Profactor IX propeptide and glutamate substrate binding sites on the vitamin K-dependent carboxylase identified by site-directed mutagenesis.

The vitamin K-dependent carboxylase, a constituent of the endoplasmic reticulum membrane, catalyzes the conversion of reduced vitamin K to vitamin K epoxide and the concomitant conversion of glutamic acid to γ-carboxyglutamic acid. To study structure-function relationships in the enzyme, seventeen clusters of charged residues of the bovine γ-glutamyl carboxylase were substituted with alanines using site-specific mutagenesis. Wild-type and mutant carboxylase species were expressed in Chinese hamster ovary cells with an immunodetectable octapeptide inserted at their amino-terminal ends. Out of 17 mutant carboxylase species that contain a total of 41 charged residue to alanine substitutions, K217A/K218A (CBX217/218), R234A/H235A (CBX234/235), R359A/H360A/K361A (CBX359/360/361), R406A/H408A (CBX406/408), and R513A/K515A (CBX513/515) had impaired carboxylase activity compared with the wild-type enzyme. The vitamin K epoxidase activities of these mutants were reduced in parallel with the carboxylase activities. CBX217/218 appears to be inactive. High propeptide concentrations were required for stimulation of carboxylation of FLEEL by CBX234/235, CBX406/408, and CBX513/515, suggesting defects in the propeptide binding site. CBX359/360/361 showed normal affinity for the propeptide, FLEEL, proPT28, and vitamin K hydroquinone but exhibited a low catalytic rate for carboxylation. These results suggest that residue 217, residue 218, or both are either critical for catalysis or for maintaining the structure of a catalytically active enzyme. Regions around residues 234, 406, and 513 define in part the propeptide binding site, while the regions around residue 359 are involved in catalysis.

Vitamin K-dependent ␥-glutamyl carboxylase is a membrane-associated endoplasmic reticulum resident enzyme that catalyzes the posttranslational conversion of specific glutamic acids of the vitamin K-dependent proteins to Gla. The vitamin K-dependent blood-clotting and regulatory proteins have clusters of 9 -12 Gla residues near their N termini. A Ca 2ϩ -stabilized three-dimensional structure of the Gla domain is essential for binding of the protein to phospholipid vesicles or cell membranes (1).
Carboxylase activity has been found in such diverse tissues as liver, testis, skin, lung, and kidney. The Gla-containing proteins are found not only in blood plasma but also in calcified tissue, spermatozoa, urine, and lung surfactant (2). Protein S and a homologous protein, Gas6 (3), have been identified as ligands for Tyro 3 and Axl, respectively, members of a family of receptor tyrosine kinases (4,5) although the role of protein S as a receptor tyrosine kinase ligand has been questioned (6). Cellular responses to activation of these receptors by protein S (4) and Gas6 (5) requires vitamin K, suggesting a requirement of ␥-carboxylation of the ligands. Thus, the vitamin K-dependent carboxylase appears to play important roles in many physiologic processes in addition to blood clotting.
The carboxylase, a bifunctional enzyme, catalyzes the conversion of glutamic acid and vitamin K hydroquinone (vitamin KH 2 ) 1 to Gla and vitamin K epoxide in the presence of CO 2 and O 2 . Studies using the crude enzyme showed that the epoxidation and the carboxylation activities are stoichiometrically coupled when CO 2 is saturating (7). Characterization of purified bovine liver carboxylase has confirmed that these reactions are carried out by the same protein (8). According to the basicity enhancement model proposed recently (9), the driving force for the carboxylation of glutamic acid derives from enzymatic oxygenation of vitamin KH 2 .
The propeptide, linked at the N-terminal end of the Gla domain of the precursor forms of the vitamin K-dependent blood proteins, directs carboxylation by mediating the interaction of the substrate protein and the carboxylase. The carboxylation recognition site (␥-CRS) located near the amino terminus of the propeptide functions as a docking element for the carboxylase (10 -12), while the free propeptide functions as a stimulatory element for carboxylation of small Glu-containing peptides (13,14). The bovine carboxylase has been purified (15,16) utilizing the affinity of carboxylase for the ␥-CRS of the propeptide of vitamin K-dependent proteins, and the fulllength human cDNA (17) and bovine cDNA (18) have been cloned. The hydropathy analysis of the primary amino acid sequence predicted that the enzyme has a short N-terminal hydrophilic region (residues 1-50), a hydrophobic region (residues 51-314), and a long C-terminal hydrophilic region (residues 315-758). The hydrophobic region contains three to five putative transmembrane domains. The C-terminal hydrophilic region includes a sequence with homology to soybean seed lipoxygenase. Seven (bovine) or eight (human) potential Nlinked glycosylation sites are located in this hydrophilic region (16). We have recently localized the binding site of the carboxylase for the factor IX propeptide (19) and a small carboxylatable peptide substrate (20) to the hydrophobic region of the enzyme using affinity-labeling reagents. We have also shown that epoxidase activity is affected by truncation of the C terminus of the enzyme, a modification that does not alter binding of propeptide or a carboxylatable peptide substrate (21).
To further analyze structure-function relationships within the bovine carboxylase, we adopted a modification of alaninescanning mutagenesis. We focused on 10 hydrophilic regions of the enzyme; within these regions, 17 clusters of charged residues were mutated to alanines, and the carboxylase-specific activity of the mutants was assessed. While twelve mutant proteins had normal carboxylase activity, five mutant enzymes with significantly decreased specific activity were identified and characterized.

EXPERIMENTAL PROCEDURES
Materials-Vent DNA polymerase and restriction endonucleases were purchased from New England Biolabs. pGEM-7Zf(ϩ) was from Promega Corp. pED and CHO-Dukx-B11 were gifts from Genetics Institute. Lipofectin Reagent was purchased from Life Technologies, Inc. ECL-Western blotting detection reagents and Hyperfilm-ECL were purchased from Amersham Corp. The anti-FLAG M2 monoclonal antibody and FLAG-bacterial alkaline phosphatase were obtained from Eastman Kodak Co. Peroxidase-conjugated goat anti-rabbit immunoglobulin and peroxidase-conjugated goat anti-mouse immunoglobulin were obtained from DAKO. Vitamin K1 (10 mg/ml) obtained from Abbott was chemically reduced with 8 mg of NaBH 4 . FLEEL and L-␣phosphatidylcholine (type V-E) were purchased from Sigma. NaH[ 14 C]O 3 was from Amersham. ProPT28 and proFIX18 were synthesized using Fmoc (N-(9-fluorenyl)methoxycarbonyl)/N-methylpyrrolidone) chemistry on an Applied Biosystems model 430A peptide synthesizer. The crude deprotected peptide was purified by HPLC. Partially purified bovine liver carboxylase was prepared as described (12).
Preparation of the Expression Vector for Wild-type Carboxylase-The cDNA of bovine ␥-glutamyl carboxylase was excised from pBLII/bCbx (22) and inserted into a cloning vector pGEM-7Zf(ϩ) (pG7/CBX). The sequence for FLAG epitope (DYKDDDDK) was introduced between the codon for the initiator Met of the carboxylase and the second codon for Ala employing the megaprimer method of polymerase chain reaction (PCR) mutagenesis (23). A 60-base pair forward primer (5Ј-GAC GTC GCA TGC GTC GAC ATG gac tac aag gac gat gac gat aag GCG GTC TCC GCT CGG-3Ј) was synthesized on an Applied Biosystems 381A DNA synthesizer. The FLAG sequence, shown in lowercase type, was located 5Ј to a 15-nucleotide sequence complementary to nucleotides 18 -33 of the carboxylase cDNA. An oligonucleotide, 5Ј-AGC ATG ACG TAG GGG-3Ј, which is complementary to nucleotides 911-926, was used as a reverse primer. The PCR reaction was performed using 1 unit of Vent DNA polymerase (2 unit/l) and the buffer provided by the manufacturer in the presence of 200 M dNTP, 50 pmol of each primer, 0.5 g of plasmid DNA in a final volume of 100 l for 30 cycles (94°C for 1 min, 56°C for 1 min, 72°C for 2 min) and an additional extension step (72°C for 7 min). The PCR product was inserted into pG7/CBX using SphI and EcoRI (pG7/FLAG-CBX). Using SalI and SmaI sites, the carboxylase cDNA excised from pG7/CBX and the FLAG-carboxylase cDNA from pG7/FLAG-CBX were inserted into an expression vector, pED (24), producing pED/CBX and pED/FLAG-CBX, respectively. The sequences within the regions amplified by PCR were confirmed by DNA sequence analysis.
Alanine-scanning Mutagenesis-Each of the charged amino acid residues within 17 clusters of charged residues was replaced by alanine. Site-directed mutagenesis by overlap extension using PCR (25) was employed to introduce alanine substitutions into the pED/FLAG-CBX. The nucleotide sequences within the regions amplified by PCR were confirmed by DNA sequence analysis.
Cell Culture, Transfection, and Cell Line Selection-Dihydrofolate reductase-deficient CHO cells, CHO-Dukx-B11 (26) were employed to express the wild-type carboxylase and mutant carboxylase species. The cells were transfected with the expression plasmids, pED/CBX, pED/ FLAG-CBX, and pED/mutant FLAG-CBXs by the Lipofectin method as described by the manufacturer.
Cells from selected colonies were grown to confluence in 12-well plates, and the level of expression of recombinant carboxylase was assayed by Western blotting. Cells were harvested, washed with 1 ml of PBS (2.7 mM KCl, 1.5 mM KH 2 PO 4 , 137 mM NaCl, and 6.5 mM Na 2 HPO 4 ), resuspended in 50 l of PBS with protease inhibitor complex (PIC; 2 mM dithiothreitol, 2 mM EDTA, 0.5 g/ml leupeptin, 1 g/ml pepstatin A, 2 g/ml aprotinin), and subjected to Western blotting. Three independent clones that expressed recombinant carboxylase were selected for further analysis.
Preparation of Cell Lysate-Confluent cells in two 140 ϫ 20-mm plates were suspended with PBS, 5 mM EDTA, washed once with 10 ml of PBS, and resuspended in 400 l of PBS, 20% glycerol, PIC. Cell number was determined, and the cell suspension was mixed with 400 l of PBS, 1% CHAPS, 0.2% phosphatidylcholine, 20% glycerol, PIC and sonicated on ice for 5 s twice with an Ultrasonic processor W-220 fitted with a microprobe (Heatsystems-Ultrasonics, Inc.). After centrifugation (16,000 ϫ g, 10 min, 4°C), the solubilized cell supernatant was stored at Ϫ80°C.
Western Blotting-Monoclonal anti-FLAG M2 antibody was used to detect wild-type FLAG-CBX and mutant FLAG-CBX species in cell lysate. A monospecific rabbit polyclonal antipeptide antibody directed against residues 86 -99 of bovine carboxylase, anti-CBX-(86 -99) (22), was employed for detecting wild-type CBX in cell lysate and bovine liver carboxylase in solubilized bovine liver microsome. Western blots were prepared by electroblotting proteins separated in a 10% SDS-polyacrylamide gel onto a polyvinylidene difluoride membrane. After blocking with 5% nonfat dry milk in PBS with 0.05% Tween 20 (PBS/T), membranes were incubated with anti-FLAG M2 monoclonal antibody (1.5 g/ml) or anti-CBX-(86 -99) antibody (1 g/ml) in PBS/T at 4°C overnight. Membranes were then incubated with peroxidase-conjugated goat anti-mouse immunoglobulin (0.2 g/ml) or peroxidase-conjugated goat anti-rabbit immunoglobulin (0.05 g/ml) in PBS/T at room temperature for 1 h. Bound antibodies were detected with the ECL detection kit (Amersham). The light emission produced was detected by autoradiography on Hyperfilm ECL (Amersham).
Quantitation of Wild-type FLAG-CBX, Mutant FLAG-CBX Species, Wild-type CBX, and Liver Microsomal Carboxylase-Quantitation was performed from Western blots prepared as described above. The concentration of wild-type FLAG-CBX in a CHO cell lysate was determined from Western blots using amino-terminal FLAG-bacterial alkaline phosphatase of known concentration as a primary standard. This preparation of FLAG-CBX was used as the secondary standard for quantitation of wild-type or mutant FLAG-carboxylase in other cell lysates. Six different aliquots of standard FLAG-CBX lysate containing from 4 to 31.5 ng were used to generate a standard curve. The developing antibody was either the anti-FLAG M2 antibody or anti-CBX-(86 -99) antibody. Autoradiographs were analyzed with an UltraScan XL laser densitometer (Pharmacia Biotech Inc.).
Assay of Carboxylase Activity-The amount of [ 14 C]O 2 incorporated into exogenous peptide substrates (3.6 mM FLEEL or 5 M proPT28) over 30 min by the recombinant enzyme contained in cell lysate or by partially purified bovine carboxylase was measured in reaction mixtures of 125 l containing 25 mM MOPS (pH 7.0), 0.5 mM NaCl, 0.16% CHAPS, 0.16% phosphatidylcholine, 8 mM dithiothreitol, 888 M chemically reduced vitamin KH 2 , and 1.4 mM NaH[ 14 C]O 3 (10 Ci; Amersham). When FLEEL was used as a substrate, ammonium sulfate (0.8 M) and proFIX18 (16 M) were included in the assay. All the assay components except carboxylase were prepared as a master mixture. The reaction was initiated by adding the master mixture to a cell lysate or liver microsomal carboxylase. Incorporated [ 14 C]O 2 was assayed as described previously (12).
Kinetic Studies-In order to determine kinetic constants for three substrates, FLEEL, proPT28, and vitamin KH 2 , the initial rate of [ 14 C]O 2 incorporation was determined at six or more different concentrations of substrate up to 2-fold K m . Since FLEEL and vitamin KH 2 showed apparent substrate inhibition as previously reported (8), concentrations of FLEEL and vitamin KH 2 were used below the inhibitory concentrations. Fixed concentrations of FLEEL, proPT28 of 3.6 mM and 5 M, respectively, were used when kinetic parameters for vitamin KH 2 were determined. Vitamin KH 2 was generally used at 222 M when determining kinetic parameters for FLEEL and proPT28. However 56 M vitamin KH 2 was used in assays for CBX513/515. Kinetic constants were determined by nonlinear regression analysis using the Michaelis-Menten equation (Deltagraph PRO 3, DeltaPoint).
Assay of Vitamin K Epoxidase Activity-Carboxylase assays were performed in a 125-l mixture as described above except that NaH[ 14 C]O 3 was replaced with the same concentration of NaHCO 3 (final concentration, 1.4 mM). Vitamin K epoxide formation was deter-mined as described previously (21). Briefly, upon completion of the 30-min incubation at 25°C in sealed tubes, the reaction mixture was extracted with 250 l of ethanol and then 750 l of hexane. The organic and aqueous phases were separated by centrifugation at 1000 ϫ g for 10 min. The organic phase was removed, and the solvent evaporated to dryness. The residue was redissolved in 200 l of methanol. Half of this solution was injected onto a reverse-phase C18 HPLC column (HYPER-SIL ODS, 5 m, 4.6 ϫ 250 mm, Custom LC, Inc.). The column was developed with a mobile phase of 10% dichloromethane, 90% methanol, which had been saturated with nitrogen. The flow rate was 1 ml/min. Vitamin K derivatives were detected at 226 nm, and vitamin K epoxide was quantitated using a purified standard.

Expression of Wild-type CBX and Wild-type FLAG-CBX in
CHO Cells-Wild-type CBX and wild-type FLAG-CBX were expressed in CHO cells. The proteins in cell lysates from either transfected or untransfected cells were separated by SDS-polyacrylamide gel electrophoresis and analyzed by Western blotting using the anti-CBX-(86 -99) polyclonal antibody and the anti-FLAG M2 monoclonal antibody. Anti-FLAG M2 antibody identified only wild-type FLAG-CBX, which migrated as a single band of 99 kDa, and did not cross-react with endogenous CHO cell carboxylase, wild-type CBX, or bovine liver carboxylase (Fig. 1A). Anti-CBX-(86 -99) antibody identified wild-type FLAG-CBX, wild-type CBX, and bovine liver carboxylase but did not detect endogenous CHO cell carboxylase (Fig. 1B). Assuming the same specific activity for hamster and bovine carboxylase, about 0.6 ng of hamster carboxylase was loaded on the gel. This is within the detectable range for bovine carboxylase in our Western blotting system, suggesting that hamster carboxylase at these concentrations does not cross-react with anti-bovine CBX-(86 -99). Thus, wild-type bovine carboxylase and wild-type bovine FLAG-CBX could be quantitated in the cell lysate by Western analysis and densitometry without interference from the endogenous CHO cell carboxylase. However, endogenous CHO cell carboxylase is detectable in assays of enzyme activity (Table I). Thus, in kinetic analyses the endogenous CHO cell carboxylase activity was subtracted from the activity of transfected CHO cells prior to evaluation of kinetic constants.
Recombinant carboxylase in cell lysate that includes 0.5% CHAPS, 0.1% PC, 20% glycerol, and PIC was stable to multiple freeze-thaw cycles. The kinetic constants for carboxylation of FLEEL by wild-type FLAG-CBX were compared with those of the bovine liver carboxylase. Bovine liver carboxylase and wildtype FLAG-CBX exhibited K m values for FLEEL of 1.0 Ϯ 0.5 mM and 1.5 Ϯ 1.0 mM, respectively. The k cat values for the bovine carboxylase and wild-type FLAG-CBX were 0.7 s Ϫ1 and 0.6 s Ϫ1 . The stability of the enzyme in the cell lysate and the similarity of the kinetic parameters for bovine carboxylase and wild-type FLAG-CBX indicate that expression of mutant carboxylases in CHO cells provides an adequate system for evaluating structure-function relationships.
Expression and Screening of Carboxylase Mutants-Because of their presumed exposure to solvent, 10 hydrophilic regions in the bovine carboxylase were selected (Fig. 2) for evaluation by site-directed mutagenesis. Seventeen clusters of charged residues were mutated to alanines by overlap extension PCR (Table II). The mutant carboxylase species were expressed in CHO cells, and the specific activities of the mutant enzymes were determined in cell lysates.
The concentrations of the carboxylase species were established from Western blots of SDS-gels run under reducing conditions. Expression levels varied widely among the mutants and among clonal cell lines expressing the same mutant. Levels of expression did not appear to correlate with preservation or loss of function. Some mutant carboxylase species with specific activities comparable with wild-type enzyme expressed poorly on average, while some mutants with defective specific activity expressed at high levels on average. A single protein was identified in cell lysates for each species. As shown in Fig. 3, mutants CBX217/218 and CBX234/235 had a slightly faster mobility than wild-type FLAG-CBX. These differences were confirmed by co-analysis of mutant carboxylase cell lysate mixed with FLAG-CBX cell lysate. Mutants CBX217/218 and CBX234/235 yielded two bands, while CBX359/360/361, CBX406/408, and CBX513/515 migrated with wild-type FLAG-CBX. These differences may be due to alterations in glycosylation.
The specific carboxylase activities of 17 mutants are shown in Fig. 4 and compared with the specific activities of bovine liver carboxylase, wild-type CBX, and wild-type FLAG-CBX. The specific activities of bovine liver carboxylase, wild-type CBX, and wild-type FLAG-CBX were similar to that observed for purified bovine liver carboxylase (15). Twelve mutant carboxylase species, which together include 30 mutations of charged residues to alanines and included all mutations C-terminal of residue 566, had specific activities indistinguishable from that of wild-type carboxylase. Five mutant carboxylase species, CBX217/218, CBX234/235, CBX359/360/ 361, CBX406/408, and CBX513/515, exhibited significantly decreased specific activity. These five mutant carboxylase species were further characterized.
The expression levels of the wild-type FLAG-CBX and the five mutant carboxylase species with decreased specific activity ranged from 0.5 fmol to 17 fmol/10 6 cells ( Table I). Expression of wild-type FLAG-CBX in CHO cells at 11.5 fmol/10 6 cells led to a 140-fold increase in carboxylase activity over untransfected CHO cells. The mutants are expressed at levels that would permit detection of carboxylase activity above the level of endogenous CHO cell carboxylase activity if the mutant species had the same specific activity as the wild-type

FLAG-CBX.
Comparison of Epoxidase and Carboxylase Activity of the Five Mutant Carboxylase Species-Specific epoxidase activity and specific carboxylase activity were determined for each of the five mutant carboxylase species. Since the endogenous CO 2 concentration in the reaction mixtures is approximately equal to the exogenously added CO 2 (27)(28)(29), the true specific carboxylase activity is approximately 2-fold that recorded in Table III. Thus, the ratio of carboxylation to epoxidation of wild-type enzyme is about 1. The ratios of the carboxylase mutants CBX234/235, CBX359/360/361, and CBX406/408 were similar to that of wild-type FLAG-CBX. Mutants CBX217/218 and CBX513/516 have slightly decreased ratios of carboxylation to epoxidation, although the low specific activities of CBX217/218 make that distinction more difficult to discern.
Stimulation of Carboxylase Activity by the Propeptide of Factor IX-The propeptide of the vitamin K-dependent proteins functions as a recognition element for the carboxylase (10, 11) as well as a stimulator of carboxylase activity (13,14). To investigate the stimulatory effect of the propeptide on the mutant carboxylases, we studied their ability to carboxylate FLEEL at varying concentrations of the factor IX propeptide, proFIX18 (Fig. 5A). ProFIX18 had no effect on the activity of CBX217/218. The activation of CBX359/360/361 followed a similar dependence on proFIX18 concentration as that of wild-type FLAG-CBX, with half-maximal stimulation at about 0.06 M. However, the maximal stimulation of CBX359/360/361 was reduced from 5-fold to 2-fold. This suggests that wild-type FLAG-CBX and CBX359/360/361 have similar affinities for proFIX18. In contrast, proFIX18 concentrations yielding halfmaximal stimulation of CBX234/235 and CBX406/408 were 10-fold higher (0.7 and 0.8 M, respectively.) CBX234/235 and CBX406/408 approached maximal stimulation comparable with the wild-type carboxylase, but much higher concentrations of propeptide were required. Although CBX513/515 had    low carboxylase activity, it could be stimulated about 49-fold with 500 M proFIX18. These data suggest that CBX234/235, CBX406/408, and CBX513/515 exhibit decreased affinity for proFIX18 but are sensitive to its stimulatory activity. In contrast, CBX359/360/361 exhibits reduced sensitivity with regard to the amplitude of stimulation. CBX217/218 is insensitive to propeptide stimulation and may be totally inactive.
Effect of Vitamin KH 2 Concentration on Carboxylation-Carboxylation of FLEEL by bovine liver carboxylase is inhibited by high concentrations of vitamin KH 2 (8). It has been shown that this inhibitory effect is not caused by solvent and lipid in the vitamin K preparation (8) and that vitamin K (up to 400 M) and vitamin K epoxide (up to 1 mM) are not inhibitory (30). We thus investigated the effect of vitamin KH 2 concentration on the carboxylase activity of the wild-type FLAG-CBX and the mutant carboxylase species (Fig. 6). Carboxylation of FLEEL by the wild-type FLAG-CBX and CBX359/360/361 was maximal at 222 M vitamin KH 2 and inhibited by higher concentrations of vitamin KH 2 in a dose-dependent manner. This is similar to the inhibition observed for purified bovine carboxylase (8). In contrast, CBX234/235, CBX406/408, and CBX513/ 515 were more susceptible to inhibition by vitamin KH 2 than wild-type enzyme. CBX234/235 and CBX406/408 exhibited maximal activity at 111 M vitamin KH 2 and CBX513/515 at 56 M vitamin KH 2 .
In order to investigate the relationship between propeptide stimulation of carboxylase activity and vitamin KH 2 concentration, we studied proFIX18 stimulation of FLEEL carboxylation by wild-type FLAG-CBX at concentrations of vitamin KH 2 of 222 M, 888 M, and 1776 M (Fig. 5B). Increasing concentrations of vitamin KH 2 blunt the stimulatory effect of the propeptide. Thus, high concentrations of vitamin KH 2 may indirectly inhibit carboxylation by reducing the affinity of the carboxylase for the propeptide. Alternatively, vitamin KH 2 may compete for the propeptide binding site. In either case, the apparent reduced affinities of CBX234/235, CBX406/408, and CBX513/515 for the propeptide may be related to the increased sensitivity of the mutant enzymes to inhibition by vitamin KH 2 .
Kinetic Analysis-Values of K m and k cat for carboxylation of FLEEL by wild-type FLAG-CBX and the mutant enzymes were determined under standard assay conditions in the presence of 16 M proFIX18. The wild-type FLAG-CBX, CBX234/235, CBX359/360/361, and CBX406/408 were studied in the presence of 222 M vitamin K. CBX513/515 was studied in the presence of 56 M vitamin K. Kinetic parameters of CBX217/ 218 could not be determined because of its low activity. The K m values of the wild-type FLAG-CBX and the mutant carboxylase species for FLEEL are between 1.1 Ϯ 0.2 and 1.9 Ϯ 0.8 mM (Table IV). CBX234/235 and CBX406/408 had k cat values similar to those of the wild-type enzyme, but the k cat for CBX359/ 360/361 was 3-4-fold lower and that for CBX513/515 was about 10-fold lower than wild-type enzyme.
Initial analyses were performed at proFIX18 concentrations that approach that required for maximal stimulation for the wild-type enzyme and therefore may not be sensitive to mutations that affect propeptide binding. In order to more accurately determine the effects of the mutations, these analyses were performed at concentrations of proFIX18 of 0, 0.16, 1.6, and 160 M. Only mutants CBX234/235 and CBX406/408 were analyzed, since the activities of the other mutants were too low to be measurable at low propeptide concentrations. As the concentration of propeptide was increased, K m values for FLEEL of the wild-type FLAG-CBX, CBX234/235 and CBX406/ 408 decreased slightly in a dose-dependent manner (Table V). While the maximal decrease in K m was observed at 1.6 M for the wild-type enzyme, CBX234/235 and CBX406/408 required higher concentrations of proFIX18 to realize comparable K m values. In addition, the wild-type FLAG-CBX achieved the maximal increase in k cat at lower concentration of proFIX18 than did CBX234/235 and CBX406/408 (Table V). Final k cat values for the two mutants were approximately equal to the k cat of the wild-type enzyme, implying that the catalytic apparatus of these mutants is intact.
Since the natural substrates of the carboxylase include the recognition site in the propeptide covalently linked to the carboxylatable glutamate residues, we also studied proPT28 as substrate. This peptide includes the propeptide of prothrombin linked to the first 10 residues of prothrombin including two glutamic acid residues. The K m of the wild-type FLAG-CBX for proPT28 is 2.3 Ϯ 0.2 M (Table IV), equivalent to the values of 2.2 and 3.6 M previously reported using bovine liver microsomes (12,31). CBX234/235, CBX406/408, and CBX513/515 all have elevated K m values relative to wild-type enzyme. These three mutants have k cat values 2-3-fold higher compared with that of the wild-type FLAG-CBX. CBX359/360/361 showed a slightly increased K m for proPT28 and a 10-fold decrease in k cat . CBX359/360/361 shows a decreased k cat relative to wildtype enzyme for both FLEEL and proPT28. This may suggest a defect in catalysis or may be related to the failure of this mutant to achieve similar levels of stimulation by the propeptide as those of wild-type carboxylase. CBX513/515 was also poorly stimulated by propeptide; however, the k cat for this mutant with FLEEL is depressed, while the k cat for CBX513/ 515 for proPT28 appears to be greater than that for wild-type enzyme. It has been previously observed (12) that while substrates containing covalently bound propeptides have decreased K m values for carboxylase relative to substrates lacking the propeptide, the V max for propeptide-containing substrates is generally depressed relative to substrates lacking the propeptide when measured under the same conditions. The kinetic behavior of CBX513/515 may be a manifestation of the same phenomenon.
The kinetic constants for vitamin KH 2 were determined for the wild-type FLAG-CBX and the mutants. The K m of the wild-type FLAG-CBX for vitamin KH 2 , determined at 3.6 mM FLEEL and 16 M proFIX18, is 73.7 Ϯ 14.0 M (Table VI). This is within the range of K m values of 10 -100 M previously reported (8,32,33). The four mutants, CBX234/235, CBX359/ 360/361, CBX406/408, and CBX513/515, have similar K m values to that of the wild-type enzyme. While CBX234/235 and CBX406/408 have k cat values similar to the wild-type enzyme, CBX359/360/361 and CBX513/515 have decreased k cat values with FLEEL but comparable with proPT28 as noted below. The low k cat observed for CBX513/515 is likely the result of 16 M proFIX18 being suboptimal for stimulation of FLEEL carboxylation by this enzyme. When 5 M proPT28 is used as a substrate, wild-type FLAG-CBX has a K m for vitamin KH 2 of 4.0 Ϯ 0.4 M, an 18-fold lower value than that observed using FLEEL. It has been reported that the K m for vitamin KH 2 is much lower in the presence of a propeptide-containing substrate than a substrate lacking the carboxylation recognition site (8,33). CBX234/235 and CBX406/408 showed similar decreases in K m for vitamin K in the presence of proPT28. CBX513/515 is not characterized by this pattern perhaps due to its pronounced decrease in affinity for proPT28.

DISCUSSION
In this study, a series of bovine mutant carboxylase species in which clusters of charged residues were replaced with alanines was expressed in CHO cells to study structure-function relationships. The carboxylase has been expressed in active form in mammalian cells (15,18) and insect cells (22). Since it was not practical to express a large number of mutants in insect cells, despite an advantage of the absence of endogenous carboxylase activity (22), we used a mammalian expression system. To distinguish the recombinant carboxylase species from endogenous CHO cell carboxylase, we placed a FLAG epitope at the N-terminal end of the carboxylase. The expression levels of the recombinant carboxylase antigens are sufficiently high relative to the endogenous carboxylase that the endogenous carboxylase antigen and activity could be distinguished in our analysis. In addition, kinetic studies showed that the wild-type FLAG-CBX is functionally indistinguishable from the liver microsomal carboxylase.
Early studies using selective solubilization of rat liver microsomes (34) showed that the carboxylase activity is facing the lumen of the endoplasmic reticulum. The cDNA sequence of the carboxylase predicts that the C-terminal half of the enzyme is exposed to the lumen because this region has 7 putative Nlinked glycosylation sites. In addition, amino acid residues 468 -663 within this region have 19.3% homology to soybean seed lipoxygenase (17). However the binding sites for a small peptide substrate and for the propeptide have been recently localized within the N-terminal third of the enzyme by affinity labeling. Kuliopulos et al. (20) showed that 125 I-labeled bromoacetyl-FLEELY, an irreversible inhibitor of carboxylase as well as a substrate, binds covalently within the first 218 amino acid residues of the enzyme. Yamada et al. (19) showed that proFIX18 in which Phe is replaced by the photoactivable crosslinking group, benzoylphenylalanine, covalently cross-links to the region between residues 184 and 225 in the carboxylase. In the current study CBX217/218 had almost undetectable carboxylase activity. The mutations in this enzyme overlap the region identified as a binding site for both FLEELY and the propeptide. CBX217/218 may have lost the ability to bind propeptide and/or a Glu-containing peptide substrate, although we cannot rule out the possibility that CBX217/218 is not folded properly. The single mutants at residues 217 and 218 will warrant attention. CBX234/235, CBX359/360/361, CBX406/408, and CBX513/515 all maintain enzyme function albeit at reduced levels. By adjustment of the standard assay conditions these mutants can be induced to behave more like the wild-type enzyme. While we cannot ascribe specific functions to the mutated residues and some of these mutations may lead to local conformational changes in the enzyme, it is unlikely that they are globally misfolded. The mutations in CBX234/235 are located close to the predicted binding region for the propeptide. On the other hand, in the primary sequence the mutations in CBX406/408 and CBX513/515 are distant from the predicted propeptide-binding site and the mutations in CBX359/360/361 are distant from the predicted peptide-substrate binding site. It is possible that these three clusters of charged residues may be in close proximity to the predicted binding sites in the three-dimensional structure. Deletion of the C-terminal 82 residues of carboxylase resulted in an enzyme with low affinity for vitamin KH 2 and suggested that the hydrophobic region from residues 684 to 710 presents an important site for vitamin KH 2 binding (21). In contrast, deletion of the hydrophilic C-terminal 47 residues did not affect enzyme activity. Similarly, our mutations in this region, CBX678/679/680 and CBX687/688, did not alter enzyme activity. We did not obtain any mutants that affected vitamin KH 2 -binding, probably because our study was focused on hydrophilic regions of carboxylase.
The vitamin K-dependent carboxylase has two distinct functions, vitamin K epoxidation and ␥-glutamyl carboxylation. These two reactions are stoichiometrically coupled except under conditions where either CO 2 or Glu substrate is not saturating (7,29). With low CO 2 concentration or with low Glu substrate concentration, the ratios of mol vitamin K epoxide formed to CO 2 fixed were reported to increase from 1 to 9.7 (7) and 3.5 (29). However, further studies of mechanisms of epoxidation and carboxylation are difficult due to the existence of endogenous substrates in the case of partially purified microsomal enzyme preparations or propeptide in the case of the enzyme purified on the basis of affinity chromatography using propeptide-containing elution systems. This raises the question of whether a mutation localized within the domains essential for CO 2 fixation might produce an uncoupling of vitamin K epoxidation and carboxylation. We determined the vitamin K epoxidase activity of the five mutants with low carboxylase activities. CBX217/218, CBX234/235, CBX359/360/361, CBX406/408, and CBX513/515 all demonstrated vitamin K epoxidase activities decreased in parallel with ␥-glutamyl carboxylation activities. These results suggest that these mutations affect a site that is important for vitamin K epoxidation as well as ␥-carboxylation.
Kinetic studies of CBX234/235, CBX406/408, and CBX513/ 515 suggest that they bind propeptide with lower affinity than wild-type enzyme. First, these mutant enzymes have slightly lowered affinity for FLEEL relative to wild-type enzyme. This defect can be overcome by increased concentration of free propeptide. In addition, studies exploring proFIX18 stimulation of FLEEL carboxylation show that CBX234/235 and CBX406/408 require higher concentrations of proFIX18 for the stimulation than the wild-type enzyme. Second, they have decreased affinity for proPT28 compared with wild-type enzyme. A significant portion of the binding energy for proPT28 interaction with carboxylase is likely derived from the propeptide region, since the K m for proPT28, 2.2-3.6 M (12, 31), is much lower than that for FLEEL, 0.7 mM (12), and carboxylation of proPT28 is inhibited by the prothrombin propeptide, proPT18, with a K i of 3.5 M (12). Wild-type carboxylase has a K m for a propeptide-containing substrate that is 500-fold lower than the K m for FLEEL; this decrease in K m is associated with a 35-fold decrease in k cat . In contrast, for mutants CBX234/235, CBX406/408, and CBX513/515 K m values for the propeptidecontaining substrate relative to FLEEL decrease by 135-fold, 320-fold, and 103-fold, respectively, while the k cat values for these mutants decrease 10-fold, 17-fold, and 1.3-fold, leaving them all with k cat values higher than that for wild-type enzyme. Finally, it is interesting that epoxidase activities of the mutant enzymes are decreased in parallel with carboxylase activity, perhaps suggesting a requirement for propeptide binding for both epoxidation and carboxylation. The behavior of CBX359/360/361 is distinct from that of CBX234/235, CBX406/408, and CBX513/515. CBX359/360/361 has a similar dependence on proFIX18 stimulation of FLEEL carboxylation as the wild-type enzyme. However, in the presence of saturating concentrations of proFIX18, CBX359/360/ 361 does not achieve the same catalytic rate for carboxylation as that of wild-type enzyme. This is in contrast to CBX234/235 and CBX406/408, which, at saturating concentrations of pro-FIX18, achieve the same catalytic rates for carboxylation as wild-type enzyme. While the K m values of CBX359/360/361 for FLEEL, proPT28, and vitamin KH 2 are the same as those for wild-type enzyme, k cat values for FLEEL and proPT28 carboxylation and vitamin KH 2 epoxidation were significantly lower compared with the wild-type enzyme. Since CBX359/360/361 appears to bind to the propeptide normally, there are two possible explanations for the defect in this mutant. Binding of propeptide may not lead to the alteration in the enzyme, most likely conformational, that leads to stimulation of carboxylation, or residues 359, 360, and/or 361 include important sites for catalysis directly.
In sum the pairwise or triple replacements of adjacent charged residues by alanine mutagenesis indicates lysine 217 or lysine 218 may be key for substrate and/or propeptide recognition or for catalysis. The basic triad arginine 359, histidine 360, lysine 361 in turn may be involved in secondary roles that affect k cat . Single mutants will allow dissection of the contributions of individual residues to the properties of this unusual carboxylation/epoxidation catalyst.