Thermodynamics of Binding of Calcium, Magnesium, and Zinc to theN-Methyl-d-aspartate Receptor Ion Channel Peptidic Inhibitors, Conantokin-G and Conantokin-T*

The binding isotherms of the divalent metal cations, Ca2+, Mg2+, and Zn2+, to the synthetic γ-carboxyglutamic acid-containing neuroactive peptides, conantokin-G (con-G) and conantokin-T (con-T), have been determined by isothermal titration calorimetry (ITC) at 25 °C and pH 6.5. We have previously shown by potentiometric measurements that con-G contains 2–3 equivalent Ca2+ sites with an average K d value of 2800 μm. With Mg2+ as the ligand, two separate exothermic sites are obtained by ITC, one of K d = 46 μmand another of K d = 311 μm. Much tighter binding of Zn2+ is observed for these latter two sites (K d values = 0.2 μm and 1.1 μm), and a third considerably weaker binding site is observed, characterized by a K d value of 286 μm and an endothermic enthalpy of binding. con-T possesses a single exothermic tight binding site for Ca2+, Mg2+, and Zn2+, with K d values of 428 μm, 10.2 μm, and 0.5 μm, respectively. Again, in the case of con-T, a weak (K d = 410 μm) endothermic binding site is observed for Zn2+. The binding of these cations to con-G and con-T result in an increase in the α-helical content of the peptides. However, this helix is somewhat destabilized in both cases by binding of Zn2+ to its weakest site. Since the differences observed in binding affinities of these three cations to the peptides are substantially greater than their comparative K d values to malonate, we conclude that the structure of the peptide and, most likely, the steric and geometric properties imposed on the cation site as a result of peptide folding greatly influence the strength of the interaction of cations with con-G and con-T. Further, since the Zn2+ concentrations released in the synaptic cleft during excitatory synaptic activity are sufficiently high relative to the K d of Zn2+ for con-G and con-T, this cation along with Mg2+, are most likely the most significant metal ion ligands of these peptides in neuronal cells.

The N-methyl-D-aspartate (NMDA) 1 subtype of glutamate receptor is a ligand-gated ion channel that displays high permeability for Ca 2ϩ . The marked excitotoxicity of glutamate is generally regarded as ascribable to its persistent interaction with the NMDA receptor (NMDAR), resulting in the establishment of neurodegenerative glutamatergic loops defined by uncontrolled elevations of intracellular Ca 2ϩ , followed by cell lysis and death. Because the Ca 2ϩ -mediated neuronal cell death, which is attendant to both acute (e.g. ischemia) and chronic (e.g. epilepsy, Parkinson's disease) neurodegenerative disorders, can be ameliorated by antagonists specific for the NMDAR (1)(2)(3)(4)(5), extensive biochemical characterization of drugreceptor interactions focused on this receptor is a widely studied topic.
Isolated from the venom ducts of predatory snails of the genus Conus, the conantokins-G (con-G) and -T (con-T) are potent and selective inhibitors of NMDAR function (6 -8) and are the only peptide antagonists of this receptor subtype described to date. More specifically, this antagonism derives from a noncompetitive inhibitory effect of the polyamine agonist site of the receptor (6). The physiological responses elicited following intracranial injections in mice include a sleep-like state in neonatal mice and a hyperactive response in older animals (9).
An unusual feature of con-G and con-T is their high abundance of ␥-carboxyglutamic acid (Gla) (10,11). Prior to its discovery in the conantokins, the presence of this amino acid in polypeptides and proteins had been noted only in certain bone proteins and in various blood coagulation factors. In these contexts, the ability of Gla to bind to Ca 2ϩ plays an integral role in the adoption of functional protein conformers (12). The interaction of Gla with Ca 2ϩ has been established for the conantokins, in which cases it has been demonstrated that both con-G and con-T adopt a significant degree of ␣-helicity in the presence of this cation (13). This ␣-helical induction is particularly profound in the case of con-G, which is essentially structureless in its apo form but assumes a full end-to-end ␣-helical conformation when Ca 2ϩ or Mg 2ϩ is fully bound to the peptide (13)(14)(15). In contrast, con-T manifests appreciable ␣-helicity in the apo form which increases slightly in the divalent cationbound state (13,16,17). Because the smaller population of conformers associated with the more structurally rigid metalbound forms of these peptides would be expected to afford more discriminate binding to the NMDAR site, metal complexation to these peptides may be necessary for full bioactivity. Of the many Conus peptides currently characterized, all of which appear to target neuronal or muscle cell receptors, only con-G and con-T lack intramolecular disulfide bridges. This observation lends support to the idea that conformational rigidity, either covalently or noncovalently imposed, is an important element of receptor recognition among members of the Conus-derived peptide family. In the case of the conantokins, metal ions other than Ca 2ϩ have also been shown to effect ␣-helix induction. Of these, Ca 2ϩ , Mg 2ϩ , and Zn 2ϩ are the most physiologically relevant metal cations in brain cells (14). Through CD-monitored titrations of con-G, we have found that the affinity of these divalent metal ions for the peptide is significantly greater for Mg 2ϩ and Zn 2ϩ than for Ca 2ϩ (14). Although CD-monitored titrations represent a convenient approach for assessing relative metal ion affinities, their corresponding K d values cannot be accurately determined from this method since the estimated K d values fall in the range of working peptide concentrations. This situation also complicates determination of the stoichiometry of metal binding. To gain a comprehensive quantitative assessment of these parameters, we have employed isothermal titration calorimetry (ITC) for monitoring the heat changes that accompany metal ion binding to the peptides. In addition to K d and stoichiometry, values for ⌬H and ⌬S can also be extracted from an ITC profile. With these data in hand, the nature of metal ion binding to both con-G and con-T can be more rigorously analyzed. The purpose of the current communication is to elaborate the thermodynamic properties of metal ion-conantokin binding.
Isothermal Calorimetry-The binding isotherms of Ca 2ϩ , Mg 2ϩ , and Zn 2ϩ to the conantokins and malonate were determined by ITC 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 Mes, 100 mM NaCl, pH 6.5. Peptide samples ranging in concentrations from 0.23-1.0 mM in a total volume of 1.4 ml were placed in the reaction cell. After equilibration, an appropriate concentration of CaCl 2 , MgCl 2 , or Zn(OAc) 2 (typically 30 -50ϫ higher in concentration than the peptide solution) in matching buffer was delivered at discrete intervals. The observed heat was measured after each injection. The total observed heat effects were corrected for the heat of dilution of ligand by performing control titrations in the absence of peptide. The resulting titration curves were deconvoluted for the best-fit model using the ORIGIN for ITC software package supplied by Microcal.
Circular Dichroism-CD titrations of con-G and con-T as a function of metal ion concentration were performed on an AVIV model 62DS spectrometer at 222 nm using a 0.1-cm path length cell thermostatted at 25°C. The peptides were dissolved in 10 mM Mes, 100 mM NaCl, pH 6.5, to a final concentration of 0.5 mM. Mean residue ellipticities were calculated by using a mean residue molecular mass of 133 Da for con-G and 128 Da for con-T. The fractional ␣-helical content was determined from mean residue ellipticities at 222 nm using the empirical relationship f (␣-helix) ϭ (Ϫ[] 222 Ϫ 2340)/30300 (18).
Sedimentation Equilibrium-Experiments were conducted using a Beckman XL-I analytical ultracentrifuge operating at 20°C in absorbance mode at 280 nm at rotor speeds of 45,000 and 52,000 rpm. The buffer used was 10 mM Mes, 100 mM NaCl, pH 6.5. The partial specific volume of con-T (0.72 ml/g) was calculated from its amino acid composition by assigning the Gla residue the v of glutamate (0.66 ml/g). The density of the buffer was determined to be 1.005 g/ml. Sedimentation data were analyzed using the single ideal species model included in the Beckman XL-I software. Baseline offset values were constrained to zero for all data sets. Calculated molecular weights represent the average of the results from three separate scans.

RESULTS
Calorimetric titrations of con-G, con-T, and malonate with various divalent metal cations were performed at 25°C, pH 6.5. Examples of the heat changes accompanying the binding of incremental additions of Zn 2ϩ to con-G and con-T are shown in the top panels of Fig. 1. The binding isotherm corresponding to a plot of integrated heats as a function of the molar ratio of Zn 2ϩ /peptide is displayed in the lower panels. These Zn 2ϩ titrations represent the more complicated isotherms of the Ca 2ϩ , Mg 2ϩ , and Zn 2ϩ data sets with 3 and 2 enthalpic transitions resulting from the complexation of Zn 2ϩ with con-G and con-T, respectively. The deconvolution of the data for con-G was achieved by fixing a multiple site model to an n ϭ 3, thereby allowing only K d and ⌬H values to float during the iterative nonlinear least squares minimization. For con-T, n 1 was con- strained to be equal to n 2 . The same constraint was employed in the case of Mg 2ϩ binding to con-G, wherein an apparent stoichiometry of 2 was obtained. For those titrations involving an n of approximately 1, all parameters (n, K d , ⌬H) were allowed to float during the minimization process. Excellent fits were obtained for stoichiometries corresponding to integer or near-integer values. This strongly suggests that neither metalinduced aggregation of peptide monomer nor metal-induced dissociation of an apo aggregate occurred under the conditions of the calorimetric titrations. An exception to this general observation occurred in the case of the Zn 2ϩ titration for con-T. For these data, a second acceptable model was generated corresponding to K d 1 ϭ 60 nM, n 1 ϭ 0.45; K d 2 ϭ 760 nM, n 2 ϭ 0.53; K d 3 ϭ 440 M, n 3 ϭ 0.95. Insofar as these parameters might indicate Zn 2ϩ -induced peptide aggregation, further exploration of this possibility was pursued, as described below.
In addition to the additional peptide binding sites that can be occupied by Zn 2ϩ , the thermodynamic signature of Zn 2ϩ binding to the conantokins was also unique compared with Mg 2ϩ and Ca 2ϩ in that a late endothermic transition occurs in the titration of both peptides. This is significantly more pronounced in the case of con-T (Fig. 1, Table I). As can be seen from the summary of the data in Table I, all other experimentally determined enthalpic changes for both peptides are exothermic. Also to be noted from Table I are the positive entropies attending occupation of each titratable site, with the exception of the entropy associated with the weaker of the two Mg 2ϩ sites in con-G.
Comparison of the K d values for Ca 2ϩ , Mg 2ϩ , and Zn 2ϩ for con-G and con-T shows a dramatic increase in affinity for Zn 2ϩ when compared with Mg 2ϩ which, in turn, binds to the peptides much tighter than Ca 2ϩ . For con-G, the tightest of the 3 Zn 2ϩ sites (K d ϭ 0.2 M) displays 200-fold more avid binding than the tightest of the 2 Mg 2ϩ sites (K d ϭ 46 M) and a 12,000-fold increase in affinity relative to the 2-3 eq Ca 2ϩ sites. For con-T, this discrimination is somewhat less pronounced, with the strength of the interaction for the tight Zn 2ϩ site being approximately 40-and 1000-fold greater than that associated with the Mg 2ϩ and Ca 2ϩ sites, respectively. In contrast, malonate manifests only a very modest selectivity for Zn 2ϩ as compared with Mg 2ϩ and Ca 2ϩ (2.3-and 6.5-fold, respectively). In addition, the absolute K d values that characterize the binding of these cations to malonate range from 1 to 4 orders of magnitude higher than the same K d values associated with their complexation to the conantokins (Table I). Both ⌬H and ⌬S values for metal ion binding to malonate are positive. The stoichiometry of approximately 0.6 determined for the association of Zn 2ϩ with malonate suggests that some degree of dimerization (or possibly higher order aggregation) may accompany binding of this particular metal ion to malonate.
In an attempt to address the phenomena underlying the endothermic transitions observed for con-G and con-T, CDmonitored titrations of these peptides with Zn 2ϩ were performed under conditions that paralleled those implemented in the ITC experiments. As can be seen for con-G ( Fig. 2A), an essentially linear increase in ␣-helicity attends the addition of Zn 2ϩ , up to an apparent plateau occurring at a metal/peptide ratio of 2:1 (m:m). The linear phase of this titration, with a midpoint of 1:1 (m:m), reflects extremely tight binding up to 2 eq of Zn 2ϩ , such that virtually all added metal ion exists in the peptide-bound form. The inset of Fig. 2A reveals that above a Zn 2ϩ /peptide molar ratio of 2, a small but defined decrease in ␣-helicity is induced. This phenomenon was not noted at similar metal/peptide ratios of Ca 2ϩ and Mg 2ϩ (data not shown). The CD-monitored titration of con-T with Zn 2ϩ is illustrated in Fig. 2B. High affinity binding of Zn 2ϩ , as with con-G, is seen in the early portion of the profile, while a significant decrease in ␣-helicity accompanies occupation of the second Zn 2ϩ site. Again, this decrease in ␣-helical content was not observed in similarly conducted Ca 2ϩ and Mg 2ϩ titrations. Because the CD-titratable diminutions in ␣-helical content correlate closely with the endothermic phases of the ITC profiles, it appears that occupancy of the weakest Zn 2ϩ site of both con-G and con-T effects a partial unwinding of the ␣-helix, which is considerably more pronounced for the latter peptide.
As indicated above, calorimetric analysis of the Zn 2ϩ titration profile of con-T was suggestive of a more complex model of metal binding involving peptide association. Zn 2ϩ -induced peptide dimerization (or higher order aggregation) could also account for the decrease in con-T and con-G ␣-helicity observed at higher Zn 2ϩ /peptide molar ratios. To address this issue, sedimentation equilibrium analysis of con-T was performed in both the apo state and at various Zn 2ϩ concentrations. As shown in Fig. 3, the apparent molecular weight of con-T increases from 3160 in its uncomplexed state to 3390 at a Zn 2ϩ /peptide ratio of 1:1 (m:m), and 3650 at a cation/peptide ration of 5.6:1 (m:m). Although these results imply some degree of self-association, the residuals for the fitted data do not display any systematic deviation (Fig. 3). Furthermore, the molecular weights determined at 0.65 mM and 0.9 mM con-T in the presence of 5 mM Zn 2ϩ were calculated to be 3710 and 3510, respectively, opposite to the trend expected if the peptide were undergoing self-association. DISCUSSION The multiple metal cation sites detected for con-G (Table I) are consistent with the multiple cation sites proposed from modeling by a genetic algorithm/molecular dynamics simulation method using the coordinates provided from the NMRderived structures of the Ca 2ϩ -and Mg 2ϩ -bound peptides (15,19). The NMR structure of the Mg 2ϩ -complexed form is notable for the spatial proximity of the side chain carboxylates of Gla 10 and Gla 14 , which are optimally positioned for metal ion coordination. This proximity would be unlikely in the absence of a stabilizing cationic bridge. A cluster comprised of Gla 3 , Gla 4 , and Gla 7 represents a second potential coordination site. The viability of these side chains as metal binding loci was supported by the simulated docking of Mg 2ϩ to the Mg 2ϩ -loaded NMR-derived structure using the genetic algorithm/molecular dynamics simulation method (15). This approach also identified Gla 7 as another likely site for Mg 2ϩ in con-G. ITC experiments failed to detect this putative third locale which may bind  13. The binding of Ca 2ϩ to con-G is too weak to lend itself to ITC methodology since prohibitively high concentrations of peptide would be necessary for probing such an interaction.
Mg 2ϩ too weakly to be amenable to this calorimetric approach. However, a third, albeit relatively weak site, was detected during the Zn 2ϩ titration of con-G and may reflect binding at this proposed Gla 7 site. Synthetic peptide variants containing individual Gla to Ala substitutions strongly support a model wherein the tight Mg 2ϩ site is maintained by Gla 10 and Gla 14 , with the weaker site being comprised of Gla 3 , Gla 4 , and Gla 7 (15). The orientation of Gla 10 and Gla 14 as determined from the Ca 2ϩ -bound NMR solution structure of con-T is strikingly similar to that found in con-G and, by analogy, can be considered a likely binding locus for metal ions (16). Docking of Ca 2ϩ into this structure using the aforementioned genetic algorithm/molecular dynamics simulation yielded a lowest energy structure showing the carboxylates of Gla 10 and Gla 14 as coordination site donors. A second site appears maintained by the two carboxylates of Gla 3 and the side chain amide carbonyl of Gln 6 . A second binding site was not experimentally detected for Ca 2ϩ and Mg 2ϩ , but was observed with Zn 2ϩ , a more avid ligand. Titrations with Zn 2ϩ of individual [Gla 10 Ala]con-T and [Gla 14 Ala]con-T variants suggest that the high affinity site of con-T includes these residues. 2 Previous studies have demonstrated that no detectable higher order species are present upon Ca 2ϩ loading of either con-G or con-T (13). In the presence of saturating concentrations of Mg 2ϩ , the monomeric molecularity of both peptides has also been established. 3 However, with Zn 2ϩ , the calorimetric titration of con-T yielded data for which a fit consistent with peptide dimerization could be generated. In addition, from the CD titration data, one interpretation of the unusual trend of decreasing con-T helicity coinciding with increasing Zn 2ϩ concentration could include Zn 2ϩ -induced peptide aggregation upon weak-site occupancy, which leads to helix unwinding. To test the possibility of Zn 2ϩ -induced con-T aggregation under the conditions employed in both the calorimetry and CD experiments, we opted for sedimentation equilibrium analysis of the apo and Zn 2ϩ -complexed peptide states (Fig. 3). Inspection of the residuals of the fits to the single ideal species model for uncomplexed and Zn 2ϩ -chelated con-T reveals the random distribution of points diagnostic of ideal solute behavior. Insofar as the distribution of residuals is considered a rigorous indicator of analyte ideality, the data are consistent with strictly monomeric solution behavior of con-T. However, when compared with the calculated sequence-based molecular weight of 2680, the higher apparent weight-average molecular weight of this peptide determined from sedimentation equilibrium (3160) is suggestive of association. Furthermore, the increase in apparent molecular weight in the presence of Zn 2ϩ also appears to be inconsistent with the contention of solute ideality. We contend that these disparate results can be addressed in terms of an artificially high partial specific volume of the peptide (0.72 ml/mg) that was employed in the molecular weight determination. Because the value of for Gla has not been determined, we assigned to it the glutamate value of 0.66 ml/mg in calculating the of con-T. This is almost certainly an overestimate of the for this residue when considering that the presence of an 2 S. E. Warder, unpublished data. 3 M. Prorok, unpublished data. additional carboxylate in Gla should effectively lower the associated with this residue by virtue of its increased hydrophilicity and capacity for hydration. In addition, in the presence of Zn 2ϩ the intimate and preferential interaction of the metal ion with the Gla side chains may profoundly perturb the of Gla. When considering the relatively high weight percentage of Gla in con-T, its contribution to the value of the peptide is significant. The observation that an increase in molecular weight does not attend increased peptide concentration at high Zn 2ϩ concentrations is additional compelling evidence in support of the nonassociative behavior of con-T.
The exotherms associated with peptide-metal binding correlate in the expected fashion with the degree of conformational change effected by metal binding. In the case of Mg 2ϩ complexation to con-G, wherein the peptide undergoes a dramatic change in ␣-helicity upon Mg 2ϩ binding (14), the total change in the ⌬H for binding of 2 eq of ligand is Ϫ8.7 kcal/mol. The enthalpic change for con-T, which undergoes a distinctly smaller increase in ␣-helicity upon saturation of its metal ion sites with Mg 2ϩ , is Ϫ4.4 kcal/mol. Clearly, bond formation occurs to a greater extent in the con-G metal-induced transition than in the con-T system. It seems unlikely that electrostatic components contribute to these favorable ⌬H values since metal ion binding to malonate, which occurs strictly through electrostatic interactions, possesses an unfavorable ⌬H (Table   I), leaving entropic forces to drive the binding event. The effective neutralization of clustered, destabilizing negative charge that occurs upon chelation of metal ions to the Gla head groups of the peptide may simply allow favorable intrapeptide hydrogen bonds to prevail.
The positive, albeit small, entropies that accompany the exothermic transitions are somewhat surprising considering the significant degree of order that metal ion binding imposes on these peptides. Factors with favorable entropies, such as release of the metal ion from its coordination sphere and release of structured water surrounding polar and non-polar residues, are likely contributing in the peptide systems. From the positive T⌬S values for metal complexation to malonate, it is probable that such contributions take entropic precedence, despite the absence of some elements of rotational bond freedom.
The positive ⌬H values observed for occupancy by Zn 2ϩ of the weak sites of both con-G and con-T correlate with a small degree of ␣-helix unwinding as implied from parallel CD-monitored titrations (Fig. 2). The relatively large positive entropy values linked to weak site Zn 2ϩ binding for both peptides is also consistent with the notion of increased structural disorder. The nature of this destabilization in con-G is difficult to state with certainty because the results of simulated Mg 2ϩ docking to con-G (15) point toward Gla 7 as a third binding locus. Because all five Gla residues of con-G exist on the same face of the helix FIG. 3. Representative sedimentation equilibrium scans and calculated fits for apparent molecular weight for con-T (bottom panels). Experiments were performed at a rotor speed of 45,000 rpm at a final peptide concentration of 0.5 mM. Distribution of residuals for the indicated fits is shown in the top panels. A, con-T in the absence of metal cations. B, con-T at 0.5 mM Zn(OAc) 2 . C, con-T at 2.8 mM Zn(OAc) 2 . The buffer was 10 mM Mes, 100 mM NaCl, pH 6.5, 20°C. in the metal-loaded structure (15), inclusion of an additional divalent cation would be expected to impart increased stabilization to the peptide. We expect that ITC titrations with Zn 2ϩ employing various Gla to Ala variants of con-G will address the nature of this third site and these studies are currently in progress. For con-T, we contend that the greater degree of ␣-helix unwinding that occurs upon Zn 2ϩ loading can be explained in terms of the capping potential of Gla 4 . The NMR solution structures of both apo-and Ca 2ϩ -loaded con-T indicate that the side chain of Gla 4 is an apparent capping residue in both structures. This contention is supported by CD data on the [Gla 4 Ala]con-T variant which displays compromised ␣-helical content in both its apo-and Ca 2ϩ -loaded forms (17). In the presence of high Zn 2ϩ concentrations, we suggest that Gla 4 is recruited from its side-chain to main-chain interaction role to share with Gla 3 in the maintenance of a weak metal-binding site, thus destabilizing the ␣-helix in the N-terminal vicinity.
Finally, the current study raises important issues concerning the identity of the metal ion that predominates in the bioactive conformation of these peptides. It has been established in numerous studies with Gla-containing bone and blood proteins that Ca 2ϩ mediates their functional aspects. In the case of prothrombin, factor IX, and protein C, which contain 10, 12, and 9 Gla residues, respectively, in their N-terminal domains, no major preference for Mg 2ϩ versus Ca 2ϩ has been noted from intrinsic fluorescence titration experiments (20 -23). Only a slight (2.5-fold) preference is observed for Mg 2ϩ over Ca 2ϩ in terms of their K d values for malonate (Table I). Despite the smaller ionic radius for Mg 2ϩ (86 pm) as compared with Ca 2ϩ (114 pm), this disparate charge to radius ratio fails to impart appreciable selectivity differences in these cases. However, for con-G, the Mg 2ϩ tight site and weak site affinities in comparison with Ca 2ϩ are 175-and 9-fold, greater, respectively. For con-T, Mg 2ϩ displays a 42-fold increase in binding strength compared with Ca 2ϩ . Both con-G and con-T display distinctly lower K d values for Zn 2ϩ than for either Ca 2ϩ or Mg 2ϩ . A pronounced (ca. 10,000-fold) greater affinity for Zn 2ϩ over Ca 2ϩ has recently been reported for a Gla-containing de novo designed peptide, which undergoes a metal-induced helical transition (24). These investigators maintain that the greater covalent character of electrostatic interactions involving Zn 2ϩ , ascribable to its filled 3d orbitals as well as higher charge density for this cation, are the major contributors to this preferential binding. However, the primacy of these factors is not borne out by our malonate data, which indicate that Zn 2ϩ manifests only a modest (6.5-fold) increase in affinity for malonate compared with Ca 2ϩ . This would suggest that the binding sites of those peptides, which avidly bind Zn 2ϩ , are particularly amenable to this smaller metal ion simply because their induced binding site cation pockets are intrinsically small. Hence, the strength of metal-ion binding in these molecules may rely more on proper steric and geometric fit than on purely electrostatic considerations.
The consideration that Zn 2ϩ is the dominant metal ion effector of conantokin function is supported by studies that indicate that transiently high local concentrations of Zn 2ϩ (ca. 300 M) can be attained in the synaptic cleft during excitatory synaptic activity (25,26). Intracerebral injections of the conantokins into pre-2-week-old mice have been shown to induce a prolonged sleep-like state in these subjects at an estimated final brain concentration of 15 M (13). Assuming a free Zn 2ϩ concentration of 300 M, this would result in essentially complete occupancy of the two tight Zn 2ϩ sites in con-G and approximately 50% occupancy of the weak site. In the case of con-T, saturation of the high affinity Zn 2ϩ site and 40% occupancy of the low affinity Zn 2ϩ site would be attained at these peptide and ligand concentrations. At a cerebrospinal fluid Mg 2ϩ concentration of 1.5 mM, tight and weak site loading of con-G would be 97% and 80%, respectively. For con-T, essentially 100% of the peptide molecules would be complexed with Mg 2ϩ . At a cerebrospinal fluid Ca 2ϩ concentration of 1.5 mM, Ca 2ϩ loading would be considerably less than Mg 2ϩ and Zn 2ϩ , with 30% and 77% occupancy of the metal sites in con-G and con-T, respectively. We therefore propose that Mg 2ϩ , as well as Zn 2ϩ , are more plausible candidates for the role of in vivo metal ions for the conantokins. However, it should be noted that the high coordination numbers and irregular coordination geometry displayed by Ca 2ϩ confer upon Ca 2ϩ -bound proteins the ability to bridge phospholipid membranes and/or other proteins. This particular feature of Ca 2ϩ -complexed biomolecules may mediate the functionality of the conantokins, despite the compromised binding capacity displayed by Ca 2ϩ compared with Mg 2ϩ and Zn 2ϩ . Dissecting the roles of these metal ions in the conantokins with respect to the NMDAR is complicated by their involvement in receptor function, namely the Ca 2ϩ permeability associated with the receptor (see above), voltage-dependent Mg 2ϩ blockage of the receptor (27,28), and allosteric voltageinsensitive and -sensitive blocks by Zn 2ϩ (29,30). Elucidating the precise role of these metal ions in terms of conantokin antagonism of the receptor is essential for a full pharmacological profile of the these peptides and represents one of the major investigative challenges associated with future work in this area.