A New Carboxylation Reaction THE VITAMIN K-DEPENDENT INCORPORATION OF H”CO,m INTO PROTHROMBIN*

The bovine plasma zymogen prothrombin contains a number ofy-carboxyglutamic acid residues which are not found in an abnormal prothrombin produced when cattle are given the vitamin K antagonist dicoumarol. These modified glutamic acid residues appear to be formed post-translationally by a reaction which requires vitamin K. It has been shown that postmitochondrial supernates from vitamin K-deficient rats incorporate added H’“CO,-into microsomal proteins upon the addition of vitamin K. This incorporation is dependent upon the presence of the prothrombin precursor in the microsomal preparations, and upon factors which are present in the postmicrosomal supernatant. Most of the radioactive protein which can be obtained from the microsomal pellet by extraction with 0.25% Triton X-100 has been identified as prothrombin and it can be shown that all of the radioactivity is in the amino-terminal activation fragment of prothrombin. This portion of the protein has previously been shown to contain the y-carboxyglutamic acid residues. Hydrolysis of the purified radioactive prothrombin resulted in a loss of 50% of the radioactivity and subsequent chromatography of the amino acid hydrolyzate demonstrated that the remaining

The bovine plasma zymogen prothrombin contains a number ofy-carboxyglutamic acid residues which are not found in an abnormal prothrombin produced when cattle are given the vitamin K antagonist dicoumarol.
These modified glutamic acid residues appear to be formed post-translationally by a reaction which requires vitamin K. It has been shown that postmitochondrial supernates from vitamin K-deficient rats incorporate added H'"CO,-into microsomal proteins upon the addition of vitamin K. This incorporation is dependent upon the presence of the prothrombin precursor in the microsomal preparations, and upon factors which are present in the postmicrosomal supernatant.
Most of the radioactive protein which can be obtained from the microsomal pellet by extraction with 0.25% Triton X-100 has been identified as prothrombin and it can be shown that all of the radioactivity is in the amino-terminal activation fragment of prothrombin. This portion of the protein has previously been shown to contain the y-carboxyglutamic acid residues. Hydrolysis of the purified radioactive prothrombin resulted in a loss of 50% of the radioactivity and subsequent chromatography of the amino acid hydrolyzate demonstrated that the remaining radioactivity was entirely in glutamic acid. These results are consistent with the hypothesis that all of the H'"CO,-was incorporated into the carboxyl groups of y-carboxyglutamic acid residues.
Vitamin K is required for the synthesis of four blood-clotting zymogens: prothrombin, factor X, factor IX, and factor VII. The vitamin appears to function post-translationally (1) by modifying a precursor protein. This precursor has been identified (2) in microsomal preparations from anticoagulant-treated rats, and has now been isolated and partially characterized (3,4). Although it is inactive in prothrombin bioassay systems, this precursor is activated to thrombin by several snake venoms (4), suggesting that the vitamin K-dependent modification is required for the physiological activation of prothrombin rather than in the activity of the thrombin generated. The liver precursor is in many ways similar to the biologically inactive form of prothrombin (abnormal prothrombin) which appears in the plasma of the bovine following administration of the vitamin K antagonist, dicoumarol. Unlike prothrombin, the abnormal prothrombin does not bind Ca2+ ions (5, 6) and this defect is presumably responsible for its failure to activate in the bioassay. As it was possible to isolate (7) a low molecular weight calcium-binding peptide from normal, but not abnormal, prothrombin, it appeared that the vitamin-dependent alteration involved a chemical modification of a specific region of the polypeptide chain. The chemical difference in the abnormal and normal prothrombin has been shown by Stenflo et al. (8) to be the presence of a number of y-carboxyglutamic acid residues in normal prothrombin but not in abnormal prothrombin. This residue has also been identified by Nelsestuen et al. (9) and the characterization has been confirmed by Magnusson et al. (10). These observations suggest that vitamin K functions as part of the metabolic system responsible for the y-carboxylation of specific glutamic acid residues of the liver prothrombin precursor.
We have recently described (11) an in vitro system which converts the rat liver microsomal precursor protein to biologically active prothrombin in response to the addition of vitamin K. This system should serve to test the hypothesis that the vitamin K-dependent, post-translational modification of the precursor involves the carboxylation of glutamic acid residues.

MATERIALS AND METHODS
Treatment ofAnimals-Male 250-g rats of the Holtzman strain were housed in coprophagy-preventing cages (12) and fed a diet low in to obtain a postmitochondrial supernatant which was incubated under the conditions previously described (11) for the in vitro synthesis of prothrombin.
Cycloheximide (100 pg/ml) and H'"CO,-(5 bCi/ml of 59.5 mCi/mmol of Na"CO, (Amersham/Searle)) were included in the incubation medium, and prothrombin synthesis was initiated by the addition of vitamin K, (20 @g/ml). After incubation for 15 min at 37", the suspension was cooled, and the microsomes were removed by centrifugation at 105,000 x g for 60 min. The microsomal pellet was extracted with calcium-free Krebs-Ringer bicarbonate buffer containing 0.015 M potassium oxalate and 0.25% Triton X-100, and the unsolubilized debris was removed by centrifugation as described above. The Triton extract was adsorbed with BaSO, (25 mg/ml). The BaSO, was removed by centrifugation and this pellet was washed and eluted as described earlier (14). Determination of Radioactiuity-Bovine serum albumin (2 mg) was added to 0.2 ml of the Triton X-100 microsomal extract, or 0.2 ml of the BaSO, adsorbed extract and the proteins were precipitated by the addition of 5 ml of 10% trichloroacetic acid. The BaSO, eluate (0.1 ml) was precipitated after the addition of 4 mg of albumin. The precipitates were held at 4' for 30 min and then collected by centrifugation at 3000 x g for 20 min. The supernatant was discarded, the pellet was dissolved in 1 ml of 0.2 M Na,CO, and reprecipitated with 5 ml of 10% trichloracetic acid. After 30 min at 4', the suspension was centrifuged as before, the supernatant was discarded, and the pellet was dissolved in 1 ml of NCS (Amersham-Searle) before transferring the sample to 10 ml of Econofluor (New England Nuclear). The distribution of radioactivity in sodium dodecyl sulfate electrophoretic gels was determined following combustion of the dried gel slices. Radioactivity was determined in a liquid scintillation spectrometer with a counting efficiency of 82% for the protein samples and 22% for the cornbusted gels. To determine the location of the radioactivity in purified "C-labeled prothrombin, the protein was dialyzed against 2 mM potassium phosphate buffer, pH 5.8, concentrated to dryness, dissolved in 6 N HCl and hydrolyzed at 105" in U~CUO for 22 hours. The hydrolyzate was concentrated to dryness and dissolved in 500 ~1 of pH 2.2 sample diluting buffer, and 400 ~1 of the sample was applied to a Beckman model 120C amino acid analyzer equipped with a coIumn of Eeckman UR30 resin. One-minute fractions (180) of column eluate was collected in glass counting vials at a flow rate of 68 ml/hour. Aquasol (10 ml) was added to each vial and radioactivity determined in a liquid scintillation spectrometer at a "C efficiency of 35% and a 3H efficiency of 14% for the double-label samples and a "C efficiency of 68% for the single-label samples. DL-[2-sH]glutamic acid (Amersham/Searle) and [1-"Clnorleucine (New England Nuclear) were added prior to acid hydrolysis to unambigously define the elution position of glutamic acid.
Isolation of Clotting Factor and Bioassay-Factor Xa and factor V were prepared as described previously (15). Phospholipid was fraction II prepared as described by Folch (16). Prothrombin was assayed by the two-stage method of Ware and Seegers as modified by Shapiro and Waugh (17). Factor X was assayed by the one-stage method of Bachman et al. (18).

Correlation of Prothrombin Synthesis and
Carboxylation-The possible existence of a vitamin K-dependent protein carboxylation was investigated by incubating (11) postmitochondrial supernatants prepared from livers of vitamin K-deficient rats in the presence of vitamin K, in the absence of vitamin K and in the presence of both vitamin K and a vitamin K antagonist, chloro-K. During these incubations, de nouo protein synthesis was inhibited with cycloheximide. The data (Table I) indicated that vitamin K was required for optimal H"CO,-incorporation into the Triton X-lOO-extractable microsomal proteins as well as for prothrombin synthesis. Furthermore, chloro-K, which inhibited prothrombin synthesis, also inhibited H'%O,-incorporation. Since the o n 1 y generally recognized requirement for vitamin K is the synthesis of the four blood-clotting proteins, much of the protein-bound radioactivity should have been incorporated into these proteins. BaSO, is a specific adsorbant for the vitamin K-dependent clotting factors, and adsorption of the 4745  (17). c 20 rg/ml. Triton extract with BaSO, indicated that a high percentage of the radioactivity was incorporated into the BaSO,-adsorbable proteins. Adsorption of the extract with BaSO, removed 56% of the radioactive protein from the Triton X-100 microsomal extract and elution from the BaSO, resulted in an increase in specific activity of Y! protein from 65 dpm/A,,, in the microsomal extract to 44,390 dpm/A,,, in the BaSO, eluate. These results suggest a highly specific incorporation of bicarbonate into the vitamin K-dependent clotting proteins. The nature of the non-BaSO, adsorbable proteins has not yet been determined.
If bicarbonate was being incorporated primarily into protein precursors of the vitamin K-dependent clotting factors, then conditions which prevent synthesis of the active clotting factors from these precursors should also prevent H'"CO,incorporation into protein. Previous experiments in our laboratory have indicated that microsomal prothrombin synthesis requires some factor(s) present in the soluble portion of the cell. When microsomes were prepared from the postmitochondrial supernate and resuspended in buffered sucrose without the addition of cytosol (Table II, A), both H'"CO,-incorporation and prothrombin synthesis were inhibited. If the protein carboxylation observed does represent the specific postribosomal incorporation of H'"CO,-into the vitamin K-dependent clotting factors, it should require the presence of the precursors of these proteins. Microsomes from normal, vitamin K-sufficient rats contain very little prothrombin precursor, when compared to microsomes from vitamin K-deficient rats, and because of the low precursor level, microsomes from these rats would be expected to incorporate less H"CO,-than microsomes from the vitamin K-deficient rats. When the extent of in vitro carboxylation was compared in systems derived from normal or vitamin K-deficient rats (Table II, B), the data indicated that the system prepared from normal rat livers incorporated less bicarbonate than that prepared from vitamin K-deficient rats. Although the administration of Warfarin in uiuo blocks prothrombin synthesis, the in vitro synthesis of prothrombin in liver microsomes is not significantly inhibited by Warfarin (11). However, as indicated in Table II, B, mitochondrial  supernatants derived from Warfarin-treated rats form less prothrombin in uitro than do supernates from vitamin K-deficient rats. A further correlation between prothrombin synthesis and carboxylation is provided by the observation (Table II, B) that H'"CO,-incorporation as well as prothrombin synthesis is reduced in systems derived from Warfarin-treated rats when compared to systems derived from vitamin K-deficient rats. These data (Table II, B) also indicate that prothrombin synthesis under these conditions may have been inhibited more than H"CO,-incorporation into the Triton extract, and suggest that there may be enhanced carboxylation of some protein other than prothrombin in the presence of Warfarin. The amount of radioactivity which was incorporated in the cytosol was investigated as well as that incorporated into the microsomal pellet. Approximately the same amount of radioactivity was incorporated into the cytosol proteins as into the microsomal extract from the vitamin K-treated system. The amount of radioactivity in this fraction was, however, not dependent on the presence or absence of the vitamin in the incubation medium. Identification of Prothrombin as Radioactive Protein-The degree of correlation between the radioactive proteins of the BaSO, eluate and the vitamin K-dependent clotting proteins was examined by ion exchange chromatography (Fig. 1). Most of the protein in the eluate eluted before any of the radioactivity, but a small amount of protein eluted at the same position as prothrombin which was detected by both two-stage activity and venom activation.
Prothrombin activity and radioactivity appeared to co-chromatograph except for a reproducible shoulder of radioactivity on the trailing edge of the prothrombin peak. This fraction of the radioactivity elutes in the position expected for rat factor X, but bioassay failed to detect any factor X activity. The similar chromatographic behavior of the prothrombin activity and the radioactivity, and the small amount of protein in this region of the chromatogram suggests that much of the radioactivity has been incorporated into prothrombin.
The properties of radioactive protein which eluted from the QAE (quaternary aminoethyl) column with prothrombin were studied further by sodium dodecyl sulfate gel electrophoresis. In this electrophoretic system, rat prothrombin has an apparent molecule weight of 85,000. When the chromatographically purified radioactive protein was subjected to electrophoresis under these conditions, most of the radioactivity was associated with the gel slice corresponding to this molecular weight (Fig. 2). Activation of prothrombin with factor Xa leads to formation of thrombin and two large activation peptides (fragment 1, M, = 23,000 and fragment 2, M, = 13,000) (15,20). All of the y-carboxyglutamic acid residues of prothrombin are reported to reside in the amino-terminal activation peptide, fragment 1 (10). Therefore, if the H'"CO,-was specifi- be subjected to random decarboxylation by acid hydrolysis, and a 50% loss of the label should be observed. The data in Table III demonstrate that about 53% of the radioactivity associated with the prothrombin isolated from the in uitro system was lost following hydrolysis in 6 N HCl. Further, ion exchange chromatography of this hydrolysate revealed the presence of only one 14C peak in the eluate from the amino acid analyzer (Fig. 3), which corresponded to the elution position of glutamic acid, the decarboxylation product of y-carboxyglutamic acid. Addition of [3H]glutamic acid and ['4C]-norleutine to the sample before hydrolysis resulted in a co-elution of 3H and '"C in the glutamic acid position of the chromatogram which was verified by its position on the chromatogram relative to the standard norleucine.

DISCUSSION
The data presented above are consistent with the hypothesis that the vitamin K-dependent step in prothrombin synthesis involves the carboxylation of glutamyl residues in a liver prothrombin precursor protein. There was a rapid, vitamin K-dependent incorporation of H'"CO,-into protein, even when de nouo protein synthesis was blocked. A significant amount of this radioactivity was associated with the vitamin K-dependent clotting factors, primarily prothrombin, and the radioactivity incorporated into prothrombin was located exclusively in the NH,-terminal activation fragment (fragment 1) of prothrombin.
Acid hydrolysis of the in vitro labeled prothrombin resulted in a loss of 50% of the radioactivity, and the remaining radioactivity was associated with glutamic acid residues, which would be consistent with the presence of radioactivity in the carboxyl groups of y-carboxyglutamyl residues.