Solution structure and fluctuation of the Mg²⁺-bound form of calmodulin C-terminal domain

Abstract

Calmodulin (CaM) is a Ca²⁺-binding protein that functions as a ubiquitous Ca²⁺-signaling molecule, through conformational changes from the “closed” apo conformation to the “open” Ca²⁺-bound conformation. Mg²⁺ also binds to CaM and stabilizes its folded structure, but the NMR signals are broadened by slow conformational fluctuations. Using the E104D/E140D mutant, designed to decrease the signal broadening in the presence of Mg²⁺ with minimal perturbations of the overall structure, the solution structure of the Mg²⁺-bound form of the CaM C-terminal domain was determined by multidimensional NMR spectroscopy. The Mg²⁺-induced conformational change mainly occurred in EF hand IV, while EF-hand III retained the apo structure. The helix G and helix H sides of the binding sequence undergo conformational changes needed for the Mg²⁺ coordination, and thus the helices tilt slightly. The aromatic rings on helix H move to form a new cluster of aromatic rings in the hydrophobic core. Although helix G tilts slightly to the open orientation, the closed conformation is maintained. The fact that the Mg²⁺-induced conformational changes in EF-hand IV and the hydrophobic core are also seen upon Ca²⁺ binding suggests that the Ca²⁺-induced conformational changes can be divided into two categories, those specific to Ca²⁺ and those common to Ca²⁺ and Mg²⁺.

Introduction

Calmodulin (CaM) is a ubiquitous intracellular Ca²⁺-binding protein that functions predominantly in the integration of Ca²⁺ in signal transduction processes. It is composed of two homologous globular domains (N- and C-terminal domains),^1–4 each consisting of two Ca²⁺ binding motifs, called EF-hand motifs.⁵ The EF-hand possesses a helix-loop-strand-helix structure and binds to Ca²⁺ via a 12-residue binding sequence. The binding sequence, which is often referred to as the EF-hand loop, is structurally composed of a loop, a strand and the N-terminal part of an α-helix. Ca²⁺ is chelated by seven oxygens from six residues, at positions 1, 3, 5, 7, 9, and 12 in a pentagonal bipyramidal fashion. The residues at positions 1, 3, and 5 are located in the loop, those at positions 7 and 9 are located in the strand, and the residue at position 12 is located in the helix structure. Ca²⁺ binding to the EF-hand of CaM induces the exposure of the hydrophobic core that provides the interaction site for target molecules, including calcium channel, myosin light chain kinase, CaM kinase, calcineurin, and nitric oxide synthase.^1–9

Mg²⁺ can also bind to all of the EF-hands of CaM, and shares the same binding site with Ca²⁺.^10–14 The Mg²⁺ association constants range between 10² and 10⁴ M⁻¹.¹⁰ Since Mg²⁺ exists in the cytosol at a millimolar concentration,¹⁵ the high affinity sites (EF-hands I and IV) can be converted into the Mg²⁺-bound forms. The folded structure of CaM is stabilized by Mg²⁺ binding.^13,16 The Mg²⁺-bound form of CaM is more stable than the apo form of CaM against denaturation by chemicals, such as urea and guanidine, and by thermodynamic effects.^17,18 Mg²⁺ increases the denaturation temperature of each domain of CaM by more than 20°C¹⁸ (Supporting Information Table S1). Mg²⁺ binding seems to be important to stabilize the structure of CaM.

Ca²⁺ and Mg²⁺ cause distinct conformational changes of CaM.^10,12,13 In the Ca²⁺-bound form, the hydrophobic core is exposed.¹ However, it has been proposed that the hydrophobic core is not exposed in the Mg²⁺-bound form.¹⁰ The Mg²⁺-bound structure of the N-terminal domain of soybean calmodulin isoform 4 (sCaM4), which shares about 78% sequence identity with bovine CaM,¹⁹ has been investigated by the backbone residual dipolar coupling approach.¹⁶ The apo form of the N-terminal domain of sCaM4 is unfolded. In the presence of Mg²⁺, the N-terminal domain of sCaM4 folds into an α-helix-rich structure.¹⁶ This is an excellent example of stabilization by Mg²⁺ binding. The Mg²⁺-bound form of sCaM4 adopts a closed conformation. Since the structure was calculated by the backbone assignments and the backbone RDC restraints, only the information about the orientation of the EF-hand helices was experimentally determined. Other details, including the Mg²⁺ coordination, the side chain orientations and the arrangement of the hydrophobic core, have remained unknown. To further clarify the details of the Mg²⁺-induced conformational changes and the features of the Mg²⁺-bound state, the tertiary structure of the Mg²⁺-bound form at atomic resolution is needed.

Here, we performed a structural analysis of the Mg²⁺-bound form of the CaM C-terminal domain (residues 78–148, CaMC) by multidimensional NMR spectroscopy. The two domains (N-terminal domain and C-terminal domain) fold independently, and the Ca²⁺-binding affinity and the Ca²⁺-induced structural changes of these domains are both equivalent to those of intact CaM.^18,20 Mg²⁺ preferentially binds to EF-hands I and IV of CaM.¹⁰ In EF-hand I, the Mg²⁺-induced chemical shift changes are localized in the N-terminal part of the binding sequence.¹⁴ However, in EF-hand IV, the Mg²⁺-induced changes occur throughout the binding sequence. The NMR signals and the fluorescence intensity of Y138, located on the C-terminal part of the EF-hand IV binding sequence, change upon Mg²⁺ binding.^12,21 Therefore, larger structural changes are anticipated for CaMC.

In an NMR study, the HSQC spectra of the apo and Ca²⁺-bound forms of CaM displayed well-dispersed signals. On the other hand, the NMR spectra of the Mg²⁺-bound form of CaM showed signal broadening for most of the residues, indicating that a slow conformational fluctuation, which is an exchange between multiple conformations, occurs around the core of the molecule. NMR signal broadening upon Mg²⁺ binding is a common feature among EF-hand proteins.^22,23 NMR is an appropriate method to monitor the exchange process. The signal broadening problem with the Mg²⁺-bound form of CaMC was solved by double point mutations. The solution structure of the Mg²⁺-bound form of CaMC was determined by multidimensional NMR spectroscopy, using the mutant. We observed the structural changes upon Mg²⁺ binding. Although the overall structure was the closed form, like the apo form, most of the structural changes in the binding site and the core region were similar to those induced by Ca²⁺.

Results

Mg²⁺ induces hydrogen bond interactions in the EF-hand IV loop structure

The glycine residue at position 6 of the binding sequence forms a hydrogen bond between its amide proton and the side chain of the first residue in the binding sequence [Fig. 1(A)].¹ The hydrogen bond is thought to contribute to the formation of the chelating structure. However, the hydrogen bond of EF-hand IV is formed only in the Ca²⁺-bound state, and is absent in the apo state. Hydrogen bond formation can be monitored by the down-field shift of the amide proton signals in NMR spectra.²⁴ The signal of G134 (position 6 in EF-hand IV) is detected in the normal shift region in the apo state and shows a large down-field shift upon Ca²⁺ binding.²⁴ The large down-field shift indicates that Ca²⁺ binding enhances the hydrogen bonding.²⁴

Figure 1.

Mg²⁺-induced and Ca²⁺-induced changes of CaMC. (A) Primary structure of the EF-hand metal-binding sequences of CaMC (EF-hand III and EF-hand IV). The Ca²⁺ and Mg²⁺ coordinating ligands are indicated by the coordination numbers. The loop structure, the β-strand and the α-helix are indicated as a wavy line, an arrow and a box, respectively. (B) Selected regions of the ¹H-¹⁵N HSQC spectra of CaMC without Ca²⁺ or Mg²⁺ at 23°C (upper left), with 10 mM Mg²⁺ at 23°C (upper right), with 10 mM Mg²⁺ at 0°C (lower right) and with 10 mM Ca²⁺ (lower left) at 23°C. (C) Hydrophobic interaction column chromatography. The samples were eluted by a concentration gradient of (NH₄)₂SO₄ from 2M to 0M. 0 mM (no Mg²⁺ or Ca²⁺ added; green, thin trace), 10 mM Mg²⁺ (blue, thin trace), 100 mM Mg²⁺ (dark blue, thick trace), 5 mM Ca²⁺ (orange, thick trace).

To investigate the hydrogen-bond formation in EF-hand IV upon Mg²⁺ binding, the ¹H-¹⁵N HSQC spectra of CaMC with and without 10 mM Mg²⁺ were recorded. Only the resonance of G98 was detected in the down-field region in the apo state [Fig. 1(B), upper left]. At 10 mM Mg²⁺, the G134 signal of the apo-form disappeared, and a weak down-field signal appeared at a different position from G98 in the apo state [Fig. 1(B), upper right]. The chemical shift of the signal was almost identical to that of G134 in the Ca²⁺-bound form [Fig. 1(B), lower left]. Therefore, the down-field signal was assigned to G134. This indicates that the hydrogen bond involving the amide proton of G134 is formed upon Mg²⁺ binding. It seems that EF-hand IV is mostly converted to the Mg²⁺-bound form, although the signal is very weak. (The signal broadening is described below.) In contrast, the signal from G98 disappeared [Fig. 1(B), upper right], suggesting that EF-hand III underwent a conformational exchange between the Mg²⁺-free and Mg²⁺-bound states. As previously reported by Ohki et al.,¹⁰ EF-hand IV was preferentially occupied by Mg²⁺, when compared with EF-hand III. In the spectra obtained at a lower temperature, to reduce the exchange rate, two down-field signals appeared in addition to the weak down-field signal of G134 [Fig. 1(B), lower right]. One was the apo-form of G98, and the other was the Mg²⁺-bound form. This confirmed the assignment of G134 and suggested that the Mg²⁺ is incorporated at least into the position 1 site bonded to position 6. The formation of the hydrogen bond is one of the stabilizing effects of Mg²⁺-binding.

The hydrophobicity of CaMC is not affected by Mg²⁺ binding

Ca²⁺ binding to CaM induces the exposure of the hydrophobic core on the protein surface, thus increasing the hydrophobicity of the molecule. The effects of Mg²⁺ and Ca²⁺ on the hydrophobicity of CaMC were evaluated by hydrophobic interaction chromatography. As shown in Figure 1(C), the apo form was eluted at [(NH₄)₂SO₄] = 1.08M, and the Ca²⁺-bound form of CaMC displayed a peak at [(NH₄)₂SO₄] = 0.35M in the presence of 5 mM Ca²⁺, indicating that Ca²⁺ binding increased the hydrophobicity of the protein. CaMC was eluted from the column at (NH₄)₂SO₄ concentrations of 1.14M and 1.2M in the presence of 10 mM and 100 mM Mg²⁺, respectively [Fig. 1(C)]. Since the Mg²⁺ binding did not increase the hydrophobicity of the protein, the hydrophobic core is probably not exposed by the Mg²⁺ binding.

Conformational exchange among the Mg²⁺-bound forms of CaMC

The two-dimensional ¹H-¹⁵N HSQC spectra of the apo and Ca²⁺-bound forms of CaMC displayed well-dispersed peaks with sharp line widths at 23°C (Fig. 2, upper panels). With 10 mM Mg²⁺, the spectrum showed broadened peaks and fewer signals (Fig. 2, lower left). In the presence of excess (100 mM) Mg²⁺, although some signals were detected, about 30 percent of the signals were missing in the spectrum (Fig. 2, lower right). These results reflect the exchange process between the apo and Mg²⁺-bound forms, and the additional conformational exchange among the Mg²⁺-bound structures. It should be noted that the Mg²⁺-induced broadening of the NMR signals does not represent the instability of the Mg²⁺-bound form of CaM. As indicated by the thermal unfolding experiment (Supporting Information Table S1), Mg²⁺ binding stabilizes the folding of CaM. The NMR signals of the Mg²⁺-bound form became sharper at higher temperatures [Fig. 3(A)]. Elevating the temperature increases the exchange rate and provides favorable conditions for observing the Mg²⁺-bound form by NMR. Although the NMR signals of the Mg²⁺-bound form became sharper at 50°C, the number of peaks in the HSQC spectrum was still slightly smaller than the expected number, and most of the signals were somewhat broadened.

Figure 2.

Mg²⁺-induced exchange broadening. The ¹H-¹⁵N HSQC spectra of CaMC alone (upper left), with 10 mM Ca²⁺ (upper right), with 10 mM Mg²⁺ (lower left), and with 100 mM Mg²⁺ (lower right) at 23°C.

Figure 3.

¹H-¹⁵N HSQC NMR spectra of Mg²⁺-bound forms of CaMC wild type and DD mutant. (A) HSQC spectra of CaMC wild type (left) and DD mutant (right) with 100 mM Mg²⁺ at 50°C. I130, V136, N137, and Y138 are labeled. (B) Secondary shifts of I130, V136, N137, and Y138 calculated with Cα (left) and CO (right) chemical shifts. Open bars correspond to the shift of the apo form of the wild type, blue bars represent the shift of the Mg²⁺-bound form of the wild type and green bars indicate the shift of the Mg²⁺-bound form of the DD mutant. Asterisks show residues with uncalculated secondary shifts, due to the lack of assignment.

Conformational exchange converged by point mutations

To improve the behavior of the Mg²⁺-bound state of CaMC, the E104D, E140D, and E104D/E140D (DD) mutants were constructed. The rationale for introducing these particular mutations was based on the Ca²⁺/Mg²⁺ binding properties of the EF-hands. E104 and E140 are located at position 12 in each of the EF-hand III and IV binding sequences, respectively [(Fig. 1(A)]. The glutamate is highly conserved at position 12 of the binding sequences. The substitution of aspartate for glutamate in other EF-hand proteins is known to diminish the binding selectivity of Ca²⁺ versus Mg²⁺.^25–27

The ¹H-¹⁵N HSQC spectra of all three mutants are similar to that of the wild type in the apo state at 23°C (data not shown), indicating that the proteins share a similar conformation. The stabilizing effect by Mg²⁺ was also conserved among the mutants (Supporting Information Table S1). Unexpectedly, the HSQC spectra of the apo forms of the E104D and DD mutants were well dispersed even at 50°C, whereas those of the wild type and the E140D mutant were crowded around the middle of the spectrum (Supporting Information Fig. S1). In the apo state, the S101 signals of the E104D and DD mutants were shifted down-field in the ¹H-¹⁵N HSQC spectra (Supporting Information Fig. S2). The down-field shift can be interpreted as the strengthening of the hydrogen bonding between the backbone amide of S101 and the side chain of E104D. S101 is located just before the helix and is involved in capping of helix F.² The N-terminus of helix F is fitted with the capping box motif.²⁸ The side chain of S101 forms a hydrogen bond with the backbone of E104 and, reciprocally, the side chain of E104 forms a hydrogen bond with the backbone of S101. The capping structure may stabilize helix F, which is known as a fragile helix in the apo state.¹⁸

The HSQC spectra of the Mg²⁺-bound form of the DD mutant at 50°C exhibited the best-resolved peaks among the wild type and mutant proteins. The backbone resonances of the Mg²⁺-bound forms of the wild type and the DD mutant were assigned, and the chemical shifts were compared. The amide chemical shift differences of I130, V136, N137, and Y138 in the EF-hand IV binding sequence were relatively larger than those of the other residues [Fig. 3(A) and Supporting Information Table S2]. To assess whether the DD mutation changes the backbone structures of these residues, the secondary shifts were calculated from the resonances of ¹³Cα, ¹³CO, and compared between the Mg²⁺-bound forms of the wild type and the DD mutant [Fig. 3(B) and Supporting Information Table S2]. These resonances were consistent between the wild type and the DD mutant, indicating that the backbone structures of these residues were essentially retained. The differences in the amide chemical shifts between the wild type and the DD mutant may be explained by the effects of the side chains. In the apo form of the wild type,³ these resonances differed significantly from those of the Mg²⁺-bound form, suggesting that Mg²⁺ binding caused the changes in the backbone structures of these residues in the wild type, which were also retained in the DD mutant [Fig. 3(B)]. In other regions, the average chemical shift difference of the Mg²⁺-bound forms between the wild type and the DD mutant was 0.08, based on the equation described in Figure 4. This is a much smaller difference, when compared with 0.23 for the average chemical shift difference between the apo form and the Mg²⁺-bound form of the DD mutant. In the wild type, Mg²⁺-induced down-field shifts of G98 and G134 were observed [Fig. 1(B)]. These down-field shifts were also observed in the DD mutant (data not shown), suggesting that the Mg²⁺ ions were similarly coordinated in EF-hands III and IV of the DD mutant and the wild type. In the ¹³C NOESY-HSQC spectrum of the wild type, the NOEs were insufficient for the internal residues, including I100 and V136 on the β-sheet. Less broadening of the main chain and side chain signals of I100 and V136 and the surrounding residues was detected for the DD mutant. More NOE interactions were found, and the total number of NOEs was increased by 35%. A comparison of the ¹³C NOESY-HSQC spectra of the wild type and the DD mutant revealed that the NOE patterns of the wild type were conserved in the mutant, indicating that the conformation of the mutant predominantly resembled that of the wild type. The wild type may also adopt a minor conformation, and the exchange between them may cause the NMR signal broadening.

Figure 4.

Mg²⁺ binding to the CaMC DD mutant. (A) Titration with Mg²⁺. Superimposition of the ¹H-¹⁵N HSQC spectra of the CaMC DD mutant with Mg²⁺ concentrations ranging from 0 to 100 mM. Each spectrum was acquired at 50°C. (B) Titration curve for Mg²⁺ binding to the CaMC DD mutant. Amide proton chemical shifts assigned to G96 (EF-hand III, circles) and G132 (EF-hand IV, squares), plotted as a function of the Mg²⁺ concentration. (C) Calculated populations of the different states of the protein, as a function of the Mg²⁺ concentration. The populations were calculated for K_III = 29.4 mM, K_IV = 2.67 mM, and 1 mM protein. (D) Backbone amide chemical shift changes induced by Mg²⁺ binding. Black bars represent the chemical shift changes between 0 and 10 mM Mg²⁺, and magenta bars indicate the chemical shift changes between 10 and 100 mM Mg²⁺. The values were calculated with the following equation: . The signals of E83, I85, N111, E114, V142, and Q143 were overlapped. The chemical shift changes of the D95, N97, G98, I100, V108, R126, A128, D131, D133, G134, and V136 residues were too broad, and are not included in the plot. The binding sequences of the EF-hands are indicated with a line. The helices are indicated by boxes with their names.

The substitution of glutamate with aspartate affects the Ca²⁺ binding potential

The binding potential of the DD mutant for Ca²⁺ was far less than that of the wild type. The Ca²⁺ titration, monitored by NMR, indicated that the DD mutant had a Ca²⁺-binding constant in the submillimolar range (data not shown). Considering that the wild type binds to Ca²⁺ with a K_d of ∼10⁻⁶ M, the substitution of glutamate with aspartate decreases the binding potential for Ca²⁺. The substitution affected the Ca²⁺-induced change in the hydrophobicity of the protein. The hydrophobic interaction chromatography revealed that the DD mutant with 5 mM Ca²⁺ eluted at a similar concentration of (NH₄)₂SO₄ to that with either Mg²⁺ or EDTA (Supporting Information Fig. S3). This indicates that the hydrophobic cavity of the DD mutant is not exposed at Ca²⁺concentrations below 5 mM.

Mg²⁺ binding to EF-hand IV has a major impact on the conformational changes of CaMC

The Mg²⁺ titration was performed by recording the ¹H-¹⁵N HSQC spectra as a function of increasing Mg²⁺ concentration, ranging 0 to 100 mM, in 100 mM KCl at 50°C for the DD mutant. The NMR signals of the amide protons of G96 (EF-hand III) and G132 (EF-hand IV) during the titration are shown in Figure 4(A). Two glycine residues exist at position 4 in each of the binding sequences and are good probes for Mg²⁺-binding to these sequences.

The amide chemical shifts of G96 and G132 were plotted as a function of the Mg²⁺ concentration [Fig. 4(B)]. The microscopic Mg²⁺-binding constants (K_III and K_IV) for EF-hands III and IV were calculated using the G96 and G132 data, respectively. EF-hands III and IV bind to two Ca²⁺ ions with positive cooperativity. In contrast, the Mg²⁺-induced chemical shift changes agreed with the theoretical curve based on the model in which two Mg²⁺ ions bind to the protein independently. The binding manner of Mg²⁺ to the protein was not cooperative. The calculated microscopic Mg²⁺ dissociation constants were K_III = 29.4 mM and K_IV = 2.67 mM. Roughly speaking, Mg²⁺ binds to the EF-hands IV and III sequentially, since the binding constants extensively differ. In the presence of 100 mM Mg²⁺, the population of the two Mg²⁺ form is significantly predominant [Fig. 4(C)].

To separate the effects of the binding of the first and second Mg²⁺ ions to the protein, the chemical shift differences between 0 and 10 mM Mg²⁺ and those between 10 mM Mg²⁺ and 100 mM Mg²⁺ were plotted in Figure 4(D). The changes up to 10 mM Mg²⁺ mainly reflect the effects of binding to EF-hand IV, and those beyond 10 mM Mg²⁺ mainly reflect the binding to EF-hand III. Upon Mg²⁺ binding to EF-hand IV, large chemical shift changes were observed throughout the EF-hand IV binding sequence. The chemical shifts in the EF-hand III binding sequence were predominantly changed upon Mg²⁺ binding to EF-hand III. In all other regions, larger chemical shift changes were observed between 0 and 10 mM Mg²⁺. Even the signals of the F-G linker changed upon Mg²⁺binding to EF-hand IV, suggesting that either helix F or helix G moved. These results indicated that the first Mg²⁺ binding event to EF-hand IV has a major impact on the structural change.

Solution structure of the Mg²⁺-bound form of CaMC

We determined the solution structure of the Mg²⁺-bound form of CaMC, using the DD mutant. Over 90% of the sequence-specific resonance assignments for the polypeptide backbone and the side chains of the Mg²⁺-bound form of the DD mutant were obtained, through analyses of the 2D and 3D heteronuclear NMR spectra recorded for the uniformly ¹³C, ¹⁵N-labeled protein. The 3D structure of the Mg²⁺-bound form of the DD mutant was calculated using 840 restraints derived from 1,159 NOEs from 3D NOESY spectra, supplemented with 84 dihedral angles (φ, ψ, and χ¹ angles) and 23 hydrogen bonding restraints. Automatic NOE assignments and structure calculations were performed using CYANA 2.2.^29–31 Over 98% of the distance restraints derived from the NOEs were assigned, and the R.M.S.D.s of the backbone atoms of the folded part of the calculated structures were ∼0.5 Å. The structural statistics are summarized in Supporting Information Table S3, and the ensemble of 20 structures of the Mg²⁺-bound DD mutant is shown in Figure 5(A).

Figure 5.

Solution structures of the Mg²⁺-bound CaMC DD mutant. (A) Ensemble of the backbone heavy atoms of the 20 lowest target functions of the Mg²⁺-bound form of the CaMC DD mutant. The α-helices are sequentially labeled with letters. Helices E and F form EF-hand III, while G and H form EF-hand IV. (B) Superimposition of the apo form of wild type CaMC (green, PDB ID: 1F71) and the Mg²⁺-bound form of the CaMC DD mutant (blue). Helices E and F of the apo form are superimposed against the corresponding helices of the Mg²⁺-bound form. (C) EF-hand IV binding sequence of the apo form of the wild type (left, green), the Mg²⁺-bound form (center, blue), and the Ca²⁺-bound form of the wild type (right, yellow, PDB ID: 1J7P) of CaMC. The side chains of A128, I130, V136, and F141 are shown. In the Ca²⁺-bound form, the side chains of the chelating residues D129, D131, D133, N137, and E140 are shown, and in the Mg²⁺-bound form, the side chain of the chelating residue D129 is also depicted. The Ca²⁺ ion is shown in black. (D) Ribbon representation of the apo form of the wild type (left, green), the Mg²⁺-bound form of the DD mutant (center, blue) and the Ca²⁺-bound form of the wild type (right, yellow) of CaMC. The side chains of F89, I130, Y138, and F141 are also shown.

The secondary structure elements of the Mg²⁺-bound form of the DD mutant are essentially identical to those of the apo form and the Ca²⁺-bound form of the wild type [Fig. 5(B)]. Helices E and F form EF-hand III, and helices G and H form EF-hand IV. The short antiparallel β-sheet connecting EF-hands III and IV is also conserved. The first six residues in the EF-hand III binding sequence and I130-G134 in EF-hand IV were not defined well, due to the lack of medium- and long-range NOEs.

Discussion

Mg²⁺-induced conformational changes in the EF-hand IV

The Mg²⁺-induced conformational changes occurred in wide area of EF-hand IV, while the closed conformation was retained. The global fold of the Mg²⁺-bound form was similar to that of the apo form [Fig. 5(B)]. In particular, the EF-hand III structures were almost identical. For EF-hand III, the backbone R.M.S.D. between the Mg²⁺-bound and apo forms was 0.85 ± 0.13 Å.

Mg²⁺ binding induced conformational changes in EF-hand IV. Initially, the N-terminal part of the EF-hand IV binding sequence, which directly interacts with Mg²⁺, was rearranged. In the apo form, the side chain of I130 is buried and interacts with the hydrophobic A128, V136 and F141 residues [Fig. 5(C), left]. In the Mg²⁺-bound form, D129 occupies this position and forms a hydrogen bond with G134. As a result, I130 was extruded [Fig. 5(C), center]. D129 is an important residue for the coordination to Mg²⁺ or Ca²⁺ on the inside of the molecule [Fig. 5(C), center and right]. As previously mentioned, the hydrogen bond formation by G134 was indicated by the amide signal position [Fig. 1(A)]. This was also observed in the Ca²⁺-induced changes. The rearrangement of the N-terminal part of the EF-hand IV binding sequence shifted the C-terminus of helix G toward the β-sheet. The backbone of A128 was converted into the regular α-helical conformation in the Mg²⁺-bound form [Fig. 5(C)]. In association with the movement of D129, A128 was also oriented toward the inside of the molecule, resulting in the extension of the α-helix. As compared to the apo form, the averaged φ and ψ angles of A128 changed from −89.3° to −63.9° and from 153.0° to −47.3°, respectively. These changes were also observed in the Ca²⁺-bound form. Accordingly, helix G was tilted toward the open orientation by approximately 10 degrees, and helix H was also tilted in a similar direction to helix G [Fig. 5(B)]. Consequently, the inter-helical angle between helices G and H in the Mg²⁺-bound form was close to that in the apo form (Supporting Information Table S4).

Conformational change in the hydrophobic core

The Mg²⁺-induced changes also occurred in another region, in addition to the postulated Mg²⁺ position. Around the C-terminal end of the EF-hand IV binding sequence, Y138 and F141 also changed their interaction partners. New aromatic interactions between F89, Y138, and F141 were created in the Mg²⁺-bound form [Fig. 5(D)]. This may be attributed to the movement of helices G and H. In the apo form, the aromatic ring of Y138 is oriented toward the C-terminus of helix E and the β-sheet.^2–4 In the Mg²⁺-bound form, the ring is away from the β-sheet and faces F89 vertically [Fig. 5(D)]. The up-field ring current shift observed in the δ proton of Y138 can be attributed to the interaction with F89. The orientation of the ring of F141 changed from helix G to helix F, and was close to V108 on helix F and the aromatic ring of F89 on helix E [Fig. 5(D)]. These conformational changes are also observed in the presence of Ca²⁺. However, the extents of the structural changes are larger in the Ca²⁺-bound form, and those of the Mg²⁺-bound form are around the mid points, as seen in Figure 5(D).

In contrast to the effects of Ca²⁺ binding, the Mg²⁺-induced structural changes exerted only a minor influence on the inter-helical angles (Supporting Information Table S4). The hydrophobic core, which provides the binding surface for the target molecules, remained buried. The conformation of EF-hand III was minimally affected by Mg²⁺ binding. The Ca²⁺-induced movements of Y138 and F141 were more significant, as compared to those induced by Mg²⁺. Taken together, the structure of the Mg²⁺-bound form can be considered as an intermediate state of the structural changes caused by Ca²⁺ binding.

The incorporation of Mg²⁺ in the binding sequence leads to the formation of ionic interactions between Mg²⁺ and the negatively charged residues. These ionic interactions contribute to the thermal stability of the Mg²⁺-bound form of CaMC. The hydrogen bond is an additional stabilizing factor. The extension of the C-terminal end of helix G increases the hydrogen bond network, in addition to the hydrogen bond between G134 and D129.

Mg²⁺ coordination

The geometry of the Ca²⁺ coordination can be described as a pentagonal bipyramid, in which six of the seven Ca²⁺ ligands are provided from the residues at positions 1, 3, 5, 7, and 12 (a bidentate ligand). The remaining Ca²⁺ ligand is the residue at position 9, via a water molecule. The Ca²⁺-induced rearrangements of positions 1, 2, and 6 of the EF-hand IV binding sequence were also found in the Mg²⁺-bound form, suggesting that the accommodation of Mg²⁺ in this region of the EF-hand IV binding sequence is similar to that of Ca²⁺, though NMR signal broadening shows dynamic behavior of positions 3 and 5.

The binding of Ca²⁺ or Mg²⁺ induced the shift of the nitrogen resonance of E139 (Supporting Information Table S5), which is one of the largest chemical shift changes. One explanation for the nitrogen resonance shift is the rearrangement of the main-chain structure, but the φ and ψ backbone dihedral angles of E139 are minimally affected by Ca²⁺ or Mg²⁺ binding, since E139 is within a helical structure. Another possible reason is hydrogen bond formation. The disruption of the hydrogen bond affects the nitrogen resonance and induces an up-field shift. The crystal structure of the Ca²⁺-bound form of CaM revealed that no atom is involved in a hydrogen bond with the main-chain amide of E139.¹ In the ensemble of NMR structures of the apo form, the side chain of N137 is located in the vicinity of the E139 main-chain amide, and could potentially form a hydrogen bond.^2–4 Although the crystal structure of the apo form of CaM has not been solved, the corresponding hydrogen bond exists in the crystal structure of the Ca²⁺-free form of the EF-hand protein, aequorin.³² Accordingly, a hydrogen bond may exist between the main-chain amide of E139 and the side-chain oxygen of N137 in the apo form of CaM. The hydrogen bond disappears upon Ca²⁺ or Mg²⁺ binding. N137, at position 9 in the binding sequence, uses its side chain to coordinate to Ca²⁺ via a water molecule. As in Ca²⁺ binding, Mg²⁺ binding may involve N137. The disruption of the hydrogen bond to helix H may be related to the neighboring conformational changes, as described above.

Additional features in the wild type Mg²⁺-bound form

In the structural determination of the Mg²⁺-bound form of the DD mutant, the structural changes were observed only in EF-hand IV, while in the wild type, the Mg²⁺-induced broadening ranged from the β-sheet to the core, including EF-hand III. The resonances of the hydrophobic residues at position 8 (I100 and V136) in the binding sequences of EF-hands III and IV were severely broadened in the wild type. The residue at position 8 is located at the center of the β-strand, and links the paired EF-hands structurally. The exchange broadening was also observed in the residues of the hydrophobic core, in the vicinity of the β-sheet region, and the residues between helices E and F (Fig. 6). One explanation for the broadening of the Mg²⁺-bound form of the wild type is the multiple conformations in the β-sheet region and the surrounding region, including the hydrophobic core. Fewer conformational exchanges in the β-sheet region occurred in the DD mutant. Replacing glutamate with aspartate at position 12 places the carboxylate group farther away from the bound metal, decreasing the population coordinated with the residue at position 12. The DD mutation may have excluded such low-population structures. Mg²⁺ binding only to EF-hand IV can lead to the simultaneous movement of both EF-hands III and IV, since a low Mg²⁺ concentration produces overall broadening. This implies that a critical mechanism underlying the cooperativity of the two EF-hands is reflected in the Mg²⁺-induced broadening.

Figure 6.

Backbone trace of the Mg²⁺-bound form of the CaMC DD mutant. The hydrophobic residues with side-chains that undergo exchange broadening in the presence of Mg²⁺ in the wild type are colored grey.

In summary, the Mg²⁺-induced conformational change mainly occurs on the helix G and helix H sides of the binding sequence of EF-hand IV, and involves the rearrangement of the hydrophobic interactions. The features of the Mg²⁺-induced conformational changes of CaMC are included in the Ca²⁺-induced conformational changes. The structure of the Mg²⁺-bound form can be regarded as an intermediate state of the structural changes caused by Ca²⁺ binding. Therefore, the Ca²⁺-induced conformational changes of CaMC can be divided into two categories, those specific to Ca²⁺ and those common to Ca²⁺ and Mg²⁺.

Materials and Methods

Protein expression and purification

For the recombinant expression and purification of CaMC (residues 78-148), the cDNA encoding this region of Xenopus CaM was cloned into the pET11a vector (Novagen). The N-terminal methionine was removed naturally. All of the CaM mutants were generated by site-directed mutagenesis, using a QuikChange kit (Stratagene). All unlabeled proteins were expressed in Escherichia coli Rosetta2(DE3)pLysS, grown in LB medium at 37 °C. The uniformly ¹⁵N or ¹⁵N, ¹³C-labeled proteins were expressed in M9 minimal medium, supplemented with either ¹⁵NH₄Cl or ¹⁵NH₄Cl and [¹³C₆]glucose as the sole nitrogen and carbon sources, respectively. The cells were disrupted by sonication, and after clarification by centrifugation, the supernatant was applied to a HiTrap Q FF or HP column (GE Healthcare), in 20 mM Tris-HCl (pH 7.5) containing 5 mM CaCl₂, which was eluted with a NaCl gradient. Fractions containing the protein were applied to a HiTrap Phenyl FF or HP column (GE Healthcare) equilibrated with 20 mM Tris-HCl (pH 7.5), containing 5 mM CaCl₂ and 2.0M (NH₄)₂SO₄. The column was washed with a buffer containing 20 mM Tris-HCl (pH 7.5), 5 mM EDTA and 2.0M (NH₄)₂SO₄. The protein was eluted with a (NH₄)₂SO₄ gradient. For dialysis and further sample preparation, Chelex 100-treated 10 mM MES (pH 6.5) buffer containing 100 mM KCl was used. The fraction containing the protein was dialyzed against 10 mM MES (pH 6.5) buffer, containing 100 mM KCl. The protein concentration was determined using a molar absorption coefficient of ɛ = 2980 M⁻¹ cm⁻¹.

Hydrophobic interaction chromatography

Hydrophobic interaction chromatography was performed using a 1 mL RESOURCE PHE column (GE Healthcare). The column was equilibrated with an appropriate equilibration buffer, containing 2M (NH₄)₂SO₄ in 20 mM Tris-HCl (pH 7.8). For the apo and Ca²⁺-loading cases, 5 mM EDTA and 5 mM CaCl₂ were included in the equilibration buffer, respectively. For the Mg²⁺-loading, either 10 mM or 100 mM MgCl₂ was included in the equilibration buffer. In all experiments, 0.16 mg of the protein was loaded on the column. All experiments were performed at 10°C. Elution was done by decreasing the ammonium sulfate concentration to 0M in 20 mM Tris-HCl, pH 7.8. The protein eluted under each condition was spectrophotometrically monitored.

NMR measurements and spectral assignments

NMR spectra were measured at 50°C on a Bruker AVANCE600 spectrometer, equipped with pulsed field gradients and triple resonance probes. The samples used in the magnesium titrations initially contained 1 mM of ¹⁵N-labeled protein, in a buffer consisting of 10 mM MES and 100 mM KCl in 90% H₂O/10% D₂O, pH 6.5. Small aliquots of a 100 mM (1 mM to 10 mM) or 1M (10 mM to 100 mM) MgCl₂ solution were added to the protein solution, and two-dimensional ¹H-¹⁵N HSQC spectra were recorded.

Resonance assignments were accomplished by analyses of multidimensional heteronuclear NMR spectra, 2D ¹H-¹⁵N HSQC, 2D ¹H-¹³C HSQC, 3D HNCO, HN(CA)CO, C(CO)NH, HNCACB, CBCA(CO)NH, HBHA(CO)NH, H(CCO)NH, HCCH-COSY, HCCH-TOCSY and CCH-TOCSY, acquired on a 1 mM protein sample (uniformly labeled with ¹³C and ¹⁵N), in buffer containing 10 mM MES-d₁₃ (pH 6.5), 100 mM KCl and 100 mM MgCl₂ in 90% H₂O/10% D₂O. All spectra were processed using NMRPipe³³ and analyzed using NMRview³⁴ and the integrated KUJIRA module.³⁵

The ¹³Cα, ¹³CO secondary shifts were calculated by the table of random coil carbon chemical shifts provided by Wishart and Sykes.³⁶

Structure calculations

NOE restraints were obtained from 3D ¹⁵N-edited NOESY and ¹³C-edited NOESY spectra with mixing times of 160 ms. Backbone φ and ψ torsion angle restraints were derived from an analysis of the Hα, ¹³Cα, ¹³Cβ, ¹³CO and backbone ¹⁵N chemical shifts using TALOS,³⁷ including 23 hydrogen bond constraints. The χ¹ angle restraints were derived from the HNHB, HN(CO)HB and NOESY spectra. Automated NOE cross-peak assignments and structure calculations with torsion angle dynamics were performed using CYANA 2.2.^29–31 A total of 100 structures were independently calculated. The 20 conformers with the lowest final CYANA target functions were used. Figures were generated with MOLMOL. The coordinates for the Mg²⁺-bound CaMC E104D/E140D mutant have been deposited in the Protein Data Bank (PDB ID: 2EQC).

Glossary

Abbrevations

COSY

correlation spectroscopy

EDTA

ethylenediaminetetraacetic acid

HSQC

heteronuclear single quantum correlation

NMR

nuclear magnetic resonance

NOE

nuclear Overhauser effect

NOESY

nuclear Overhauser effect spectroscopy

RDC

residual dipolar coupling

R.M.S.D.

root mean square deviation

TOCSY

total correlation spectroscopy