The T-superfamily of Conotoxins*

We report the discovery and initial characterization of the T-superfamily of conotoxins. Eight different T-superfamily peptides from five Conusspecies were identified; they share a consensus signal sequence, and a conserved arrangement of cysteine residues (- -CC- -CC-). T-superfamily peptides were found expressed in venom ducts of all major feeding types of Conus; the results suggest that the T-superfamily will be a large and diverse group of peptides, widely distributed in the 500 different Conusspecies. These peptides are likely to be functionally diverse; although the peptides are small (11–17 amino acids), their sequences are strikingly divergent, with different peptides of the superfamily exhibiting varying extents of post-translational modification. Of the three peptides tested for in vivo biological activity, only one was active on mice but all three had effects on fish. The peptides that have been extensively characterized are as follows: p5a, GCCPKQMRCCTL*; tx5a, γCCγDGW+CCT§AAO; and au5a, FCCPFIRYCCW (where γ = γ-carboxyglutamate, W+ = bromotryptophan, O = hydroxyproline, T§ = glycosylated threonine, and * = COOH-terminal amidation). We also demonstrate that the precursor of tx5a contains a functional γ-carboxylation recognition signal in the −1 to −20 propeptide region, consistent with the presence of γ-carboxyglutamate residues in this peptide.

Cone snails (genus Conus) are perhaps the most successful genus of marine invertebrates, with over 500 species, all of which are venomous (1,2). These predatory marine snails have evolved a highly sophisticated neuropharmacological strategy based on small peptides (10 -35 amino acids) in their venoms (3,4). Most Conus peptides potently affect ion channel function; these are widely used pharmacological reagents in neuroscience, and several are being directly developed as diagnostic and therapeutic agents. Most Conus peptides are highly disulfide-rich; generically, Conus peptides with multiple disulfide cross-links have been referred to as conotoxins. It has become apparent in recent years that there are tens of thousands of different conotoxins in Conus venoms. Because of the remarkably rapid interspecific divergence of peptide sequences, each Conus species has its own distinct repertoire of between 50 and 200 different venom peptides (5).
A major simplification in understanding this complex array of Conus venom peptides is that most of the ϳ50,000 different molecular forms can be grouped into just a few superfamilies. Peptides in the same superfamily share both a conserved pattern of disulfide connectivity and a highly conserved signal sequence (when prepropeptide precursor sequences of the peptides are compared) (5,6). Three large superfamilies of conotoxins are well characterized: the O-superfamily, comprising several distinct pharmacological families including the -, -, ␦-, and O-conotoxins (7); the A-superfamily, to which the ␣-conotoxins belong (8); and the M-superfamily, to which the -conotoxins belong. In this paper, we describe the T-superfamily, a previously uncharacterized group of Conus peptides that exhibit a novel disulfide pattern and share a conserved signal sequence.
The data presented in this report suggest that T-superfamily peptides are a major group of Conus peptides, and that considerable diversity will exist within the superfamily. Eight members of the T-superfamily have been identified in the venom ducts of four different cone snails, including fish-hunting, snail-hunting, and worm-hunting Conus. Although the molecular targets of T-superfamily peptides have not yet been identified, this report provides a clear roadmap for a systematic exploration of this diverse, yet coherent, group that may encompass ϳ1,000 distinct pharmacologically active peptides.

MATERIALS AND METHODS
Extraction and Fractionation of Crude Venom-The venom of Conus purpurascens was obtained by milking specimens maintained in aquaria as described previously (9). The collection from ϳ90 milkings (ϳ0.5 ml) was diluted with 50 ml of 0.1% trifluoroacetic acid in water (buffer A) then fractionated on a Vydac C 18 preparative column (22 mm ϫ 25 cm, 15-m particle size, 300-Å pore size, 20 ml/min flow rate). Venom components were eluted by a gradient with limiting buffers consisting of 0.1% trifluoroacetic acid and 60% acetonitrile (CH 3 CN) in 0.092% trifluoroacetic acid (buffer B60) or 90% acetonitrile in 0.08% trifluoroacetic acid (buffer B90). The absorbance at 220 nm was monitored, and fractions were collected at 30-s intervals.
Lyophilized Conus aulicus venom (550 mg) obtained from the Philippines was extracted with 40 ml of 40% CH 3 CN in 0.5% trifluoroacetic acid. The suspension was homogenized at low speed with three strokes of a glass/Teflon homogenizer attached to a drill press and then centrifuged at 100,000 ϫ g for 10 min. The supernatant was diluted with 10 volumes of 0.1% trifluoroacetic acid and then fractionated on a preparative C 18 HPLC 1 column as described above.
Lyophilized Conus textile venom (400 mg) obtained from the Philip-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF167164 -AF167168.
§ These authors contributed equally to this work. § § To whom correspondence should be addressed.
pines was extracted sequentially with 10 ml each of 0%, 20%, 40%, and 60% CH 3 CN. The mixture was sonicated for three 30-s periods while immersed in ice water, centrifuged at 5,000 ϫ g for 5 min, and the combined supernatant was stored at Ϫ20°C. The extract was fractionated in several runs on a Vydac C 18 semi-preparative column (10 ϫ 250 mm, 5-m particle size) and an analytical column (4.6 ϫ 250 mm), both eluted with a 0 to 28% gradient of CH 3 CN (0.45%/min) at a flow rate of 5 ml/min. Corresponding fractions were pooled for further purification.
Purification of the T-superfamily Peptides-The peptides of C. purpurascens ( Fig. 1) and C. textile (Fig. 2) were purified from relevant fractions of the venom by analytical reverse phase chromatography (4.6 ϫ 250 mm, 5 m, 300 Å, Vydac C 18 or Microsorb MV). Gradient elution was done using the same buffer systems as for preparative columns. Peptides from C. aulicus (Fig. 5) were purified from preparative fractions using a sulfonic-based strong cation exchange-HPLC column (Vydac 400VHP575, 5 mm, 7.5 ϫ 50 mm), followed by a run on a reverse-phase C 18 column. The strong cation exchange column was eluted by a gradient with limiting buffers consisting of 10 mM phosphate, pH 2.5, in 50% acetonitrile and 0.25 M NaCl, 10 mM phosphate, pH 2.5, in 50% acetonitrile. The active peak from this column was concentrated and desalted before application on the analytical C 18 column. Other details of the purification procedures are described in the legends of Figs. 1, 2, and 5.
Peptide Sequencing-Due to the limited amount of peptide p5a from C. purpurascens, it was sequenced on an ABI model 477A peptide sequencer without reduction and alkylation of potential cysteine residues. For determination of Cys residues in tx5a, au5a, and au5b, the peptides were reduced with dithiothreitol and alkylated with 4-vinylpyridine as described below. Approximately 20 -80 pmol of the peptides were used. The alkylated peptides were sequenced by Edman degradation using an Applied Biosystems model 492 Sequenator (DNA/ Peptide Facility, University of Utah). The 3-phenyl-2-thiohydantoin derivatives were identified by HPLC. Predicted masses for each sequence were verified by mass spectrometry, as described below.
Reduction and Alkylation of the Purified Peptide-The C. textile peptide (tx5a) and the C. aulicus peptides (au5a and au5b) were reduced with dithiothreitol and alkylated with 4-vinylpyridine. Prior to reduction, the peptide solution was adjusted to pH 8 with 0.5 M Tris base and 10 mM dithiothreitol was added. The solution was flushed with nitrogen gas, incubated at 65°C for 15 min, and then cooled to room temperature. After adding 4-vinylpyridine (5 l/ml of solution), the mixture was left in the dark at room temperature for 25 min. The mixture was diluted with 500 l of 0.1% trifluoroacetic acid prior to purification of the reduced peptide on an analytical reverse-phase HPLC column.
Mass Spectrometry-Matrix-assisted laser desorption (MALD) (10) mass spectra were obtained using a Bruker REFLEX (Bruker Daltonics, Billerica, MA) time-of-flight (11) mass spectrometer. The sample (in 0.1% trifluoroacetic acid) was applied with ␣-cyano-4-hydroxycinnamic acid. Electrospray (ESI) mass spectra were obtained using an Esquire ion trap mass spectrometer (Bruker Daltonics). The HPLC-purified sample, collected in 0.1% trifluoroacetic acid and acetonitrile, was diluted with 1% acetic acid in methanol, transferred to a fused silica capillary, and infused at approximately 250 nl/min. Liquid secondary ionization (LSI) mass spectra were measured on a Jeol HX110 double focusing magnetic sector mass spectrometer (Jeol, Tokyo, Japan). The sample in a glycerol matrix was bombarded with high energy (25 keV) Cs ϩ ions. The mass accuracy was typically better than 1000 ppm for the time-of-flight instrument, 200 ppm for the ion trap instrument, and 50 ppm for the magnetic sector instrument.
Asp-N Digestion-The tx5a peptide was digested with endoproteinase Asp-N as directed in the procedure provided by Roche Molecular Biochemicals. The lyophilized peptide was dissolved in 100 l of 50 mM sodium phosphate buffer, pH 8.0. Endoproteinase Asp-N (1 or 5 g of enzyme/20 g of peptide) in 10 mM Tris-HCl, pH 7.5, was added and the mixture was incubated for 17 h at 37°C. The reaction was stopped with 500 l of 0.1% trifluoroacetic acid, and the digest was fractionated by HPLC with a linear gradient of 0.9% acetonitrile per ml/min. The intact masses of the digestion fragments were analyzed using MALD-MS and ESI-MS prior to chemical sequencing.
Chemical Synthesis-The peptides p5a and au5a were synthesized on Rink amide resin using Fmoc (N-(9-fluorenyl)methoxycarboxyl) chemistry and standard side chain protection except on the cysteine residues. For p5a synthesis, the first and third cysteine residues were S-trityl protected, while the second and fourth were protected with S-acetamidomethyl (acm) groups. Two possible disulfide-bonded forms of au5a were synthesized. In one isomer (S1), the first and third cysteine residues were S-trityl-protected, while the second and fourth were protected with acm groups. In the other isomer (S2), the first and fourth cysteine residues were S-trityl-protected, whereas the second and third cysteine residues were acm-protected.
Peptides were removed from resin and precipitated as described previously (12,13). A two-step oxidation protocol was used to selectively fold the peptide as detailed elsewhere (14,15) with the modifications described below.
Following preparative purification of the linear peptide by HPLC with a 10 -50% gradient of buffer B60, the appropriate HPLC fraction was added dropwise over several min to an equal volume of 20 mM K 3 Fe(CN) 6 solution in 0.1 M Tris buffer, pH 7.7, and stirred for 30 min. An Alltech "extract-clean" syringe containing C 18 silica (1 g of silica, 100-m particle size, 60-Å pore size) was wetted by gravity perfusion with buffer B60 for ϳ30 min followed by 1-2 min of perfusion with buffer A. The peptide oxidation mixture was diluted at least 2-fold with buffer A and passed through the silica under vacuum (flow rate ϳ50 ml/min). The silica was washed with ϳl liter of buffer A under vacuum, and peptide was then eluted by gravity perfusion with 20 ml of buffer B60. Removal of acm protection and closure of the final disulfide was done by oxidation with 5 mM iodine in 5% trifluoroacetic acid for 5 min. Fully oxidized peptide was purified by preparative HPLC using a 20 -50% gradient of B60. Synthesis was confirmed by LSI-MS analysis and HPLC co-elution.
A second batch of au5a was synthesized on a 357 ACT Peptide Synthesizer (Advanced Chemtech, Louisville, KY) using standard Fmoc chemistry. The disulfide bonds were formed by a random folding strategy in the presence of 1 mM reduced and 0.5 mM oxidized glutathione (pH adjusted to 7.5). The major product (50% of the mixture) that had the desired folding pattern was purified by reverse phase HPLC and then lyophilized.
Biological Assay-Mice were injected intracranially or intraperitoneally with peptides in 15-20 l of saline or with saline alone, and observed for behavioral changes. Siamese fighting fish were similarly injected with 10 l of sample in the dorsal muscle and observed for suppression of reactions to self-observation following placement in front of a mirror. In control fish, behaviors typically include a "gill display" (downward extension of the gill flap); extension of dorsal, ventral, and pectoral fins; and, sometimes, agitated swimming and rubbing against the fish's reflection in the mirror. This behavior is similar to that produced by the presence of another fighting fish. The effect of peptide injection on gill display and extension of fins was used as a measure of activity.
A second hallmark symptom elicited by higher doses of T-superfamily peptides was an abnormal dorsal fin. These fish have long dorsal fins, which they can greatly extend in display, but on injection of T-superfamily peptides, the fins droop far below their usual resting position. Observing this symptom does not require putting a mirror in front of the fish.
Analysis of Tx5.1 and Gm5.1 Clones by the Expressed Sequence Tag Method-First-strand synthesis of complementary DNA was primed from oligo(dT) extension at the PstI site of a linearized modified pUC13 plasmid using polyadenylated mRNA isolated from C. textile and Conus gloriamaris venom ducts. The products were size-fractionated by gel electrophoresis and used to transform Escherichia coli MC1061 to produce cDNA libraries (16). Expressed sequence tags were identified from single colonies randomly selected from Ampicillin-LB plates plated with Conus cDNA libraries (17). Insert sizes of the clones were analyzed by single colony PCR (18) with vector-specific oligonucleotides flanking the insert region (500 nM amount of each oligonucleotide, 2.5 mM MgCl 2 , 20 mM Tris-HCl, pH 8.4, 50 mM KCl, 250 g/ml bovine serum albumin, 125 M amount of each dNTP, and 0.5 unit of Taq DNA polymerase). Reaction mixtures were amplified (50 cycles of 25 s at 94°C, 25 s at 54°C, 2 min at 72°C) using a 1605 Air Thermo-Cycler TM (Idaho Technology, Idaho Falls, ID). Amplification products were analyzed by gel electrophoresis (1.5% agarose, 0.5ϫ TBE buffer). Clones containing insert sizes larger than 400 base pairs in length were selected for sequencing. Templates were prepared using QIAprep Spin Miniprep kit (Qiagen, Valencia, CA) and submitted for fluorescent sequencing primed by oligonucleotides M13R and subsequently M13U (19) at the Health Sciences Center Sequencing Facility, Eccles Institute of Human Genetics, University of Utah. All molecular biology techniques were as described by Sambrook et al. (20), unless otherwise specified. Sequence analysis was performed with SeqMan (DNASTAR, version 2.55; DNAS-TAR, Inc., Madison, WI) and Pileup (21).
Cloning of Tx5.2, P5.1, and Im5.1-The DNA sequence of clone Tx5.1 was analyzed, and the following oligonucleotide primers containing EcoRI endonuclease sites were designed to screen cDNA libraries of other species: first strand primer 1 (5Ј-GGA ATT CGG AAG CTG ACT ACA AGC AGA-3Ј) and second strand primer 2 (5Ј-GGA ATT CCA AAT GAT GTA ATT ACT GAC-3Ј). The cDNA libraries were then screened using PCR. The reaction mixture contained standard PCR reagents and the following: 500 nM primer 1, 500 nM primer 2, 100 ng/l cDNA library. Reaction mixtures were then amplified in thin-walled PCR tubes for 35 cycles at 94°C for 25 s, 52°C for 35 s, and 72°C for 2 min using the Air Thermo-Cycler. PCR products were digested with EcoRI, ligated into Bluescript SK(Ϫ) vector (Stratagene, CA), and used to transform E. coli DH5␣.
Cloning of Gm5.2-The cloning of Gm5.2 was identical to the above procedure except for the following alterations: oligonucleotide primer 3 (5Ј-AGC TCT AGA GGA AGC TGA CTA CAA GCA-3Ј) designed from 5Ј-untranslated region of Tx5.1, and oligonucleotide primer 4 (5Ј-CAC AAG CTT TAG GTC ATC CAG TTC C-3Ј) designed from the consensus 3Ј-untranslated region of several cloned fragments obtained through expressed sequence tag screening of a C. textile cDNA library, were used instead of primer 1 and primer 2. 100 ng/l cDNA library was amplified as described above, and the amplicon was ligated into Bluescript SK(Ϫ) vector using blunt-end ligation.
␥-Glutamyl Carboxylase Assays-The peptide pro(Ϫ20 to Ϫ1).FLEELamide, PLSSLRDNLKRTIRTRLNIR. FLEEL-NH 2 (which contains the propeptide sequences Ϫ20 to Ϫ1 of Tx5.2 covalently linked to FLEELamide at the amino terminus) was synthesized by Dr. Bob Schackmann of the DNA/Peptide Facility, Huntsman Cancer Center (supported by Grant NCICA 42014), University of Utah. The identity of the peptide was verified by ESI-MS.
Partially purified ␥-glutamyl carboxylase was prepared by the following procedure. Microsomes of C. textile were prepared as described by Stanley et al. (22). Microsomes were suspended in buffer containing 2.0 M NaCl, 0.1 M MOPS, pH 7.0, 0.8% CHAPS, 0.8% phosphatidyl choline, and incubated at 4°C for 1 h, then centrifuged for 1 h at 125,000 ϫ g. The supernatant containing ␥-glutamyl carboxylase was adjusted to 66% saturation in ammonium sulfate. The enzyme recovered in the precipitate was dissolved in buffer containing 0.1 M NaCl, 0.025 M MOPS, pH 7.0, 0.1% CHAPS, 0.1% phosphatidylcholine, distributed into aliquots, quick frozen in liquid N 2 , and stored at Ϫ80°C. Fresh aliquots of enzyme were thawed individually for enzyme assays.
Carboxylase assays were performed using 0.5 g of partially purified enzyme according to methods described by Stanley et al. (22). FLEEL and Tx5.2 pro(Ϫ20 to Ϫ1).FLEEL-amide were used as substrates in the carboxylase reaction. Experiments were done in triplicate, and the data were fitted to a single-site binding model and analyzed using Graph Pad Prism from GraphPad Software, Inc. (San Diego, CA).
Nomenclature-In this report, we adopt a nomenclature that is based on conventions used for naming ion channels, the likely molecular targets of these peptides. Putative sequences deduced from clones will be named as follows: 1) letters (one for fish-hunting species, two for non-fish-hunting species) designate the Conus species source of the clone, 2) a number represents the disulfide framework, and 3) a second number, separate by a decimal, indicates the order of clone identification. For example, P5.1 is from C. purpurascens, has disulfide framework "5" (the number assigned to the -CC--CC-pattern), and is the first clone from the species with this framework. To distinguish clones from peptides that have been isolated from Conus venom, in the latter case all of the species letter(s) are small, the disulfide framework is repre-sented by an arabic numeral and the order of discovery is indicated by a letter, starting with "a"; thus, clone P5.1 may encode peptide p5a isolated from venom. Likewise, the clones from C. aulicus, Au5.1 and Au5.2, correspond to the venom-purified peptides au5a and au5b.
T-superfamily peptide clones will be given the numerical designations 5.1, 5.2, etc., in the order in which they are discovered. The peptides purified from venom will be named as described previously (8); hence, au5a and au5b.
Finally, when a molecular target is assigned to a toxin, the name will be prefixed by the appropriate pre-existing or newly assigned Greek letter (-for calcium channels, ␣for acetylcholine receptors, and so forth). Thus, if a member of the T-superfamily from C. textile is determined to have a novel physiological mechanism, it might be called -conotoxin TxVA (which might, for example, be encoded by clone Tx5.2; -TxVA would then permanently replace tx5a). This nomenclature gives information about physiological mechanism, structure, and the natural source from which the peptide was originally obtained.
We note that peptides of the conantokin family were referred to using "V" (conantokin-G was initially described as GV). Since this is not a multiply disulfide-bonded peptide, we now refer to all peptides of the conantokin family by a letter designation for the species: thus, conantokin-G, conantokin-R, etc. We will reserve the provisional designation "5" and the Roman numeral "V" for members of the T-superfamily having the cysteine arrangement ---CC---CC---.

Discovery of a Novel Class of Conotoxin Clones-
We have carried out a systematic characterization of clones present in cDNA libraries of Conus venom ducts. By analyzing different clones using an expressed sequence tag strategy, we identified different classes of cDNAs encoding Conus peptides. Although most of these fall into previously identified groups (such as the O-superfamily), a number of clones that were frequently encountered in some cDNA libraries do not belong to previously characterized superfamilies of Conus peptides.
One class of cDNAs encoded a novel group of Conus peptides characterized by a long 3Ј-untranslated region (ϳ600 base pairs), which exhibited no sequence homology to conserved 3Ј-untranslated regions of characterized superfamilies. The prepropeptide precursor associated with this 3Ј-untranslated region predicted a small mature peptide with a unique pattern of four cysteine residues, ---CC------CC-. This novel class of cDNA clones was encountered at a frequency of over 20% in a cDNA library prepared from C. textile venom ducts. The sequence of the first such cDNA clone for which a complete open reading frame was deduced is shown in Table I. Like all Conus peptides, the predicted translation product has a prepropeptide organization, with a disulfide-rich mature conotoxin sequence present in a single copy at the COOH-terminal end; this clone is designated Tx5.1. Since the signal sequence encoded by clone Tx5.1 and the arrangement of Cys residues in the predicted  TGC TGT CTC CCA GTG TTC GTC ATT CTT CTG CTG CTG ATT GCA TCT GCA CCT AGC GTT  M  C  C  L  P  V  F  V  I  L  L  L  L  I  A  S  A  P  S  V  61/ 10 20s 30 40 250 60 MCCLPVFVILLLLIASAPSVDAQPKTKDDVPLAPLHDNAKSALQHLNQRCCQTFYWCCVQ* mature tx5a peptide are novel, these define a novel group of Conus peptides (which we designate the T-superfamily of conotoxins).
Evidence That T-superfamily Peptides Are Broadly Distributed and Diverse in Conus-Given the apparent high frequency of clones encoding precursors belonging to this new superfamily of Conus peptides in C. textile venom, we used a PCR approach to identify related peptides in C. textile and other Conus cDNA libraries (see "Materials and Methods"). Peptides clearly belonging to the T-superfamily were identified from cDNA libraries made from venom ducts of C. textile, C. gloriamaris, C. purpurascens, and Conus imperialis. The deduced prepropeptide sequences of these peptides are shown in Table II.
These results strongly suggest that this new superfamily of Conus peptides will be widespread in the genus, since peptides belonging to the superfamily appear to be expressed in venom ducts of all major Conus feeding types (C. purpurascens is a fish-hunting species, C. imperialis specializes on polychaete worms, while C. textile and C. gloriamaris are snail-hunting Conus species). C. textile and C. gloriamaris each expressed two widely divergent peptide sequences belonging to the T-superfamily.
Purification and Characterization of p5a, a Peptide Belonging to the T-superfamily-In addition to a definition of the T-superfamily by cDNA cloning, three different peptides that clearly belong to the T-superfamily were directly isolated from venom. One of these was purified from venom obtained by milking C. purpurascens (9); the peptide from venom is clearly encoded by the cDNA clone P5.1 from a C. purpurascens cDNA library (Table II).
The sequence assignment was confirmed by synthesis of a peptide with the above sequence and specific disulfides (Cys 1 -Cys 3 ; Cys 2 -Cys 4 ). This synthetic peptide co-eluted with the natural material (see Fig. 1E). We give this peptide the provi-sional designation p5a, which is the mature peptide encoded by clone P5.1. The COOH terminus is presumably processed by conventional mechanisms to yield the amidated COOH-terminal Leu residue.
The peptide showed no obvious symptomatology when injected intracranially or intraperitoneally into mice. However, when injected into male specimens of the Siamese fighting fish, Betta splendens, a clear deviation from normal behavior was observed. An immediate aggressive display is normally elicited

MRYLPVFVILLLLIASIPSDTVQLKTKDDMPLASFHGNGRRILRMLSNKR.LCCVTEDWCCEWW
Im5.1  1. Purification of p5a. A, the components of milked venom from C. purpurascens were fractionated by preparative RP C 18 HPLC column. Peptides were eluted using a linear gradient of 0 -100% buffer B90 over 100 min. B, the fraction indicated in A was repurified by analytical HPLC using a gradient of 25-55% buffer B60 over 30 min. The arrow indicates the peak corresponding to p5a. C-E, co-elution of native p5a with synthetic material. Purified native peptide (C), synthetic peptide (D), and both combined (E) were chromatographed using a 20 -50% gradient of buffer B60 over 30 min. In each case, a single homogeneous peak eluted at exactly 18.56 min. in these fish in response to their reflection when placed in a mirrored aquarium; injection of relatively high levels of the peptide suppressed this behavior (ED min ϳ8 nmol).

MRCLPVFVILLLLIASAPSVDAQPKTKDDVPLASLHDN-K-LQ---------CC-----CC
Purification and Characterization of a Peptide Belonging to the T-superfamily of Conotoxins from C. textile Venom-A peptide that belongs to the T-superfamily was purified from C. textile venom using reversed phase HPLC, as shown in Fig. 2. This venom component caused hyperactivity and other excitatory behavior upon intracranially injection into mice. The purified material also potently affected fish (ED min ϳ0.2 nmol using the Betta gill display assay).
The results of an Edman sequence analysis detected no phenylthiohydantoin derivatives; consequently, the peptide was reduced and alkylated, and the Edman sequence analysis of the modified peptide is shown in Table III. There are still four positions (1, 4, 7, and 10) that could not be assigned, but nine other residues could be identified unambiguously.
The sequence obtained closely corresponds to that predicted by clone Tx5.2 (see Tables II and III). However, the Glu residues predicted by the clone at positions 1 and 4 could not be assigned; examination of the Edman analysis revealed that a small yield of Glu was in fact detected in both of these cycles. This is a characteristic noted in previous Edman analyses of peptides containing ␥-carboxyglutamate (Gla).
The two remaining blanks in the Edman sequence analysis, at positions 7 and 10, were predicted from the cDNA clone to be Trp and Thr, respectively. We have previously shown that Trp residues in Conus peptides can be modified to 6-bromotryptophan (which could account for the blank obtained for residue 7) (23). Recently, we demonstrated that in a novel peptide from C. geographus, contulakin-G, a threonine residue was O-glycosylated (24); an O-glycosylated threonine could account for the blank at position 10. Thus, ␥-carboxylation of Glu 1 and Glu 4 to Gla, bromination of Trp 7 to 6-Br-Trp, and O-glycosylation of Thr 10 would explain the Edman sequencing results shown in Table III. We also note that no further phenylthiohydantoin derivatives were obtained in Edman steps beyond residue 13, despite the prediction from the nucleic acid sequence of clone Tx5.2 of two additional amino acid residues (see Table III).
The hypothesis that the peptide is post-translationally modified as proposed above is strongly supported by mass spectrometry data. ESI-MS analysis revealed a m/z 964. In the ESI-MS/MS spectrum of the m/z 965 negatively charged precursor, loss of one or two molecules of CO 2 (m/z 942.8 and 920.1) predominated, indicative of two Gla residues. The MALD-MS analysis indicated the presence of both a Hex-HexNAc moiety and the Gla residues. Based on this evidence for the presence of a glycosylated residue, two Gla residues, a bromotryptophan residue, and the cDNA clone obtained, we proposed the sequence: Gla-Cys-Cys-Gla-Asp-Gly-Trp*-Cys-Cys-Thr † -Ala-Ala-Pro-OH, where Gla ϭ ␥-carboxyglutamatic acid, Trp* ϭ bromotryptophan, and Thr † ϭ Hex-HexNAc-Thr. The observed mass of the C. textile peptide (1929.4 Da) was consistent with the calculated mass (1929.42 Da). Comparison of the proposed sequence with the clone obtained indicates that the Leu-Thr dipeptide has been cleaved from the COOH terminus of the peptide.  Thus, in contrast to p5a, tx5a, the first T-superfamily peptide isolated and characterized from C. textile venom, exhibited a high degree of post-translational modification. The p5a peptide from C. purpurascens is unmodified except for amidation of the COOH terminus.
Evidence for a Functional ␥-Carboxylation Recognition Sequence in the Tx5.2 Prepropeptide-The discovery that two glutamate residues in tx5a were ␥-carboxylated suggested the presence of a ␥-carboxylation recognition signal in the "pro" region of the precursor. Recently, it was established that the Ϫ1 to Ϫ20 region of the ␥-carboxylated conantokins contains recognition signals that confer a higher affinity when present NH 2 -terminal to a target sequence (25). However, there is no obvious sequence homology between the Ϫ1 to Ϫ20 regions of the conantokins and Tx5.2.
In order to test whether the C. textile Tx5.2 prepropeptide does indeed contain a ␥-carboxylation recognition sequence in its Ϫ1 to Ϫ20 region, a peptide was synthesized with the Ϫ1 to Ϫ20 region from Tx5.2 attached to a standard ␥-carboxylation target sequence, FLEEL. The ␥-carboxylation of FLEEL was assessed in the presence and in the absence of the Ϫ1 to Ϫ20 region of Tx5.2.
As shown in Fig. 4, the presence of the Ϫ1 to Ϫ20 Tx5.2 region does indeed increase the affinity by over 2 orders of magnitude for the targeted FLEEL sequence. The estimated EC 50 values in the presence and absence of propeptide are 0.59 and 140 M, respectively. It should be noted that maximum activity in the presence of saturating amounts of FLEEL was not achieved and so the EC 50 of 140 M is probably a lower estimate. Thus, not only is ␥-carboxyglutamate present in the mature peptide region, but a carboxylase recognition signal is present immediately NH 2 -terminal to the targeted glutamate residues, in the Tx5.2 prepropeptide. There may also be recognition signals in the prepropeptide for bromination and Oglycosylation enzymes. Thus, Tx5.2 and other members of the T-superfamily may provide good model substrates for studying post-translational modification of Conus peptides.
Purification of T-superfamily Peptides from C. aulicus Venom-Two peptides belonging to the T-superfamily were purified from C. aulicus venom as shown in Fig. 5. Edman sequencing of the two peptides showed that they had the Cys pattern of the T-superfamily. The two purified peptides, designated au5a and au5b, have the amino acid sequences shown in Table II.
The amino acid sequences were confirmed by LSI-MS (observed monoisotopic [M ϩ H] ϩ values for au5a and au5b are m/z 1436.6 and 1388.6, respectively; cf. calculated values of 1436.5 and 1388.5 Da).
No obvious symptomatology was elicited when 5 nmol of the au5a peptide was injected intracranially into mice. However, using the Betta gill display assay, the peptide was active at an ED min of ϳ0.2 nmol (Table IV).
A preliminary attempt to identify the molecular target of peptide au5A has been initiated. The peptide was iodinated at the tyrosine residue; the monoiodo-derivative was active (0.3 nmol of the monoiodo-derivative suppressed gill display). Since this derivative was biologically active, radiolabeled 125 I-au5a peptide was prepared for binding assays. These experiments were technically difficult, given the hydrophobicity of this peptide and high nonspecific binding background routinely observed. No measurable specific binding could be detected when either mouse brain or fish brain membranes were used. These results are consistent with either the molecular target of peptide au5a not being in neurons, or with rapid dissociation of radiolabeled peptide from the target receptor. DISCUSSION We describe the characterization of a novel group of peptides found in Conus venoms, designated the T-superfamily of conopeptides. Eight peptides belonging to this superfamily have been identified; three were isolated from venom and biochemically characterized, and two have been chemically synthesized (p5a from C. purpurascens and the au5a from C. aulicus). It seems probable that the members of this superfamily will be pharmacologically diverse, with a variety of different molecular targets (in much the same way that members of the O-superfamily target different sites on a diverse set of voltagegated ion channels). Considering that the T-superfamily peptides identified so far fall into a size range of only 10 -17 amino acids, the four Conus species examined express a remarkable In the y axis, 14 CO 2 incorporated into the substrates is expressed as percentage of maximal incorporation at the highest substrate concentration used in the experiment. The assay used partially purified carboxylase from C. textile microsomes as described under "Materials and Methods." The data were fitted to a single-site binding model. diversity of T-superfamily peptides.
The three peptides isolated from venom differ dramatically in the extent of post-translational modification found. In contrast to p5a and au5a, the tx5a peptide, which appears to be encoded by clone Tx5.2, has an exceptionally high density of post-translational modifications. The peptide contains two ␥-carboxylated glutamate residues, one O-glycosylated threonine, one hydroxylated proline, and one brominated Trp. This is the first peptide in which these diverse modifications have been observed together, although each has been described previously in other Conus peptides. In addition, there may be an unusual proteolytic cleavage at the COOH terminus, although we cannot be absolutely certain whether this is physiological, a polymorphism, or an artifact of storage.
Using a partially purified C. textile vitamin K-dependent ␥-glutamyl carboxylase, we demonstrated the presence of a recognition signal sequence in the Ϫ1 to Ϫ20 propeptide region of the tx5a precursor (deduced from clone Tx5.2). It is noteworthy that this Tx5.2 recognition sequence, which provides a Ͼ100-fold increase in apparent affinity for the carboxylase enzyme, shows no obvious sequence homology to the only other Conus peptide recognition sequence that has been functionally demonstrated, that of conantokin-G (25). The characteristic signature of T-superfamily peptides is the presence of two pairs of cysteine residues; most T-superfamily peptides identified so far have five amino acids between the cysteines (except tx5a, which has four). For the two peptides that have been synthesized (p5a and au5a), directed synthesis of specific disulfide-bonded forms was carried out. The disulfide bonding pattern of the native peptides is Cys 1 -Cys 3 , and Cys 2 -Cys 4 . Since a conserved arrangement of cysteine residues generally implies a conserved disulfide configuration, it seems highly likely that the disulfide pattern of all T-superfamily peptides in Table II will be the same as the p5a and au5a peptides.
In this work, we have described eight different members of the T-superfamily of conotoxins; for two of these, both the cDNA clone and the actual venom peptide have been identified. Two of the peptides were isolated from venom, but corresponding clones have not yet been analyzed. For four of the peptides, an amino acid sequence can be predicted from the cDNA clone, but the extent of post-translational modification has not yet been specified. For some of these peptides, considerable posttranslational modification may very well occur. Thus, in tx5a, Glu, Thr, and Trp residues are modified to ␥-carboxyglutamate, O-glycosylated threonine, and 6-bromotryptophan, respectively. The same amino acids are present in the mature toxin region of clone Gm5.1; whether or not these will have similar modifications must be confirmed by characterizing the biologically active peptide from the venom of this species. We note that the most heavily modified peptide, tx5a, was the only one of the three T-superfamily conotoxins that was active in mice. This peptide may offer an unusual opportunity to evaluate the effects of different post-translational modifications on biological activity.
The identification of eight different T-superfamily conotoxins from our relatively small sample (five Conus species, approximately 1% of the genus) suggests that the T-superfamily will be large and diverse. These peptides are among the smallest of the multiply disulfide-bonded conotoxins, with four of the amino acids being highly conserved Cys residues. Except for the polymorphic variation in the au5a peptides, the amino acid sequences are remarkably divergent; several have very unusual distribution of amino acids (such as Gm5.1, with over 50% of the residues being Trp or Cys). We have demonstrated that the degree of post-translational modification of the small sample of FIG. 5. Purification of peptides from C. aulicus venom and comparison of natural au5a with synthetic peptides. Lyophilized C. aulicus venom was fractionated by preparative HPLC. A, the peak that eluted at 15-16 min from the preparative column was fractionated on an strong cation exchange-HPLC column by elution with a gradient of 0 -0.25 M NaCl in 10 mM phosphate 50% CH 3 CN, pH 2.5, over 100 min. B, the peak indicated by an arrow in A was applied on an analytical RP C 18 column and eluted with a gradient of 0 -90% CH 3 CN in 0.1% trifluoroacetic acid over 60 min at 1 ml/h. The broad arrow indicates au5a and the thin arrow, au5b. C-E, separate analytical runs and co-elution of natural au5a (N) and synthetic au5a with a 1-3, 2-4 Cys bonding pattern (S1). The C 18 column was eluted with a gradient of 18 -36% CH 3 CN in 0.1% trifluoroacetic acid over 30 min. F, analytical run of a combination of natural au5a and synthetic peptide with au5a sequence but a 1-4, 2-3 Cys bonding pattern (S2). The gradient used was 33-39% CH 3 CN in 0.1% trifluoroacetic acid over 30 min. peptides so far characterized from the T-superfamily also differs dramatically. Thus, there is every reason to expect many hundreds of different peptides belonging to the T-superfamily of conotoxins in Conus venoms. The work described in this report provides the defining characterization of this potentially large and diverse group of biologically active peptides.