Modifications of Bovine Prothrombin Fragment 1 in the Presence and Absence of Ca(I1) Ions LOSS OF POSITIVE COOPERATIVITY IN Ca(I1) ION BINDING FOR THE MODIFIED PROTEINS*

Chemical modification of bovine prothrombin fragment according to the 27,4946-4952 and provided a series of fragment 1 derivatives in which various nitrogen-containing side chains were N-acetylated and/or N-2,4,6-trinitro-phenylated. In addition the des-[Ala-l,Asn-2]- and des- [Ala- 1 ,Asn-2,Lys-3]-fragment 1 derivatives were pre-pared by limited enzymatic hydrolysis of fragment 1 using cathepsin C and plasmin, respectively. Quanti-tative studies on the Ca(I1) binding of these proteins have been accomplished using 4SCa(II) equilibrium dialysis. Binding of these fragment 1 derivatives to phosphatidylserine/phosphatidylcholine (PS/PC) vesi- cles (25:76) in the presence of Ca(I1) ions has been studied using the light-scattering technique. Acylation of the 6 lysine residues of fragment 1 by the action of acetic anhydride (500-fold molar excess) in the presence of 76 mM Ca(II), pH 8.0, results in loss of positive cooperativity in Ca(I1) binding (Scatchard plot) and an increase in the number of Ca(I1) ions bound. The


LOSS OF POSITIVE COOPERATIVITY IN Ca(I1) ION BINDING FOR THE MODIFIED PROTEINS*
(Received for publication, September 11,1991) David J. Weber, Paula Berkowitz, Mary G. Panek, Nam-Won Huh, Lee G . Pedersen, and Richard G. Hiskey Chemical modification of bovine prothrombin fragment 1 according to the procedure of D. J. Welsch and G . L.  [Biochemistry 27,4946-4952 and earlier papers] provided a series of fragment 1 derivatives in which various nitrogen-containing side chains were N-acetylated and/or N-2,4,6-trinitrophenylated. In addition the des-[Ala-l,Asn-2]-and des-[Ala-1 ,Asn-2,Lys-3]-fragment 1 derivatives were prepared by limited enzymatic hydrolysis of fragment 1 using cathepsin C and plasmin, respectively. Quantitative studies on the Ca(I1) binding of these proteins have been accomplished using 4SCa(II) equilibrium dialysis. Binding of these fragment 1 derivatives to phosphatidylserine/phosphatidylcholine (PS/PC) vesicles (25:76) in the presence of Ca(I1) ions has been studied using the light-scattering technique.
Acylation of the 6 lysine residues of fragment 1 by the action of acetic anhydride (500-fold molar excess) in the presence of 76 mM Ca(II), pH 8.0, results in loss of positive cooperativity in Ca(I1) binding (Scatchard plot) and an increase in the number of Ca(I1) ions bound. The Ca(I1)-dependent PS/PC binding of the acylated protein is reduced. Removal of 2 and 3 residues from the amino terminus likewise leads to loss of positive cooperativity in Ca(I1) binding and reduced binding affinity to PS/PC vesicles. The important role of the amino-terminal 1-10 sequence is discussed. We conclude that positive cooperativity in Ca(I1) binding is not a prerequisite for the Ca(I1)-dependent binding of bovine prothrombin fragment 1 to PS/PC vesicles.
The vitamin K-dependent carboxylation of specific glutamic acid residues in the amino-terminal region of several coagulation and anticoagulation proteins forms a negatively charged domain that contains the essential y-carboxyglutamic acid (Gla)' residues (1)(2)(3)(4)(5)(6). These proteins require Ca(I1) for physiological function, and several essential coagulation steps * This work was supported by Grants HL-27995 (to L. G. P.), HL-20161 (to R. G. H.), and HL-06350 (to R. G. H. and L. G. P.) from the National Institutes of Health. 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.
Recently  reported a series of studies involving the action of acetic anhydride (pH 8.5) on bovine fragment 1. In the absence of metal ions the a-NH, group at Ala-1 and the t-NH, groups of the 5 lysine residues at positions 3, 11, 44, 57, and 97 were acylated.
Under these reaction conditions asparagine 101 was converted to the imide derivative by acetylation of the nitrogen atom of the side-chain carboxamide group. This protein did not bind to PS/PC vesicles. When the acetylation was conducted in the presence of Ca(I1) ions, the a-NH, of Ala-1 and Asn-101 were protected from acylation. This protein derivative displayed slightly diminished PS/PC binding. Acetylation in the presence of Mg(I1) protected Asn-101 from acylation but did not protect the a-NH, of Ala-1. This protein did not bind to PS/PC vesicles. Welsch and Nelsestuen (19) report that His-96 is irreversibly modified by acylation in the presence or absence of Ca(I1) ions. Thus this modification has no effect on Ca(I1)-dependent binding of the protein to PS/PC vesicles. The reaction of bovine fragment 1 with TNBS was also evaluated in these studies. Surprisingly, the TNP-F-1 exhibited almost normal binding to PS/PC vesicles despite the attachment of a TNP group at the NH2-group of Ala-1. In order to investigate the role of the amino-terminal region we have examined several modifications at this portion of the fragment 1 molecule.   (23). The data are tabulated in Table 11. Bovine fragment 1 (entry 1, Table 11) binds seven Ca(I1) ions when measured under these conditions (31). The data are best fit to a model which involves three cooperative and four equivalent noninteracting Ca(I1) sites (23).
(u' = moles of protein bound/moles of phospholipid). The phospholipid binding isotherms are shown in Figs. 4 and 6.
Values for N (moles of protein bound/mole of phospholipid) are also listed for each modified protein (Table 11). Error bars indicate the 95% confidence level as obtained by SAS NLIN option analysis of the reciprocal plots from separate determinations. Acylation of fragment 1 in the presence of Ca(I1) ions (entry 3, Table 11, and Fig. 2, inset) results in the modification of the NH: of the 5 lysine residues. The positive charge of the protein at pH 7.4 is reduced, and the molecule becomes more negatively charged at this pH. The modified protein, H2Na-Ala-l-(N'Ac-Lys)S-F-l, binds an increased number of Ca(1I) ions relative to fragment 1. Positive cooperativity of Ca(I1) binding is abolished; however the affinity of the nine equivalent, noninteracting sites is relatively high (kSie = 1270 M-'). The modified protein retains the ability to bind to PS/PC vesicles although the Kd value is reduced relative to fragment 1. Acylation of the NHg of Ala-1 (entry 4) and the lysine sidechains reduces Ca(I1) affinity relative to entry 3 and abolishes positive cooperativity in Ca(I1) binding. The modified protein does not bind to PS/PC vesicles. Acetylation of fragment 1 in the absence of metal ions at pH 8.0 (entry 5, Table 11, and Fig. 2) possibly removed a Ca(I1) site, (relative to entry 3 or 4), the positive cooperativity in Ca(I1) binding, the PS/PC binding, and the metal ion promoted quenching of the intrinsic fluorescence. It is probably fair to say that the measured Ca(I1) stoichiometry generally has an error of f l ; thus differences in stoichiometry of +1 are not necessarily significant.
Acylation of TNP-F-1 in the presence of Ca(I1) ions was expected to lead to little change in the properties of the product. However, the resulting product (entry 13, Table 11) exhibited an apparently reduced Ca(I1) stoichiometry and a reduced PS/PC affinity relative to entry 3 or 8. As expected the reaction of TNP-F-1 with acetic anhydride in the absence of Ca(I1) (entry 14, Table 11) abolished PS/PC binding. The Ca(I1) stoichiometry and kSit, values of entry 5 and 14 were similar. Thus acylation of fragment 1 (entry 5) or the TNP-F-1 derivative (entry 14, Table 11) in the absence of Ca(I1) ions yields proteins that are quite similar in Ca(I1) binding properties to reduced and carboxymethylated fragment (compare entries 5 and 14 to entry 2, Table 11).

DISCUSSION
The principal focus of this study was to quantitatively evaluate the effects of chemical modification at the amino groups of bovine fragment 1 on Ca(I1) and PS/PC binding. Several conclusions regarding the "Ca conformation" required for Ca(I1)-dependent binding of fragment 1 to PS/PC vesicles emerge from this study.
We have previously suggested that the positive cooperativity observed in Ca(I1) binding to bovine fragment 1 and other vitamin K-dependent proteins was a prerequisite for the Ca(I1)-conformation required for Ca(I1)-mediated binding to PS/PC vesicles. We based this argument on: 1) the absence of observed cooperativity in "Mg(I1) binding to fragment 1 (23) and the lack of PS/PC binding by the fragment l-Mg(II) complex; and 2) on the properties of 7,8,. The latter studies indicated that the "Mg(I1) binding to the 7,8,33-yMGlu-F-l was identical to "Mg(I1) binding to fragment 1. The metal ion-promoted quenching of the intrinsic fluorescence was retained by the 7,8,33-Gla-modified protein; however this protein had lost positive cooperativity in Ca(I1) binding, exhibited reduced Ca(1I) affinity and, importantly, did not bind to PS/PC vesicles. The acylation studies of ) that lead to H2Na-Ala-1-(N-Ac-Lys),-F-l upon acylation in the presence of Ca(I1) ions and N"-Ac-Ala-l-(N-Ac-Lys),-F-l in the presence of Mg(I1) ions also support the idea of different conformational populations induced by the interaction of different ions (calcium and magnesium with different charge densities, coordination states, and ligand preferences) with fragment 1. Nevertheless, it is clear from the data reported in Table I1 that positive cooperativity in Ca(I1) binding to the various fragment 1 derivatives is not a precondition reflective of, or required for, PS/PC binding. Furthermore, the Ca(I1) affinity of the particular protein does not appear to be a good predictor of PS/ PC binding. For example, des-[Ala-l,Asn-2]-F-l (entry 6, Table 11) exhibits a kSi, value of 330 M" which is quite similar
However, the former protein binds to PS/PC vesicles, whereas the latter does not. Thus the ability of fragment 1 to exhibit Ca(I1)-mediated phospholipid binding must depend on conformational features that are mediated by the Ca(I1) binding process. Positive cooperativity in Ca(I1) binding to bovine fragment 1 involves three of the seven Ca(I1) sites (23). The positive cooperativity is abolished when the amino-terminal region of the protein is modified. For example, positive cooperativity of Ca(I1) binding is abolished by: removal of 2 or 3 residues from the amino terminus; modification of Gla residues 7 and 8; introduction of a N"-TNP or N"-acyl group at the amino terminus; or acetylation of the N-amino groups of the 5 Lys residues. Since positive cooperativity is not observed in Mg(I1) binding to the protein, the phenomena must involve some property of Ca(I1) ions as well as some property of the ligand (fragment 1).
Hodgson et al. (32,33) have suggested that the major reason for the presence of Gla rather than Glu in the family of Ca(I1) binding coagulation proteins is the availability in the former of the additional carboxylate group. Hodgson et al. note that the additional carboxylate permits polymeric arrays of Ca(I1) ions in the crystallographic structures of malonate complexes with Ca(I1) or Ba(I1). Malonate complexes with smaller metal ions such as Mg(I1) or Be(I1) are invariably monomeric. Polymeric arrays were also observed by Zell et al. (37) in their Ca(I1) a-ethylmalonate crystal study.
One might expect to observe positive cooperativity in the formation of a polymeric system involving two or more Ca(I1) ions that interact through a network of carboxylate ligands. The positive cooperativity observed with Ca(I1) binding to fragment 1 might reflect the formation of such a polymeric matrix involving three Ca(I1) ions. Each Ca(I1) ion would be bound to two or more Gla carboxylate groups in hepta-or octacoordinate arrays as observed by Hodgson et al. (32,33). Each Ca(I1) ion would interact with the negative ligands ( i e . Gla carboxylates) positioned by the previous Ca(I1) ion leading to the observed positive cooperativity associated with Ca(I1) binding by the native protein.
The negatively charged Gla carboxylates in the protein can be stabilized by Ca(I1) ions and/or by salt bridges with available amino groups. Chemical modification of the amino groups will affect salt bridge formation and might also affect interactions between the bound Ca(I1) ions. Thus, acetylation of fragment 1 in the presence of Ca(I1) ions might be expected to yield a modified protein (entry 3, Table 11) which retained the essential Ca(I1)-promoted interactions between Ca(I1) ions, amino groups, and the Gla carboxylate groups. Viewed in this way, the a-NH, group of Ala-1 must be an integral feature of the polymeric Ca(I1)-Gla domain matrix. The involvement of the a-NH, group of Ala-1 in the positive cooperativity of the Ca(I1) binding process is evident from a comparison of fragment 1 (entry 1, Table 11) with either des-[Ala-l,Asn-2]-F-l (entry 6, Table 11) or des-[Ala-l,Asn-Z,Lys-31-F-1 (entry 7, Table 11). Removal of Ala-1 abolishes the cooperativity of Ca(I1) binding. The involvement of the t-NH, groups of Lys residues in the Ca(I1) binding process is seen by comparison of the Ca(I1) binding stoichiometry of the W-acetyl-or N-TNP fragment 1 derivatives. Modification of the t-NH, groups leads to an increase in Ca(I1) ions bound by Gla carboxylate groups as the Gla-t-NH, salt bridges are abolished. Cooperativity in Ca(I1) binding is lost in all cases suggesting that Gla-Lys salt bridges are involved in this event as well.
antibody, H-11, which binds to a conserved epitope involving Phe-5 in the bovine prothrombin sequence. Binding of H-11 is inhibited by Ca(II), Mg(II), and Mn(I1) ions. Church et al. (34,35) conclude that Phe-5 is buried and thus inert to H-11 upon divalent metal ion binding to bovine fragment 1. Thus the amino-terminal 1-10 sequence must adopt an ordered conformation as a result of divalent ion binding. This proposal is supported by the present studies which indicate that subtle changes in the 1-10 region such as modification of Ala-1, acetylation of des-[Ala-1,Asn-21-F-1 or des-[Ala-l,Asn-2,Lys-31-F-1, and conversion of Gla-7,8 to y-methyleneglutamyl residues can abolish or modify PS/PC binding by the protein.
We wished to demonstrate that the PS/PC binding exhibited by the TNP derivatives of the various fragment 1 derivatives was in fact a Ca(I1)-dependent process. Two control experiments indicate the Ca(I1) requirement. In the first experiment EDTA was added to the Ca(I1). TNP-F-1. PS/PC binding complex. The complex immediately dissociated. The second control utilized the modified protein, (14) in which all Gla residues are modified. This protein does not bind to PS/PC vesicles in the presence of Ca(I1) ions. Treatment of this protein with excess TNBS, isolation of the TNP-10-y-MGlu-F-1, and incubation with Ca(I1) and PS/PC vesicles gave no indication of protein-phospholipid binding. Thus the binding observed by incorporation of a TNP residue at Ala-1 or Lys-3 or Gly-4 in the appropriate fragment 1 derivative must reflect the importance of the Ca(I1):Gla interactions in establishing the phospholipid binding conformation. The effect of a dinitro-or trinitrophenyl substituent at the amino terminus on PS/PC binding also suggests the presence of an ordered conformation in the 1-10 region of the protein. Recent studies by Schwalbe et al. (36) indicate that Ca(I1) but not Mg(I1) protected the amino terminus of the amino-terminal peptides from human I1 (residues 1-41 and 1-44, 60:40, v/v), bovine X (residues 1-44), and bovine IX (residues 1-42). The introduction of a TNP group at the amino-terminal residue of these peptides also promoted Ca(I1)-dependent binding of the peptides to PS/PC vesicles.
Acylation of fragment 1 in the presence of Mg(I1) or Ca(I1) protected the P-amido nitrogen atom of Asn-101 from imide formation. Acylation in the absence of metal ions leads to the production of the Asn-101 imide and perhaps modification at other sites. Acylation under these conditions results in the loss of the intrinsic fluorescence transition and PS/PC binding. Evidence suggesting a role for a non-Gla domain metal ion binding site in this region of the kringle domain is presented in the accompanying manuscript (30). binder 1/N = 71.42; K, = 2.5 X loe M"; K, = 0.4 p~.