VITAMIN K , PROTHROMBIN, AND 7-CARBOXYGLUTAMIC ACID

By JOHAN STENFLO, Malmo, Sweden

CONTENTS

I. Introduction 11. Vitamin K-Dependent Proteins in Blood Coagulation

111. Abnormal Plasma Prothrombin Induced by the Vitamin K Antagonistic

IV. The Liver Precursor of Prothrombin V. yCarboxyglutamic Acid in Prothrombin

VI. Vitamin K-Dependent Carboxylation

Coumarin Anticoagulants

Acknowledgments References

1 3

7 14 17 23 26 26

I. Introduction

Prothrombin is the zymogen of the proteolytic enzyme thrombin, which by limited proteolysis converts soluble fibrinogen to insoluble fibrin, ultimately forming a blood clot. Vitamin K is required for normal biosynthesis of prothrombin (factor 11) * and three other plasma proteins (factors VII, IX, and X), which are also zymogens of pro- teolytic enzymes involved in blood coagulation. The vitamin K-de- pendent modifications are required for the in vivo conversion of these proteins to the enzymatically active forms. The four vitamin K-de- pendent proteins have now been both purified and chemically charac- terized (1) . The complete amino acid sequence of prothrombin has been determined by Magnusson et al. (2) and that of factor X by Enfield et al. (3) and Titani et al. (4). The mechanisms of in vitro activation of bovine prothrombin and factors IX and X have been outlined in considerable detail (5-1 7). Our understanding of the regulation of blood coagulation in uivo, however, is still incomplete.

Vitamin K activity is exhibited by P-methyl-l,4-naphthoquinones substituted at the 3 position with a phytyl group (phylloquinone or

*The nomenclature for the coagulation factors is that recommended by an international committee [Wright, I., J. Am. Mrd. Ass., 170, 325-328 (1959)l.

I

Advances in Enzymology and Related Areas of Molecular Biology, Volume 46 Edited by Alton Meister

Copyright © 1978 by John Wiley & Sons, Inc.

2 JOHAN STENFLO

vitamin K,) or with a multiprenyl side chain (the menaquinone series). Vitamin K was discovered by Dam (18) in the 1930s and the vitamin K-antagonist dicoumarol [3,3’-methylenebis(4-hydroxycou- marin)] was isolated and characterized by Campbell and Link in 1941 (19). Until the 1960s little progress was made in the under- standing of the mechanism of action of the vitamin or the antagonist in relation to the biosynthesis of blood coagulation proteins. At this time experiments were initiated in which inhibitors of protein bio- synthesis like actinomycin D and cycloheximide were used to study the regulatory effect of the vitamin on coagulant protein biosynthesis. Although early in vivo experiments suggested that vitamin K was in- volved in prothrombin biosynthesis by regulating DNA transcription, later experimental evidence favored the hypothesis that the vitamin was active in the conversion of a liver prothrombin precursor to bio- logically active prothrombin (20,2 1). This controversy was resolved when an intrahepatic precursor of prothrombin was identified in the livers of vitamin K-deficient rats (21,22) and when abnormal, bio- logically inactive prothrombin was isolated from blood plasma ob- tained from cows treated with the vitamin K-antagonistic coumarin anticoagulants (23-25). Structural comparisons of normal bovine pro- thrombin with abnormal, biologically inactive prothrombin led to the identification of the vitamin K-dependent structure, y-carboxyglu- tamic acid in prothrombin (26) . The identification of this previously unknown amino acid improved our understanding of the interactions of the vitamin K-dependent proteins with calcium ions and phospho- lipid. Furthermore a vitamin K-dependent carboxylase, which has now been partially characterized, was identified in rat liver (27-30). The rapid growth in our knowledge of blood coagulation biochemistry, particularly with regard to the chemistry and biochemistry of the vitamin K-dependent proteins, and our improved understanding of the involvement of vitamin K in the biosynthesis of these proteins, justify a brief review of this field.

This chapter deals mainly with the recent developments in the vitamin K and prothrombin field that led to the characterization of biologically inactive prothrombin from cows treated with the vitamin K-antagonistic coumarin anticoagulants, the identification of the in- trahepatic prothrombin precursor in vitamin K-deficient rats, and the characterization of the carboxylated glutamic acid residues in pro- thrombin. The functional significance of these residues and the char-

VITAMIN K, PROTHROMBIN, AND 7-CARBOXYGLUTAMIC ACID 3

acteristics of the vitamin K-dependent carboxylating enzyme system also are reviewed. Early metabolic studies on the involvement of vitamin K in the biosynthesis of blood coagulation proteins have been carefully reviewed (20,21) and are not cited here, nor do we discuss the information on the chemical propcrties of vitamin K and its antagonists (31) or more clinically oriented studies on vitamin K and prothrombin. For a broader perspective on blood coagulation, the reader is encouraged to consult a recent review (I) .

11. Vitamin K-Dependent Proteins in Blood Coagulation

Blood coagulation is a series of consecutive conversions of zymogens of proteolytic enzymes to the corresponding active forms (1). This ultimately leads to the generation of thrombin, with subsequent con- version of soluble fibrinogen to insoluble fibrin. The product of each reaction in a sequence is the active enzyme in the following reaction. To describe such a sequence of reactions, the term “cascade” or “waterfall” reaction was introduced by Davie and Ratnoff (32) and by Macfarlane (33). In the simplified cascade scheme (intrinsic system) of Figure 1, the emphasis is on the reactions involving the vitamin K-dependent proteins. Factor X can be activated to factor Xa (the “a” is used to denote the enzymatically active forms of the coagulation proteins) either by factor IXa (1,5,34,35) in the so-called intrinsic system (Fig. 1) or in the so-called extrinsic system by factor VIIa (1,5,7-9) and a chemically uncharacterized tissue factor. The latter way of activation of factor X bypasses the earlier reactions in the intrinsic system. Both systems are triggered by cellular damage. Of the four vitamin K-dependent proteins, three (i.e., factors IX, X, and prothrombin) are active in the intrinsic system, whereas factor VII is the extrinsic factor X activator. Factors VIIa and IXa cleave the same peptide bonds in factor X (1,5,7,35).

The active forms of the vitamin K-dependent proteins are serine endopeptidases. In factors IXa, Xa, and thrombin (factor IIa) the amino acid sequences around the active sites are homologous with the corresponding sequences in trypsin and the other pancreatic serine endopeptideases (1). Trypsin cleaves almost any lysyl or arginyl bond, whereas the blood coagulation enzymes have a very restricted sub- strate specificity. Thus activation of prothrombin by factor Xa in- volves the cleavage of only two peptide bonds in prothrombin (1,9-17).

4 JOHAN SmNFLO

FACTOR XI1 FACTOR XI

1

I

- - - - - - - _ - - - - - - - I

1 I

I r - - - - - - - FACTOR IX FACT 0 R 1x0

I I I I

I

I I I

I I I I I I I I

FACTOR Vlll’ FACTOR VWI I I I I

I

co-, PI.

[FA C TOR I X 0 - FACT0 R WllL CO * - PL ]

FACTOR X a - - - - - - - FACTOR X

I I I I

FACTOR V I

I

[FACTOR Ca x,-FACTOR *, PL I v‘-Ca FACT!? * -PL] I I I I I I I I

PROTHROMBIN THROMBIN - - - - - - - - - - - -’ Fig. 1. Initiation of blood coagulation in the intrinsic system according to the “cascade” scheme. Emphasis is on the reactions involving vitamin K-dependent proteins; PL denotes phospholipid. Modified from Davie and Fujikawa (1).

Each zymogen activation in the blood coagulation cascade is the result of such limited proteolysis. When the system is triggered by tissue damage, the sequential arrangement of the individual blood coagula- tion reactions gives rise to consecutive limited proteolytic events, thereby producing an enormous potential for enzymatic amplification. The reaction rate in the cascade is influenced by accessory factors, such as phospholipid and the two high molecular weight proteins, factors VIII and V, which have no known enzymatic activities (1).

VITAMIN K, PROTHROMBIN, AND 7-CARBOXYCLUTAMIC ACID 5

Factors V and VIII accelerate the rates of activation of prothrombin and factor X respectively (1).

The accessory factors that interact directly with the vitamin K-de- pendent parts of the coagulation proteins are calcium ions and phos- pholipid. Prothrombin and factors IX and X have been shown to bind Ca2+ ions in solution (1 7,36-40), whereas the biologically inactive forms of these proteins, which are synthesized after administration of the vitamin K-antagonistic coumarin anticoagulants do not (see be- low). It is clear from Figure 1 that the reactions in the cascade where the vitamin K-dependent proteins participate require calcium ions, and the activation of factor X and prothrombin, also require phos- pholipid. In the presence of calcium ions, prothrombin binds to phos- pholipid, whereas in the absence of calcium ions no binding is demon- strable (13,4149). The part of prothrombin engaged in the calcium and lipid binding is the NHZ-terminal, fragment 1 part (43, Fig. 2). By comparison, the carboxyterminal part of the prothrombin molecule (prethrombin l ) , from which thrombin is derived, does not bind to phospholipid, nor does enzymatically active thrombin (43), whose substrate is circulating as well as platelet-bound fibrinogen (1). An important difference between thrombin and factor Xa is that pro-

Th FX FX I I 0 0

F r a g . 1 "Frog. 2, A B r. n +

4 - + PRETHROMBIN 1

4 + Fra g . l-2 Y THROMBIN

Fig. 2. Schematic diagram of the prothrombin molecule: COO-, ycarboxyglu- tamic acid residues: CHO, a carbohydrate side chain; Su, the active site scrine residue; Th, the single peptide bond in prothrombin cleaved by thrombin; FX, the two peptide bonds cleaved by factor X; A, the A or light chain of thrombin; B, the B or heavy chain. The nomenclature used to denote the peptide fragments generated from prothrombin by thrombin and factor X is that proposed by a task force of the International Committee on Thrombosis and Hemostasis. References for a comparison of this nomenclature to those used previously can be found in Myrmel et al., Biochis fry , 15, 1767-1773 (1976).

6 JOHAN STENFLO

thrombin activation to thrombin involves cleavage at two peptide bonds with subsequent removal of the vitamin K-dependent part of the molecule from the enzymatically active part (1,2,11-17), whereas in factor Xa the vitamin K-dependent part is linked to the enzymat- ically active part by disulfide bridges (1,2,5-10). Thus unlike thrombin, factor Xa binds calcium ions and phospholipid. The increase in the rate of activation of prothrombin caused by the accessory factors has been illustrated in vitro in experiments that demonstrated that pro- thrombin activation in the complete system (factor Xa, Ca2+, factor V, and phospholipid) was some 20,000-fold faster than the rate of acti- vation in an identical mixture from which factor V and phospholipid were omitted (14,17).

The importance of the lipid binding in uiuo has been proved in ex- periments (50,51) where factor Xa was infused into an experimental animal, either with or without calcium ions and phospholipid. In these experiments thrombus formation required that factor Xa be in- fused together with phospholipid (thromboplastin). The plasma con- centration of the vitamin K-dependent proteins is low (i.e., prothrom- bin, 100-200 mg/liter; factors IX and X , 5-10 mg/liter; factor VII < 1 mg/liter) (1,8,52-53). I t is thus apparent that the membrane affinity of these proteins is of vital importance for the blood coagulation cascade, which otherwise would be stifled because of dilution of the reactants and the inhibitory effect of circulating protease inhibitors, notably antithrombin 111. This has been demonstrated experimentally with the abnormal prothrombin induced by coumarin anticoagulants (see below).

The close phylogenetic and functional relationship of the vitamin K-dependent proteins is reflected in their similar behaviors during purification. Several purification procedures are available for the vita- min K-dependent proteins, and most of them depend on the initial adsorption to and elution from barium citrate or some similar salt of divalent cations (54). This very effective purification step is usually followed by ammonium sulfate fractionation and chromatography on DEAE cellulose or DEAE Sephadex (Fig. 3), after which various forms of afinity chromatography are used to obtain the pure proteins (1,8,53). The proteins are assayed during purification either by co- agulation bioassays (54,55) or with the aid of monospecific antisera (24,56). Because of the low plasma concentration of factor VII, puri- fication of this protein is far more demanding than purification of the

VITAMIN K, PROTHROMBIN, AND 7-CARBOXYGLUTAMIC ACID 7

100

8 0

60

4 0

20

0 500 700

E f f l u e n t v o l u m e I m l l

- ?O r

3

* - e e

5 - X LL

0

0

u '0 .c

?

x

- P

Y

5 i 0 E ? f ? a

0 900

Fig. 3. Chromatogram of vitamin K-dependent proteins on a column of DEAE- Sephadex A-50 (2.5 X 40 crn). The column was equilibrated with 0.1 M phosphate buffer, 1 mM benzamidine, pH 6.0. Elution was accomplished with a linear gradi- ent of NaCl (0.15-0.55 M). Curves: 0-0, absorbance at 280 nm; H, C1- concentration. Factor VII was eluted with the peak between prothrombin and protein C. The concentrations of the vitamin K-dependent proteins were deter- mined immunochemically. Modified from Stenflo (56).

other vitamin K-dependent proteins. To obtain approximately 2 mg of pure factor VII, 50 liters of plasma must be processed (8,53).

111. Abnormal Plasma Prothrombin Induced by the Vitamin K-Antagonistic Coumarin Anticoagulants

The first observation that dicoumarol treatment leads to the bio- synthesis of abnormal prothrombin was made by Hemker et al. (57), who indirectly demonstrated the occurrence of an inhibitor of pro- thrombin activation in plasma from patirnts treated with the anti-

Fig. 4. Crossed immunoelectrophowis (61) of bovine plasmas containing abnor- mal prothrombin induced by administration of coumarin anticoagulants. The electrophoresis was performed at pH 8.6 in 0.075 M barbital buffer containing 2 mM calcium lactate. A monospecific antiserum against bovine prothrombin was used. (u) Immunoprecipitate obtained before administration of dicoumarol to the cow. (b)-(d) Precipitates given by samples obtained 2, 4, and 7 days after the beginning of dicoumarol administration. The prothrombin activity in the

8

VITAMIN K , I’ROTHROMBIN, AND 7-CARBOXYGLUTAMIC ACID 9

coagulant dicoumarol. They also suggested that the inhibitor was a precursor of prothrombin. Direct evidence of the presence of abnormal prothrombin in plasma from patients treated with the vitamin K-an- tagonistic coumarin anticoagulants was provided with immunochemical techniques by Ganrot and Nil4hn (58,59) and by Josso et al. (60). Crossed immunoelectrophoresis (61,62) made it possible to compare the abnormal prothrombin with normal prothrombin before fraction- ation of plasma, since both prothrombins have the same main anti- genic determinants. Three important differences were noted (59); un- like normal prothrombin, the abnormal one neither bound calcium ions nor was adsorbed to barium citrate. Furthermore, when citrate plasma from a dicoumarol-treated individual was coagulated by the addition of calcium chloride, the abnormal prothrombin was not ac- tivated to thrombin as normal prothrombin is but appeared to remain unchanged. These results have since been extended and the dicou- marol-induced human prothrombin has been partially purified (63- 68). Since purified dicoumarol-induced prothrombin does not inhibit the activation of normal prothrombin (24), the relationship of the inhibitor initially observed by Hemker et al. (57) to the immuno- chemically demonstrated abnormal prothrombin is not clear.

Figure 4 shows the appearance of abnormal prothrombin during dicoumarol treatment of a cow. Like human dicoumarol-induced pro- thrombin, the bovine counterpart has the same main antigenic deter- minants as the normal protein (23-25). Crossed immunoelectrophoresis has also been successfully used to demonstrate abnormal, dicoumarol- induced forms of factors IX and X in the bovine species (56,69). Like dicoumarol-induced prothrombin they do not bind calcium ions. In the terminology used by Hemker and coworkers (69), the coumarin anticoagulant induced forms of the vitamin K-dependent proteins are termed PIVKA (proteins induced by vitamin K-absence; i.e., PIVKA 11, VII, IX, and X).

The differences between normal and dicoumarol-induced prothrom- bin observed with immunochemical methods suggested that calcium binding and biological activity are related to vitamin K action. A comparison between the two prothrombins was undertaken to reveal

samples relative to an arbitrary standard is given to the right. The electrophoretic pattern of bovine plasma in agarose gel is given below as a reference for the elcctro- phoretic mobilities of normal (cathodal) and abnormal (anodal) prothrombin. From Stenflo, and Ganrot (24).

10 JOHAN STENFLO

the chemical basis of these differences. Since biosynthesis of abnormal, biologically inactive prothrombin could be induced in cattle by ad- ministration of coumarin anticoagulants, procedures based on con- ventional chromatographic methods for the purification of the bovine dicoumarol-induced prothrombin were devised by Stenflo and Ganrot (23,24) and Nelsestuen and Suttie (25). These purification procedures are far more cumbersome than those used for normal prothrombin because unlike biologically active prothrombin, the dicoumarol-in- duced prothrombin cannot be adsorbed to barium citrate. Recently a simplified method utilizing adsorption to immobilized antibodies as an essential purification step was described by Wallin and Prydz (70). Access to plasma from coumarin-treated cattle made possible the large-scale purification process necessary to obtain enough ma- terial for structural studies on the purified proteins. Furthermore, abnormal bovine prothrombin was advantageous for investigation be- cause it has been more carefully characterized than human prothrom- bin (54). Its amino arid sequence has been determined (2), and a large body of information is available on the mechanism of activation of bovine prothrombin (1,ll-17).

The purified dicoumarol-induced prothrombin is not activated to thrombin under normal circumstances, nor does it inhihit the activation of normal prothrombin (24,25). Structural comparisons of the intact proteins established that the molecular weights, as judged by acrylamide gel electrophoresis in sodium dodecyl sulfate, the sedimentation co- efficients, the amino acid compositions after acid hydrolysis, the carbo- hydrate compositions and the NHI- and COOH-terminal amino acid residues, are identical (25,71). Equilibrium dialysis showed that the dicoumarol-induced prothromt)in binds at most one calcium ion per mole (36,37), whereas normal prothrombin binds 10-12 calcium ions per mole ( I 7,37-39) (Fig. 5). At low calcium concentrations there is a positive cooperativity in the binding of calcium ions to normal prothrombin according to all reports ( 1 7,37,39,72) but one (38).

Although the dicoumarol-induced prothrombin has no prothrombin activity, it can be activated to thrombin by nonphysiological prothrom- bin activators (25,60). Thus staphylocoagulase, an exoprotrin from Stukhylococcus aureus, generates the same amount of thrombin activity from both normal and dicoumarol-induced prothrombin (60). In con- trast to activation of prothrombin with its physiological activator, factor Xa ( l ) , activation with staphylocoagulase is due not to pro-

VITAMIN K, PROTHROMBIN, AND 7-CARBOXYCLUTAMIC ACID 1 1

4 4

MOLES C~'*BOUND/MOLE PROT.

Fig. 5. Scatchard plot of CaPf binding of normal and abnormal, dicoumatol- induced prothrombin: 0, normal prothrombin; 0, abnormal prothrombin. From Stcnflo, and Canrot (37).

teolysis but to the formation of a complex of prothrombin and staphylo- coagulase that has thrombin activity (73). Nelsestuen and Suttie (25) showed that both normal and dicoumarol-induced prothrombins can be activated by trypsin and by the venom from the snake Echis carinatus and that the same amount of thrombin activity was generated from both. These experiments indicated that the COOH-terminal (throm- bin) portion (Fig. 2) of the dicoumarol-induced prothrombin molecule was intact and that the dicoumarol-induced prothrombin was inactive in viva and in the standard prothrombin bioassays because of a defect in the activation mechanism. Additional evidence that the difference between the two prothrombins is in the NH2-terminal parts of the molecules was provided when the proteins were cleaved with thrombin and the two fragments, fragment 1 and prethrombin 1, obtained from normal and dicoumarol-induced prothrombin were compared (74). Prethrombin 1 from both prothrombins had identical electro- phoretic and immunochemical properties. Fragment 1 from normal prothrombin bound Ca2+ ions; but that from dicoumarol-induced pro-

12 JOHAN STENFLO

thrombin did not. Differences were demonstrated in the peptide maps prepared from thermolysin digests of reduced and carboxymethylated fragment 1 from the two prothrombins. By digestion of normal bovine prothrombin with trypsin and subsequent adsorption of the digest with barium citrate, a calcium-binding peptide was isolated from the fragment 1 region of prothrombin (75). When dicoumarol-induced prothrombin was subjected to the same treatment, no such peptide could be isolated. I t was thus firmly established that the vitamin K-dependent structures responsible for calcium binding were in the NH2-terminal (fragment 1) part of prothrombin, and that the COOH- terminal (thrombin) part of the molecule was not modified by the action of vitamin K. Simultaneous studies on the phospholipid binding of prothrombin and fragment 1 clearly demonstrated that i t was the fragment 1 portion of prothrombin that bound to phospholipid and that this binding involved Ca2+ ions (43). Stenflo (76) isolated and sequenced tryptic peptides obtained from digests of the IVH2-terminal cyanogen bromide fragment (residues 1-72) of the two prothrombins. The peptide containing residues 4-10 from normal prothrombin had the same amino acid sequence as a tryptic peptide isolated previously by Magnusson (77) and shown to have an anomalously high negative charge. This result was confirmed, and it was also shown that the peptide that consists of residues 4-10 in dicoumarol-induced prothrom- bin had lower negative charge, even though the amino acid composi- tions of these two peptides after acid hydrolysis were identical.

Prothrombin activation results from interaction of the protein with phospholipid, CaZ+ ions, factor V, and factor Xa (Fig. 1). Since calcium is required for binding of prothrombin to phospholipid (41-49), it seemed as if the abnormal prothrombin that did not bind calcium ions would not be bound to a phospholipid surface. These assumptions have since been verified experimentally by Esmon et al. (78), who showed that the purified bovine dicoumarol-induced prothrombin did not bind to phospholipid in either the presence or absence of calcium ions (Fig. 6). Furthermore, addition of phospholipid to a mixture of factor Xa, calcium ions, and normal prothrombin resulted in an im- mediate drastic increase in the rate of prothrombin activation. No such increase in the rate of activation was observed in an identical experiment in which dicoumarol-induced prothrombin was used. Simi- lar experiments in which the factor V-catalyzed increase in the rate of prothrombin activation was studied, indicated that the factor

VITAMIN K, PROTHROMBIN, AND 7-CARBOXYGLUTAMIC ACID 13

15 . . l . , l l 6 n I

Fig. 6. Binding of normal and abnor- - - - 2 mal, dicoumarol-induced prothrombin

to phospholipid (PL). Rothrombin, normal or abnormal (finai concentra- tion 0.2 mg/ml) was incubated for 30 min with 0.75 mg/ml of PL (equimolar mixture of ["C] dioleoyl phosphati- dylcholine and dioleoyl phosphatidyl glycerol) in 0.04 M Tris-HCI, 0.07 M NaCI, pH 7.5, in the presence or ab- sence of 10 mM CaCIz. After incubation,

20 1 0.4 rnl samples were chromatographed on a Bio-gel A 0.5 M column (0.9 X 29

10 7 cm) equilibrated with buffer (see above) with or without 10 mM CaClr. The flow rates were 10 ml/hr and the frac- tion volumes were 0.55 ml. (n) Normal

- 4 - prothrombin with CaClz but without 20

PL. (b) Normal prothrombin with 10 - 2 \ = CaC12 and with PL. (c) Abnormal pro- 3 0 ' 6 0 thrombin with CaC12 and with PL. " (d) Abnormal prothrombin and PL but

without CaC12. (e) Abnormal prothrom- bin without PL but with CaC12. Curves:

- - 0, thrombin activity; A-A, PL, *4C radioactivity. From Esmon et al. (78), with permission.

6

5 .

7 10 - - 4 B

. - C

E 2 3 0 . '

b

- '

5; c - -

' ' '

10 20 30 40 FRACTION NUMBER

V-binding part (fragment 2) in dicoumarol-induced prothrombin was intact .

It has been reported that vitamin K is involved in the glycosylation of prothrombin and that there are differences in carbohydrate com- position between prothrombin isolated before and during dicoumarol administration (67,79,80). This report has not been verified (81-83), and it is in disagreement with the results obtained by comparing normal and dicoumarol-induced bovine prothrombin in which no sig- nificant difference in carbohydrate composition could be observed (25,7 1). Furthermore, asialo- and aglycoprothrombin retain biological activity (82) and bind calcium ions (23), indicating that vitamin K is

14 JOHAN STENFLO

not required for the biosynthesis of the carbohydrate prosthetic groups and that these groups arc not rcquircd for biological activity.

Long-term treatment of cows with coumarin anticoagulants, in ad- dition to inducing the biosynthesis of abnormal, biologically inactive prothrombin that neither binds calcium ions nor adsorbs to barium citrate, has been reported (84) to lead to the biosynthesis of prothrom- bin having low biological activity, although it does bind to barium citrate. This abnormal prothrombin was claimed to represent an inter- mediate form between the previously characterized abnormal pro- thrombin and normal prothrombin. The thrombin portion of such prothrombin was normal, whereas the Ca'' ion binding of fragmcnt 1 was reported to be intermediate between that of fragment 1 from normal prothrombin and that of the chemically characterized abnor- mal prothrombin. Results of immunochemical and electrophoretic ex- periments on plasma from a cow on long-term dicoumarol treatment have not supportcd the idea that there are intermediate forms of abnormal prothrombin (24, and unpublished results from the author's laboratory). The experiments that werc performed in buffer containing calcium ions showed only two species of prothrombin, that is, the nor- mal and the dicoumarol-induced prothrombin that did not bind cal- cium ions or adsorb to barium citratc. KO forms with intermediate electrophoretic mobilities were observed, as would have been expected if there were a series of prothrombin molecules with different calcium ion afinities. O n the other hand, other data (85,86) seem to support the idea that a series of bovine prothrombin molecules with varying calcium ion affinities is synthesized during dicoumarol administration.

IV. The Liver Precursor of Prothrombin

Early studies on the mode of action of vitamin K in intact rats, perfused rat livers, and rat liver slices could not reveal whrther the vitamin regulates de nuuu synthesis of the vitamin K-dependent coagu- lation factors or whether it is involved in the conversion of an intra- hepatic prothrombin precursor to biologically active prothrombin (20, 2 1) . Later the experimental evidence, particularly the identification of abnormal prothrombin in plasma from cows trcatcd with coumarin anticoagulants (23-25), favored the opinion that the vitamin is in- volved in the conversion of a prothrombin precursor to biologically active protein. I t was demonstrated that administration of vitamin K

VITAMIN K, PROTHROMBIN, AND 7-CARBOXYGLUTAMIC ACID 15

to vitamin K-deficient hypoprothrombinemic rats led to rapid ap- pearanre of prothrombin in plasma and that this process was virtually uninhibited by prior administration to the rats of the protein synthesis inhibitor cycloheximide (87-89). Convincing evidence in favor of the precursor hypothesis was obtained by Shah and Suttie (21), who ad- ministered cycloheximide to vitamin K-deficient rats and subsequently gave them radioactively labeled amino acids. Administration of vita- min K to these rats resulted in a prompt burst of unlabeled prothrombin to the plasma. These experiments provided strong evidence for the existence of an intrahepatic pool of prothrombin precursor in the liver of vitamin K-deficient rats. The prothrombin precursor in vitamin K-deficient rats and from rats given vitamin K antagonists was then demonstrated directly in livcr microsomal extracts by Suttie (90). Prothrombin artivity was measured in the extracts both with standard methods and after activation with Echis carinatus venom. The pro- thrombin activity in the microsomal extracts from warfarin-treated animals was lower than in rontrol animals, whereas the thrombinlike artivity liberated by the Echis carinatus venom had increased by a factor of 4-5 compared to the untreated controls. The thrombinlike activity generated in microsomal extracts by Echzs carinatus venom was inhibited by the low molecular weight protein hirudin (91), which is a specific inhibitor of thrombin (92). After injection of vitamin K, the concentration of the precursor in the liver decreased rapidly (91). The partially purified liver precursor reacted with an antiserum raised against rat plasma prothrombin. The liver precursor, unlike plasma prothrombin, was not adsorbed to barium sulfate.

The prothrombin precursor has now been purified to homogeneity from the liver of warfarin-treated rats (93,94), and its properties have been compared to those of rat prothrombin (95). In the procedure used by Esmon et al. (93) the precursor activity was solubilized from the microsomes by treatment with Triton X-100 followed by ammo- nium sulfate fractionation of the Triton X-100 extract. Thereafter gel permeation chromatography, on Biogel A-5 m and A-0.5 m followed by chromatography on QAE-Sephadex and heparin-agarose were used. The final product (100-150 l g from 20 rat livers) was elec- trophoretically homogeneous. In the purification procedure of Moris- sey et al. (94) the precursor was purified by adsorption to agarose- insolubilized antibodies against rat prothrombin. The adsorbed pro- tein was eluted with sodium dodecvl sulfate and was shown to be

16 JOHAN STENFLO

electrophoretically homogeneous. The apparent precursor purified by Esmon et al. (93,95) had a lower negative charge (p l = 5.8) than biologically active plasma prothrombin (p l = 5.0). The fragments obtained by activating the purified liver precursor with factor X a and with the venom from the Taipan snake (Oxyuranus scutellatus scutelfatus) were identical with those obtained when biologically active rat plasma prothrombin was activated as judged by acrylamide grl electrophoresis in buffer containing sodium dodecyl sulfate. The rate of activation of the precursor by factor X a was not increased by the addition of phospholipid, whereas the rate of activation of normal prothrombin rose drastically when the phospholipid was added. The results of the comparison of the rat plasma prothrombin and the hepatic prothrombin precursor induced by coumarin anticoagulants are identi- cal with those obtained by comparing bovine plasma prothrombin and dicoumarol-induced prothrombin. The two anticoagulant-induced, bio- logically inactive proteins are both immunoreactive and the thrombin parts of the abnormal molecules are identical to those of the bio- logically active proteins. The abnormal molecules, however, do not bind calcium ions or phospholipid. Immunoreactive prothrombin has also been synthesized in a heterologous system using rat liver messenger RNA and rabbit reticulocytes (96). Recently a second apparent pro- thrombin precursor with a pZ of 7.2 was isolated from rat liver micro- soma1 preparations (97). The increased basicity of this protein is a property of the fragment 1 part of the molecule; but the chemical differencr between the p l = 5.8 and the p l = 7.2 precursors is not known. We do not know whether one or both of these proteins is a prothrombin precursor, or whether they are altered (degraded?) pre- cursors.

The effects of coumarin anticoagulants on the amount of hepatic prothrombin precursor and the amount of abnormal prothrombin in blood plasma have been systematically studied in different species (98). TO this end, thrombin activity generated by Echis carinatus venom was determined in blood plasma from the anticoagulant-treated ani- mals (rat, cow, mouse, hamster, guinea pig, rabbit, dog, and chick) both before and after removal of biologically active prothrombin by barium citrate adsorption. The amount of prothrombin precursor ac- tivity (i.e., thrombin activity generated by Echis carinatus venom) was determined in Triton X-100 solubilized liver microsomes from the anticoagulant-treated animals. The bovine was the only species studied

VITAMIN K, PROTHROMBIN, AND y-CARBOXYGLUTAMIC ACID 17

that did not accumulate prothrombin precursor activity in the liver during anticoagulant treatment; moreover, only in cows could sig- nificant amounts of thrombin activity be generated in plasma by the venom after barium sulfate adsorption. By contrast, the other species seemed to respond to the anticoagulant treatment by accumulation of prothrombin precursor in the microsomes but had no abnormal prothrombin in their plasma. Warfarin administration lowers the con- centration of normal plasma prothrombin in the rat; but in contrast to the bovine and human species, no abnormal, immunochemically observable prothrombin appears in plasma. This might be due to the absence of antigenic determinants in an abnormal prothrombin; but since the intrahepatic prothrombin precursor reacts with antirat pro- thrombin (91), any abnormal plasma prothrombin would presumably be immunochemically reactive. It has, however, been postulated (79, 99,100) that there is an abnormal prothrombin in rat plasma, but these reports have not been verified (21,101). I t thus appears safe to conclude that in humans and in the bovine species the prothrombin precursor-or, which is more likely, an altered precursor-is readily excreted, whereas in the other species it accumulates in the liver.

V. y-Carboxyglutamic Acid in Prothrombin

The vitamin K-dependent structure in bovine prothrombin, y-car- boxyglutamic acid (Gla), was first identified in the peptide-containing residues 4-10 in prothrombin by Stenflo et al. (26). After degradation of this peptide with aminopeptidase M and carboxypeptidase B, a tetrapeptide (residues 6-9) Leu-Glx-Glx-Val was isolated, which had a negative charge too high to be consistent with its amino acid com- position. After acid hydrolysis, two glutamic acid residues were ob- tained. Proton nuclear magnetic resonance spectroscopy and mass spectrometry showed that each of the two glutamic acid residues in the peptide had one extra carboxyl group on their y-carbon atoms. Figure 7 schematizes the structure of the tetrapeptide, which thus has two residues of y-carboxyglutamic acid, a previously unknown amino acid. The corresponding peptide from dicoumarol-induced prothrom- bin had two glutamic acid residues in positions 7 and 8, as judged from the electrophoretic mobility and amino acid analysis of the pep- tide (76). The new amino acid is a malonic acid derivative. Therefore, after heating of the tetrapeptide from prothrombin after lyophilization

18 JOHAN STENFLO

KJC, ;I-$ HOOC\ ,!XIOH HOOC COOH

FH FH \$I

$5 FH2 FH2 I Y - C H - CO -NH-CH-CO-NH-CH-CO-NH- CH- COOH

Fig. 7. From Stenflo et al. (26).

Structure of tetrapeptide (residues 6-9) isolated from normal prothrombin.

in the acid form, it was decarboxylated; its electrophoretic mobility was then identical with that of synthetic Leu-Glu-Glu-Val (26). The presence of Gla in other peptides from the same part of prothrombin was immediately confirmed by Nelsestuen et al. (102), who charac- terized a glutamylserine dipeptide (residues 33 and 34) by mass spec- trometry and showed that the glutamic acid residue was carboxylated on the y-carbon atom, and by Magnusson et al. (1031, who reported the amino acid sequence of the first 42 amino acid residues in pro- thrombin. Mass spectra and electrophoretic mobility data ( 103) showed that all 10 glutamic acid residues were substituted with a 7-carboxyl group (Fig. 8). The positions of the carboxylated glutamic acid resi- dues have also been determined by sequenator degradation of a pep- tide containing residues 12-44, in conjunction with mass spectrometric identification of the thiohydantoin derivatives of y-carboxyglutamic acid (104). These experiments verified the previous assignment of the positions of the Gla residues in prothrombin. Detailed mass spectro- metric studies of acetylated and permethylated peptides from this region of the protein led to the same result (105). These experiments, determination of y-carboxyglutamic acid residues following alkaline hydrolysis (104), and carboxyl group quantitation (106), also indi- cated that there is no “undercarlmxylation” of glutamic acid but that each of the 10 glutamic acid residues is quantitatively carboxylated to 7-carboxyglutamic acid. Based on sequence data on normal prothrom- bin, structural comparison between fragments from normal and dicou- marol-induced prothrombin, and determinations of y-carboxyglutamic acid as dihydroxyleucine after 3H-diborane reduction (107), it now seems safe to conclude that the 10 y-carboxyglutamic acid residues in the NHz-terminal part of prothrombin constitute the entire vitamin K-dependent modification in this protein. In the positions where bio- logically active prothrombin has y-carboxyglutamic acid, dicoumarol- induced prothrombin has glutamic acid (76,104). y-Carboxyglutamic

1

5

10

15

2

0

25

Pro

thro

mbi

n V

al

Arg

Ly

s C

ly A

sn L

eu G

la A

rg C

la C

ys Le

u G

la C

la P

ro C

ys

Ser

Agr

C

la

Fac

tor

X

Ala

Asn

S

er -

Phe

Leu

C

la C

la -

Val

L

ys G

ln C

ly A

sn Leu C

la A

rg C

la C

ys

Leu

Cla

Gla

Ala

Cys

S

er Leu

Cla

P

rote

in C

A

la A

sn

Ser

-

Phe

Leu

Gla

Cla

-

Leu

Arg

Pro

Gly

Asn

Val

C

la A

rg G

la C

ys

Ser

Cla

Cla

Val

C

ys

9 P

he

Gla

F

acto

r IX

T

yr A

sn S

er C

ly L

ys

Leu

Cla

Gla

Phe

V

al

Arg

-

Cly

Asn

Leu

F

acto

r V

II

Ala

Asn

-

Cly

Phe

Leu

?

7 Le

u Leu

- P

ro G

ly S

er Leu

Ala

Asn

Ly

s C

ly P

he

Leu

Gla

Cla

-

30

35

- P

roth

rom

bin

Cla

Ala

Ph

e C

la A

la L

eu

Gla

Ser

Leu

Ser

Ala

Thr

ID

F

acto

r X

Cla

Ala

Arg

Cla

Val

Ph

e G

la A

sp A

la C

la C

ln T

hr

Pro

tein

C

Cla

Ala

A

rg C

la I

le P

he

P A

sn T

hr

? P

Thr

Fig

. 8.

Am

ino

acid

seq

uenc

e ho

mol

ogie

s fo

r bo

vine

pro

thro

mbi

n, f

acto

r X

(li

ght

chai

n),

new

vit

amin

K-d

epen

dent

pro

tein

(pr

otei

n C

, li

ght

chai

n),

fact

or

IX,

and

fa

ctor

VII

. P

osit

ions

of

y-ca

rbox

ygul

tam

ic

acid

(G

la)

that

ha

ve

been

ex

peri

men

tall

y de

term

ined

ar

e in

dica

ted.

T

he

num

beri

ng i

s th

at o

f pr

othr

ombi

n.

Das

hes

refe

r to

spa

ces

that

ha

ve b

een

inse

rted

to

bri

ng t

he

prot

eins

in

to

alig

nmen

t fo

r gr

eate

r ho

mol

ogy,

Ref

eren

ces

to t

he s

eque

nces

are

giv

en i

n th

e te

xt w

ith

the

exce

ptio

n of

th

e se

quen

ce o

f fa

ctor

V

II (

151)

.

20 JOHAN STENFLO

acid residues adjacent to arginine residues make the latter inaccessible to trypsin. Thus the peptide bonds between Arg 16 and Gla 17 and between Arg 25 and Gla 26 are not cleaved by trypsin, whereas in dicoumarol-induced prothrombin, which has Glu in positions 17 and 26, trypsin readily cleaves these bonds (76). Gla Residues appear both singly and in pairs, and there does not seem to be any homology of sequence around these residues.

The structural studies performed on normal and dicoumarol-induced bovine prothrombins were confirmed and extended by experiments on rats. 7-Carboxyglutamic acid in rat prothrombin was thus elegantly demonstrated by Esmon et al. (27) by showing vitamin K-dependent in vitro incorporation of HL4C03- into the endogenous liver prothrom- bin precursor. After extraction of the microsomes with Triton X-100, biologically active “C-labeled prothrombin was isolated. The labeled prothrombin was cleaved by factor Xa, whereby the two activation fragments (fragment 1, mol. wt. 23,000, and fragment 2, mol. wt. 4300) were formed. Acrylamide electrophoresis in sodium dodecyl sulfate established that all the radioactivity in the prothrombin was in the NHs-terminal fragment 1. Acid hydrolysis of the labeled pro- thrombin resulted in loss of approximately 50% of the radioactivity due to decarboxylation. The remaining radioactivity was in glutamic acid residues. These experiments formed the foundation for subsequent work to characterize the vitamin K-dependent carboxylase. The dem- onstration of an in vivo incorporation of H14C03- into prothrombin has also been claimed (108).

7-Carboxyglutamic acid has been found in bovine factors IX and X and in a new vitamin K-dependent protein whose function is unknown. Quantitation of carboxyl groups in a peptide containing residues 5-43 from the light chain of factor X indicated the presence of approxi- mately one extra carboxyl group on each of the glutamic acid residue in the peptide (109). Sequenator degradation of the light chain of factor X with mass spectrometric identification of Gla established that the first 11 glutamic acid residues in the light chain of this protein are carboxylated (110). In the new vitamin K-dependent protein (protein C), nine Gla residues have been found in the NH2-terminal part of the light chain ( I 10). In both bovine ( 1 10) and human (1 1 1 ) factor IX, Gla has been identified in positions 7 and 8. All these carboxylated glutamic acid residues except the eleventh in factor X appear in positions homologous to the positions of Gla in prothrombin (Fig. 8).

VITAMIN K, PROTHROMBIN, AND 7-CARBOXYGLUTAMIC ACID 2 1

The standard conditions for acid hydrolysis of proteins (6 M HC1 at 110°C for 20 hr) cause rapid decarboxylation of 7-carboxyglutamic acid to glutamic acid. Free Gla and the dipeptide Gla-Gla have, however, been isolated in low yields after partial acid hydrolysis (6 M HCl at 45°C for 24 hr) of bovine prothrombin (2,105). Methods have now been developed that allow sensitive qualitative and quanti- tative determination of y-carboxyglutamjc acid in protein hydrolysates and the identification of Gla residues in sequence work. Following alkaline hydrolysis of peptide-bound Gla (104,112,113), the acid can be determined quantitatively on the amino acid analyzer (Table I).

TABLE I

Amino Acid Composition of Peptides from Normal Prothrombin.

Residues 6-9 Residues 12-44

Acid Alkaline Acid Alkaline Peptide hydrolysis hydrolysis hydrolysis hydrolysis

Lysine Histidine Arginine Half-cystine 7-Carboxyglutamic acid Aspartic acid Threonine Serine Glutamic acid Prolinc Glycine Alanine Valine Methionine Isoleucine Leucine Tyrosine Phen ylalanine Tryptophan

1 . O

2.1 2 . I b

1.9 0 .9 2 .5

2.0 0.4 8.5 0 .8 0 .9 4.3

1.9

1' l C

1 .o 0.7 4"

2 . 0 + - -

2.8

8.6 2.4

0.3 0.4 0.9 3.5 4.6

40

2 . 0 +

.From Fernlund et al. (104).

c The values given are molar ratios relative to 1 mol of valine for the peptides with residues 6-9 and relative to 4 mol of leucine for the peptide with residues 12+. Values below 0.3 were omitted. No corrections were made for destruction or in- complete hydrolysis.

Determined as S-carboxymethylcysteine.

22 JOHAN STENFLO

Under standard conditions used for analyzing protein hydrolysates, Gla is eluted before aspartic acid from the amino acid analyzer. Sensitive qualitative identification of y-carboxyglutamic acid in pro- teins was obtained by Zytkovicz and Nelsestuen (107) after [3H]di- borane reduction with subsequent acid hydrolysis and identification of the reduction product 5-5” 3H]dihydroxyleucine in the amino acid analyzer effluent. The positions of y-carboxyglutamic acid residues in peptides have been successfully established by mass spectrometry (26, 102,103,105). Detailed mass spectrometric data on acetyl permethyl and other derivatives of peptides containing y-carboxyglutamic acid from prothrombin have been published by Morris et al. (105). With this procedure the positions of 7-carboxyglutamic acid residues could be determined even in a peptide containing 17 amino acid residues. An alternative approach has been taken by Fernlund et al. (104), to identify y-carboxyglutamic acid during sequenator degradation of peptides and proteins. To minimize the decarboxylation of y-carboxy- glutamic acid during conversion of the 2-anilino-5-thiazolinone dcriva- tives to phenylisothiocyanate derivatives, less drastic conditions than those employed in standard procedures were developed. The methyl- esterified phenylthiohydantoin derivative of y-carboxyglutamic acid was identified by mass spectrometry. Since some decarboxylation in- variably occurs in the sequenator during degradation of the peptide, positions at which 7-carboxyglutamic acid was identified also contained glutamic acid. As the degradation proceeded, the amount of glutamic acid increased. The phenylthiohydantoin derivative of y-carboxyglu- tamic acid has also been identified by high-voltage paper electro- phoresis ( 1 1 1). Carboxyl group activation with a water-soluble carbo- diimide and subsequent measurement of the incorporation of glycine ethylester into peptides has been used successfully (106,109) to measure the number of y-carboxyglutamic acid residues in peptides. A reduc- tion in anodal electrophoretic mobility at pH 6.5 of a peptide that has been briefly heated to 150°C relative to that of the unheated peptide has also been used to demonstrate the new amino acid (2,26, 103).

Five procedures for the chemical synthesis of the racemic form y-carboxyglutamic acid have been published (104,114-1 17). In one of them (104) malonic acid diethyl ester was coupled to P-acetamido- acrylic acid ethyl ester by Michael addition, forming 3-acetamido- 1 , 1,3-propanetricarboxyIic acid triethyl ester (I). Depending on the

VITAMIN K, PROTHROMBIN, AND 7-CARBOXYGLUTAMIC ACID 23

method used to isolate the product, either (I) or I-acetyl-2-pyrroli- done-3,5-dicarboxylic acid diethyl ester (11) was obtained. On alkaline hydrolysis both (I) and (11) gave y-carboxyglutamic acid, which was crystallized as the monoammonium salt. The crystal structure of the monoammonium salt of y-carboxyglutamic acid has been determined

Recently free y-carboxyglutamic acid was isolated from human urine (1 19). The urinary excretion in nine persons was between 19 and 42 jimole/24 hr. From the prothrombin turnover in man (2.4 mg/kg/24 hr) a 70 kg person would contribute 24 pmole/24 hr from the catabolism of prothrombin. Since prothrombin is quantitatively dominant among the vitamin K-dependent blood coagulation pro- teins, these data suggest that the bulk of 7-carboxyglutamic acid is not metabolized further in the body but is excreted unaltered in the urine.

( 1 18).

VI. Vitamin K-Dependent Carboxylation

The development of an in vitro system to study the biosynthesis of blood coagulation proteins has been slow. Although cell-free systems that showed an increase in vitamin K-dependent proteins on incuba- tion have been described, most of them did not respond to the i n vitro addition of vitamin K (21). One of these systems synthesized radio- immunologically detectable amounts of prothrombin (1 20,12 1) but did not respond to the in vitro addition of the vitamin. However, progress has been rapid since the discovery that prothrombin, like many other proteins made for export, is synthesized in precursor form and that conversion of the precursor to biologically active protein requires vitamin K-dependent carboxylations. The first system to pro- duce prothrombin in response to in vitro addition of vitamin K was described by Shah and Suttie (1 22). Postmitochondrial supernates from vitamin K-deficient rats responded to the addition of vitamin K by producing a significant amount of biologically active prothrombin. After the intrahepatic prothrombin precursor had been characterized and the vitamin K-dependent step in prothrombin synthesis had been shown to be the formation of 7-carboxyglutamic acid residues, Esmon et al. (27) demonstrated that the same postmitochondrial supernatant would catalyze a vitamin K-dependent incorporation of H14C0 3- into the endogenous microsomal precursor. Radioactive prothrombin was

24 JOHAN STENFLO

isolated from this system following incubation, and i t was shown that the radioactivity was present as 7-carboxyglutamic acid residues in the fragment 1 portion of prothrombin. The vitamin K-dependent carboxylase is remarkable in that its substrate is a precursor of a plasma protein. Both enzyme and substrate are present in liver microsomes. Studies on the enzyme have been complicated because no low mo- lecular weight substrate has been available until very recently, when Suttie et al. (123) showed that the pentapeptide Phe-Leu-Glu-Glu-Val was carboxylated by the solubilized microsomal enzyme system. Since no other peptides have been tried so far, the minimal sequence require- ment is not known. In prothrombin the first 10 glutamic acid residues in the NH2-terminal part of the molecule are carboxylated, and they occur both singly and in pairs (Fig. 8).

The requirements of the vitamin K-dependent carboxylase has been studied (28-30) in washed liver microsomes from vitamin K-deficient rats (i.e., microsornes containing the prothrombin precursor). The ac- tivity has been shown to require the presence of the precursor, 02, vitamin K, and HC03-, and to be stimulated by an energy source and factor(s) present in the postrnicrosomal supernatant. A major factor in the supernatant is a protein(s) acting as a NAD+ (NADPf) reductase. The requirement for reducing equivalents from pyridine nucleotides in the systems has been shown to be largely a requirement for the reduced form of vitamin K. Dithiothreitol (29,30) can be used as the source of reducing equivalents for the reaction, and this reducing agent might also protect an essential sulfhydryl group in the enzyme system. In this system vitamins with a geranyl or farnesyl group a t the 3 position of the vitamin are considerably more active than the phytyl derivative (phyloquinone) (30). The carboxylase activity in this micro- soma1 preparation is inhibited by warfarin, and this inhibition can be overcome by high concentrations of the vitamin (28).

The vitamin Kdependent carboxylase has now been solubilized in various detergents (29,124,125), and the solubilized preparation retains many of the properties of the membrane-associated system. The solu- bilized system is still stimulated by dithiothreitol and inhibited by merruricals. It has been reported that the solubilized system is in- hibited by the spin-trapping agent 5,5-dimethyl- 1 -pyroline-N-oxide and that 0 2 is not required in the solubilized system when the reduced form of the vitamin is used (124). The latter finding, however, is not supported by other data showing the requirement for 0 2 in such a

VITAMIN K , PROTHROMBIN, AND 7-CARBOXYGLUTAMIC ACID 25

system (125). I t is particularly noteworthy that the solubilized system is not inhibited by warfarin but is still sensitive to a direct vitamin K antagonist such as 2-chloro-3-phytyl- 1,4-naphthoquinone (1 25). Fur- thermore incubation in the absence of ATP and in the presence of an ATP inhibitor AMPP(NH)P does inhibit the membrane-bound car- boxylase activity (28), but not the solubilized system (125). These data may suggest that the energy to drive the carboxylation comes from the reoxidation of the reduced vitamin. I t has been proposed (124) that the semiquinone is the active form of vitamin K, but how it would participate in the carboxylation reaction is unknown. The semiquinone may also be merely an intermediate form in hydroquinone reoxidation. I t has recently been claimed that the hydroxylation of an alanine to a serine residue is a vitamin K-dependent step (126).

Originally it was assumed that 7-carboxyglutamic acid residues occur only in coagulation proteins, but now a new plasma protein (protein C) that contains y-carboxyglutamic acid residues has been found (56,127, 128). A recent significant finding has been the isolation of a low mo- lecular weight protein from bone by Hauschka et al. (112) and by Price et al. (113). The amino acid sequence of the XH2-terminal region of this y-carboxyglutamic acid containing protein is not homolo- gous to the NH2-terminal sequences in the vitamin K-dependent plasma proteins (1 13). According to preliminary reports, Gla residues have now also been demonstrated in renal tissue (129) and in kidney stones ( 1 30). Evidence that the carboxylation of the bone protein is vitamin K-dependent, has been presented (131), and vitamin K-de- pendent carboxylation has been demonstrated in kidney cortex (129), although the product has not been identified.

Any theory of the mechanism of action of vitamin K must take into consideration the possibility that the formation of the 2,3-epoxide of the vitamin is involved in the reaction (132). Bell and Matschiner demonstrated that warfarin administration increased the level of this metabolite in liver tissue, resulting in an increased ratio of vitamin K-oxide to vitamin K (133,134). I t was postulated that the 2,3-epoxide is a competitive inhibitor of the action of the vitamin and that warfarin exerts its anticoagulant effect by inhibiting an enzyme that reduces the 2,3-epoxide (135,136). Although this theory was supported by the observation that the epoxide reductase was less sensitive in warfarin- resistant rats (137,138), it has not been shown to be tenable (139-141). More rerently, Willingham and Matschiner (142) have postulated

26 JOHAN STENFLO

that the formation of the epoxide (“epoxidase” activity) is a nccrssary step in the action of the vitamin in promoting prothrombin biosyn- thesis. This hypothesis was based on observations that “epoxidase” activity increased in liver under various treatments in much the same manner as the concentration of the prothrombin precursor (143--144).

The flavoprotein DT-diaphorase, which catalyzcs the oxidation of NADH and KADPH by redox dyes and quinones, has been investi- gated by Ernster et al. (1455147). The enzyme, which is probably identical with the vitamin K-reductase described by Martius et al. (148-1 50), is highly sensitive to dicoumarol and related anticoagulants. The possibility that the enzyme may be involved in the metabolism of vitamin K in liver microsomes, perhaps with the 2,3-~poxide of the vitamin as an intermediate, has been considered (147). The involve- ment of this enzyme in the hepatic metabolism of vitamin K is hypo- thetical, and the mechanism whereby thr coumarin anticoagulants in- hibit the action of vitamin K is still unknown. The vitamin K-de- pendent carboxylase system should, however, serve as a useful tool in the investigation of these problems.

Acknowledgments

Work by the author cited in this chapter was generously supported by grants from the Swedish Medical Kesearch Council and from Direktor Albert Pihlssons Stiftelse.

The author wishes to thank Dr. J. W. Suttie and his coworkers for many valuable discussions on vitamin K and prothrombin (152).

References

1. Davie, E. W., and Fujikawa, K., A n n . RCD. Eiochcm., 44, 799-829 (1975). 2. Magnusson, S., Petersen, T. E., Sottrup-Jcnsen, L., and Claeys, H., in

Protcascs and Biological Control, R. Reich, D. B. Riflain, and E. Shaw, Eds., Cold Spring Harbor Laboratory, New York, 1975, pp. 123-149.

3. Enfield, D. L., Ericsson, L. H., Walsh, K. A., Neurath, H., and Titani, K., Proc. Nat. Acod. Sci. (U.S.), 72, 16-19 (1975).

4. Titani, K., Enfield, D. L., Ericsson, L.. H., Walsh, K . A,, and Nrurath, I l , , Proc. N a f . Acad. Sci. ( U S . ) , 72, 3082-3086 (1975).

5. Fujikawa, K., Coan, M. 14.. Lcgay, M. E., and Davie, E. IV., Bzochcmisfry,

6. Fujikawa, K., Titani, K., and Davie, E. W., Proc. Nu!. h a d . Sci. ( U S . ) , 72, 13, 5290-5299 (1974).

3359-3363 (1975).

VITAMIN K , I'ROTHROMBIN, AND 7-CARBOXYGLUTAMIC ACID 27

7. Jesty, .J., Spencer, A. K., and Nernerson, Y., J. Biol. Chem., 240, 5614-5622 (1974).

8. Kadcliff, R., and Nernerson, Y., J. Biol. Chcm., 250, 388-395 (1975). 9. Morita, T., Iwanaga, S., Suzuki, T., and Fujikawa, K., FebsLcrt., 36,313-317

( 1 973). 10. Furie, B. C., Furie, B., Gottlieb, A., and Williams, W. J., Biochim. Biophys.

1 I . Esmon, C. T., and Jackson, C. M., J. Biol. Chrm., 249, 7782-7790 (1974). 12. Esmon, C. T., and Jackson, C. M., J. Biol. Chcm., 249, 7791-7797 (1974). 13. Esmon, C. T., Owen, W. G., and Jackson, C. M., J. Biol. Chem., 249, 7798-

7807 (1974). 14. Jesty, J., and Esnouf, M. P., Biochem. J., 131, 791-799 (1973). 15. Kisiel, W., and Hanahan, D. J., Biochem. Biophys. Res. Commun., 59, 57&577

(1 974). 16. Butkowski, K. J., Bajaj, S. P., and Mann, K. G., J. Biol. Chcm., 249,6562-6569

(1974). 17. Bajaj, S. P.,Rutkowski, R. J., and Mann, K. G., J. Biol. Chem., 250,215&'2156

( 1 975). 18. Darn, H., Vitam. Horm., 24, 295-306 (1966). 19. Link, K. P., Harucy LEcf., 39, 162 (1943). 20. Olson, R. E., Vitam. Horm., 32, 483-51 1 (1974). 21. Suttie, J. W., Vitam. Horm., 32, 463-481 (1974). 22. Shah, D. V., and Suttie, J. W., Proc. Nut. tlcad. Sci. (U.S.), 68, 1653-1657

( 1 97 1 ). 23. Stenflo, J., Acta Chem. &and., 24, 3762-3763 (1970). 24. Stenflo, J., and Ganrot, P. O., J. Biol. Chem., 247, 8160-8166 (1972). 25. Nelsestuen, G. L., and Suttie, J. W., J. Biol. Chem., 247, 8176-8182 (1972). 26. Stenflo, J., Fernlund, P., Egan, W., and Roepstorff, P., Proc. Naf . Acad. Sci.

(US.) , 77, 273@2733 (1974). 27. Esmon, C. T., Sadowski, J. A., and Suttie, J. W., J. Biol. C k m . , 250, 4744-

4748 (1975). 28. Sadowski, J. A., Esmon, C. T., and Suttie, J. W., J. Biol. Chem., 251, 277&

2775 (1976). 29. Mack, D. O., Sum, E. T., Girardot, J. M., Miller, J. A., Dclaney, R., and

Johnson, B. C., J. Biol. Chem., 251, 3269-3276 (1976). 30. Friedman, P. A., and Shia, M., Biochim. Biophys. Res. Commun., 70, 647-654

(1976). 31. Weber, F., and Wiss, O., in The Vitamins, Vol. 3, W. H. Sebrell and R. S.

Harris, Eds., Academic Press, New York, 1971, pp. 457466. 32. Davie, E. W., and Katnoff, 0. D., Sciencc, 145, 13 10-1 3 12 (1 964). 33. Macfarlane, K. G., Nafurc, 202, 498-499 (1 964). 34. Fujikawa, K., Legaz, M. E., Kato, H., and Davie, E. W., Biochemisfry, 73,

450845 16 ( 1974). Fujikawa, K., Titani, K., and Davie, E. W., Proc. Nut. Acad. Sci. (U.S.), 72, 3359-3363 (1 975).

36. Nelsestuen, G. I,., and Suttie, J. W., Biochemisfry, 11, 4961-4964 (1972).

Acta, 365, 121-132 (1974).

35.

28 JOHAN STENFLO

37. Stenflo, J., and Ganrot, P. O., Biochcm. Biophys. RCS. Commun., 50, 98-104 11973).

38. Benson, B. J., Kisiel, W., and Hanahan, D. J., Biochim. Biophys. Acfa, 329,

39. Henriksen, R. A,, and Jackson, C. M., Arch. Biochem. Biophys., 770, 149-159 ( I 975).

40. Furie, B. C., Mann, K. G., and Furie, B., J. Biol. Chrtn., 251, 3235-3241 ( I 976).

41. Papahadjopoulos, D., and Hanahan, D. G., Biochim. Biophys. Acfa, 90,436-439

42. Bull, R. K., Jevons, S., and Barton, P. G., J. Biol. Chcm., 247, 2747-2754 ( I 972).

43. Gitel, S. N., Owen, W. G., Esnion, C. T., and Jackson. C. M., Proc. Nat. Acad. Sci. (US.), 70, 1344-1348 (1973).

44. Nelsestuen, G. L., Broderius, M., Zytkovicz, T. H., and Howard, J. B., Biochem. Biophys. Rcs. Commun., 65, 233-240 (1975).

45. Benson, B. J., and Hanahan, D. J., Biochcmisfry, 74, 3265-3277 (1975). 46. Bajwa, S. S., and Hanahan, D. J., Biochim. Biophys. Acla, 444, 118-130 (1976). 47. Subbaiah, P. V., Bajwa, S. S., Smith, C. M., and Hanahan, D. J., Biochim.

Biophys. Acfa, 444, 131-146 (1976). 48. Jackson, C. M., Esmon, C. T., Gitel, S. N., Owen, W. G., and Henriksen,

R. A., in Boerhauc Series, Vol. 10, H. C. Heniker and J. J. Veltkamp, Eds., Leiden University Press, Leiden, 1975, pp. 59-88.

49. Nelsestuen, G. L., J. Biol. Clum., 257, 5648-5656 (1976). 50. Wessler, S., and Yin, E. T., J. Lob. Clin. hlcd. , 72, 256-259 (1968). 51. Barton, P. G., Yin, E. T.. and Wessler, S., J . Lipid. Rcs., 77, 87-95 (1970). 52. Davie, E. W., and Kirby, E. P., Curr. Top. G11. Rcgul., 51-86 (1973). 53. Kisiel, W., and Davie, E. W., Biochcmisfry, 74, 49284934 (1975). 54. Magnusson, S., in The Enzymes, VoI. 3, P. D. Boyer, Ed., Academic Press,

New York, 1971, pp. 277-321. 55. Denson, R. W. E., in Human Blood Coagulafion Hacmosfasis and Thrombosis,

R. Biggs, Ed., Blackwell Scientific Publications, Oxford, 1972, pp. 602-675. 56. Stenflo, J., J. Biol. Chcm., 257, 355-363 (1976). 57. Hemker, H. C., Veltkamp, J. J., Hensen, A., and Loeligcr, E. A., Nafurc, 200,

589-590 ( I 963). 58. Nilthn, J.-E., and Ganrot, P. O., &and. J. Clin. Lab. Invest., 22, 17-22 (1968). 59. Ganrot, P. O., and NilChn, J.-E., Scand. J. Clin. Lab. Inucsf., 22, 23-28 (1968). 60. Josso, F., Lavergne, J. M., Gouault, M., Prou-Wartelle, O., and Soulier,

J. P., Thrombos. Diafhcs. Hacmorrh., 20, 88-98 (1968). 61. Laurell, C.-B., Anal. Biochcm., 70, 358-361 (1965). 62. Ganrot, P. O., Scand. .I. Cfin. Lab. Znvcsf., 29 (Suppl. 124), 3 9 4 7 ( I 972). 63. Hemker, €1. C., Muller, A. D., and Loeliger, E. A,, Thrombos. Dinthrs.

Haemorrh., 23, 633-637 (1970). 64. Denson, K. W. E., Brif. J. Hacmafol., 20, 643-648 (1971). 65. Brozovic, M., and Curd, L. J., Brif. J. Hacmalol., 24, 579-588 (1973). 66. Cesbron, N., Boyer, C., Guillin, M.-C., and Menache, D., Thrombos. Diafhcs.

Hatmorrh., 30, 437450 (1 973).

81-87 (1973).

( 1 964).

VITAMIN K, PROTHROMBIN, AND y-CARBOXYGLUTAMIC ACID 29

67. Morrison, S. A., and Esnouf, M. P., Nafurc N e w Biol., 242, 92-94 (1973). 68. Guillin, M.-C., Menache, D., Aronson, D., Amar, M., and Schlegel, N.,

.4bsfracts, Vth Congress OJ Thrombosis and Haemosfasis, July 21-26, 1975, p. 200. 69. Reekers, P. P. M., Lindhout, M. J., Kop-Klaassen, B. H. M., and Hemker,

H. C., Biochim. Biophys. Acta, 317, 559-5612 (1973). 70. Wallin, R., and Prydz, H., Biochem. Biophys. Rcs. Commun., 62, 398-406 (1975). 71. Stenflo, J., J. Biol. Chem., 247, 8167-8175 (1972). 72. Bjork, J., and Stenflo, J., F E B S Lcff . , 32, 343-346 (1973). 73. Hemker, H. C., Bas, B. M., and Muller, A. D., Biochim. Biophys. Acfa, 379,

74. Stenflo, J., J. Biol. Chem., 248, 6325-6332 (1973). 75. Nelsestuen, G. L., and Suttie, J. W., Proc. N a f . Acad. Sci. (U.S.), 70, 3366-3370

(1973). 76. Stenflo, J., J. Biol. Chem., 249, 5527-5535 (1974). 77. Magnusson, S., Thrombos. Diafhes. Haemorrh. (Suppl.) , 54, 31-35 (1973). 78. Esmon, C. T., Suttie, J. W., and Jackson, C. M., J. Biol. Chem., 250, 4095-

4099 (1975). 79. Pereira, M., and Couri, D., Biochim. Biophys. h a , 237, 348-355 (1971). 80. Johnson, H. V., Martinovic, J., and Johnson, B. C., Biochem. Biophys. Rcs.

Commun., 43, 1040-1049 (1971). 81. Pereira, M. A., and Couri, D., Biochim. Biophys. Acfa , 267, 375-378 (1972). 82. Nelsestuen, G. L., and Suttie, J. W., Biochem. Biophys. Rcs. Commun., 45,

83. Henriksen, A,, Christensen, T. B., and Helgeland, L., Biochim. Biophys. Acfa,

84. Prowse, C. V., Mattock, P., Esnouf, M. P., and Russell, A. M., Biochim. Biophys. Acfa , 432, 265-279 ( I 976).

85. Malhotra, 0. P., and Carter, J. R., J. Biol. Chem., 246, 2665-2671 (1971). 86. Malhotra, 0. P., Nafure New Biol., 239, 59-60 (1972). 87. Hill, R. B., Gaetani, S., Paolucci, A. M., Rama Rao, P. B., Alden, R.,

Ranhotra, .G. S., Shah, D. V., Shah, V. K., and Johnson, B. C., J. Biol. Chem., 243, 393C3939 (1968).

88. Bell, R., and Matschiner, J. T., Arch. Biochem. Biophys., 735, 152-159 (1969). 89. Suttie, J. W., Arch. Biochem. Biophys., 141, 571-578 (1970). 90. Suttie, J. W., Science, 179, 192-194 (1973). 91. Shah, D. V., Suttie, J. W., and Grant, G. A,, Arch. Biochem. Biophys., 759,

483491 (1973). 92. Markwardt, F., Hoppe-Zeylcr’s 2. Physiol. Chem., 308, 147-156 (1957). 93. Esmon, C. T., Grant, G. A., and Suttie, J. W., Biochemisfry, 14, 1595-1600

(1 975). 94. Morrissey, J. J., Jones, J. P., and Olson, R. E., Biochem. Biophys. Rcs. Commun.,

95. Grant, G . A,, and Suttie, J. W., Arch. Biochem. Biophys., 776, 650-662 ( 1 976).

96. Nardacci, N. J., .Jones, J. P., Hall, A. L., and Olson, R. E., Biochcm. Biophys. Res. Commun., 64, 51-58 (1975).

97. Grant, G. A.. Ph.D. thesis, Universiy of Wisconsin, Madison, 1975.

180-188 (1975).

198-203 (1971).

427, 348-352 (1976).

54, 1075-1082 (1973).

30 JOEIAN STENFLO

98. Carlisle, T. L., Shah, D. V., Schlegel, K., and Suttie, J. W., Proc. SOC. Exp. B i d . M e d . , 148, 140-144 (1975).

99. Johnson, H. V., Martinovic, J., and Johnson, B. C., Biochcm. Biophys. Rcs. Commun., 43, 1040-1048 (1971).

100. Johnson, H. V., Boyd, C., Martinovic, J., Valkovich, G., and Johnson, B. C. : Arch. Riochem. Biophvs., 748, 431-442 (1972).

101. Morriuey, J. J., Kipfer, K. K., and Olson, R. E., Proc. A m . Soc. Exp. Biol., 3 1 , 218 (1972).

102. Nelsestuen, G. L., Zytkovicz, T. Il . , and Howard, J. B., J. B i d . Chcm., 249, 6347-6350 (1 974).

103. Magnusson, S., Sottrup-Jensen, L., Petenen, T. E., Morris, H. K., and Dell, A., FERS Ltf., 44, 189- 193 (1 974).

104. Fernlund, P., Stenflo, J., KoepstorfT, R., and Thomsen, J., J . Biol. Chcm.,

105. Morris, H. R., Dell, A,, Petenen, T. E., Sottrup-Jensen, L., and Magnusson, S., Biochem. J., 153, 663--679 (1976).

106. Howard, J. B., and Nelsestuen, G . L., Hiochrm. Riophys. Rcs. Commun., 59,

107. Zytkovicz, T. H., and Nelsestuen, G . L., J. H i d . Chcm., 250,2968-2972 (1975). 108. Girardot, J.-M., Delaney, R., and Johnson, B. C., Biochrm. Hiophys. Hes.

Commun., 59, 1197-1203 (1974). 109. IIoward, J. B., amd Nelsestuen, G. L., Proc. Nof . Acad. Sci. ( U S . ) , 72, 1281-

1285 (1975). 110. Bucher, D., Nebelin, E., Thornsen, J., and Stenflo, J., FEBS Lcf f . , 68,293-296

( I 976). 1 11. Fryklund, L., Borg, H., and Anderson, I..-O., FEBS Ltf., 65. 187-189

(1976). 112. Hauschka, P. V., Lian, J. B., and Gallop, P. M., Proc. N a t . Acad. Sci. ( U S . ) ,

113. Price, P. A,, Otsuka, A. S., Poser, J. W., Kristaponis, J., and Raman, N., Proc. N a f . h a d . Sci. ( U S . ) , 73, 1447-1451 (1976).

114. Morris, H. K., Thompson, M. R., and Dell, A,, Biochcm. Biophys. Res. Cornmiin., 62, 856-861 (1975).

115. Marki, W., and Schwyzer, K., Heir. Chim. Ada, 58, 1471-1477 (1975). 116. Prestidge, K. L., Harding, D. R. K., Battenby, J. E., and Hancock. W. S.,

117. Bajusz, S., and juhhz , A,, Acfa Chim Acad. Sci. flung., 88, 161-165 (1976). 118. Satyshur, H. A,, Rao, S. T., Stenflo, J., and Suttie, J. W., presented at the

American Crystallographic Association Meeting at Northwestern University, Evanston, Ill., 1976.

119. Fernlund, P., Clin. Chim. Acfa, 72, 147-155 (1976). 120. Johnston, M. F. M., and Olson, R. E., J . B i d . Chem., 247. 3994-4000 (1972). 121. Johnston, M. F. M., and Olson, R. E., J. Bioi. C'hcm., 247, 4001 -.4007 (1972). 122. Shah, D. V., and Suttie, J. W., Biochem. Biophys. Rrs. Conrmun., 60, 1397-1402

(1974). 123. Suttie, J. W., Hageman, J. M., Lehrman, S. K., and Rich, D. H., J. Riol.

Chrm., 251, 5827-5830 (1976).

250, 6 125-6 I33 (1 975).

757-763 (1974).

72, 3925-39119 (1975).

J. Org. C h m . , 40, 3287-3288 (1975).

VITAMIN K , I’ROTHROMBIN, AND 7-CARBOXYG1,UTAMIC ACID 3 1

124. Girardot, J. M., Mack, D. O., Floyd, R. A,, and Johnson, B. C. , Biochem. Biophys. Res. Commun., 70, 655-662.

125. Esnion, C. T., and Suttie, J. W., J . B i d . Chem., 257, 6238-6243 (1976). 126. Matthes, K. J., Mcd. Well., 26, 1777-1782 (1975). 127. Kisiel, W., Ericsson, L. H., and Davie, E. W., Biochemisfry, 75, 48934900

(1976). 128. Seegers, W. H., Novoa, E., Henry, K. L., and Hassovna, H. I., Thrombos. Rcs.,

129. IIauschka, P. V., Friedman, P. A., Traverso, I I . P., and Gallop, P. M., Biochrm. Biophys. Rcs. Cornmun., 71, 1207-1213 (1976).

130. Lian, J. B., and Prien, E. L., Jr., Fed. Proc., 35, 1763 (Abstr.) (1976). 131. Hauschka, P. V., Reid, M. L., I ian, .J. B., Friedman, P. A., and Gallop,

P. M., Fed. Proc., 35, 1354 (Abstr.) (1976). 13’2. Matschincr, J. T., Bell, K. G., Amelotti, J. M., and Knauer, T. E., Biochim.

Biophys. Ada , 207, 309 3 15 (1 970). 133. Bell, K . G., and Matschiner, J. T., Arrh. Biochem. Biophys., 141,473-476 (1970). 134. Bell, K. G., and Matschiner, J. T., Nature, 237, 32-33 (1972). 135. Bell, R. G., Sadowski, J. A,, and Matschiner, J. T., Biochcmisfry, 7 7 , 1959-

1961 (1972). 136. Ken, P., Laliberte, K. E., and Bell, K. G., Mof. Pharmacol., 10,373-380 (1973). 137. Bell, R. G., and Caldwell, P. T.. Biochcmislry, 72, 1759-1762 (1973). 138. Zimmerman, A., and Matschiner, J. T., Biochcm. Pharmacol., 23, 1033-1040

(1974). 139. Sadowski, J. A., and Suttie, J. W., Biochcmisfry, 73, 3696-3699 (1974). 140. Goodman, S. R., Houser, R. M., and Oslon, R. E., Riochcm. Biophys. Res.

Commun., 61, 250-257 (1974). 141. Caldwell, P. T., Ken, P., and Bell, R. G., Biochcm. Phnrmacol., 23, 3353-3362

(1974). 142. Willingham, A. K., and Matschiner, J. T., Biochcm. J., 740,435-441 (1974). 143. Bell, K. G., Caldwell, P. T., and Holm, E. E. T., Biochcm. Pharmacol., 2.5,

1067-1070 (1 976). 144. Willingham, A. K., Laliberte, K. E., Bell, K. G., and Matschiner, J. T.:

Riochrm. Phnrmacol., 25, 1063-1066 (1976). 145. Ernster, L., nanielson, L., and Ljunggren, M., Bzochem. Biophys. Ada, 58,

146. Ernster, L., Melhods in Enzymology, Vol. 10, K. W. Estabrook and M. E. Pullman, Eds., Academic Press, New York, 1967, pp. 309-317.

147. Emster, L., Lind, C., and Rase, B., Eur., J . Biochem., 25, 198-206 (1972). 148. Martius, C., and Strufe, R., Biochem. Z . , 326, 24-25 (1954). 149. Marki, F., and Martius, C., Biochcm. Z . , 333, 11 1-135 (1960). 150. Marki, F., and Martius, C., Riochcm. Z. , 334, 293-303 (1961). 151. Kisiel, W., and Davie, E. W., Riochcmidry, 14, 49284934 (1975). 152. Stenflo, J., and Suttie, J. W., .4nn. RPZJ. Biochcm., 46, (1977).

8, 543-552 (1976).

171-188 (1962).