The interaction of prothrombin and factor X with membranes of widely varying composition is reported. These protein-membrane interactions require the presence of acidic phospholipid; phosphatidylserine is the most effective from the standpoint of protein-membrane affinity. The maximum protein binding capacity is proportional to the phosphatidylserine content of the membrane up to about 15% and the stoichiometry is 9.2 +/- 1.3 phosphatidylserine residues per prothrombin molecule and 5.2 +/- 1.5 per factor X molecular. The protein apparently causes clustering of the phosphatidylserine residues in these membranes of low phosphatidylserine content. Above about 20% phosphatidylserine, the factor limiting protein-membrane interaction appears to be protein packing density on the membrane surface. The maximum protein binding capacity at 50% phosphatidylserine is 1.2 g of protein per g of phospholipid. Phosphatidic acid is similar to phosphatidylserine in its ability to bind these proteins except for somewhat larger dissociation constants. Phosphatidylethanolamine and phosphatidylglycerol are less effective in all respects for promoting these protein-membrane interactions. Small amounts of phosphatidylserine mixed with these latter phospholipids, however, have a major effect on protein-membrane binding so that the dissociaction constants are more characteristic of membranes of high phosphatidylserine content.

Membrane-binding properties of human and bovine forms of vitamin K-dependent proteins Z, S, and C were characterized. Each of these proteins showed unique properties and interspecies differences that correlated with specific amino acid sequence variations in the amino-terminal 45 residues. Protein Z showed 100-fold slower membrane binding and dissociation kinetics relative to other vitamin K-dependent proteins that have been tested. This property seemed to correlate with an extra gamma-carboxyglutamic acid (Gla) residue at position 11 of protein Z. The interspecies difference for protein Z consisted of a higher packing density for the bovine protein on the membrane and a 9-fold slower dissociation rate. Higher affinity correlated with Asp at position 34 of bovine protein Z, where the human protein contains Asn. While both protein S species showed high affinity for the membrane, it was significantly greater for the human protein versus bovine protein S. Again, higher affinity correlated with an Asp (vs Asn) at position 34. Protein C was characterized by binding affinities that were 100-1000-fold lower than the other proteins. Low affinity appeared to be related to loss of Gla-32 (homologous to Gla-33 of protein Z). Interspecies differences of protein C appeared to be related to proline at position 10 (homologous to position 11 of protein Z) of bovine protein C, which produced at least 10-fold lower affinity than the human protein. Comparable substitutions at positions homologous to 11, 33, and 34 of protein Z may also underlie membrane binding behaviors of other vitamin K-dependent proteins. The three-dimensional structure of strontium-prothrombin fragment 1 [Seshadri et al. (1994) Biochemistry 33, 1087] shows that these positions are clustered on the protein surface near strontium-8, another possible candidate for membrane contact. A membrane contact mechanism consisting of an isolated protein-lipid ion pair is proposed. Comparison of naturally occurring vitamin K-dependent proteins has provided possible bases for divergent membrane binding and suggested future approaches to determine biological function.

The study of prothrombin- and factor X-membrane interaction by light-scattering intensity measurements at 90 degrees is reported. This technique, which uses a fluorometer as a light-scattering photometer, can be applied to measurement of free and membrane-bound protein concentrations, from which equilibrium constants can be obtained. The following equilibria adequately describe the observed properties of prothrombin-membrane interaction (formula: see text) where P and PL are protein and phospholipid, PiCa and PLjCa are the calcium complexes, P'iCa is the protein after undergoing a calcium dependent transition, and P'-PLi+j+mCa is the protein-membrane complex. Several lines of evidence indicate that i, j, and m are interrelated and m decreases to 0 when i and j are saturated. In agreement with this, direct calcium binding measurements indicate m values of 3.2 +/- 1.5 and 1.1 +/- 1.5 at 0.5 and 1.2 mM calcium, respectively. The total number of functional calcium ions in the complex (i + j + m) is 6 to 9 based on Hill coefficients for the reactions and direct calcium binding measurements. In reaction 3, the maximum stoichiometry of calcium per acidic phospholipid is 1:2. While the details of factor X-membrane binding were not determined in quite as great detail, the equilibria (identified) appear the same but a major difference is the calcium concentration needed to initiate protein-membrane binding.