Sequence Requirements for theN-Methyl-d-aspartate Receptor Antagonist Activity of Conantokin-R*

Conantokin-R (con-R), a γ-carboxyglutamate-containing 27-residue peptide, is a natural peptide inhibitor of the N-methyl-d-aspartate (NMDA) subtype glutamate receptor. Synthetic analogs of con-R were generated to evaluate the importance of the individual structural elements of this peptide in its NMDA receptor antagonist activity, measured by inhibition of the spermine-enhanced binding of the NMDA receptor-specific channel blocker, [3H]MK-801, to rat brain membranes. Progressive C-terminal truncations of the 27-residue peptide revealed stages of severe activity loss. These occurred at con-R[1–11] and con-R[1–7], corresponding to the deletions of Leu12–Pro27 and Met8–Pro27 respectively. A second set of analogs featured single Ala substitutions in the fully active con-R[1–17] fragment. The replacement of Met8 and Leu12 by Ala resulted in approximate 20- and 55-fold decreases of inhibitor potency, respectively. In addition to these two residues, the only other positions where a single Ala substitution led to substantial losses (from 11-fold to >1000-fold) of activity were those of the first five N-terminal amino acids. Based on the above findings, the binding epitope of con-R was localized to the N-terminal turn of the helix and other residues on one face along two subsequent turns. This contribution pattern of the side chains in activity closely resembles the results obtained with another member of this peptide family, conantokin-T. The secondary structure and metal ion binding properties of the con-R variants were also evaluated using circular dichroism spectroscopy. Divalent cation-dependent increases of α-helix content were observed in most analogs. However, analogs with replacement of Gla11 and Gla15, as well as truncation fragments shorter than 15 residues, lost the ability to be stabilized by metal ions. These results confirmed the location of the primary divalent cation binding locus at Gla11 and Gla15. Additional interactions were indicated by the reduced α-helix stability in the Ala analogs of Gla4, Lys7, and Arg14.

The conantokins are a family of neuroactive peptides isolated from venoms of various species of marine snails of the genus Conus. The three well characterized members of this group include con-G, 1 from Conus geographus (1), con-R, from Conus radiatus (2), and con-T, from Conus tulipa (3). The conantokins posses a high content of Gla, a post-translational modification that has also been observed in several other Conus-derived venom peptides (4 -9). The presence of Gla was first reported in vertebrate blood and bone proteins (10,11), where it arises from the ␥-carboxylation of glutamic acid by a vitamin K-dependent process (12). The responsible enzyme, a vitamin K-dependent ␥-carboxylase, has been purified from bovine liver (13), and its cDNA has been cloned, sequenced, and expressed (14 -16). A similar vitamin K-dependent enzyme activity was found in the microsomal fraction of C. radiatus venom ducts (17). The cloning of the cDNA for con-G allowed examination of the sequence requirements of the Conus-derived carboxylase, which showed substantial differences between the carboxylation signal sequence specificities of mammalian and Conus carboxylases, despite the similar cofactor requirements (18).
con-R contains 27 amino acids, exceeding by 10 and 6 residues the sizes of con-G and con-T, respectively. Other features that distinguish this peptide from con-G and con-T include a single amino acid insertion at Ala 10 and a five-member disulfide loop near the C terminus. con-R also possesses a free ␣-carboxylate at its C terminus, unlike con-G and con-T, both of which feature amidated C termini. Several amino acids are rigidly conserved among the conantokins. These include the N-terminal Gly-Glu sequence, four Gla residues, and an Arg residue preceding the Gla closest to the C terminus of the peptides. In addition, Lys 7 , Met 8 , Leu 12 , and Lys 19 are present in both con-R and con-T. It is also relevant that a recently obtained clone of the con-G cDNA coded for Val 5 , which is the same residue as in con-R (18).
In addition to the occurrence of the post-translationally modified Gla, an additional remarkable structural feature of the conantokins is their stable end-to-end ␣-helical secondary structure (19 -21). In con-G, the ␣-helical conformation is dependent on the presence of divalent metal ions (22,23), whereas con-T is ␣-helical in the absence of cations (24). The primary metal ion binding site has been localized to Gla 10 and Gla 14 in both con-G and con-T (23,25,26). Metal ion chelation by these residues leads to a strong stabilization of the ␣-helical fold. The examination of the secondary structure and cation binding properties of con-R revealed that residues 1-20 form a moderately stable ␣-helix, whereas the C-terminal disulfide loop segment is in a nonhelical conformation (27). Additionally, full metal ion binding capacity of this peptide was retained by a 17-residue N-terminal fragment (27).
The common receptor target of these peptides is the NMDAR (3,28). Evidence in support of NMDAR antagonism by the conantokins was obtained by both electrophysiological (2,29,30) and pharmacological (31) approaches. Structure-activity relationships of con-G (32-34) and con-T (25,35) have been examined, and critical residues were identified in each peptide. In this communication, a similar analysis of con-R is reported. Chemical modifications were utilized to evaluate the functional importance of structural elements in con-R. Several functionally important residues were identified, and this information was compared with the results that are available on the other two peptides.

EXPERIMENTAL PROCEDURES
Peptide Synthesis, Purification, and Characterization-Syntheses of con-R (NH 2 -GE␥␥V 5 AKMAA 10 ␥LAR␥ 15 NIAKG 20 CKVNC 25 YP-COOH) and its analogs were performed by automated solid phase peptide synthesis on an Applied Biosystems (Foster City, CA) model 433A peptide synthesizer (19). The Fmoc protection strategy was used with the following side chain protecting groups: tert-butyl ester for Gla, Glu, Tyr; trityl for Asn, Cys; 2,2,5,7,8-pentamethyl-chroman-6-sulfonyl for Arg; and butyloxycarbonyl for Lys. The derivatized Gla was synthesized as described previously (36). con-R, containing a C-terminal free acid, was synthesized on a preloaded H-Pro-2-chlorotrityl resin support (No-vaBiochem, San Diego, CA), as described (27). All of the truncated variants were synthesized as the peptide amides, using 5-(4-Fmocaminomethyl-3,5-dimethoxyphenoxy)-valeric acid-p-methylbenzylhydrylamine resin support (PerSeptive Biosystems, Framingham, MA). Peptides were cleaved with 88% trifluoroacetic acid in the presence of the radical scavengers dithiothreitol and triisopropylsilane. Peptide purification was typically achieved by anion exchange chromatography on an fast protein liquid chromatography Bioscale DEAE20 (Bio-Rad) column, followed by desalting on a Sephadex G-15 column (1.5 ϫ 100 cm) that was equilibrated and eluted with 0.1% NH 4 OH. The peptides were characterized by reverse-phase high performance liquid chromatography and matrix-assisted laser desorption ionization mass spectrometry, as described (19). Concentrations of peptide stock solutions were determined by quantitative amino acid analysis. In Vivo Mouse Injection Assays-Two age groups of mice were injected with peptide or buffer intracranially as previously described (19). Mice Circular Dichroism Spectroscopy-CD spectra of con-R analogs were collected at 25°C in 10 mM sodium borate, 100 mM NaCl, pH 6.5, on an Aviv (Lakewood, NJ) model 200SF spectrometer using a 1-cm-pathlength cell. The peptide solutions and buffers used were treated with Chelex-100 resin prior to the experiments to ensure that they contained no multivalent metal ions. CD spectra of the peptides were acquired in 100% aqueous buffer with the above composition or with the addition of 20 mM CaCl 2 , 1.5 mM MgCl 2 , or 1.5 mM ZnCl 2 . The peptide concentration was 35 M. The ␣-helical content was determined from mean residue ellipticities at 222 nm using the empirical relationship, f␣ ϭ (Ϫ[⌰] 222 -2340)/30,300 (37).
NMR Studies-Peptide samples (ϳ2 mM) were dissolved in a solution containing 10 mM sodium borate, 100 mM NaCl, 20 mM MgCl 2 , pH 6.5, 10% 2 H 2 O. Two-dimensional total correlation spectroscopy (spin-lock interval, 60 ms) and nuclear Overhauser enhancement spectroscopy (mixing time, 150 ms) spectra were recorded at 5°C at a spectral window of 6200 Hz on a three channel Varian UnityPlus spectrometer (599.89 MHz, 1 H) employing standard pulse sequences (38,39) using the hypercomplex phase sensitive method (39). The H 2 O signal was suppressed by presaturation. Relaxation delays of 1.6 and 2.8 s (including presaturation) were used in the total correlation spectroscopy and nuclear Overhauser enhancement spectroscopy experiments, respectively. Time domain data (t 2 and t 1 ) were recorded as 2048 ϫ 256 complex matrices with 16 and 56 scans per t 1 increment for the total correlation spectroscopy and nuclear Overhauser enhancement spectroscopy spectra, respectively.
Linear prediction to the 512 complex data points was applied to the fids in the t 1 domain, and zero filling in both t 2 and t 1 domains was used (40). Final 4K ϫ 2K complex time domain data sets were Fourier transformed with Gaussian apodization functions applied in both dimensions. A first order base-line correction was applied in F 2 domain. All spectra were processed identically using Varian software that accompanied the spectrometer. The spectra were referenced relative to the signal of DSS (␦ ϭ 0 ppm).
Isothermal Titration Calorimetry Analyses-The binding isotherms of Mg 2ϩ to the con-R variants were determined by measurements of the heat changes accompanying titration of the metal ions into solutions of the relevant sample. The titrations were performed with an OMEGA titration calorimeter (Microcal Inc., Northhampton, MA) at 25°C in a buffer containing 10 mM Na-Mes, 100 mM NaCl, pH 6.5. The exact protocols used as well as the methods for data analyses have been published previously.

RESULTS
To elucidate the detailed structure-activity relationships in the con-R sequence, a number of analogs were synthesized. Progressive C-terminal truncations were introduced in a series of variants. In addition, in a second set of con-R variants, individual side chain residues were replaced with Ala. Members of this latter group were 17 amino acids in length, based on the fully active con-R [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17] sequence. The NMDAR antagonist activity of each analog was determined by the [ 3 H]MK-801 binding assay ( Fig. 1) and compared with the potency of wild-type con-R.
The in vivo effects of con-R variations were also tested by intracranial injections of mice with peptide solutions. The 15-g dose of full-length con-R used for the 10-day of age group resulted in a sleep-like inactive state within 15 min post-injection, which lasted 6 -8 h. Similar effects were seen with the positive control group, which was injected with the same concentration of con-T. To a second, 30-day of age group, a quantity of 45 g of con-R was administered. These mice also exhibited a sleep-like condition for 2-6 h, but unlike the age 10-day group, these animals awakened periodically and showed considerable disorientation, impaired movement, e.g., dragging hind legs, and vocalized repeatedly. At 6 -8 h post-injection, members of this group showed some signs of hyperactivity, such as constant walking and cage climbing. The 30-day-old positive control group injected with con-T underwent a shorter inactive sleep period (Ͻ1 h) and exhibited disoriented/hyperactive behavior 2-8 h after injection. Animals from both age groups appeared normal in behavior after 24 h. The mouse intracranial injection experiments were also performed with con-R [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17]. The 10-day-old mice receiving the same dose of con-R [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17] as wild-type con-R and con-T slept for 2.5-4 h, a significantly shorter period than the 6 -8 h observed for the other two peptides. The 30-day-old animals exhibited an impaired state for up to 2 h, after which normal activity was resumed, with no signs of disorientation or hyperactivity. Again, in this case, although similar syndromes were observed as in the groups receiving an equivalent dose of con-R or con-T, the duration of the effects was shorter.
The overall conformation of each analog was examined by CD spectroscopy in the absence of multivalent cations and in the presence of Ca 2ϩ or Mg 2ϩ , respectively. These secondary structure observations revealed that con-R-derived peptides of 15 amino acids or longer displayed the cation-dependent increase of ␣-helical character. Truncation analogs shorter than this length had progressively less ␣-helical content in the apoform and a slight decrease of ␣-helicity upon loading with high concentrations of either Ca 2ϩ or Mg 2ϩ . Fig. 2 illustrates the CD spectra of con-R[1-15] and con-R [1][2][3][4][5][6][7][8][9][10][11][12][13][14], in both the absence of metal ions and in the presence of Mg 2ϩ . The above described differences in cation-dependent conformational transitions are clearly observed in these two analogs. Additionally, the ␣-helicity of the apo-peptide is higher for the shorter analog. This can be explained by the deletion of Gla 15 and, therefore, the elimination of charge repulsion between Gla 11 and Gla 15 in the absence of complex forming cations. In the Ala replacement  series, some peptides showed altered conformational properties. Analogs of Gla 4 , Lys 7 , and Arg 14 had lower ␣-helicity, both in the apo-and the metal ion-bound states. Similarly, the Lys 7 3 Ala modification resulted in reduced overall ␣-helicity. The double replacement of Gla 11 and Gla 15 increased the apo ␣-helix content, with Ala being more ␣-helix promoting than Glu. However, metal ion binding resulted in a decrease rather than an increase of ␣-helical conformation for these analogs.
To further investigate whether possible conformational alterations influenced the activity of these variant peptides, 1 H NMR analyses were performed on three variants of con-R [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17] in their Mg 2ϩ -bound forms, viz. con-R17[E 2 A], con-R17[K 7 A], and con-R17[M 8 A], that showed a nearly full loss of NMDAR inhibitory activity, no loss in activity, and partial loss of activity, respectively. The ␣H chemical shift index values of these three variants, relative to those of wild-type con-R [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17], are illustrated in Fig. 3. Only minor relative changes are noted for these peptides up to residue Gla 11 , which for con-R17[E 2 A] and con-R17[K 7 A] displays relatively large differences from wild-type con-R [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17]. However, this residue shows anomalous conformational behavior in con-G (23), con-R (26), and con-T (24) and appears to be greatly influenced by metal binding. In any case, the largest changes at this location, which are also in parallel, are in the two variants, con-R17[E 2 A] and con-R17[K 7 A]. Because opposite effects were observed on the activity of these variants, this alteration does not appear to be meaningful in terms of NMDAR inhibitory activity. To determine whether the changes in chemical shift index indices at Gla 11 for these latter two variants indeed did not reflect macroscopic alterations in metal binding to the variant peptides, isothermal titration calorimetry analysis of the binding of Mg 2ϩ to both the Glu 2 3 Ala and Lys 7 3 Ala variants were conducted. The K d values obtained were 14 and 11 M, respectively (data not shown). These values are comparable with the K d of 18 M previously obtained for Mg 2ϩ /con-R17 binding (27), and indicate that the Mg 2ϩ binding affinities are similar for the three peptides. Similarly, other deviations in the chemical shift index values at residues that are near the C terminus of the peptide occur to the same degree with these same two peptides. This also indicates that such changes are not important with regard to the NMDAR activity of these peptides. DISCUSSION con-R is the most recently isolated member of the conantokin class of neuroactive conopeptides, and it is perhaps the most interesting example of these peptides, based on the modest amount of information that is available on this peptide at present. con-R is at least 4-fold more potent in the in vitro [ 3 H]MK-801 binding NMDAR assay than either con-G or con-T. It is also significant that this peptide shows some selectivity to NMDAR containing the NR2B and NR2A subunits (2). In addition to the subunit selectivity and perhaps as a consequence, con-R also shows high potency, as well as a favorable protective index against behavioral toxicity in anticonvulsant models when compared with NMDA inhibitors with different mechanisms, e.g. ifenprodil and MK-801 (2).
The primary structure of con-R also possesses features that are unique in its class, such as the C-terminal extension that contains a single disulfide loop. A conformational analysis of full-length con-R, along with two variants lacking the disulfide region, revealed that this latter segment is a major factor in governing the conformational behavior of the peptide (27). Although the receptor antagonist and metal ion binding properties of the analogs lacking the disulfide bridge were comparable with the full-length peptide, the secondary structure of the variants was shown to consist of a more stable ␣-helix. This is due to the elimination of ␣-helix breaker sequence elements, such as Gly 20 and the (i, iϩ4) disulfide linkage. In full-length con-R, this structure perpetuates the secondary structure disruption to the N-terminal region, which otherwise possesses a higher intrinsic ␣-helix propensity (27).
This contribution of the side chains has some similarities to the results obtained with con-G and con-T variants. The Nterminal turn of the ␣-helix and residues on one face along two subsequent turns were found to be important contributors to the receptor antagonist activity of the latter peptides. In con-R, replacement of Gly 1 , Glu 2 , Gla 3 , or Gla 4 by Ala resulted in a 10-fold to Ͼ1000-fold diminution in their activities. This segment of the sequence is 100% conserved among the three conantokins. Analyses of the structure-activity relationships of both con-G (32-34) and con-T (25,35) showed that the Nterminal four amino acids are major contributors to the NMDAR inhibitory activity of both peptides. These four residues are equally crucial in con-R. Three additional analogs, Val 5 3 Ala, Met 8 3 Ala, and Leu 12 3 Ala, suffered 20 -60-fold reductions in inhibitor potencies. Some of these side chains probably contribute the indicated binding energies to ligandreceptor interactions. Others may be involved in effecting a productive binding by inducing a conformational change in the receptor that is necessary for inhibition.
Lys 7 is identical to the amino acid found in con-T at this position. In contrast, con-G has a Gla in the analogous position. The Lys 7 3 Gla modification of con-T resulted in a peptide with diminished apo ␣-helicity (26). This showed that the radically different side chains in the two homologous sequences confer significantly disparate conformational behavior. This residue does not seem to be involved in interactions, because both Ala and Lys variants retain full activity in con-G, as does the con-R17[K 7 A] analog of con-R. Therefore, it is expected that, as in con-T, Lys 7 in con-R serves to stabilize the ␣-helical conformation of the peptide. Met 8 of con-R is again identical to this residue of con-T. An Ala replacement of the latter caused a 16-fold activity decrease. 2 In contrast, the corresponding analog of con-R had an approximate 57-fold potency loss. Despite this difference, Met 8 seems to play an important role in the function of both. Residue 9 is not conserved among the three conantokins. Gln 9 is required for the NMDAR inhibitory activity of con-G, whereas Leu 9 of con-T can be replaced by Ala without loss of potency. Ala 9 in con-R most likely occupies a nonimportant position in the sequence. Ala 10 constitutes a single amino acid insertion, considering that Gla 11 and Gla 15 correspond to Gla 10 and Gla 14 in the other two conantokins. Gla 11 can be replaced by Ala in all three of the conantokins without consequence to NMDAR antagonism. However, the tight metal ion binding site is compromised after this alteration. Ile 12 in con-G was found to be required for the NMDAR antagonist activity of that peptide (34). Leu 12 of con-T also contributes in this regard, although to a lesser extent. 2 Leu 12 is similarly involved in the inhibition of the NMDAR by con-R, despite the altered side chain and the metal ion binding site compared with the other two peptides. Because of the insertion at Ala 10 , Ala 13 has no obvious counterpart in the other two sequences and is probably not important for activity. Arg 14 is analogous to Arg 13 found in both con-G and con-T. An Ala replacement of this residue had a minor effect on NMDAR inhibition in con-G and was without consequence in con-T. Likewise, no adverse effect of that modification resulted in con-R. This conserved side chain plays a role in the secondary structure stabilization in each molecule, because Ala substitutions decrease the ␣-helix content of each. Because the fragment con-R [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15] displays full receptor antagonist potency, further modifications beyond residue 15 were not made. Upstream of Gla 15 , con-R has no homology with con-G and con-T, except for Lys 19 , also found in con-T.
The deletion variant con-R16[⌬A 10 ] displayed a 28-fold decrease in inhibitory activity. The modification in this peptide caused the metal ion binding site formed by Gla 11 and Gla 15 to shift upstream by one position. This is identical to the positions of these residues in both con-G and con-T. The metal ioninduced conformational change in this variant is similar to that in con-R [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17], which indicates that the cation site is still functional. The likely reason for the reduced activity is the shifting of Leu 12 , an important contributor to activity, to sequence position 11. The position change of this side chain results in a spatial displacement and also a change in orientation, viz. a 100°turn in relation to the ␣-helix register, which may prevent a positive interaction between the peptide and the receptor.
There are several Ala residues in the sequence of con-R, and their locations are of interest in light of the Ala scan results of the three peptides. Ala 6 , Ala 9 , Ala 10 , and Ala 13 appear to be positioned where the different residues found in one or both of the other conantokin sequences could be substituted by Ala without functional consequences.
The structure-activity relationships of con-R are more similar to those of con-T than to con-G (Fig. 4). In fact, there is just one amino acid of importance in con-R and con-T that is not identical. Even this residue occupies the same position, viz. sequence number 5, which is a Val in con-R and a Tyr in con-T. The shortest segment of con-R with measurable activity was con-R [1][2][3][4][5][6][7][8]. In con-G, residue segment (1-13) is the shortest fragment that displays detectable activity (34). con-T is more similar to con-R in this respect as well, because its residue (1-8) fragment is still active, but the (1-6) fragment is inactive. 2 These functional similarities are underscored by the extensive primary structure homology that exists between con-T and the first 20 residues of con-R. There is 50% identity be- tween these two sequences, after accounting for the insertion at Ala 10 in con-R.
The results of the in vivo mouse injection experiments reveal distinct differences for con-R [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17] compared with full-length con-R. Both the 10-day-old and the age 30-day-old groups exhibited comparable syndromes for the two peptides, nevertheless the effects lasted for 2-4 h with con-R[1-17] and 6 -8 h with con-R. Considering that the in vitro NMDAR antagonist potencies of the two compounds are the same, their different bioavailability/turnover is the likely reason for these differences. Although the C-terminal disulfide loop segment is without apparent secondary structure (27), this motif may convey properties that increase the half-life of the peptide in vivo.
In light of structural studies of con-G and con-T (20, 21, 23-25), the secondary structure results for con-R analogs show that a commonality exists. The divalent cation-dependent increases of ␣-helix content were observed in all of the analogs that contained both Gla 11 and Gla 15 . The analogs with double replacement of these residues, as well as the truncation fragments that were shorter than 15 residues, did not display this phenomenon. This is the direct consequence of removing the helix stabilizing (i, iϩ4) metal ion binding site. Additional interactions were indicated by the helix stability changes in the Ala analogs of Lys 7 and Arg 14 . Both of these peptides had a decreased secondary structure. These positive residues are placed (i, iϩ3) from Gla residues, which can provide electrostatic stabilization between those side chains. The Ala analog of Gla 4 was also destabilized significantly. The NMR-derived solution structure of con-T (24) indicated that the side chain of Gla 4 is involved in N-terminal capping interactions with both the terminal amine and the backbone amide nitrogens in the first residues of the ␣-helix. It is very likely that Gla 4 in con-R fulfills a similar function.
In conclusion, the results presented herein on the structureactivity relationships of con-R provided additional insight into the NMDAR antagonism of this peptide. The data also allowed comparisons to be made between con-R and the other two well characterized conantokins. The three peptides have similar mechanisms of action at the NMDAR, although subtle differences are also noted. con-R is more closely related to con-T than to con-G, with regard to both the secondary structure properties and the receptor antagonism.