Conantokin-G Precursor and Its Role in γ-Carboxylation by a Vitamin K-dependent Carboxylase from a ConusSnail*

Conantokin-G isolated from the marine snailConus geographus is a 17-amino acid γ-carboxyglutamate (Gla)-containing peptide that inhibits theN-methyl-d-aspartate receptor. We describe the cloning and sequence of conantokin-G cDNA and the possible role of the propeptide sequence. The cDNA encodes a 100amino acid peptide. The N-terminal 80 amino acids constitute the prepro-sequence, and the mature peptide is derived from the remaining C-terminal residues after proteolysis, C-terminal amidation, and a unique post-translational modification, γ-carboxylation of glutamate residues to Gla. Mature conantokin-G peptide containing Glu residues (E.Con-G) in place of Gla is a poor substrate for the vitamin K-dependent γ-glutamyl carboxylase (apparentK m = 3.4 mm). Using peptides corresponding to different segments of the propeptide we investigated a potential role for the propeptide sequences in γ-carboxylation. Propeptide segment −20 to −1 covalently linked to E.Con-G or the synthetic pentapeptide FLEEL increased their apparent affinities 2 orders of magnitude. These substrates are not efficiently carboxylated by the bovine microsomal γ-glutamyl carboxylase, suggesting differences in specificities between the Conus and the mammalian enzyme. However, the role of propeptide in enhancing the efficiency of carboxylation is maintained.

The vitamin K-dependent ␥-carboxylation of glutamate residues was originally discovered as a novel post-translational modification in the blood coagulation cascade (1); some of the key clotting factors such as prothrombin must be ␥-carboxylated in order for proper blood clotting to occur. Somewhat later, this post-translational modification was also found in certain bone proteins (2). This modification was restricted to these rather specialized mammalian systems until a very unusual peptide, conantokin-G, was described from the venom of the predatory marine snail, Conus geographus (3). Conantokin-G is a 17-amino acid peptide that inhibits the N-methyl-D-aspartate receptor (4). Unlike most Conus peptides, which are multiply disulfide-bonded, conantokin-G has no disulfide cross-links but has 5 residues of ␥-carboxyglutamate residues; this remains the highest density of ␥-carboxyglutamate found in any functional gene product characterized to date.
Most of the biologically active components of the Conus venom are multiply disulfide bonded peptides (the conotoxins). These have been shown to be initially translated as prepropeptide precursors, which are then post-translationally processed to yield the mature disulfide-cross-linked conotoxin. Conantokin-G differs strikingly from most conotoxins not only in having ␥-carboxyglutamate residues, but also because it has no disulfide cross-links. We report below an analysis of a cDNA clone encoding the conantokin-G precursor. Furthermore, we establish the probable function of one region of the precursor that is excised during the maturation of the functional conantokin-G peptide.
The presence of ␥-carboxyglutamate in a non-mammalian system was initially controversial because vitamin K-dependent carboxylation of glutamate residues had primarily been thought to be a highly specialized mammalian innovation. However, we have found that conantokin-G is only one member of a family of peptides; a variety of other conantokins have been found including conantokin-T and conantokin-R from two other fish-hunting cone snails (5,6). All three peptides have a high content of ␥-carboxyglutamate (4 -5 residues). ␥-Glutamyl carboxylase has been purified from mammalian sources (7,8) and has been expressed both in mammalian and insect cell lines (9,10). Recently it was shown that, as is the case in the mammalian system, the carboxylation reaction in Conus venom ducts absolutely requires vitamin K, and the net carboxylation increases greatly in the presence of high concentrations of ammonium sulfate. In these respects, the mammalian and the Conus ␥-carboxylation venom systems are very similar (11).
The propeptides of vitamin K-dependent blood coagulation proteins share extensive sequence similarity. This sequence is believed to interact with the carboxylase and constitutes the ␥-carboxylation recognition sequence (␥-CRS). 1 In this report, we analyze the conantokin-G precursor sequence for potential ␥-CRS sequences. The results described below identify a sequence present in the Ϫ1 to Ϫ20 region of the conantokin-G prepropeptide, which when covalently linked to the N-terminal of the substrate stimulates carboxylation by the Conus enzyme.

MATERIALS AND METHODS
Conus radiatus venom ducts were obtained from Dr. L. J. Cruz (University of the Philippines). Vitamin K (phytonadione) was from Abbot Laboratories, and NaH 14 CO 3 55 mCi/mmol from NEN Life Science Products. Bovine microsomes were a gift from Dr. D. W. Stafford (University of North Carolina, Chapel Hill, NC).
Conus microsome preparation from frozen venom ducts of C. radiatus was performed as described by Stanley et al. (11). Carboxylase assay using 1 g of Conus microsomal protein per assay was performed as follows: Conus microsomes were solubilized in 0.7% CHAPS/0. Ci of NaH 14 CO 3 , 6 mM dithiothreitol, 222 M reduced vitamin K (prepared as described by Ref. 12). Substrate and inhibitor concentrations are indicated in the legends to the figures and tables. For experiments with bovine microsomes, 380 g of microsomal protein was present in each reaction. Reaction mixtures were incubated at 25°C for 30 min and were quenched by the addition of 75 l of 1 N NaOH. 160 l of the quenched reaction mixture was transferred to 1 ml of 5% trichloroacetic acid and boiled to remove unincorporated 14 CO 2 . After cooling, 5 ml of Ecolite (NEN Life Science Products) was added, and the 14 CO 2 incorporated was determined in a Beckman LS 9800 counter. The amount of microsomal proteins present in the various experiments are indicated under "Results." All reported values are averages of three independent determinations.
Peptides were synthesized on a 357ACT peptide synthesizer (Advanced ChemTech) using Fmoc (N-(9-fluorenyl)methoxycarbonyl) chemistry strategy. The peptides were cleaved from the solid support by treatment with trifluoroacetic acid/phenol/ethanedithiol/thioanisole (90/5/2.5/2.5 by volume) and purified by reverse-phase high pressure liquid chromatography. The integrity of the peptides were verified by electrospray mass spectroscopy. Experiments were done in triplicate, and the data were analyzed using Graph Pad Prism from GraphPad Software, Inc. (San Diego, CA).
Isolation of conantokin-G cDNA was as follows. A cDNA library from C. geographus, in a pUC plasmid derivative, was plated out in duplicate and probed with end-labeled degenerate oligonucleotides corresponding to the mature toxin sequences. Hybridization was done in 3 M tetramethyl ammonium chloride (Aldrich), 0.1 M Na 2 HPO 4 , 0.001 M EDTA, 5ϫ Denhardt's solution, 0.6% SDS, 100 g/ml sheared salmon sperm DNA for 24 h at 48°C. Washes were done at room temperature in 3 M tetramethyl ammonium chloride, 0.05 M Tris-HCl, pH 8, 0.2% SDS for 15 min and 1 h at 50°C in a solution of the same composition. The filters were washed twice in 2ϫ SSC, 0.1% SDS at room temperature for 15 min each. The filters were then exposed to x-ray films. Double stranded DNA from purified clones were sequenced by dideoxynucleotide chain terminating method (13).

Characterization of the Conantokin-G Precursor-A
cDNA clone encoding conantokin-G was identified as described under "Materials and Methods." The clone was sequenced in both strands, and the nucleotide sequence obtained is shown in Fig.  1. The predicted amino acid sequence of the open reading frame encoded by this cDNA clone and the post-translational modifications that take place to yield the mature peptide are also indicated in the figure. The cDNA sequence predicts the presence of a valine residue at position 5 of the mature conantokin-G sequence. However, active peptide isolated from the venom contains leucine at this position. Using oligonucleotides corresponding to the 5Ј and 3Ј sequences of the cDNA as primers, we have also isolated the leucine containing cDNA by polymerase chain reaction amplification of total venom duct cDNA. We have used leucine containing conantokin-G peptides in our experiments.
The sequencing data reveal that the conantokin-G precursor is generally organized in a typical Conus prepropeptide motif, with the mature peptide encoded at the C-terminal end. As expected, all ␥-carboxyglutamate residues are encoded by the codons for glutamate. At the C-terminal end of the predicted open reading frame are codons encoding -Asn-Gly-Lys-Arg-Stop, which signals processing to a C-terminal Asn-NH 2 .
Immediately adjacent to Gly 1 of the mature conantokin-G is a typical proteolytic signal (Ala-Arg) for Conus prepropeptides. Proteolysis after the Arg residue would excise the N-terminal prepro-region of 80 amino acids containing a signal sequence of 21 amino acids at the N terminus. (However, at this point, we do not know the order of cleavage of the signal sequence and propeptide sequences.) In contrast to previously described signal sequences for conotoxins, the signal sequence for the conantokin-G precursor lacks a Cys residue (14).
A noteworthy feature of the prepropeptide is the relatively long intervening pro-region (59 amino acids). Among precursors of small Conus peptides (under 30 amino acids in length) sequenced to date, this is the longest pro-region that has been characterized so far. As we establish below, this 59-amino acid pro-region contains a ␥-carboxylation recognition sequence that presumably plays a critical role in the conversion of Glu to Gla.
Identification of ␥-CRS of Conantokin-G-In the case of mammalian proteins that undergo ␥-carboxylation, the ␥-CRS is contained in a 18-amino acid sequence (16-amino acid sequence in the case of bone peptides) immediately N-terminal to the mature peptide sequence (15)(16)(17). We investigated the ability of peptides shown in Table I to serve as substrates or affect the activity of the Conus carboxylase.
The concentration dependence for the carboxylation of peptides 1 and 3 (Table I) are shown in Fig. 2. The apparent K m for E.Con-G is 3400 Ϯ 215 M, and that for Ϫ20.Pro-E.Con-G is 28 Ϯ 3 M. The results clearly indicate that covalent linkage of the propeptide sequence Ϫ20 to Ϫ1 makes E.Con-G into an efficient substrate (note that the Ϫ20 to Ϫ1 sequence has no Glu residues that would be substrates for carboxylation). Similar observations were made when the Ϫ20 to Ϫ1 peptide was covalently linked to FLEEL (Table I). The apparent K m for peptide 3 is less than that of peptide 2, suggesting that propeptide sequences between Ϫ20 and Ϫ11 also interact with the carboxylase (Table I).
We then investigated the effect of the addition of pro (Ϫ20 to Ϫ1) in trans, on the carboxylation of FLEEL, E.Con-G, Ϫ10.Pro-E.Con-G, and Ϫ20.Pro-E.Con-G. We also determined the effect of addition of Pro(Ϫ20 to Ϫ11) and Pro(Ϫ11 to Ϫ1) on the carboxylation of Ϫ10.Pro-E.Con-G.
Inspection of the K m values in Table I clearly indicates that substrates in which the propeptide sequences Ϫ20 to Ϫ1 are covalently linked to the N terminus are efficient substrates for carboxylation. In the case of both E.Con-G and FLEEL, the K m decreases by 2 orders of magnitude. Pro(Ϫ20 to Ϫ1) stimulates the carboxylation of both FLEEL and E.Con-G when added in FIG. 1. A, nucleotide sequence of cDNA of conantokin-G and the expected translation product, prepro-conantokin-G. B, sequence of prepro-conantokin-G (L5V) and the mature toxin sequence. The arrow shows expected proteolytic processing site for the release of mature toxin product. ␥ represents ␥-carboxyglutamate residues, and # indicates the amidated C-terminal residue. The initial cDNA isolated was of conantokin-G (L5V). We have since isolated conantokin-G coding-cDNA, which has a sequence identical to our initial isolate except for Leu instead of Val at position 5 of the mature toxin sequence.
trans, but the effects are quite small: ϳ25% enhancement in the case of E.Con-G and 40% for FLEEL. Pro(Ϫ40 to Ϫ21) and Pro(Ϫ60 to Ϫ41) have no effect on the carboxylation of FLEEL or E.Con-G (data not shown).
Propeptide sequence Ϫ20 to Ϫ1 is an inhibitor for the carboxylation of Ϫ10.Pro-E.Con-G ( Fig. 3 and Table I) and Ϫ20.Pro-E.Con-G. The IC 50 of Pro(Ϫ20 to Ϫ1) is very similar to Pro(Ϫ30 to Ϫ1), suggesting that the interaction of the propeptide with the carboxylase in the region between Ϫ30 and Ϫ1 is probably limited to Ϫ20 to Ϫ1. Pro(Ϫ40 to Ϫ21) and Pro(Ϫ60 to Ϫ41) did not inhibit carboxylation of Ϫ10.Pro-E.Con-G even at concentrations 100-fold greater.
We also investigated the effect of alanine substitutions of the basic amino acids in the propeptide. Ϫ20 GKDRLTQMKRIL-KQRGNKAR Ϫ1 GEEELY-NH 2 , a 26-amino acid peptide containing propeptide sequences Ϫ20 to Ϫ1 of conantokin-G was used as the wild type substrate. Individual peptides in which KR at positions Ϫ12 and Ϫ11, KQR at positions Ϫ8, Ϫ7, and Ϫ6, NK at positions Ϫ4 and Ϫ3, and NK-R at positions Ϫ4, Ϫ3, and Ϫ1 were substituted by alanine were used as substrates in the carboxylation reaction. Carboxylation reactions were done at a substrate concentration of 40 M, the apparent K m of carboxylation for the wild type substrate. Alanine substitutions in the context of the remaining propeptide sequences had little effect (Ͻ10%) on carboxylation.
We examined the ability of bovine microsomes to use Ϫ20.Pro-E.Con-G as substrate for ␥-carboxylation. As shown in Table II, although this peptide is an excellent substrate for the Conus enzyme, it is extremely poor for the mammalian enzyme. DISCUSSION We have demonstrated that the ␥-carboxylated 17-amino acid Conus peptide conantokin-G is initially translated as a prepropeptide of 100 amino acids. In general, the organization of the conantokin-G precursor is similar to that previously reported for disulfide-rich conotoxins from Conus venoms. The mature peptide is found in a single copy at the C-terminal end of the precursor. Before maturation of the peptide, a number of post-translational processing events have to take place (Fig. 1). These events include the ␥-carboxylation of five glutamate residues, C-terminal amidation of asparagine-17 following excision of the C-terminal tripeptide, and a proteolytic event between Arg Ϫ1 and Gly 1 .
One notable feature of the conantokin-G prepropeptide is the length of the intervening region between the signal sequence and the mature peptide. The 59 amino acids in the intervening pro-region is the longest so far reported for any Conus venom peptide. We have demonstrated one potential function of this extended region: the presence of a ␥-carboxylation recognition sequence for the Conus venom duct ␥-glutamyl carboxylase in the Ϫ1 to Ϫ20 region.
In mammalian blood coagulation and bone Gla proteins, ␥-carboxylation of glutamate residues is carried out by a vitamin K-dependent carboxylase. A conserved motif (16) ␥-carboxylation recognition sequence in the propeptide sequence binds the ␥-carboxylase and is required for a polypeptide substrate to be a high affinity target for the ␥-carboxylase. In the experiments described above, we carried out an analysis using segments of the conantokin-G prepropeptide to identify potential sites that might serve as ␥-carboxylase recognition signals for the Conus enzyme. The results reveal that a ␥-carboxylation  G as substrates (B). The data were fitted to a hyperbola in a single-site binding model. The apparent K m for E.Con-G was determined from a Lineweaver-Burk plot and for Ϫ20.Pro-E.Con-G from the best fit to a hyperbola.

FIG. 3. Kinetic analysis of the inhibition of carboxylation of
؊20.pro-E.Con-G by Pro(؊20 to ؊1). The data were fitted to singlesite competitive binding model. 50 M Ϫ20.Pro-E.Con-G was used in the carboxylation reaction and concentration of Pro (Ϫ20 to Ϫ1) was varied.
a Represents apparent IC 50 values. For experiments 6 -9, carboxylation reactions were carried out using a 50 M concentration of Ϫ20.Pro-E.Con-G and varying concentrations of competing peptide. The results were fitted to a single-site competition model.
b No effects of these peptides were observed in the carboxylation reaction at a concentration of 1 mM.
recognition sequence is included in the Ϫ1 to Ϫ20 region of the conantokin-G prepropeptide. This appears to increase the affinity of the Conus carboxylase by approximately 2 orders of magnitude under the assay conditions used. Knobloch and Suttie (18) and others (19) found that the propeptide sequences of Factors IX and X at micromolar concentrations stimulated the carboxylation of oligopeptide substrates, suggesting a probable positive allosteric effector role. In addition, the propeptide at micromolar concentrations acted as a competitive inhibitor of carboxylation of a substrate whose sequences were based on residues Ϫ18 to ϩ10 of prothrombin (20). Similarly, the Conus propeptide (Ϫ20 to Ϫ1) inhibits the carboxylation of propeptide-containing substrates, (i.e. Ϫ10.Pro-E.Con-G and Ϫ20.Pro.E.Con-G). However, we were unable to observe equivalent, strong stimulation of carboxylation of FLEEL and E.Con-G by the Conus propeptide.
The relevant propeptide sequence (Ϫ20 to Ϫ1) for conantokin-G, human Factors IX (21) and X (22), and prothrombin (23) are shown in Table III. Except for the presence of a hydrophobic residue at position Ϫ16, the conantokin-G sequence does not appear to share the conserved features of the mammalian propeptides (16). Phenylalanine is present at position Ϫ16 in all the propeptide sequences except in gas6 and protein S, in which it is leucine as in the case of Con-G. In addition, isoleucine is present at position Ϫ10 of Con-G in place of a conserved alanine residue. Besides, positions Ϫ6 and Ϫ7, which are always hydrophobic for the mammalian propeptide sequences, are basic and polar residues, respectively, for Con-G. This suggests differences between the recognition specificities of the mammalian and Conus enzymes. A preliminary characterization of the amino acids important for recognition by the Conus enzyme has been carried out. The individual basic amino acids in the propeptide region Ϫ20 to Ϫ1 can be replaced by alanine without significantly affecting the apparent K m for carboxylation. The importance of hydrophobic residues in the mammalian ␥-CRS suggests that they may also be important for the Conus carboxylase, and this role needs to be investigated.
What is the role of the ␥-CRS Ϫ? Does it merely tether the substrate to the carboxylase, or does it play additional roles in the ␥-carboxylation reaction? How do multiple carboxylations occur on a single substrate molecule? Does the ␥-CRS set a limit to the distance at which ␥-carboxylation will take place? Is the ␥-CRS sufficient for the fidelity of carboxylation, or are additional membrane components necessary? These are some of the questions that need to be addressed to understand the mechanism of action of ␥-glutamyl carboxylase.
The orientation in which a Glu presents itself to the active site of the carboxylase may determine whether it will be carboxylated. In the case of Con-G not all the Glu residues are ␥-carboxylated (Glu 2 is not carboxylated, whereas Glu 3 and Glu 4 are carboxylated). The solution structures of Con-G and Con-T as determined by CD and NMR spectroscopy (24, 25) are a mixture of ␣ and 3 10 helices. Rigby et al. (26) also determined the structure of the metal-free conformer of conantokin-G by NMR spec-troscopy. In all these structures, the Gla residues are on the same side of the conantokin structure; this would allow a membranebound enzyme to carry out efficient carboxylation of Glu residues oriented in the same direction with optimum stereochemistry.
In the discussion above, we have emphasized the differences in the ␥-carboxylation recognition signal sequences in the mammalian and Conus systems. These differences may be due to the large evolutionary distance between the two species. However, there is an underlying general similarity between the two enzymes: the catalytic reaction they carry out, their cofactor requirements, and a recognition signal (albeit differing in sequence) in the Ϫ1 to Ϫ20 propeptide region. It will be important to purify the Conus enzyme and characterize it to determine the relationship between the two enzymes.  peptides Pro-Con-G is the propeptide sequence (Ϫ20 to Ϫ1) of conantokin-G; hFIX (21), hFX (22), and hPT (23) are the propeptide sequences of human Factors IX and X and prothrombin. Pro-Con-G GKDRLTQMKRILKQRGNKAR hFIX TVFLDHENANKILNRPKR hFX SLFIRREQANNILARVTR hPT HVFLAPQQARSLLQRVRR Conantokin-G Precursor and Its Role in ␥-Carboxylation 5450