Magnesium(II) is a crucial constituent of the blood coagulation cascade. Potentiation of coagulant activities of factor IX by Mg2+ ions.

We recently showed that not only Ca ions but also Mg ions play a crucial role in stabilizing the native conformation of coagulation factor IX. We here report that Mg ions at physiological concentrations greatly augment the biological activities of factor IX. In clotting assays with dialyzed plasma, addition of Mg ions enhanced the apparent coagulant activity of factor IXa, while that of factor Xa was scarcely affected. Activation of factor X by factor IXa in the presence of factor VIIIa, phospholipids, and Ca ions was accelerated by Mg ions. It appeared that the cation increased the affinity between factor IXa and factor VIIIa, thereby increasing the apparent catalytic efficacy of the enzyme. We also evaluated the effect of Mg ions in the coagulation pathway initiated by tissue factor and found that activation of factor IX by factor VIIa•tissue factor was accelerated by the cation. Consequently, clotting of normal plasma induced by factor VIIa•tissue factor was shortened by the cation, while no such effect was observed in plasma deficient in factor IX or VIII. These results indicate that the previously unrecognized plasma component, Mg ions, plays crucial roles in blood coagulation and, moreover, that contributions of factors IX and VIII in the coagulation cascade have been seriously underestimated in previous investigations.

The coagulation proenzyme factor IX plays a central role in hemostasis in concert with its essential procofactor factor VIII, as is apparent from the fact that genetic defects in these factors cause a life-threatening tendency to bleed, hemophilia (1). Once activated, factor IX forms a complex with the activated form of factor VIII (IXa⅐VIIIa) 1 on phospholipid membranes in the presence of Ca 2ϩ ions. The enzyme complex catalyzes efficient activation of factor X, and factor Xa in turn converts prothrombin to thrombin, which ultimately leads to successful formation of hemostatic plugs. In the classical cascade model of coagulation, factors VIII and IX are classified as components of the "intrinsic pathway" and are placed downstream of factor XI (Refs. 2 and 3; for recent review, see Refs. 4 and 5). However, it is now considered that the intrinsic pathway is insignificant in normal hemostasis, since individuals who lack initiators of this mechanism, i.e. factor XII (6), high molecular weight kininogen (7), and prekallikrein (8), are all asymptomatic. To today's best knowledge, the single mechanism that solely governs blood coagulation is the so-called "extrinsic pathway," which starts with the exposure of the extravascular component, tissue factor, to the bloodstream and formation of the active factor VIIa⅐tissue factor complex (VIIa⅐TF). The complex is able to activate factor X without a requirement for factors VIII and IX. Why, then, do hemophiliacs experience such severe hemorrhagic diatheses? This question was partly answered by the finding that VIIa⅐TF can activate not only factor X but also factor IX (9). Since then, evidence has accumulated for the indispensable roles of factors VIII and IX in the coagulation pathway initiated by tissue factor. Komiyama et al. (10) showed that factor IX is preferred to factor X as a substrate for VIIa⅐TF. More recently, Mann and co-workers (11,12) demonstrated that the VIIa⅐TF-induced thrombin generation is strongly dependent on factor VIII. These studies clearly demonstrated the significance of factors VIII and IX. With these observations in mind, we attempted to validate the importance of these proteins considering a previously unrecognized factor, Mg 2ϩ ions.
Nonetheless, we found recently that folding of the native tertiary structure of factor IX requires not only Ca 2ϩ but also Mg 2ϩ ions (13). The Ca 2ϩ -dependent binding of various conformation-specific probes was greatly enhanced by Mg 2ϩ ions at physiological concentrations even in the presence of excess Ca 2ϩ ions. This result suggests that factor IX has a specific binding site(s) for Mg 2ϩ ions that does not interact with Ca 2ϩ ions, and, moreover, that the binding of Mg 2ϩ ions promotes additional changes in the Ca 2ϩ -bound conformation of the molecule. We failed to detect such Mg 2ϩ -induced additional conformational changes in other vitamin K-dependent coagulation factors, such as factor X. Moreover, the Ca 2ϩ -dependent activation of factor IX by the intrinsic pathway activator factor XIa was accelerated by Mg 2ϩ ions with a dramatic reduction in the apparent K m (13). Thus, the Mg 2ϩ -induced additional conformational change appears to enhance the protein's ability to function. The activation by factor XIa is, however, not essential for physiological coagulation (11). Therefore, we examined the effect of Mg 2ϩ ions on the function of factor IX in the coagulation pathway initiated by VIIa⅐TF. Our results shed light on the indispensable roles of factors VIII and IX in vivo.

EXPERIMENTAL PROCEDURES
Materials-Plasma deficient in factor IX or factor VIII was from George King Biomedical. Phospholipids (phosphatidylcholine (PC) from egg yolk, phosphatidylethanolamine (PE) from egg yolk, and phosphatidylserine (PS) from bovine brain, all Ͼ98% pure), and bovine serum albumin (BSA) were from Sigma.
Proteins-The coagulation factors used were all of human origin. The following proteins were prepared by published methods: factor IX and factor X (14), factor IXa␤ (15), factor Xa (16), and ␣-thrombin (17). * This work was supported in part by Grant-in-Aid for Scientific Research 07770091 from the Ministry of Education, Science and Culture of Japan. 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.
Factor XIa was a generous gift from Dr. Yutaka Komiyama of Kansai Medical University (Osaka, Japan). Factor VIIa and placenta-derived tissue factor apoprotein were supplied by Dr. Tomohiro Nakagaki of Chemo-Sero Therapeutics Research Institute (Kumamoto, Japan). Recombinant factor VIII ("Kogenate") was a product of Bayer.
Preparation of Phospholipid Vesicles and Reconstitution of Tissue Factor-Phospholipids in chloroform were mixed at desired stoichiometry and dried under N 2 stream in a glass tube. They were suspended in 20 mM Tris-HCl, 140 mM NaCl, pH 7.5 (TBS), at 1 mg/ml, and were dispersed by sonication in a bath-type sonicator for 30 min. Incorporation of tissue factor apoprotein into the vesicles was achieved by a mixed detergent/phospholipid micelle technique (18). The stock solution of tissue factor in 2% Triton X-100 was diluted (to 0.01-1 nM) with TBS containing 5 mM Ca 2ϩ and mixed with the suspension of phospholipid vesicles (typically 800 M). The concentration of detergent was adjusted to 0.06%, and the mixture was incubated at 37°C for 30 min. The suspension of relipidated tissue factor was diluted appropriately with TBS plus 1 mg/ml BSA (TBS/BSA).
Clotting Assay-Pooled plasma was prepared by mixing citrated platelet-poor plasma from several normal healthy individuals. The plasma was dialyzed against TBS for 2 h at ambient temperature and stored frozen at Ϫ80°C. Factor VIII-or IX-deficient plasma was dialyzed similarly prior to use. One part (50 l) of the dialyzed plasma was mixed with one part of the phospholipid suspension and one part of the solution of coagulation initiator in TBS/BSA, and the mixture was incubated at 37°C for 2 min. Clotting was initiated by the addition of one part of 10 mM Ca 2ϩ (plus 4 mM Mg 2ϩ ). The time required for clot formation was measured in an Amelung Coagulometer KC 4A. For VIIa⅐TF-induced clotting, plasma was preincubated with factor VIIa, and the reaction was initiated by the simultaneous addition of the suspension of relipidated tissue factor and the Ca 2ϩ (plus Mg 2ϩ ) solution.
Chromogenic Assay for Factor Xa-The mixture of factor IXa, factor VIII, phospholipid vesicles (75% PC, 25% PS; w/w), and Ca 2ϩ (plus Mg 2ϩ ) in 80 l of TBS/BSA was placed in a 96-well assay plate and equilibrated at 37°C. Thrombin (10 l) was added to 0.1 unit/ml to activate factor VIII, and incubation was continued for 1 min. Then the reaction was started by the addition of 10 l of factor X and was terminated by the addition of EDTA after appropriate times (typically 5 min). Factor Xa was quantified by measuring its amidase activity with t-butoxycarbonyl-Leu-Gly-Arg-p-nitroanilide (Seikagaku Kogyo, Tokyo, Japan) as substrate; the substrate was added to 0.5 mM, and the initial rate of p-nitroaniline liberation was monitored at 405 nm in a kinetic plate reader (Well Reader; Seikagaku Kogyo), with pure factor Xa as the standard.
Activation of Factor IX-Factor IX (0.1 M in TBS/BSA) was incubated with 1 nM factor VIIa plus 0.1 nM tissue factor reconstituted in phospholipid vesicles (30 M 75% PC, 25% PS) in the presence of Ca 2ϩ (plus Mg 2ϩ ) at 37°C. The reaction was quenched at appropriate times by addition of 1/10 volume of 20 mM EDTA plus 1% Triton X-100. Detergent was included to attenuate tissue factor activity that would otherwise affect the assay. Factor IXa was quantified by a specific clotting assay as described earlier (13).

RESULTS AND DISCUSSION
We examined the effect of Mg 2ϩ ions on clotting. Pooled normal plasma was dialyzed to remove endogenous metal ions and the exogenously added anticoagulant, sodium citrate. It was then incubated with activated forms of coagulation factors and phospholipids. Clotting was initiated by addition of Ca 2ϩ ions (plus Mg 2ϩ ions) at concentrations close to physiological. When factor XIa was used as initiator, clotting times were reduced by Mg 2ϩ ions. Approximately 10-fold lower concentrations of factor XIa were sufficient to yield the same clotting times as seen in the absence of Mg 2ϩ ions (Fig. 1A). This result was reasonable because Mg 2ϩ ions potentiate factor IX activation by factor XIa (13). Next, we tested the effect of Mg 2ϩ ions on factor IXa-induced clotting. Shorter clotting times were again obtained with Mg 2ϩ ions (Fig. 1B). The apparent clotting activity of factor IXa was increased approximately 3-fold by Mg 2ϩ ions. By contrast, the cation had a minimal effect on factor Xa-induced clotting (Fig. 1C). It was thus clear that Mg 2ϩ ions accelerated not only factor IX activation by factor XIa but also factor X activation by factor IXa, while all the processes after the generation of factor Xa were unmodified. This result is consistent with our previous result that, among various vitamin K-dependent coagulation enzymes, only factor IX is responsive to Mg 2ϩ ions (13).
We next investigated the activation of factor X by factor IXa in the presence of factor VIIIa and phospholipids. Factor X at a physiological concentration (0.2 M) was activated in the presence of various concentrations of Ca 2ϩ ions ( Fig. 2A). The activation rate at any Ca 2ϩ concentrations was increased by Mg 2ϩ ions, while Mg 2ϩ ions alone were ineffective. When a 10-fold higher concentration of factor X was used, the amount of factor Xa generated and the effect of Mg 2ϩ ions were essentially the same (data not shown). These results indicated that the physiological level of factor X was considerably higher than the K m value so that the reaction could proceed at maximum velocity (V max ), and that Mg 2ϩ ions increased the catalytic efficacy (k cat ) of the enzyme. Indeed, the reported K m values of the IXa⅐VIIIa complex (approximately 0.06 M) are well below the plasma concentration of factor X (19,20). When we varied the concentration of factor VIIIa, a pronounced leftward shift of the required cofactor concentration was observed in the presence of Mg 2ϩ ions (Fig. 2B). No such shift was observed upon varying the concentration of phospholipids (Fig. 2C). It appeared that Mg 2ϩ ions augmented the coagulant activity of factor IXa by increasing the affinity of the enzyme for factor VIIIa rather than for phospholipids. This result reflects the observed effect on catalytic efficacy, since it is known that factor VIIIa increases the apparent k cat while phospholipids decrease the apparent K m (19,20).
We also evaluated the effect of Mg 2ϩ ions on VIIa⅐TF-initiated coagulation, the main physiological pathway. We tested the effect on factor IX activation in a purified system. Factor IX at a physiological concentration (0.1 M) was incubated with VIIa⅐TF and the factor IXa generated was quantified. The cation increased the rate of activation, and the velocity increased approximately 2-fold by Mg 2ϩ ions under the experimental condition (Fig. 3). This effect of Mg 2ϩ ions seemed rather small and, in plasma, the effect was more striking (see below). We are presently uncertain of the reason of this quantitative discrepancy but, nevertheless, it appeared that the cation also has a positive effect on factor IX activation.
We next investigated the effect of Mg 2ϩ ions on VIIa⅐TFinduced clotting. Tissue factor, reconstituted on vesicles of 75% PC, 25% PS, and factor VIIa were added to dialyzed plasma. Clotting times were again decreased by Mg 2ϩ ions in normal plasma and approximately 5 times lower concentrations of tissue factor was sufficient to yield the same clotting time (Fig.  4A). However, the cation was ineffective in factor IX-or VIII- deficient plasma (data not shown). These results suggest that the cation is indeed involved in the mechanism of physiologic coagulation through its action on factor IX. We also examined the effect of Mg 2ϩ ions with phospholipids of different composition. Gilbert and Arena (21) recently showed that PE, a constituent of natural procoagulant membranes (activated platelets), provides high affinity binding sites for factor VIII in membranes with low PS content and thereby facilitates efficient assembly of the IXa⅐VIIIa complex. We employed a preparation of phospholipids with a composition (75% PC, 20% PE, 5% PS) much closer to physiological than frequently used, and re-evaluated VIIa⅐TF-induced clotting. This preparation was reported to be almost equipotent to the preparation of 75% PC, 25% PS with respect to the assembly of the active IXa⅐VIIIa complex (21). We expected that the contribution of factors VIII and IX would be emphasized and thus the effect of Mg 2ϩ ions would be more apparent.
Indeed, the effect of Mg 2ϩ ions was very striking with this preparation (Fig. 4B), and the cation again had minimal effects in plasma deficient in factor IX (Fig. 4C) or factor VIII (data not shown). Raising the concentration of tissue factor hardly decreased the clotting time without Mg 2ϩ ions (i.e. under conditions wherein the apparent coagulant activity of factor IX was attenuated), suggesting that the rate of generation of factor Xa should be reduced considerably. By contrast, factor Xa should be produced very rapidly in the presence of Mg 2ϩ ions. It appeared that VIIa⅐TF-induced factor Xa generation is very strongly dependent on Mg 2ϩ ions and this action is solely mediated through the effects on factor IX/IXa. It is thus strongly suggested that most factor Xa is generated by IXa⅐VIIIa, not by the direct action of VIIa⅐TF, in blood plasma where both Ca 2ϩ and Mg 2ϩ ions are present.
Magnesium ions potentiated all the factor IX-dependent processes in coagulation, and the cation appeared to be a crucial constituent of the coagulation cascade. Our data also underline the physiological significance of factors VIII and IX. The effect of Mg 2ϩ ions on the activation by factor IXa of factor X is especially noteworthy. The cation augmented the apparent catalytic efficacy of factor IXa by increasing the affinity of the protease for the cofactor. The plasma concentration of factor VIII is quite low (0.7 nM) as compared with those of factor IX (90 nM) and other components (11). Moreover, the active form, factor VIIIa, readily loses its activity either through spontaneous inactivation or through degradation by activated protein C (22,23). Therefore, only minute amounts of factor VIIIa should be available during coagulation, and its availability seems to represent a critical bottleneck in the cascade. Our present observation that Mg 2ϩ ions facilitated efficient factor Xa generation with much lower amounts of factor VIIIa than those needed with Ca 2ϩ alone (Fig. 2B) effectively explains how the hemostatic process in vivo is guaranteed.
Although the cascade (or waterfall) theory (2, 3) constitutes a fundamental biochemical basis for the mechanism of coagulation and, in principle, it remains a viable theory after several revisions, erroneous descriptions in early cascade models still seriously influence the current models, found in many textbooks (e.g. Ref. 24). Because the original model was based on observations in test tubes rather than in vivo, it explains the mechanism of widely used diagnostic tests in vitro, i.e. APTT (activated partial thromboplastin time) and PT (prothrombin time). However, it does not reflect the actual hemostatic process. The most obvious discrepancy between the model and events in vivo is the role of factors VIII and IX, which have been classified as components of the intrinsic pathway. The intrinsic pathway, whose importance was stressed in the original hypothesis, now seems likely to be a mere in vitro artifact. Only the extrinsic (or tissue factor) pathway is operative in vivo.
Since the discovery of the mechanism of factor IX activation by VIIa⅐TF (9), many investigations have been undertaken to re-evaluate the roles of factors VIII and IX in the cascade. Kinetic parameters of activation of factors IX and X by VIIa⅐TF (10) provide a solid biochemical basis for the hypothesis that factor IX is the preferred substrate of the enzyme complex. Mann and co-workers (11,12) definitively demonstrated the importance of factors VIII and IX in the tissue factor pathway through elucidation of the detailed activation kinetics of each constituent in a reconstitution system containing plasma concentrations of factors V, VIII, IX, and X and prothrombin. The effects of Mg 2ϩ ions were, however, not examined in these earlier studies. Since Mg 2ϩ ions enhanced the coagulant activity of factor IX in the tissue factor pathway, the contributions of factors VIII and IX in this pathway seem to be much greater than previously assumed. In addition, the differences in lipid composition of procoagulant membranes between those used experiments in vitro (containing large amounts of PS; 20 -30%) and those available in vivo (containing rather small amounts of PS; ϳ5%) also must be considered as pointed by Gilbert and Arena (21). Thus, the relative significance of each reaction step that involves protein/phospholipid interactions might be different from that predicted from investigations with artificial membranes.
The significance of factors VIII and IX has clearly been seriously underestimated in previous laboratory investigations. We propose here a revised cascade theory of coagulation, which includes a novel component, the Mg 2ϩ ion, and is much simpler than previous models (Fig. 5). Although several impor-tant reactions (e.g. conversion of procofactors to active cofactors) are omitted from the figure for simplicity, our model illustrates the main elements of the principal chain reaction, in which an active protease activates another downstream protease zymogen. We have also omitted the mechanism of factor X activation by VIIa⅐TF for clarity. The evidence shows unequivocally that most, if not all, factor Xa is generated by the action of IXa⅐VIIIa. However, the significance of direct activation cannot be ignored. As suggested by Hoffman et al. (25), the amount of factor Xa generated directly by VIIa⅐TF at the very early phase of coagulation is, by itself, very limited but it does produce thrombin, perhaps in a minuscule amount but perhaps sufficient for activation of platelets in the vicinity and, thus, for promotion of the expression of procoagulant surfaces. This small amount of factor Xa at the early phase should be immediately neutralized by plasma protease inhibitors and is unlikely to participate in the wholesale generation of thrombin at the later propagation phase. However, it might play a role in the initiation of coagulation. Our hypothesis must be examined by detailed analyses and is now open to discussion. We believe, however, that this simple model can explain the events in our bodies much better than previous models, and it also provides a clue to the answer of the long-standing question: why do hemophiliacs hemorrhage so severely? To reproduce the precise events that occur in vivo, we must consider the following: activation of factor X by VIIa⅐TF (see text), conversion of procofactors (V and VIII) to active cofactors, inactivation of factors Va and VIIIa via the thrombomodulin/protein C pathway (23), feedback activation of factor XI by thrombin (26,27), neutralization of proteases by serpins, in particular by antithrombin III and tissue factor pathway inhibitor (27), and so on. We must also consider the effects of certain cellular mechanisms (in particular participation of platelets) and blood flow for an understanding of the dynamic hemostatic system. PL, phospholipids.