JBIC Journal of Biological Inorganic Chemistry

, Volume 15, Issue 5, pp 667–675

Metal ion determinants of conantokin dimerization as revealed in the X-ray crystallographic structure of the Cd2+/Mg2+–con-T[K7γ] complex


  • Sara E. Cnudde
    • Department of ChemistryMichigan State University
  • Mary Prorok
    • Department of Chemistry and Biochemistry, W.M. Keck Center for Transgene ResearchUniversity of Notre Dame
  • Francis J. Castellino
    • Department of Chemistry and Biochemistry, W.M. Keck Center for Transgene ResearchUniversity of Notre Dame
    • Department of ChemistryMichigan State University
Original Paper

DOI: 10.1007/s00775-010-0633-2

Cite this article as:
Cnudde, S.E., Prorok, M., Castellino, F.J. et al. J Biol Inorg Chem (2010) 15: 667. doi:10.1007/s00775-010-0633-2


Predatory sea snails from the Conus family produce a variety of venomous small helical peptides called conantokins that are rich in γ-carboxyglutamic acid (Gla) residues. As potent and selective antagonists of the N-methyl-d-aspartate receptor, these peptides are potential therapeutic agents for a variety of neurological conditions. The two most studied members of this family of peptides are con-G and con-T. Con-G has Gla residues at sequence positions 3, 4, 7, 10, and 14, and requires divalent cation binding to adopt a helical conformation. Although both Ca2+ and Mg2+ can fulfill this role, Ca2+ induces dimerization of con-G, whereas the Mg2+-complexed peptide remains monomeric. A variant of con-T, con-T[K7γ] (γ is Gla), contains Gla residues at the same five positions as in con-G and behaves very similarly with respect to metal ion binding and dimerization; each peptide binds two Ca2+ ions and two Mg2+ ions per helix. To understand the difference in metal ion selectivity, affinity, and the dependence on Ca2+ for dimer formation, we report here the structure of the monomeric Cd2+/Mg2+–con-T[K7γ] complex, and, by comparison with the previously published con-T[K7γ]/Ca2+ dimer structure, we suggest explanations for both metal ion binding site specificity and metal-ion-dependent dimerization.


CrystallographyConantokinsγ-Carboxyglutamic acidPeptide α-helix









N-Methyl-d-aspartate receptor


The conantokins are a small family of γ-carboxyglutamate (Gla)-containing Conus peptides that act as potent and specific antagonists of the N-methyl-d-aspartate receptor (NMDAR). The most well-characterized members of this group include conantokin-G (con-G) and conantokin-T (con-T), which are 17 and 21 amino acids in length, respectively [1, 2]. The peptides share sequence identity within their first four residues and contain Gla residues at positions 3, 4, 10, and 14. Despite high overall homology, a conspicuous primary structural difference between the peptides occurs at sequence position 7. A Gla residue in con-G and a Lys residue in con-T occupy this locus. Their high relative contents of Gla residues permit con-G and con-T to interact with a variety of metal cations, which in turn contribute to the stabilization of their α-helical conformations. However, in the absence of metal cations, con-G adopts a random-coil structure devoid of secondary structural elements, whereas con-T assumes a high degree of helicity [3, 4]. These different secondary structural tendencies primarily derive from the nature of the residue at position 7, which provides for a destabilizing array of Gla residues spaced at i, i + 4, i + 7 intervals in con-G (positions 3, 7, and 10) in contrast to a stabilizing Gla3, Lys7, Gla10 arrangement of the corresponding residues in con-T. More recently, a series of solution studies revealed that the ability of con-G to undergo Ca2+-mediated antiparallel dimerization, a property not shared by con-T, is ascribable to the presence of Gla7 [5]. This self-association has been demonstrated to rely on optimal intra- and interpeptide Gla–Ca2+ contacts that require an i, i + 4, i + 7, i + 11 motif of Gla spacing within the monomer. In support of this model, it has been shown that replacement of Lys7 in con-T with a Gla residue results in a peptide that can form an antiparallel dimer in the presence of Ca2+ [6]. High-resolution crystal structures of the Ca2+-bound con-G and con-T[K7γ] (where γ is Gla) antiparallel dimers have been determined and confirm the predictions arising from the solution studies [7]. Though the helices of con-G and con-T[K7γ] are virtually identical in these structures, as are the Ca2+ binding sites, the dimer interface is completely different. Paradoxically, both peptides have higher affinity for Mg2+ than for Ca2+ [8] but neither dimerizes in the presence of Mg2+. Also, we have previously demonstrated that both the monomeric and the dimeric forms of the conantokins (as shown for con-G) display NMDAR activity, albeit with different kinetic signatures for each of the four NMDAR subunit combinations tested [9]. In an effort to understand the apparent incongruity between metal affinity and dimerization, and to ascribe a molecular basis for the different NMDAR activity profiles of the monomer and dimer forms of the peptide toxin, a comparison of conantokin structures in monomeric and dimeric metal-bound states is required. Furthermore, insight into the interactions that prevail in the metal ion triggered self-assembly of peptide helices has applications across the biochemical spectrum, including metal sensoring and protein design [1013]. With these objectives in mind, we have determined the X-ray crystallographic structure of the Cd2+/Mg2+–con-T[K7γ] complex, the first crystal structure of a monomeric conantokin, which is the physiologically relevant oligomerization state for con-T. The details of this structure and the salient differences that exist in the Ca2+-coordinated versus the Mg2+-coordinated species, with emphasis on the disposition of Gla residues, are reported herein.

Materials and methods

Peptide synthesis, purification, and characterization

Con-T[K7γ] was synthesized, purified, and characterized as previously described [6]. The amino acid sequences of the peptides relevant to this communication, as well as their α-helical heptad repeat assignments, are provided later.


Lyophilized con-T[K7γ] was dissolved in 0.05 M MgCl2 and 100 mM tris(hydroxymethyl)aminomethane–HCl, pH 8.0, to a concentration of 10 mg/ml. Crystals were grown at 20 °C by the hanging drop vapor diffusion method in 1 M sodium acetate, 0.1 M N-(2-hydroxyethyl)piperazine-N′-ethanesulfonic acid, pH 7.8, and 0.05 M CdSO4. The crystals appeared after 3 days.

Data collection

The crystals were briefly soaked in a solution containing 1 M sodium acetate, 0.1 M N-(2-hydroxyethyl)piperazine-N′-ethanesulfonic acid, pH 7.8, 0.05 M CdSO4, and 30% glycerol at room temperature, and were flash-cooled by immersion in liquid nitrogen. Data were collected at the Advanced Photon Source SBC 19-BM at Argonne National Laboratory using radiation with a wavelength of 1.0 Å at a temperature of −173 °C. Data were processed and scaled using the HKL [14] suite of programs to a resolution of 1.3 Å in the hexagonal space group, P6322. The crystal-to-detector distance was 125 mm and 120° of data were collected with an oscillation of 1°. The crystal parameters and data collection statistics are listed in Table 1.
Table 1

Data collection and refinement statistics for the Cd2+/Mg2+–con-T[K7γ] complex

Data collection

 Space group


  a, b, c (Å)

57.1, 57.1, 32.8


20–1.3 (1.32–1.30)


6.6 (15.0)


37.5 (3.9)

 Completeness (%)

98.5 (83.8)




 No. of measured reflections


 No. of unique reflections




 No. of atoms







 Mean B factors


RMS deviations

 Bond lengths (Å)


 Angles (Å)


Data for the highest-resolution shell are shown in parentheses

Structure determination

The phases were determined by molecular replacement using the program Phaser [15] and the Ca2+–con-T[K7γ] structure as search model (Protein Data Bank ID 2DPR). Using one molecule per asymmetric unit [16], we calculated a Matthews coefficient of 2.2 Å/Da and solvent content of 43.8%, consistent with the existence of one molecule per asymmetric unit [17]. The rotational and translational Z scores for the solution were 9.4 and 10.4, respectively, and the log likelihood gain (a measure of how good the model is relative to a random atom distribution) was 150.

Refinement of con-T[K7γ]

Rigid-body refinement using Refmac5 (CCP4 [16] suite of programs) resulted in an Rwork of 53.2% and an Rfree of 53.2%. An anomalous difference map was calculated to distinguish Cd and Mg sites, since only Cd atoms have a significant anomalous signal at the wavelength used in data collection (Fig. 1b). A total of four Cd sites were identified in the electron density of the anomalous difference map at peaks of 24σ (Cd3), 14σ (Cd5), and 11σ (Cd1 and Cd4); however, only Cd3 is 100% occupied. Cd5 was refined to 80% occupancy and Cd1 and Cd4 were both refined to 50% occupancy (of Cd only). We were unable to definitively determine whether these partially occupied Cd sites were also partially occupied by Mg since Mg does not have an anomalous signal. However, the geometry and coordination distances suggest that this is not the case for the Cd3, Cd5, and Cd4 sites. One Mg site (Mg2) was located from the FoFc map contoured at 8σ (Fig. 1a) (the metal binding sites are numbered to correspond to the numbering of the sites in the Ca2+-bound structure [7]. Subsequent refinements of all 21 residues of con-T[K7γ] after adding four Cd2+, one Mg2+, and 28 waters produced an Rwork of 18.61% and an Rfree 20.47%. The program SHELXL [18] was employed to complete the refinement. Alternative side-chain conformations for Met8 and Val17 were modeled and waters were added, producing final Rwork and Rfree of 11.03 and 14.40% (from SHELXL [18]), respectively.
Fig. 1

The structure contains a total of three Cd ions, one mixed-occupancy Mg/Cd site, and one Mg2+. a The overall structure of Cd2+/Mg2+–con-T[K7γ] with an anomalous difference map contoured at 8σ. The four Cd2+ are shown as bluespheres and the Mg2+ is shown as a magenta sphere. Cd3 had the highest peak in the anomalous map at 24σ and is 100% occupied. Cd1 showed a peak at 14σ and is 80% occupied. Cd2 and Cd4 both had peaks in the anomalous map at 11σ and both are 50% occupied. b The overall structure of Cd2+/Mg2+–con-T[K7γ] with a FoFc map contoured at 5σ. The four Cd2+ are shown as blue spheres and the Mg2+ is shown as a magenta sphere. Mg1 is 100% occupied. The side chains are colored by atom

Isothermal titration calorimetry

Binding isotherms accompanying the titration of con-T[K7γ] with MgCl2 or CdCl2 were obtained by Isothermal titration calorimetry (ITC) using a VP-ITC calorimeter (MicroCal, Northampton, MA, USA). Peptide (0.1–0.2 mM) was dissolved in 20 mM 2-morpholinoethanesulfonic acid, 100 mM NaCl, pH 6.5, and placed over a small amount of Chelex-100 resin to ensure a metal-free species prior to the introduction of titrant. Titrations were performed at 25 °C with MgCl2 or CdCl2 (5–10 mM) dissolved in the above-mentioned buffer. The power compensations thus obtained were integrated over time and the resultant heat changes were deconvoluted for the best-fit model using the ORIGIN software program supplied by MicroCal. The parameters (n, Kd, ΔH) thus obtained represent the average of at least two experiments. Error estimates were 8% or less for replicate experiments.

Analytical ultracentrifugation

Sedimentation equilibrium experiments were performed using an Optima XL-I analytical ultracentrifuge (Beckman Instruments, Palo Alto, CA, USA) in an An-60 Ti rotor equipped with a standard two-channel cell. Con-T[K7γ] was dissolved in 20 mM 2-morpholinoethanesulfonic acid, 100 mM NaCl buffer, pH 6.5, at a concentration of 0.20 mM. The peptide samples in the absence and presence of metal ions were independently rotated at 48,000 and 52,000 rpm at 20 °C until equilibrium was attained. Absorbance monitoring was performed at 275 nm. The apparent molecular weight (Mw,app) was obtained by global fitting of multiple scans (three to five scans at each speed from two separate analyses) to a single ideal species using the sedimentation analysis software supplied by Beckman. A partial specific volume of 0.711 ml/g was assigned to con-T[K7γ] [6]. Gla residues were assigned the partial specific volume value of glutamate (0.66 ml/g). The fractional dimer content of self-associating peptides at equilibrium was calculated as previously described [5].


Structure of Cd2+/Mg2+–con-T[K7γ]

Our goal in this study was to understand the structure of the monomeric Mg2+-bound con-T[K7γ], contrast this structure with the structure of dimeric Ca2+-bound con-T[K7γ], and by comparison of the two divulge the metal ion specificity of the metal binding sites and understand how Mg2+ binding leads to monomeric species whereas Ca2+ binding results in dimeric species. However, all attempts at obtaining crystals in the presence of only Mg2+ failed, compelling us to investigate mixed-metal systems, culminating in the current report. The overall structure of Cd2+/Mg2+–con-T[K7γ] is shown in Fig. 1a. A total of three Cd2+ ions, one Mg2+, one mixed-metal Mg2+/Cd2+, and 61 waters were built into the structure. Alternative side-chain conformations were modeled for residues Met8 and Val17. Electron density was also seen for the C-terminal amide nitrogen atom. Much like Ca2+–con-G and Ca2+–con-T[K7γ] [7], Cd2+/Mg2+–con-T[K7γ] adopts an α-helical conformation over the entire length of the peptide chain (Fig. 1). Gla residues 3, 7, 10, and 14 define one face of the helix to which four metal ions (Mg/Cd1, Mg2, Cd3, Cd4) are coordinated. Mg/Cd1 is six-coordinate, chelating Gla residues 3 and 7 of the helix, Gla3 of a symmetry-related molecule, and a water molecule (W20) (Fig. 2a). Its coordination to Gla3 and Gla7 is similar to that seen for the Ca1 atoms of the Ca2+-bound con-G and con-T[K7γ] structures, but the coordination distances are much shorter in this structure than those seen in the Ca2+-bound structure [7]. Mg/Cd1 refines to only 50% Cd occupation and it is unclear as to whether this partially occupied Cd2+ site is also partially occupied by Mg2+, since Mg does not have an anomalous signal (Fig. 1a). However, the octahedral symmetry and relatively short coordination lengths (between 1.9 and 2.3 Å) are consistent with Mg2+ binding at this site, probably with higher occupancy than that of Cd since the coordination distances reflect Mg binding. It is therefore expected that this mixed-metal site is fully occupied with a combination of Mg2+ and Cd2+ ions. Cd3 is tetrahedrally coordinated by Gla4 and Gla7 of the helix and by Gla4 and Gla7 of a symmetry-related molecule. Cd3 is positioned directly on a crystallographic twofold axis. Unlike the Cd1 site, Cd3 is 100% occupied (Fig. 2b). This metal ion site is unique since both Ca2+-bound con-G and con-T[K7γ] structures do not have a metal at this position [7]. As depicted in Fig. 2c, Cd4 is octahedrally coordinated by Gla10 of the helix, Gla10 of a symmetry-related molecule, and four water molecules, two of which are symmetry-related to the other two (W22a, W22b, W51a, and W51b). Like Cd3, Cd4 is located directly on a crystallographic twofold axis and shares no homolog in the Ca2+-bound structures. One Mg2+ ion (Mg2) is also seen at this interface. Mg2 is octahedrally coordinated, making bidentate contacts to both Gla10 and Gla14 and also coordinates with W24 and W23 (W23 is in two different conformations, W23a and W23b) (Fig. 2d). Mg2 does not coordinate with a crystallographic symmetry-related molecule at this interface. The exclusive role of Mg2 is Gla-mediated chelation of a single helix. On the opposite face of the helix, a fourth Cd2+, Cd5, coordinates with both the nitrogen and the oxygen atoms of Gly1 (Fig. 2e), Glu16, and Lys19 of a crystallographically related molecule and two water molecules (W18 and W19). This binding site is also unique to this structure as none of the previous conantokin structures show metal binding at this location.
Fig. 2

Metal coordination in Cd2+/Mg2+–con-T[K7γ] (shown in green). a Mg/Cd1 is six-coordinate. A crystallographically related α-helix is colored yellow. Note the shorter coordination lengths. b Mg2 is octahedrally coordinated. Mg2 makes bidentate coordination with both γ10 and γ14 and also coordinates with waters W24 and W23. W23 has two different conformations, W23a and W23b. c Cd3 is tetrahedrally coordinated. Cd3 sits directly on a crystallographic twofold axis and coordinates with both γ4 and γ7 of the structure and its symmetry-related molecule (yellow). d Cd4 is octahedrally coordinated. Cd4 coordinates with γ10 of the structure and a crystallographically related molecule (magenta) along with waters W22a and W51a and the symmetry-related waters W22b and W51b. c Cd5 is six-coordinate. Cd5 coordinates with Glu16 and Lys19 of a symmetry-related molecule (magenta) and also coordinates with G1 and two waters, W18 and W19. All distances are shown in angstroms. Cd ions are blue spheres, the waters are red spheres, and Mg1 is a magenta sphere. The side chains are colored by atom

It is likely that Cd2+ is essential for crystallization, since no crystals grew in the absence of Cd2+, or even in decreased concentrations of Cd2+, consistent with the fact that three out of the four Cd2+ ions reside on crystallographic axes. It was not possible to determine whether the partially occupied Cd2+ sites are also partially occupied by Mg2+, since Mg2+ is significantly less electron dense than Cd2+ (46 vs. ten electrons). However, it is unlikely that Mg2+ is present at Cd3 because it is tetrahedral and Mg2+ is almost always octahedrally coordinated. The relatively hard Mg2+ cation is also not likely to coordinate nitrogen amines tightly. Further, the coordination lengths in the Cd3, Cd4, and Cd5 sites are significantly longer than those seen for Mg2+ (2.2–2.6 vs. 1.7–2.2 Å for a typical Mg2+). It would thus appear that the the Cd3, Cd4, and Cd5 sites occupied exclusively by Cd2+ and are critical for defining the crystal lattice contacts.

Comparison of Ca2+–con-T[K7γ] and Cd2+/Mg2+–con-T[K7γ]

An overlay of the Cd2+/Mg2+–con-T[K7γ] structure onto the Ca2+-bound con-T[K7γ] dimer (Fig. 3a) shows that Gla residues 3, 4, and 7 move significantly in the Cd2+/Mg2+–con-T[K7γ] structure, whereas Gla10 and Gla14 are virtually identical. The unique Gla4–Gla7 Cd2+ chelation, seen only in the Cd2+/Mg2+–con-T[K7γ] structure, causes an especially significant reorientation of Gla4. Lys18 is also significantly different in the two structures, although this is most probably due to its involvement in crystal packing interactions. Mg2 is close to the position of Ca2 of the Ca-bound structure as Ca2 is also coordinated in a bidentate fashion to both Gla10 and Gla14. Mg/Cd1 is relatively close to Ca1 and the coordination to Gla3 and Gla7 is also similar in the two structures. However, Cd3, Cd4, and Cd5 are in no way similar to any metal ions in the Ca-bound structure.
Fig. 3

Comparison of Ca2+- and Cd2+/Mg2+-bound con-T[K7γ]. Overlay of the Cd2+/Mg2+–con-T[K7γ] structure onto the Ca2+–con-T[K7γ] dimer structure. The Cd2+/Mg2+–con-T[K7γ] structure is shown in magenta and the Ca2+–con-T[K7γ] dimer structure is shown in blue. The four Cd2+ are shown as magenta spheres and the Mg2+ is shown as a green sphere. The Ca2+ are shown as blue spheres. The side chains are colored by atom. b A close-up of a in the region around the Mg2 site of Cd2+/Mg2+–con-T[K7γ] and the Ca2 site of the Ca2+–con-T[K7γ] dimer. Mg2 is shown as a green sphere, whereas Ca2 is shown as a blue sphere. Distances involving Mg2 are colored green. c A close-up of a in the region around the Ca1 site of the Ca2+–con-T[K7γ] dimer showing that cross-helix contacts are too long for chelation from the metal ion position seen in the Cd2+/Mg2+–con-T[K7γ] structure. c A close-up of a at the Ca1 binding site. Mg/Cd1 is shown as a magenta sphere and Ca1 is shown as a blue sphere. The distance between Mg/Cd1 and Ca1 is 1.67 Å. Distances involving Mg/Cd1 are colored magenta

Metal ion binding properties of con-T[K7γ]

To aid in the interpretation of the metal ion occupancy seen in the crystal structure of Cd2+/Mg2+–con-T[K7γ], the Cd2+- and Mg2+-binding properties of con-T[K7γ] were quantitated by ITC. Figure 4 depicts the results accompanying the separate titrations of con-T[K7γ] with MgCl2 and CdCl2. Mg2+ binds the peptide at a single class of sites with a stoichiometry of 1.9 mol per mole of peptide and an average Kd of 7.3 μM. These values are in good agreement with those obtained under similar conditions in a previous study [6]. The ΔH and TΔS values obtained were −5.0 and 2.0 kcal/mol, respectively. In contrast, the binding of Cd2+ to con-T[K7γ] is characterized by two classes of sites that differ markedly (approximately 300-fold) in affinity. The higher-affinity sites manifest a Kd for Cd2+ of 200 nM and bind with a stoichiometry of 2.1 mol Cd2+ per mole of peptide, whereas the lower-affinity sites bind an average of 2.7 mol Cd2+ per mole of peptide with a Kd of 60 μM. These values are compatible with the metal ion placements observed in the Cd2+/Mg2+ crystal structure in which the metal ion site maintained by Gla10 and Gla14 is fully occupied by Mg2+ and the locus formed by Gla3 and Gla7 may be partially occupied by Mg2+. Because Cd2+ does not effectively displace Mg2+ at these locations during crystal growth, it seems likely that these represent two of the three low-affinity Cd2+ sites as derived from ITC. The two high-affinity Cd2+ sites and the remaining low-affinity Cd2+ site could then be ascribed to the Cd1, Cd3, and Cd4 sites. Occupation of the high-affinity Cd2+ sites is attended by favorable enthalpy (ΔH = −4.4 kcal/mol) and entropy (TΔS = 4.7 kcal/mol). However, Cd2+ binding to the lower-affinity sites is endothermic in nature (ΔH = 2.4 kcal/mol) and is therefore entropically driven with a TΔS of 8.3 kcal/mol.
Fig. 4

Representative calorimetric titrations of con-T[K7γ] with MgCl2 and CdCl2. Increments (5–10 μl) of the ligand solutions were added to the peptide at 200-s intervals. The upper panels depict the heat changes accompanying ligand injection and the lower panels represent the binding isotherms corresponding to the integrated heat changes. Titrations were conducted at 20 °C and in buffer consisting of 20 mM 2-morpholinoethanesulfonic acid, 100 mM NaCl, pH 6.5. a Con-T[K7γ] (0.2 mM) titrated with 10 mM MgCl2. The fitted data for this experiment yielded n = 1.9, Kd = 8.3 μM, and ΔH = −5.2 kcal/mol. b Con-T[K7γ] (0.2 mM) titrated with 10 mM CdCl2. The fitted data for this experiment yielded n1 = 2.2, Kd1 = 205 nM, ΔH1 = −4.2 kcal/mol, n2 = 2.9, Kd2 = 58 μM, and ΔH2 = 2.3 kcal/mol

Solution-phase molecularity of con-T[K7γ] in complex with Cd2+and Mg2+

To determine whether saturating concentrations of Cd2+ or Mg2+ promote the formation of higher-order complexes of con-T[K7γ], sedimentation equilibrium experiments were performed (Fig. 5). In the apo form, con-T[K7γ] displays an Mw,app of 3,430 ± 240. Although this value is higher than the sequence-based molecular weight of 2,680, the tendency of monomeric con-T and con-T[K7γ] to exhibit high Mw,app values has been noted before [6] and may be attributable to the assignment of an artificially high partial specific volume to Gla, i.e., that of Glu, when calculating the partial specific volume of the peptide. The addition of 15 mM Mg2+ results in no significant change from the apo Mw,app (3,290 ± 290). However, in the presence of 5 mM Cd2+, a large increase in Mw,app (7,950 ± 440) is observed. On the basis of a Cd2+ to con-T[K7γ] ratio of 5:1 (as determined from ITC) and an \( M_{{{\text{Cd}}^{ 2+ } }} \) of 112, this value is in good agreement with the value of 7,980 calculated for a Cd2+-mediated dimer of con-T[K7γ] ([2 × Mw,app,apo] + [10 × \( M_{{{\text{Cd}}^{ 2+ } }} \)] = 7,980).
Fig. 5

Sedimentation equilibrium data corresponding to con-T[K7γ] in the apo form or in the presence of Mg2+ and/or Cd2+. a Overlay of sedimentation equilibrium scans obtained and the calculated fits for apo con-T[K7γ] (circles), con-T[K7γ] + 15 mM Mg2+ (squares), con-T[K7γ] + 5 mM Cd2+ (inverted triangles), and con-T[K7γ] + 5 mM Cd2+/60 mM Mg2+ (upright triangles). The data were collected at 20 °C at a rotor speed of 52,000 rpm in 20 mM 2-morpholinoethanesulfonic acid, 100 mM NaCl, pH 6.5. The peptide concentration was 0.2 mM. The chloride salts of the metal ions were used. The Mw,app values corresponding to the fits for the individual scans depicted were 3,320 (apo), 3,290 (15 mM Mg2+), 7,800 (5 mM Cd2+), and 4,800 (5 mM Mg2+/5 mM Cd2+). b Comparison of the Mw,app values obtained for con-T[K7γ] in the absence and presence of Mg2+ and/or Cd2+. Mw,app values were obtained through global fitting of multiple equilibrium scans acquired at rotor speeds of 48,000 and 52,000 rpm. The error bars represent the standard deviations


Identity and specificity of the metal ion binding sites

Several questions become apparent from the data presented so far: (1) Does the Cd2+ dimer share the same dimerization interface as the Ca2+ dimer? (2) Which metal binding sites constitute the “tight” binding Cd2+ sites versus the weaker sites? (3) Are the tighter or the weaker Cd2+ binding sites involved in dimerization? (4) Why is the Mg2+-bound species monomeric, whereas the Cd2+ and Ca2+ species are dimeric?

The crystal structure of the Mg2+/Cd2+-bound peptide clearly identifies the two Mg2+ binding sites as being the Mg/Cd1 and Mg2 sites. The coordination distances in the remainder of the sites are not consistent with Mg2+ binding, whereas the coordination distances in both the Mg/Cd1 and the Mg2 sites are completely consistent with Mg2+ binding. It is thus clear that Mg2+ binds exclusively to the metal 1 and metal 2 binding sites, consistent with the 2:1 Mg to peptide stoichiometry found by ITC. It is also clear that at least one of the two Mg binding sites also defines the dimerization interface in the Cd2+-bound dimer. This is shown by the Mg2+ analytical ultracentrifugation (AUC) challenge experiments, which demonstrate that addition of Mg2+ leads to dimer dissociation. Therefore, Mg2+ binding disrupts the Cd-induced dimer, meaning that the dimer interface must include at least one of the Mg2+ ion binding sites. Further, the fact that dimer dissociation occurs only when the Mg2+ to Cd2+ ratio is very high suggests that the Mg2+ binding site or the site responsible for Cd2+ dimerization must overlap with at least one of the tight Cd binding sites, as opposed to the lower-affinity sites. This is because the loose Cd2+ binding sites have affinities around tenfold lower than those of the Mg binding sites. Therefore, even stoichiometric concentrations of Mg2+ would be expected to effectively compete for the weak Cd binding sites. The fact that it requires Mg concentrations tenfold or higher to begin to disrupt dimerization means that the Cd sites that are coincident with the Mg sites must be the high-affinity Cd binding sites. This is structurally consistent as these sites have higher coordination between the peptide and metal than do the other three Cd sites (both the metal 1 and metal 2 binding sites are tetracoordinate with the peptide). It is thus clear that the Cd and Ca dimer interfaces share at least one of the metal binding sites and are therefore likely to at least partially overlap.

The crystal packing in the structure is also instructive regarding this issue. Each con-T[K7γ] helix makes metal-lattice interactions with three different symmetry mates. This is in contrast to the Ca2+-bound con-G and con-T[K7γ] dimer structures, where all of the metal coordination was between one other helix partner, making it clear that the crystalline dimer was unambiguously the dimer found in solution. Although it is not possible to rule out the probability that some of the lattice interactions seen in our structure also occur in the Cd2+-bound solution dimer, it is absolutely clear that the crystal packing we see is not fully capitulated in solution as there is no evidence of such a large oligomer in solution. It is also telling that Mg2+ binding abrogates the solution dimer and the Mg2+ binding site is the only metal binding site that does not make lattice interactions.

Magnesium-bound monomers versus calcium-bound dimers

The question of why the Mg-bound peptide does not dimerize is explained by examination of the structures of Cd2+/Mg2+-bound and Ca2+-bound con-T[K7γ]. Sedimentation equilibrium data indicate that whereas the exclusively Ca2+- or Cd2+-complexed peptide is dimeric, the Mg2+-bound species is monomeric. In solutions containing both Cd2+ and Mg2+, the Mw,app of con-T[K7γ] decreases with increasing Mg2+ content. Solution-phase Mg2+-mediated dimer collapse has also been observed with Ca2+-saturated peptide [5].

From the crystal structure presented herein, it appears that Mg2+ occupancy of the metal ion site bridged by Gla10 and Gla14 is sufficient to disrupt the dimer; although there is also partial Mg2+ occupancy of the Cd1 locus maintained by Gla3 and Gla7. Because this site is only 50% occupied by Cd2+ and the coordination distances are substantially shorter than those found for the three other Cd sites, the Cd1/Mg1 site is likely to be partially occupied by Mg2+. However, it is apparent that the Gla10/Gla14 site, which has 100% Mg2+ occupancy, is the preferred Mg2+ binding locus. Although the model providing the best fit for our calorimetric data corresponded to two identical Mg2+ sites, the Kd value for Mg2+ binding (7.3 μM) could likely reflect an average value for two nonidentical sites that differ in affinity by only twofold or threefold, with the Gla10/Gla14 locus being the preferred Mg2+ site. This Mg2 site is similar to the Ca2 site in the Ca2+-bound con-T[K7γ] dimer, wherein Ca2 at this Gla10/Gla14 site chelates Gla residues 7 and 10 of the partner helix. The Cd2 site, which may have up to 50% Mg2+ occupancy, involves coordination to Gla 3 and Gla7. In the Ca2+–con-T[K7γ] structure, this locus corresponds to Ca1 and chelates Gla14 of the partner helix. Hence, Mg2+ occupies the same con-T[K7γ] sites that are occupied by Ca2+, but cannot induce formation of the dimer. This can be ascribed to the inability of Mg2+ to participate in the cross-helix interactions made by Ca1 and Ca2 in the Ca2+-bound dimer. As shown in Fig. 3b, the distance between Mg/Cd1 and Ca1 is about 0.84 Å. Ca1 coordinates with Gla10 and Gla14 of one helix and Gla7 and Gla10 of the dimer helix, whereas Ca/Mg1 only coordinates with Gla10 and Gla14. Given this placement, Mg1 is positioned 3.08 and 2.91 Å away from Gla10 and Gla7, respectively, of the partner helix of the Ca2+ dimer. These distances are too great to be spanned by Mg2+, which generally chelates with distances between 1.7 and 2.2 Å, whereas Ca2+ coordinates in a range from 2.2 to 2.6 Å. It is likely that if the helices were brought close enough together to accommodate Mg2+ coordination distances, the result would be unacceptable steric and electrostatic repulsion. Such prohibitively long interhelical distance considerations could also apply to Mg2+ binding at its putative second locus, i.e., the Mg/Cd1 site, in which Gla3 and Gla7 are involved in malonate-type chelation of the metal ion (Fig. 3c). Hence, despite the determination, by sedimentation equilibrium, that Cd2+ alone can induce dimerization of con-T[K7γ], the presence of Mg2+ at the metal ion binding site spanned by Gla10/Gla14 (and possibly Gla3/Gla7) is sufficient to destabilize the dimer, even though Cd2+ remains bound at other loci. However, given the high binding affinity for Cd2+ in the Mg2+ binding sites (0.2 vs. 7.2 μM for Cd2+ and Mg2+, respectively), it seems inconsistent that the crystal structure shows exclusive Mg2+ binding in one of the high-affinity metal binding sites and significant binding in the other. There are many potential explanations for this phenomenon but the simplest is that the present crystal form is consistent only with the monomeric peptide. Therefore, even though the Mg2+-bound monomer may have been the minor form in solution, it could still have been the form that crystallized, especially if its solubility is significantly lower than that of the dimer. The equilibrium would then be driven to produce the monomer in its solid form.

In previous work, we showed that monomeric con-G and a covalently constrained antiparallel dimer form of con-G are biologically active insofar as they both exert in vitro effects on the NMDAR [9]. In that study, four physiologically prevalent subunit combinations of the NMDAR were utilized, namely, NR1a/NR2A, NR1a/NR2B, NR1b/NR2A, and NR1b/NR2B. Real-time electrophysiology traces revealed sharp differences in the mode of receptor activity exerted by each peptide species at the separate NMDAR combinations (e.g., the monomer is inactive at NR1a/NR2A-containing receptors, whereas the dimer potentiates at this combination), as well the kinetics of peptide action (e.g., the monomer shows slow onset and offset of inhibition at NR1b/NR2B, whereas the dimer inhibits with very fast onset and offset rates). These results suggest that distinctly different NMDAR activity profiles can be exhibited by a single conantokin, depending on the monomer–dimer distribution of the peptide in the vicinity of its NMDAR binding site. A possible reason for the differences in NMDAR activity between the monomer and the dimer can be culled from a comparison of Cd2+/Mg2+-bound and Ca2+-bound con-T[K7γ] structures (Fig. 3a), which reveals a significant reorientation of Gla4 between monomer and dimer states. This residue, which lies on the outer face of the dimeric species, is completely conserved in all ten conantokins characterized to date [1921] and has been shown to be an absolute requirement for activity in con-G, con-T, and conantokin-R [22]. We submit that the NMDAR subunit selectivity of a conantokin may be influenced by the disposition of Gla4, which, in turn, may be affected by the dimerization state of the peptide.

In conclusion, the structural analyses presented here underscore the importance of metal ion coordination geometry in influencing the molecularity of con-T[K7γ]. Additionally, these results raise the possibility that metal ion binding motifs that mimic the Gla10/Gla14 site can be engineered into other peptides or proteins in an effort to control oligomerization states in a reversible manner. Our data also show that a dimer similar to that formed by Ca2+ coordination can be formed with the much higher affinity Cd metal ion, meaning that much stronger metallozipper helical dimers can be formed than is possible using Ca.


We thank the following for financial support: GR-179 (085P100549) from the Michigan Economic Development Corporation (to J.H.G.), NIH Grant HL019982 (to F.J.C.), and NIH Grant GM0638947 (to J.H.G.). CCDC-667723 contains the supplementary crystallographic data for this paper. These data can be obtained free of charge from the Cambridge Crystallographic Data Centre via http://www.ccdc.cam.ac.uk/data_request/cif.

Copyright information

© SBIC 2010