anticoagulant activity, the tPA-induced fibrinolytic activity of APC was also tested in a cell-free system (38). The lysis time was shortened in the presence of plasmaderived APC or wild-type rAPC and VaP4 + Met mutant rAPC, but only slightly by the G1uZ0 single and double mutants (Fig.  5). The profibrinolytic effect ofAPC has been attributed specifically to inhibition of the activation of prothrombin in cell-free plasma (39). The decrease in fibrinolytic activity and the factor Va inactivation exhibited by G1uZ0 + Ala mutant APC in our studies are both consistent with this hypothesis.
All forms of PC were studied for activation by thrombin alone and by the thrombin-TM complex in both the absence and presence of phospholipid vesicles. Our results (Fig. 6 ) indicate that Factor Va cofactor activity was measured a s described under "Experimental Procedures." The 100% activity in the above system was 0.226 PM thrombidmin. The percentage cofactor activity value was the average from two independent experiments. The range of each point is smaller than the size of the point. thrombin alone activates all constructs into APC in a similarly slow pattern and that the thrombin-TM complex accelerates the activation process (data shown for only wild-type rPC and G1uZ0 + Ala mutant rPC). The presence of phospholipid vesicles had an accelerating effect on the activation process by the thrombin-TM complex for wild-type rPC (Fig. 6 A ) and Val34 + Met mutant rPC, similar t o that for plasma-derived PC (data not shown). In contrast, phospholipid vesicles had no effect on activation of G1uZ0 + Ala mutant rPC (Fig. 6 B ) and double mutant rPC (data not shown). The Caz+-dependent activation of wild-type rPC (Fig. 6 A ) and G1uZ0 + Ala mutant rPC (Fig. 6 B ) by the thrombin-TM complex is shown in Fig. 7. As indicated, activation by the thrombin-TM complex of wild-type rPC was hyperbolic as a function of CaC1, concentration in the absence of phospholipid, whereas the addition of PCPs changed the CaC1, dependence into a sigmoidal curve. Similar curves were obtained for Val34 + Met mutant rPC (data not shown). In contrast, hyperbolic curves were obtained for G1uZo + Ala mutant rPC and double mutant rPC (data not shown) in the pres- ence of phospholipid (Fig. 7). Table I1 summarizes the parameters for the CaCl, dependence of the thrombin-TM complex activation of plasma and recombinant PCs, including derived Hill coefficients (40). These data suggest that the ability of PC to undergo a Ca'+-dependent conformational change associated with phospholipid interaction has been lost upon mutation of G1uZ0.
The loss of Ca2+-induced conformational change upon G1uZo mutation is also seen by epitope mapping of PC by the calciumdependent, lipid-independent monoclonal antibody H-11 (18, 28) directed against the Gla domain of vitamin K-dependent proteins. As shown in Fig. 8, increasing concentrations of CaC1, abrogated binding of H-11 to wild-type rPC and Val34 + Met mutant rPC, but not to G1uZ0 + Ala mutant rPC and double mutant rPC. Both solid-and solution-phase assays gave similar results (Fig. 8). CaC1, concentration midpoints for H-11 binding were approximately 3.5 mM for wild-type rPC, 5 mM for Val34 + Met mutant rPC, and >20 m~ for G1uZ0 + Ala mutant rPC and double mutant rPC.

DISCUSSION
Two missense mutations (G1uZ0 + Ala and Val34 4 Met) found in a kindred (protein CVemont) with severe type I1 protein C deficiency and exhibiting a high penetrance of PC-associated familial thrombosis were the first reported naturally occurring mutations in the Gla domain of PC (13). To understand the structural and functional consequences of the mutations, both appropriate single mutant, double mutant, and wild-type PC recombinant proteins were expressed in human kidney 293 cells and purified to homogeneity. Chemical characterization indicated that the recombinant proteins appeared to be properly processed in regard t o y-carboxylation of glutamic acid residues, P-hydroxylation of aspartic acid, and proteolytic processing. SDS-PAGE also indicated that a similar amount of glycosylation had also occurred in all forms.
Abolition of clotting time prolongation in the APTT assay in the case of the G1uZo mutants and only a minor change for the VaP4 mutant indicate that the mutation G1uZ0 + Ala is primarily responsible for the dysfunction of PC in the affected family members. I n vitro studies involving the inactivation of purified human factor Va more specifically address the issue of G1uZ0 mutation. Cleavage by APC at the lipid-dependent Ar$06 site of the human factor Va heavy chain, required for full inactivation of factor Va cofactor activity, is markedly decreased for the G1uZ0 mutant (41). Finally, in a cell-free system (381, the GluZo mutants are ineffective in enhancing tPA-induced fibrinolysis. Consequently, the results from both the anticoagulant and fibrinolytic studies strongly suggest that the physiological defect in the reported naturally occurring PC variant is the lack of Glaze with subsequent loss of the Ca2+-dependent conformational change and lipid association. This interpretation is consistent with reports from Zhang et al. (19) and Zhang and Castellino (20,21) involving the systematic replacement of amino-terminal Glu residues by Asp in PC.  Our results suggest that GlaZ0 also plays an important role in the activation of PC. In our studies, all forms of PC exhibited Ca2+ concentration-dependent hyperbolic activation by the thrombin-TM complex ( Fig. 7 and Table 11) in the absence of phospholipid. However, activation in the presence of phospholipid resulted in sigmoidal curves for wild-type rPC and Val34 + Met mutant rPC and Hill coefficients significantly greater than 1.0, indicating complex conformational interactions (42). This is in contrast to retained hyperbolic curves and Hill coefficients =1 for both G1uZ0 + Ala and G1uZ0 + AlaNa134 + Met mutant rPCs. The implication of these observations is that differences in calcium-dependent activation and conformational change of naturally occurring mutant PC are due to the mutation at G1uZ0. The results (Figs. 6 and 7 ) also suggest that the lipiddependent component of PC activation by the thrombin-TM complex has been lost upon G1uZo mutation, but not the lipidindependent component. Similar curves to those presented in Fig. 7 (44). Extending these results t o our study may offer a plausible explanation for the apparent severity and high penetrance of thrombotic disease in family members with the G1uZ0 mutation. The lipid-independent binding of Gla-domainless PC (44) or G1uZo + Ala mutant PC (this study) to the thrombin-TM complex appears to be the same as for the wild-type molecule.
If this were the case in vivo under conditions in which the thrombin-TM complex is limiting relative to PC concentration, binding of mutant PC would have the effect of a competitive substrate inhibitor of wild-type PC binding and activation. Consequently, G1uZo + Ala mutant PC, in addition to being ineffective in converting membrane-bound factors Va and VIIIa to their inactive forms, would also compromise the thrombin-TM activation of normal circulating PC by acting as a competitive inhibitor. Finally, epitope mapping by calcium-dependent antibody H-11 provides additional insight into the calcium ion-induced conformational change of the Gla domain of PC. Ca2+ has previously been shown to abrogate binding of H-11 to PC, but has no effect on H-11 reactivity toward descarboxyl-PC (28). In our studies, Ca2+ reduces binding of wild-type rPC and Val34 + Met mutant rPC to H-11, whereas binding of both G1uZo mutants to H-11 is significantly less responsive to the presence of Ca". This suggests that the G1uZ0 mutants lack the Ca2+-induced, lipid-independent conformational change of the PC molecule involved in H-11 recognition; and taken with the above activation results, may represent the same conformational change required for the lipid-independent component of PC activation by the thrombin-TM complex.
Energy minimization (45,46) has been used to generate a three-dimensional model for the PC Gla domain in the presence of Ca2+ ion based upon the x-ray structure for the bovine prothrombin fragment 1-Ca2+ complex (47). The resulting structure is not dramatically different from that for bovine prothrombin fragment 1.' Based upon the derived structure for PC, it is possible to structurally interpret changes in the functional properties of the mutant molecules reported in this study. The x-ray structure for the bovine prothrombin fragment 1-Ca2+ complex indicates that GlaZ1 (residue 20 in PC) critically interacts with both surface-exposed Ca-6 and buried Ca-5 (19). Exposed Ca-6 and Ca-7 have been suggested to be involved in Ca2+-mediated, negatively charged lipid membrane surface-dependent processes (47). Buried Ca-1 through Ca-5, on the other hand, have been proposed to be involved in an essential Gladependent conformational change, which does not involve lipid membrane. The model predicts that in the case of PC, GlaZ0 is similarly involved in the binding of Ca-5 and Ca-6 and that mutation at this site would alter both lipid-dependent and -independent properties involving Ca2+ ions. Specifically, we propose that the observed changes in APTT, factor Va inactivation, fibrinolytic activity, and a portion of the activation process, all requiring negatively charged lipid membrane, are due to the loss of exposed Ca-6 binding upon mutation of G1uZ0. Furthermore, we propose that changes in H-11 monoclonal antibody binding and lipid-independent PC activation are due to the loss of GlaZ0 interaction with buried Ca-5 and the associated lipidindependent conformational change of the Gla domain. In regard to the naturally occurring mutation, also including Val34 -Met, the x-ray structure predicts that Met34 in PC resides in a loop joining two a-helixes of the Gla domain, with the R group distal from any of the Ca2+ ions (47). Consequently, this mutation would be expected t o have little effect on the structure and function of the protein, consistent with our experimental observations.
In summary, this study has investigated the functional significance of each of two naturally occurring mutations in the Gla domain of protein C in a kindred exhibiting severe type I1 familial PC deficiency and thrombosis. The results indicate that the functional abnormalities can be attributed to the G1uZ0 + Ala mutation and consequent decrease in thrombin-TM activation of PC on the vascular endothelial surface as well as diminished intrinsic APC anticoagulant and profibrinolytic activity.