Identification of Amino Acids in the y-Carboxylation Recognition Site on the Propeptide of Prothrombin *

A y-carboxylation recognition site on the propeptide of the vitamin K-dependent blood coagulation proteins directs the carboxylation of glutamic acid residues by binding to the vitamin K-dependent carboxylase. To determine residues that define this site, we evaluated the effect of mutation of certain residues in the prothrombin propeptide on the extent of carboxylation. The prothrombin cDNA modified by site-specific mutagenesis was expressed in Chinese hamster ovary cells using a system that yields functional fully carboxylated prothrombin. The cell supernatants containing recombinant prothrombin were evaluated for the extent of r-carboxylation by immunoassay. Conformation-specific anti-prothrombin:Ca(II)-specific antibodies measure native completely carboxylated prothrombin; antiprothrombin:total antibodies measure all forms of prothrombin, regardless of r-carboxyglutamic acid content. Mutation of His-” to Gly, Val-” to Ser, Leu-I6 to Gly or Asp, or Ala-” to Asp was associated with a partial (30-66%) inhibition of y-carboxylation. Mutation of Ala-l4 to Ser or SermB to Val did not inhibit ycarboxylation. From this and earlier work, residues whose mutation leads to a significant impairment of carboxylation include His-“, Val-“, Phe-“, Leu-lS, and Ala-“. Residues whose mutation does not alter the carboxylation recognition site include Ala-14, Ser-‘, Arge4, and Arg-‘. To determine the size of the recognition site, the in vitro carboxylation of propeptidecontaining synthetic peptides was compared. A 28residue peptide, based upon residues -IS to +lO of prothrombin, and a 64-residue peptide, based upon residues -IS to +36 of prothrombin, were carboxylated by partially purified bovine carboxylase with similar K, values of 2-5 j&M. These results indicate that the y-carboxyglutamic acid-rich region of prothrombin makes a minimal contribution to carboxylase binding. A molecular surface of about five amino acids located within the propeptide appears to define the carboxylation recognition site on the precursor forms of the vitamin K-dependent proteins.

A y-carboxylation recognition site on the propeptide of the vitamin K-dependent blood coagulation proteins directs the carboxylation of glutamic acid residues by binding to the vitamin K-dependent carboxylase. To determine residues that define this site, we evaluated the effect of mutation of certain residues in the prothrombin propeptide on the extent of carboxylation. The prothrombin cDNA modified by site-specific mutagenesis was expressed in Chinese hamster ovary cells using a system that yields functional fully carboxylated prothrombin.
The cell supernatants containing recombinant prothrombin were evaluated for the extent of r-carboxylation by immunoassay. Conformation-specific anti-prothrombin:Ca(II)-specific antibodies measure native completely carboxylated prothrombin; antiprothrombin:total antibodies measure all forms of prothrombin, regardless of r-carboxyglutamic acid content. Mutation of His-" to Gly, Val-" to Ser, Leu-I6 to Gly or Asp, or Ala-" to Asp was associated with a partial (30-66%) inhibition of y-carboxylation.
Mutation of Ala-l4 to Ser or SermB to Val did not inhibit ycarboxylation.
From this and earlier work, residues whose mutation leads to a significant impairment of carboxylation include His-", Val-", Phe-", Leu-lS, and Ala-". Residues whose mutation does not alter the carboxylation recognition site include Ala-14, Ser-', Arge4, and Arg-'. To determine the size of the recognition site, the in vitro carboxylation of propeptidecontaining synthetic peptides was compared. A 28residue peptide, based upon residues -IS to +lO of prothrombin, and a 64-residue peptide, based upon residues -IS to +36 of prothrombin, were carboxylated by partially purified bovine carboxylase with similar K, values of 2-5 j&M. These results indicate that the y-carboxyglutamic acid-rich region of prothrombin makes a minimal contribution to carboxylase binding. A molecular surface of about five amino acids located within the propeptide appears to define the carboxylation recognition site on the precursor forms of the vitamin K-dependent proteins.
Prothrombin, a vitamin K-dependent blood coagulation protein, is the zymogen of the enzyme thrombin. This protein * This work was supported by Grants HL42443 and HL38216 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges.
This article must therefore he hereby marked "advertisement" in accordance with 18 U.&C. Section 1734 solely to indicate this fact.
§ Recipient of a postdoctoral fellowship from the National Science Foundation. contains 10 y-carboxyglutamic acid residues near the amino terminus of the mature protein (Stenflo et al., 1974;Nelsestuen et al., 1974). Prothrombin undergoes two metal-dependent conformational transitions that lead to the expression of a biologically active conformer that binds to membrane surfaces (Nelsestuen, 1976;Prendergast and Mann, 1977;Borowski et al., 1986). These transitions have an absolute requirement for most, if not all, of the y-carboxyglutamic acid residues and for calcium ions. In the presence of calcium ions, a complex of Factor Xa, Factor Va, and membrane surfaces converts prothrombin to thrombin (Mann et al., 1982(Mann et al., , 1988. Vitamin K-dependent carboxylation is a post-translational process that takes place in the endoplasmic reticulum (Suttie, 1985;Furie and Furie, 1988;Carlisle and Suttie, 1980). The vitamin K-dependent carboxylase is a membrane protein that has recently been purified lO,OOO-fold (Hubbard et al., 1989b). This enzyme binds to a recognition element located on the propeptide of the precursor form of prothrombin, proprothrombin, and precursors of the other vitamin K-dependent proteins (Jorgensen, 1987a;Foster et al., 1987;Ulrich et al., 1988;Hubbard et al., 1989a). This recognition element, termed the y-carboxylation recognition site (Jorgensen et al., 1987a), identifies the vitamin K-dependent proteins for post-translational carboxylation of specific glutamic acid residues to generate y-carboxyglutamic acid residues. This reaction requires reduced vitamin K, molecular oxygen, carbon dioxide, the glutamyl-containing polypeptide substrate, and the carboxylase (Suttie, 1985).
Some of the residues that define the y-carboxylation recognition site have been identified. From studies of the expression of Factor IX mutants prepared by site-specific mutagenesis, phenylalanine -16 and alanine -10 are critical residues in the 18-residue propeptide (Jorgensen et al., 1987a). Synthetic peptide substrates used in in vitro carboxylation assays have confirmed the special role of phenylalanine -16 in peptides based upon the structure of proprothrombin (Ulrich et al., 1988). Another synthetic peptide based upon residues -10 to +lO of proprothrombin is poorly carboxylated in this assay, suggesting the important role of the propeptide NH,terminal region in carboxylase recognition. Although naturally occurring Factor IX propeptide mutants, Factor IX Cambridge (Arg-' + Ser) and Factor IX San Dimas (Arg+ --, Gln), are partially carboxylated (Diuguid et al., 1986;Ware et al., 1989), the carboxylation in vitro of synthetic peptides with primary structures based upon these mutants suggests that their carboxylation recognition sites are intact and that residues -1 and -4 are not part of the carboxylation recognition site (Hubbard et al., 1989a).
The extent of the carboxylation recognition site is unknown. Although the carboxylation of synthetic peptides re-quires an intact propeptide (Ulrich et al., 1988), Price et al. (1987) have noted marked homology among the structures of the Gla' regions of the vitamin K-dependent proteins. On this basis, Price has suggested that the carboxylation recognition site might extend into the Gla domain to include Glu-X-X-X-Glu-X-Cys (residues +16 to +22 in prothrombin) and may not be restricted to the propeptide as we have hypothesized (Jorgensen et al., 1987a).
To define the size and structure of the recognition site for the carboxylase, we have constructed a series of point mutations in the region of the prothrombin cDNA that encodes the propeptide.
By using an expression system that yields fully carboxylated wild-type recombinant prothrombin (Jorgensen et al., 1987b), mutations of the amino-terminal propeptide amino acids were analyzed for the effect of mutation on the extent of carboxylation of the secreted protein. We conclude that His-", Val-I", Phe-16, Leu-15, and Ala-"' compose part of the carboxylation recognition site. In addition, kinetic evidence comparing the binding of synthetic peptide substrates to the carboxylase suggests that the recognition site is contained solely within the propeptide.
We conclude that specific amino acids within the propeptide define a molecular surface, the carboxylation recognition site, that interacts directly with the vitamin K-dependent carboxylase.

EXPERIMENTAL PROCEDURES
Mutagenesis and Expression of the Mutated cDNAs-The preparation of the 2-kilobase EcoRI fragment containing the human prothrombin coding sequence has been previously described (Jorgensen et al., 1987b). This fragment was inserted into the phage M13mp18, and the prothrombin cDNA was used as a template for site-directed mutagenesis (Oostra et al., 1983). Oligonucleotides used for the seven different. mutations ( Fig. 1) were synthesized on an Applied Biosysterns 381A DNA synthesizer and purified prior to use. Phage containing the desired mutation were identified by colony hybridization using "P-labeled mutagenic oligonucleotides. The mutants were finally characterized by DNA sequencing using the dideoxy chain termination method (Sanger et al., 1977). In each case the sequence of a 0.3kilobase Hind111 fragment containing the mutation was verified in its entirety and then inserted into the Hind111 site of a plasmid PUC 12 containing a wild-type prothrombin cDNA lacking this fragment. The mutated prothrombin cDNA was excised from PUC 12 using EcoRI and inserted into the EcoRI site of an expression vector, pMT2 (Kaufman et al., 1986;Jorgensen et al., 1987b). Cell Culture, DNA Transfection, and Cell Selection-A Chinese hamster ovary cell line CHODUKX-Bll (Chasin and Urlaub, 1980) was previously cloned by limiting dilution and clone 4 selected on the basis of its ability to fully carboxylate prothrombin under the expression conditions employed (Jorgensen et al., 1987b). Clone 4 was grown and transfected as previously described with few modifications (Jorgensen et al., 1987b).
Cells (3 X 106) were suspended in 0.7 ml of maintenance medium and placed in an electroporation cuvette. DNA (10 fig in 100 ~1 of 10 mM Tris, 1 mM EDTA) was added, and the mixture was incubated at 0 "C for 5 min; then, a brief pulse of 960 microfarads at 350 V was applied to the cuvette with a Bio-Rad Gene Pulser.
The transfected cells were incubated for 5 min at 0 "C and then plated in a lo-cm* culture dish containing 10 ml of maintenance medium.
Forty-eight h later, the cells were placed in selective medium lacking nucleosides. When colonies became visible (15-30 colonies/ dish) the colonies were pooled and the cells grown until confluent. The medium was replaced, and the cells were incubated an additional 48 h prior to supernatant collection. The subcloning of the initial transformants was performed by plating at low cell density.
Culture dishes (10 cm*) were seeded at low density (approximately 10 cells/dish). When the colonies became visible, they were isolated independently with the aid of cloning rings and reestablished in cell culture. Protein Reagents-Human prothrombin was purchased from Enzyme Research Laboratories.
The purification of rabbit anti-prothrombin:total or anti-prothrombin:Ca(II)-specific antibodies was 1 The abbreviations used are: Gla, y-carboxyglutamic acid; HPLC, high pressure liquid chromatography.
Following each coupling step all uncoupled a-amino termini were acetylated.
Sequential deprotection of the a-amino termini prior to coupling was achieved with 50% trifluoroacetic acid in methylene chloride.
Prior to HF cleavage the 2,4-dinitrophenyl protecting group on histidine was removed from the resin-bound peptide using 20% a-mercaptoethanol.
The resin was then washed extensively with dimethylformamide followed by dichloromethane. The cleavage of the benzyl-protected peptide from the resin and simultaneous removal of the side chain protecting groups were performed using anhydrous hydrogen fluoride.
Extraction of cleaved deprotected peptide from resin was performed using 50 ml of 30% acetic acid in water after a rinse with 50 ml of anhydrous ethyl ether. in each reaction was determined following the procedure of Larson et al. (1981).
Mutations were prepared in M13mp18 phage by site-directed mutagenesis using mutagenic oligonucleotides (Fig. 1) and the method of Oostra (1983). For each mutant the sequence of a fragment from the Hind111 site in the polylinker of M13mp18 to the Hind111 site in proprothrombin cDNA containing the mutation was verified by dideoxy sequencing and inserted into the wild-type prothrombin cDNA lacking this fragment in the PUC vector. The resulting mutated DNA was ligated into pMT2, a mammalian expression plasmid used in previous studies (Jorgensen et al., 1987a, 198713) that is closely related to p91023 (Wong et al., 1985;Kaufman et al., 1986).
Expression of Mutated and Wild-type Prothrombin cDNAs in Chinese Hamster Ovary Cells-The wild-type and mutated prothrombin cDNAs were inserted into the EcoRI cloning site of pMT2. Introduction of the plasmids into dihydrofolate reductase-negative Chinese hamster ovary cells leads to the expression of a polycistronic message encoding both prothrombin and dihydrofolate reductase. The transfected cells were selected for their dihydrofolate reductase-positive phenotypes. Once the colonies were visible, the cells were dispersed and the polyclonal populations were grown until confluent. The conditioned media were harvested after 2 days at confluence and assayed by competition radioimmunoassay for total prothrombin antigen. 2.2 pg/ml (about 0.035-1.1 pg/lO' cells/24 h).
Post-translational Cleavage of Mutant Prothrombins-To assure that normal post-translational propeptide cleavage had occurred during the synthesis of the mutant prothrombins, the molecular size of each prothrombin form was determined and compared with plasma-derived prothrombin and wildtype recombinant prothrombin. The proteins synthesized by Chinese hamster ovary cells expressing each of the different prothrombin species were metabolically labeled with ["S] cysteine, leading to the production of 'S-labeled prothrombin in the tissue culture supernatant.
The mutant and wild-type labeled prothrombins were immunoprecipitated with antiprothrombin:total antibodies complexed to protein A-Sepharose. After elution from the matrix the proteins were analyzed by sodium dodecyl sulfate-polyacryamide gel electrophoresis. As shown in Fig. 2, autoradiography demonstrated co-migration of the mutant prothrombins and the wild-type prothrombin. These results indicate that the mutation of these specific residues within the propeptide does not interfere with either propeptide cleavage or signal peptide cleavage. Thus, the lack of immunoreactivity of any prothrombin species to the conformation-specific antibodies would be due to decreased carboxylation and not to the presence of an attached propeptide.

Effects of Propeptide
Mutations on Prothrombin Carboxylation-The extent of carboxylation of prothrombin derived from each mutant cDNA was determined by comparison of the concentrations of the native prothrombin antigen and the total prothrombin antigen. The native prothrombin antigen in the tissue culture supernatant is the fully carboxylated form of prothrombin.
Fully carboxylated prothrombin was quantitated using a competition radioimmunoassay with antiprothrombin:Ca(II)-specific antibodies which bind to fully carboxylated prothrombin in the presence of metal ions. These antibodies recognize prothrombin only when the protein is sufficiently y-carboxylated to undergo the two metalinduced conformational transitions. The concentration of native prothrombin antigen correlates closely with the coagulant activity (Blanchard et al., 1983). The total prothrombin antigen in the culture supernatants was measured by a competitive radioimmunoassay using anti-prothrombin:total antibodies that bind to all forms of prothrombin regardless of the state of carboxylation.
The extent of carboxylation of the different prothrombin species is expressed as the ratio of the native prothrombin antigen concentration to the total prothrombin antigen concentration. Under the conditions used for protein expression, wild-type prothrombin appears fully carboxylated, with a native prothrombin antigen to total prothrombin antigen ratio of 0.89 + 0.15 (Fig. 3). By comparison, plasma-derived prothrombin that is known to be fully carboxylated had a native prothrombin antigen to total pro- These results indicate that the mutation of histidine -18 to glycine, valine -17 to serine, leucine -15 to glycine or aspartic acid, or alanine -10 to aspartic acid partially impairs y-carboxylation. The degree of inhibition is less for these amino acid substitutions than that of the mutation of phenylalanine -16 to alanine (Jorgensen et al., 1987a;Rabiet et al., 1987;Ulrich et al., 1988). These results indicate that the y-carboxylation recognition site is partially disrupted by mutations at residues -18, -17, -15, and -10 and, based upon previous results, is almost completely disrupted by mutation at residue -16.
The observation of partially carboxylated prothrombin forms in the tissue culture supernatants could be due to the presence of some cells that can fully carboxylate prothrombin and other cells that are defective in carboxylation.
To demonstrate that the protein products of all cells transfected with a given plasmid generated prothrombin that was carboxylated to the same extent, four stable transfectants were subcloned. Five to eight single colonies were isolated from cell transfected with pMT2-PT/AD-10, pMT2-PT/LG-15, pMT2-PT/ LD-15, and pMTP-PT. The supernatants were analyzed as described above (Fig. 4). Although the expression level of prothrombin varied in each subclone, the extent of carboxylation was identical, within experimental error, among subclones derived from the same primary transfectant.
Extent of the y-Carboxylation Recognition Site; in Vitro Carboxylation of Peptides Containing Part of or the Entire Gla Domain-To determine the extent to which amino acids within the Gla domain contribute to the recognition of proprothrombin by the vitamin K-dependent carboxylase, we studied the in uitro carboxylation kinetics of synthetic peptide homologs based upon the amino acid sequence of prothrombin. proPT28, previously described and extensively studied (Ulrich et al., 1988;Hubbard et al., 1989), contains residues -18 to +lO of proprothrombin including two glutamic acid residues at positions +6 and +7. proPT54, spanningproprothrombin from residues -18 to +36, includes all of the amino acids encoded by Exon II in the prothrombin gene (  Degen and Davie, 1987) and the 10 glutamic acid residues that are carboxylated (+6, +7, +14, +16, +19, +20, +25, +26, +29, +32) in prothrombin.
The 54-residue peptide, proPT54, was synthesized by solidphase methods and purified by reverse-phase high performance liquid chromatography.
An average coupling efficiency of 99.02% was determined.
The yield of peptide obtained was 10% of the theoretical yield. As shown in Fig. 5, this peptide yielded a single major peak following purification. proPT28 and proPT54 were compared as substrates for the partially purified bovine vitamin K-dependent carboxylase (Hubbard et al., 1989a). As shown in Table I, the K,,, for the carboxylation of proPT54 in this set of experiments was 2 pM, compared with 5 fiM for proPT28 and 10 mM for the pentapeptide substrate FLEEL. The V,,,,, values for proPT28 and proPT54 were the same within experimental error. These results indicate that the affinities of proPT28 and proPT54 for the carboxylase are similar when measured by the indirect yet physiologically relevant kinetic parameter K,,,. Furthermore, a synthetic peptide proPT18, from -18 to -1, is a competitive inhibitor of proPT28, with a K; of 3.5 PM (Ulrich et al., 1988). Thus, lengthening the substrate to include the entire Gla domain does not enhance binding of these peptides to the carboxylase. Under typical carboxylation reaction conditions using 1 pM substrate, it was found that approximately 45% of the glutamic acid residues in proPT28 were carboxylated, and approximately 40% of the glutamic acid residues in proPT54 were carboxylated (Table II). These data reveal that more CO, molecules were incorporated into proPT54 than proPT28 on a mole to mole basis. However, these data alone are insufficient to distinguish the extent of carboxylation at any individual glutamic acid position, e.g. whether 40% of the proPT54 molecules were fully carboxylated or whether all proPT54 molecules were 40% carboxylated.

DISCUSSION
The y-carboxylation recognition site is a recognition sequence within the vitamin K-dependent proteins that directs y-carboxylation during the final stages of the biosynthesis of these proteins. This recognition element in the vitamin Kdependent blood coagulation proteins immediately follows the signal peptide that destines the protein for translocation to the endoplasmic reticulum and precedes the glutamic acidrich region that becomes y-carboxylated.
The carboxylation recognition site binds directly to the carboxylase. Synthetic peptides that contain the carboxylation recognition site competitively inhibit y-carboxylation of large peptide substrates (Ulrich et al., 1988), stimulate the carboxylation of small glutamate-containing substrates (Knobloch and Suttie, 1987), and form a complex with the vitamin K-dependent carboxylase (Hubbard et al., 1989b). This region shows marked sequence homology among the vitamin K-dependent proteins (Pan and Price, 1985). Previously, we demonstrated that phenylalanine -16 and alanine -10, both highly conserved amino acids in the vitamin K-dependent proteins, play a critical role in defining the carboxylation recognition site (Jorgensen et al., 1987a;Rabiet et al., 1987;Ulrich et al., 1988). In the current study, we have extended the analysis of the amino acids that define the carboxylation recognition site by exploring the role of other residues in the propeptide of prothrombin.
In contrast to Factor IX (Kaufman et al., 1986;Jorgensen et al., 1987a), prothrombin is fully carboxylated in the expression system that we have employed (Jorgensen et uZ., 1987b). Therefore, we have used prothrombin rather than Factor IX to determine the effect of point mutations in the propeptide on the extent of y-carboxylation.
Our choice of mutations remains somewhat arbitrary. Independent circular dichroism analyses indicate that mutations at residues -16 and at -10 do not disrupt the secondary structure of synthetic peptides containing the propeptide of prothrombin (Huber et al., 1989). The extent of y-carboxylation was estimated using an immunochemical approach. We have previously used conformation-specific antibodies to monitor the extent of carboxylation of various forms of prothrombin (Tai et al., 1980;Blanchard et al., 1981;Blanchard et al., 1983). Since the conformational transition that is monitored is dependent upon full or nearly full carboxylation of prothrombin, the expression of native prothrombin antigen correlates closely with the amount of fully carboxylated prothrombin. Since the same observations have been made for Factor IX (Liebman et al., 1985;Liebman et al., 1987), we previously employed conformation-specific antibodies to estimate the amount of fully carboxylated Factor IX in the tissue culture supernatants of cells expressing Factor IX from plasmids containing Factor IX cDNA with propeptide-based mutations (Jorgensen et al., 1987a). When these Factor IX species were subsequently purified and subjected to direct y-carboxyglutamic acid analysis, we confirmed excellent correlation between the y-carboxyglutamic acid content of each species and its immunoreactivity with the conformation-specific antibodies (Rabiet et al., 1987). The immunochemical analysis obviates the need for large scale expression, protein purification, alkaline hydrolysis, and chemical analysis. Thus, in the current work we were able to examine 6 residues in the prothrombin propeptide by site-specific mutagenesis. Taken with earlier data, our results indicate that histidine -18, valine -17, phenylalanine -16, leucine -15, and alanine -10 are components of the carboxylation recognition site (Fig. 6). The substitution of alanine -10 by aspartic acid decreased carboxylation of prothrombin by about 50%. The substitution of alanine by glutamic acid decreased immunoreactivity in Factor IX by about 80% (Jorgensen et al., 1987a). Although both results clearly demonstrate a role for alanine -10 in carboxylase recognition, it remains uncertain whether these differences reflect variations between Factor IX and prothrombin propeptides or whether the substitution of an aspartic acid or glutamic acid yields different results.
Alanine -14 and serine -8 are not part of the site. These amino acids were mutated to substantially alter the character of the amino acid side chain. Despite conversion of the hydrophobic alanine to a hydrophilic serine and a hydrophilic serine to a hydrophobic valine, no effect on the extent of y-carboxylation was noted. Neither of these residues is well conserved in the propeptides of the vitamin K-dependent proteins (Pan and Price, 1985).
We have previously hypothesized that the y-carboxylation recognition site is restricted to the propeptide region of the vitamin K-dependent blood coagulation proteins (Jorgensen et al., 1987a). Price et al. (1987) identified a concensus sequence within the vitamin K-dependent proteins, Glu-X-X-X-Glu-X-Cys (located at residues 16-22 in human prothrombin), and suggested that the carboxylation recognition site might include this region of the Gla domain in addition to sites located within the propeptide. Pottorf and co-workers (1987) prepared a synthetic 17-residue substrate correspond-ing to residues 13-29 of bovine prothrombin and found that it was not an effective substrate for in vitro carboxylation (Pottorf et al., 1987). Our in vitro carboxylation studies utilized synthetic peptides that contain the intact propeptide of human prothrombin and varying amounts of the 36-residue Gla domain of prothrombin. proPT54 includes and proPT28 does not include the consensus sequence at positions +16 to +22. That proPT28 and proPT54 exhibit similar X, and V,., values in in vitro carboxylation leads us to conclude that the carboxylase binds similarly to these substrates and that peptide sequences +ll to +36 within the Gla domain do not contribute to substrate recognition. Additional findings that the prothrombin propeptide, proPT18 (residues -18 to -l), is a competitive inhibitor of the carboxylase with a Ki of 3.5 PM (Ulrich et al., 1988) and that placement of the propeptide sequence in front of the first 14 residues of tissue plasminogen activator results in carboxylation of the glutamyl residue at position +12 (a residue not carboxylated in proTPA32*) further support the argument that the carboxylation recognition site may be the soie determinant for carboxylation of accessible and adjacent glutamic acid residues. Our studies, including successful synthesis, purification, and efficient in vitro carboxylation of proPT54, will allow us to address questions of carboxylation directionality and processivity by determining the kinetic course of 14C02 fixation at each of the glutamyl residues. Huber et al. (1989) have demonstrated that synthetic peptides that contain the carboxylation recognition site can be induced to assume a-helical character under conditions that are known to lead to secondary structure in short peptides that are helical within the context of the native protein of which they are a component (Kaiser and Kezdy, 1987). This a-helical structure suggests that the carboxylation recognition site is located within or adjacent to this helix. The examination of additional amino acids in the propeptide for their possible role in y-carboxylation and the determination of the three-dimensional structure of the propeptide by two-dimensional NMR, currently in progress, should facilitate understanding of the structural correlates required for carboxylase recognition.