Structural and Energetic Determinants of Apo Calmodulin Binding to the IQ Motif of the Na_V1.2 Voltage-Dependent Sodium Channel

SUMMARY

The neuronal voltage-dependent sodium channel (Na_v1.2), essential for generation and propagation of action potentials, is regulated by calmodulin (CaM) binding to the IQ motif in its α-subunit. A peptide (Na_v1.2_IQp, KRKQEEVSAIVIQRAYRRYLLKQKVKK) representing the IQ motif had higher affinity for apo CaM than (Ca²⁺)₄-CaM. Association was mediated solely by the C-domain of CaM. A solution structure (2KXW.pdb) of apo ¹³C,¹⁵N-CaM C-domain bound to Na_v1.2_IQp was determined with NMR. The region of Na_v1.2_IQp bound to CaM was helical; R1902, an Na_v1.2 residue implicated in familial autism, did not contact CaM. The apo C-domain of CaM in this complex shares features of the same domain bound to myosin V IQ motifs (2IX7) and bound to an SK channel peptide (1G4Y) that does not contain an IQ motif. Thermodynamic and structural studies of CaM-Na_v1.2_IQp interactions show that apo and (Ca²⁺)₄-CaM adopt distinct conformations that both permit tight association with Na_v1.2_IQp during gating.

INTRODUCTION

The voltage-dependent sodium channel type II (Na_v1.2) is required for propagation of action potentials along unmyelinated axons found primarily in neurons and muscle (Schaller and Caldwell, 2003). Na_v1.2 has one α-subunit and one or more β-subunits (Catterall, 2000). The α-subunit has four homologous, pore-forming transmembrane domains, and intracellular domains that control channel gating. Inactivation of Na_v1.2 α-subunit involves interactions of the intracellular loop connecting domains III and IV and the C-terminal tail which has been modeled to contain six α-helices (Cormier et al., 2002).

Patch clamp studies showed that deletion of the putative sixth α-helix maintains the channel in a closed, inactivated state (Cormier et al., 2002; Mantegazza et al., 2001). A sequence (residues 1901 to 1927, Na_v1.2_IQp) within that region contains a classical CaM-binding IQ motif (Mori et al., 2000) (IQxxΦBGΦxxB), where X is any amino acid, Φ is an aromatic residue, and B is Arg or Lys (Figure 1A, B, C). All ten isoforms of Na_V contain a single IQ motif that allows channel regulation by CaM (Yu and Catterall, 2003). Many IQ motifs found in myosins, neuronal growth proteins, and ion channels bind preferentially to apo (Ca²⁺-depleted) CaM (Liu and Storm, 1990; Urbauer et al., 1995). However, this preference is not universal. For example, the IQ motif in Ca_V 1.2 binds preferentially to (Ca²⁺)₄-CaM (DeMaria et al., 2001).

Figure 1. Features of Voltage-Dependent Sodium Channel IQ Motifs and CaM.

A: Consensus sequence of 10 human voltage-dependent sodium channel IQ motifs. B: Na_v1.2_IQp sequence modeled as an ideal α-helix. C: Helical wheel projection of IQ motif of Na_v1.2_IQp displaying its hydrophobic and polar faces. D-G: Four structures of CaM (residues 1-75 blue, 76-80 black, 81-148 red) shown with bound peptide (green) and Ca²⁺ (yellow spheres) as appropriate. Square icons above ribbon diagrams indicate apo (open circle) or Ca²⁺-saturated (yellow filled circle) sites in the N-domain (blue) and C-domain (red).

CaM is essential to many eukaryotic signal transduction pathways (Klee et al., 1986). As seen in Figure 1D, it is composed of two domains (N/blue and C/red) that each bind two Ca²⁺ ions cooperatively. Although the domains are similar in sequence and structure, the C-domain binds Ca²⁺ 10- to 25-fold more favorably than the N-domain (Klee et al., 1986). Ca²⁺ binding triggers opening of hydrophobic clefts in each domain of CaM, altering its free energy of association with target proteins (Crivici and Ikura, 1995). The canonical CaM-binding sequence forms a basic, amphipathic, alpha-helix (BAA motif) (Yap et al., 2000) that binds preferentially to (Ca²⁺)₄-CaM forming a compact ellipsoid (Figure 1E shows a BAA motif from CaMKII bound to (Ca²⁺)₄-CaM (1CDM)). There are few available structures of apo CaM bound to targets. In one, two apo CaM molecules are bound to a fragment of myosin V containing two IQ motifs (2IX7; (Houdusse et al., 2006), Figure 1F). The C-domain of CaM adopted a semi-open conformation having interhelical angles smaller than those of (Ca²⁺)₄-CaM. In another structure (1G4Y (Schumacher et al., 2001), Figure 1G), two CaM molecules, each having an apo semi-open C-domain but Ca²⁺-saturated open N-domain, are bound to two copies of a fragment of the SK channel that does not contain an IQ motif. CaM is widely recognized for extreme molecular plasticity.

To understand the Ca²⁺-dependent roles of each domain of CaM in binding the IQ motif of Na_V1.2, fluorescence spectroscopy and NMR were used to monitor thermodynamic and structural properties of association. We determined that the C-domain of CaM bound Na_v1.2_IQp with very high affinity (nM K_d values), while the N-domain of CaM interacted weakly with the IQ motif. Na_v1.2_IQp binding to CaM selectively reduced Ca²⁺-binding affinity of sites III and IV, without significantly affecting sites I and II. Complexes of Na_v1.2_IQp with apo CaM (the C-domain alone or full-length CaM) were unperturbed in the presence of elevated NaCl. Heteronuclear NMR studies of apo ¹³C,¹⁵N-CaM_76-148 (C-domain alone) bound to Na_v1.2_IQp were used to determine a solution structure. In it, the C-domain of CaM adopted a semi-open conformation in binding to the primarily helical peptide. Furthermore, although Na_v1.2_IQp bound tightly to both apo and (Ca²⁺)₄-CaM, it caused unique chemical shift perturbations in each case. This suggests that the binding interface between (Ca²⁺)₄-CaM and Na_v1.2_IQp is very different from that of apo CaM_76-148:Na_v1.2_IQp, consistent with the physiological function of Ca²⁺-mediated inactivation Na_v1.2. These results are discussed with respect to CaM recognition of IQ motifs in general, and regulation of sodium channels specifically.

RESULTS

CaM Binding to Na_v1.2_IQp

CaM binding to fluoresceinated Na_v1.2_IQp (Fl-Na_v1.2_IQp) was monitored by steady-state fluorescence emission to determine domain-specific differences in effects on peak intensity. As shown in Figure 2A for apo CaM, no change in Fl-Na_v1.2_IQp was observed upon addition of 3 equivalents of CaM_1-80 (N-domain). Significant increases in intensity of Fl-Na_v1.2_IQp were observed after addition of CaM_76-148 (C-domain, 15%) or CaM_1-148 (full-length, 13%). This domain-specific pattern of change for apo CaM was preserved in 10 mM Ca²⁺ (Figure 2B), although the increase observed for (Ca²⁺)₄-CaM_1-148 (25%) was larger than that for (Ca²⁺)₂-CaM_76-148 (17%). The precision in intensity was ~1% for 3 replicate trials.

Figure 2. Domain-specific Interactions of CaM with Na_v1.2_IQp.

Emission spectra of Fl-Na_v1.2_IQp in the absence (solid line) and presence (dashed line) of slight excess of (A) apo CaM_1-80 (left), apo CaM_76-148 (middle), and apo CaM_1-148 (right) or (B) Ca²⁺-saturated CaM. C: Titration of Fl-Na_v1.2_IQp with apo CaM (open circles) or Ca²⁺-saturated CaM (filled diamonds). The titrations of apo CaM_1-80 were normalized to the maximum signal observed after addition of 10 mM CaCl₂ (Newman, 2008). For CaM_76-148 (middle), and CaM_1-148 (right), binding was stoichiometric and is plotted as a function of the ratio of [CaM_total]:[peptide], with an inset showing saturation versus [CaM_total]. D: Equilibrium Ca²⁺ titrations of CaM_1-80 (left), CaM_76-148 (middle), and CaM_1-148 (right) using intrinsic fluorescence to monitor Ca²⁺ binding in the absence of peptide [sites I and II (blue) or sites III and IV (red)], or after addition of excess Na_v1.2_IQp [sites I and II (cyan) and sites III and IV (orange)]. Supported by Tables S1 and S2.

CaM Titration of Na_v1.2_IQp

The affinities of CaM binding to Fl-Na_v1.2_IQp were determined using fluorescence anisotropy. Figure 2C shows that the anisotropy of Fl-Na_v1.2_IQp increased upon binding either apo CaM (open symbols) or Ca²⁺-saturated CaM (closed symbols). CaM_1-80 bound to Na_v1.2_IQp with low affinity (left panel); the K_d was 1.64 ± 0.65 μM for (Ca²⁺)₂-CaM_1-80, and 70.8 ± 5.7 μM for apo CaM_1-80. In contrast, both CaM1-148 and CaM_76-148 (apo and Ca²⁺-saturated forms) bound stoichiometrically to Na_v1.2_IQp, with a K_d ≤10 nM. The titrations were fit to a single-site binding isotherm that accounted for total CaM (Table S1); however, only an upper limit for the K_d can be determined under conditions of limiting CaM (Newman, 2008).

Effect of Na_v1.2_IQp on Calcium Binding to CaM

To determine how Na_v1.2_IQp affects the Ca²⁺-binding properties of each domain of CaM, equilibrium titrations were performed on the isolated domains (CaM_1-80, CaM_76-148) and compared to the same sites within CaM_1-148 (Figure 2D). Changes in intrinsic fluorescence intensity of phenylalanine reported on Ca²⁺ binding to sites I and II, while changes in tyrosine intensity reported on binding to sites III and IV. Total free energies of binding are listed in Table S2. Titrations are shown in blue (sites I and II) and red (sites III and IV) in the absence of peptide, and shown in cyan (sites I and II) and orange (sites III and IV) in the presence of peptide. Figure 2D shows that Na_v1.2_IQp had little effect on sites I and II, whether in CaM_1-80 (left panel) or CaM_1-148 (right panel). However, it selectively reduced the Ca²⁺ binding affinity of sites III and IV of CaM_76-148 (middle) by a factor > 500, and CaM_1-148 (right) by a factor > 750 (Table S2).

Residue-Specific Effects of Na_v1.2_IQp on CaM

Effects of Na_v1.2_IQp on apo CaM were also manifested as conformational differences detected in the ¹⁵N/¹H-HSQC spectra. A spectral overlay of apo CaM_1-148 alone and bound to Na_v1.2_IQp showed that half of the observed amide chemical shifts were nearly identical (Figure S1). An overlay of ¹⁵N/¹H-HSQC spectra of apo CaM_1-80 (N-domain) alone and apo CaM_1-148 bound to Na_v1.2_IQp (Figure S2) showed that resonances unperturbed by peptide binding were attributable to the N-domain. In contrast, resonances corresponding to C-domain residues were dramatically affected by peptide binding. Those in the spectrum of the apo CaM_76-148:Na_v1.2_IQp complex were nearly identical to corresponding C-domain residues in apo CaM_1-148:Na_v1.2_IQp (Figure S3). An overlay of three of these ¹⁵N/¹H-HSQC spectra in Figure 3A shows that the resonances of (a) apo CaM_1-80 alone (green), and (b) apo CaM_76-148 bound to Na_v1.2_IQp (blue) closely resembled the spectrum of (c) apo CaM_1-148 bound to Na_v1.2_IQp (red).

Figure 3. 15N/¹H-HSQC Spectra of ¹³C,¹⁵N-CaM.

A: Overlapping peaks in ¹⁵N/¹H-HSQC spectra of apo CaM_1-80 (blue) and apo CaM_1-148 with equimolar Na_v1.2_IQp (red) are boxed by dashed squares, while those overlapped in the apo CaM_76-148 (green) and apo CaM_1-148 with equimolar Na_v1.2_IQp (red) are shown by solid circles. The spectra of apo CaM_1-80 alone and apo CaM_76-148:Na_v1.2_IQp complex account for > 95% of the peaks of apo CaM_1-148-Na_v1.2_IQp B: Overlay of ¹⁵N/¹H-HSQC spectra of apo CaM_76-148 (green) and (Ca²⁺)₂-CaM_76-148 (black), each with equimolar Na_v1.2_IQp. Cross peaks of apo CaM_76-148-Na_v1.2_IQp are labeled. Supported by Figures S1, S2, S3.

Figure 3B shows the effect of Ca²⁺ binding. Almost all resonances in the ¹⁵ N-HSQC spectrum of (Ca²⁺)₂-CaM_76-148 bound to Na_v1.2_IQp (black) were distinct from those of apo CaM_76-148 bound to Na_v1.2_IQp (green), indicating that most backbone amides have different chemical environments, presumably because they have adopted distinct conformations.

Analysis of ¹⁵N/¹H-HSQC spectra of apo CaM_76-148 ± Na_v1.2_IQp (Figure 4A) showed that peptide binding caused significant chemical shift perturbations (Δδ) of the CaM backbone amides (δ_average of 0.51 ppm). It also dramatically increased dispersion of ¹⁵N/¹H-HSQC cross peaks (Figure 4A) by breaking the symmetry of similar chemical environments of homologous residues in the paired EF-hands of apo CaM_76-148. Mapping Δδ onto the sequence of CaM_76-148 (Figure 4B) indicated that the greatest perturbations were observed in the region between helices F and G. For residues 108-117,Δδ_average was 1.47 ppm. Unassigned amides (for residues 76, 132, 133, 134, 135, and 141) are indicated with short gray bars below the abscissa; yellow bars indicate Ca²⁺-binding sites III and IV.

Figure 4. 15N/¹H-HSQC-Detected Changes in Apo CaM_76-148 upon Na_v1.2_IQp Addition.

A: Overlay of ¹⁵N/¹H-HSQC spectra of apo CaM_76-148 (blue) and apo CaM_76-148 bound to Na_v1.2_IQp (red). B: Quantified backbone amide chemical shift perturbations of apo CaM_76-148 upon binding Na_v1.2_Iqp. Ca²⁺-binding sites are yellow bars. Unassigned residues (76, 132, 133, 134, 135, 141) are indicated as small bars below the abscissa. C: ¹⁵N/¹H-HSQC crosspeak intensities at identical contour levels showing residues of apo CaM_76-148 bound to Na_v1.2_IQp. D: Fitted exchange curves and rates for the residues shown in C. E: Bar graph of amide protection factors. Chemical shift perturbations (F) and protection factor magnitudes (G) mapped onto the structure of apo CaM_76-148:Na_v1.2_IQp (unassigned residues are light gray; Na_v1.2_IQp is a gray rod). Supported by Table S3.

Hydrogen/Deuterium Backbone Amide Exchange

Backbone amide hydrogen/deuterium exchange rates of apo CaM_76-148:Na_v1.2_IQp were determined to identify slowly exchanging amides that are usually located in regions of persistent secondary structure. In the ¹⁵N/¹H-HSQC spectrum, 37 amide resonances were detected 30’ after addition of D₂O to a lyophilized sample. Protection factors (Table S3) were calculated from the decay of cross peak intensities (examples shown in Figure 4C) that were fit to a mono-exponential function (Figure 4D) and corrected for intrinsic exchange rates (Figure 4E) (Bai et al., 1993; Molday et al., 1972). Residues with observable exchange rates were used as H-bond restraints during subsequent structure calculations.

Structure of Apo CaM_76-148:Na_v1.2_IQp

Structural statistics of the apo CaM_76-148:Na_v1.2_IQp complex are in Table 1. This structure was determined based on 109 dihedral angular restraints and 1865 unambiguous distance restraints comprised of 457 intra-residue, 298 short, 300 medium, 331 long, 245 intra-peptide, 188 intermolecular, and 46 hydrogen bonds. The restraints are distributed evenly throughout the protein, with a slightly higher number observed in α-helical and β-sheet regions than in the empty Ca²⁺-binding sites III and IV. Figure 5A shows a superposition of the 20 lowest energy structures of the apo CaM_76-148:Na_v1.2_IQp complex that best satisfied the experimental restraints. The interhelical angles calculated by UCSF Chimera (Pettersen et al., 2004) between helices E-F and G-H of these structures were 77.4° ± 1.7 and 78.0° ± 2.3 respectively. The apo CaM_76-148:Na_v1.2_IQp complex was well defined, with an ensemble RMSD of 0.31 ± 0.05 Å (Table 1). Figure S4 shows a Ramachandran plot of the final averaged structure.

Table 1.

Structural statistics and root-mean-square deviation for 20 structures of apo CaM_76-148:Na_v1.2_IQp complex

Structural statisticsa	<SA>	${<\stackrel{‒}{\mathit{SA}}>}_{\mathrm{r}}$
RMSD from experimental distance restraints (Å)b
All (1865)	0.008 ± 0.001	0.007
CaM intra-residue (457)	0.007 ± 0.002	0.005
CaM sequential (298)	0.005 ± 0.004	0.003
CaM medium range (300)	0.009 ± 0.002	0.008
CaM long range (331)	0.006 ± 0.001	0.005
Intra-peptide (245)	0.005 ± 0.002	0.003
CaM-peptide intermolecular (188)	0.012 ± 0.002	0.011
hydrogen bond (46)	0.016 ± 0.001	0.021
RMSD from experimental torsional angle restraints (deg)c
Φ and Ψ angles (109)	0.3 ± 0.03	0.2
Ramachandran Statisticsd
Residues in most favored regions	80.2%
Residues in additional allowed regions	17.6%
Residues in generously allowed regions	2.2%
Residues in disallowed regions	0.0%
CNS Potential Energy (kcal/mol)
Etot	89 ± 4.8	80
Ebond	5 ± 0.5	1
Eang	65 ± 2.9	62
Eimp	5 ± 0.6	4
Erepel	6 ± 1.2	6
Enoe	6 ± 1.2	5
Ecdih	1 ± 0.1	1
Cartesian coordinate RMSD (Å)	N, Ca, and C’	all heavy
<SA> vs. ${<\stackrel{‒}{\mathit{SA}}>}_{\mathrm{r}}$e	0.31 ± 0.05	0.95 ± 0.11

^aWhere <SA> is the ensemble of 20 NMR-derived solution structures of CaM/peptide; ${<\stackrel{‒}{\mathit{SA}}>}_{\mathrm{r}}$ is the mean atomic structure; ${<\stackrel{‒}{\mathit{SA}}>}_{\mathrm{r}}$ is the energy-minimized average structure. The CNS F_repel function was used to simulate van der Waals interactions using a force constant of 4.0 kcal/mol Å⁻⁴ with the atomic radii set to 0.8 times their CHARMM values.

^bDistance restraints were employed with a square-well potential (F_noe = 50 kcal/mol Å⁻²). Hydrogen bonds were given bounds of 1.8-2.4 Å (H-O) and 2.7-3.3 Å (N-O). No distance restraint was violated by more than 0.3 Å in any of the final structures.

^cTorsional restraints were applied with values derived from an analysis of the C’, N, C_a, H_a, and C_b chemical shifts using the TALOS program. Force constant of 200 kcal mol⁻¹ rad⁻² was applied for all torsional restraints.

^dRamachandran statistics calculated with ProCheck. http://www.ebi.ac.uk/thornton-srv/software/PROCHECK/

^eRMSD for CaM residues 80-128 and 134-146 and peptide residues 1905-1920.

Figure 5. Solution Structure of Apo CaM_76-148:Na_v1.2_IQp.

A: Ensemble of 20 lowest energy structures of apo CaM_76-148 (black) bound to Na_v1.2_IQp (red) B: Hydrophobic interaction interfaces of 20 lowest energy structures in 2KXW. Apo CaM_76-148 and residues 1904-1922 of Na_v1.2_IQp (green backbone) with selected interfacial hydrophobic residues shown in sticks. C: Vacuum electrostatic potentials mapped on the surface apo CaM_76-148 and Na_v1.2_IQp. Surfaces are positive (blue), negative (red) or hydrophobic (white). Supported by Figure S4 and Movies S1 and S2.

Interface Between Na_v1.2_IQp and Apo CaM_76-148

The hydrophobic interface of apo CaM_76-148:Na_v1.2_IQp was well defined as judged by the constrained positions of interacting residues (Figure 5B). The buried solvent-accessible hydrophobic surface calculated using GETAREA (Fraczkiewicz and Braun, 1998) was 1393 Ų (751 Ų Na_v1.2_IQp, 642 Ų apo CaM_76-148). In apo CaM_76-148, hydrophobic residues A88, V91, F92, L112, and M145 accounted for 43% of its buried surface, while in Na_v1.2_IQp, hydrophobic residues V1911, I1912, Y1916, Y1919, and L1920 accounted for 53.5% of its buried surface area. To demonstrate how well residues at the interface were constrained, a superposition of the 20 lowest energy structures was rotated by 2° increments in PyMol and captured with ImageJ. Movie S1 shows an overlay of apo CaM_76-148 models (wireframe) bound to Na_v1.2_IQp (cartoon and sticks). Movie S2 shows the apo CaM_76-148 models with all sidechains, followed by models showing only hydrophobic sidechains located at the interface with Na_v1.2_IQp.

Figure 5C shows vacuum electrostatic potentials for coordinates of apo CaM_76-148 and Na_v1.2_IQp corresponding to their conformation in the complex as calculated using Pymol (Schrödinger, LLC.). As expected, favorable electrostatic interactions were observed between negatively charged apo CaM_76-148 and positively charged Na_v1.2_IQp., and the interface is dominated by hydrophobic surface.

Figure 6A shows all contacts between apo CaM_76-148 and Na_v1.2_IQp that are • 4.5 Å (tabulated by by Contacts of Structural Units (CSU) (Sobolev et al., 1999)). Interactions of the IQ residues (I1912, Q1913) with apo CaM_76-148 were identified on the basis of numerous NOEs detected in (a) the 3D ¹³C-edited and ¹⁵N-edited NOESY spectra and (b) the ¹³C-edited and ¹²C,¹⁴N-filtered 3D NOESY spectra. Figures 6B and 6C show NOEs for selected residues of apo CaM_76-148. As seen in the CSU analysis, I1912 inserted into the shallow hydrophobic pocket of semi-open apo CaM_76-148 (Figure 6D) while Q1913 formed hydrogen bonds with backbone atoms L112 and E114 in the turn connecting helices F and G (Figure 6E). A comparison of CaM sequences from 102 species (Ataman et al., 2007) showed that residues E114 and G113 were identical in all, while L112 was 89.2% identical.

Figure 6. Analysis of apo CaM_76-148:Na_v1.2_IQp Binding Interface.

A: Residues of apo CaM_76-148 within 4.5 Å of each residue of Na_v1.2_IQp in 2KXW are listed; four peptide residues (I1912, Q1913, Y1916, Y1919) made • 5 contacts (highlighted in red). 5 contacts (highlighted in red). B: Slices of NOESY spectra showing intermolecular NOEs observed between I1912 of Na_v1.2_IQp and apo CaM_76-148. Slices from the 3D ¹³C-edited NOESY spectra are in blue boxes, while slices from 3D ¹³C-edited and ¹²C,¹⁴N-filtered NOESY spectrum are in red boxes. C: Top panel shows Q1913 carboxamide proton shifts in the 2D ¹²C,¹⁴N-filtered NOESY spectrum, and lower panels show slices of ¹⁵N-edited NOESY spectra showing NOEs between Na_v1.2_IQp Q1913 N_εH₂ and 7 loop backbone amides in apo CaM_76-148. D: Location of apo CaM_76-148 residues shown in panel B relative to I1912. E: Location of apo CaM_76-148 residues shown in panel C relative to Q1913. Dashed lines show hydrogen bonds between the carboxamide of Q1913 and backbone atoms of L112 and E114.

Calcium Binding to Apo CaM_1-148:Na_v1.2_IQp Complex

Figure 7A shows that apo CaM_1-148 bound to Na_v1.2_IQp underwent a structural transition upon Ca²⁺ binding. Figure 7A (left panel) shows the overlay of ¹⁵N/¹H-HSQC spectra of apo CaM_1-148 (red) and (Ca²⁺)₄-CaM_1-148 (black) bound to Na_v1.2_IQp, indicating that Ca²⁺ caused significant change in the chemical environment of apo CaM_1-148 resonances within the complex. For reference, Figure 7A (right panel) compares (Ca²⁺)₄-CaM_1-148 alone and after addition of Na_v1.2_IQp. Figures 7B, C, and D show the ¹⁵N/¹H-HSQC cross peaks for selected residues of the N- and C-domains of apo CaM_1-148 when bound to Na_v1.2_IQp at increasing [Ca²⁺]. The N-domain of apo CaM_1-148 was in fast exchange over the course of the Ca²⁺ titration; representative resonances are shown in Figure 7B. Plots of the normalized Ca²⁺-dependent change in chemical shifts indicated that the N-domain titrated almost completely between 0 and 2 Ca²⁺ equivalents.

Figure 7. Ca²⁺ binding to Apo CaM_1-148:Na_v1.2_IQp.

A: ¹⁵N/¹H-HSQC spectral overlays of apo (red) and (Ca²⁺)₄-CaM_1-148 (black) when bound to Na_v1.2_IQp (left panel) and (Ca²⁺)₄-CaM_1-148 alone (blue) and bound (black) to Na_v1.2_IQp (right panel). B: ¹⁵N/¹H-HSQC cross peak intensity and position as a function of added Ca²⁺ equivalents for selected N-domain residues (E6, G40, V55) of apo CaM_1-148, selected C-domain residues (A102, E114, N137) of CaM_1-148, and three unassigned C-domain residues (X, Y, Z) of CaM_1-148 when Ca²⁺-saturated. The quantified change in chemical shift perturbations as a function of added Ca²⁺ equivalents for these residues are plotted in the lower panels. Supported by Figure S6.

Observation of Ca²⁺-induced changes in peak position and intensity of the C-domain of apo CaM_1-148 bound to Na_v1.2_IQp (Figure 7C) showed that peaks were in intermediate or slow exchange. Thus, changes in peak intensities were used to determine relative populations over the course of the titration. These peaks broadened beyond the limit of detection after addition of 2 equivalents of Ca²⁺ (Figure 7C) as has been seen for Ca²⁺ binding to melittin-bound CaM (Newman, 2008). New peaks that appeared at the midpoint of the Ca²⁺ titration and increased through the addition of 4 equivalents of Ca²⁺ (see Figure 7D) are likely to correspond to residues in the C-domain of (Ca²⁺)₄-CaM_1-148 based on correlations with effects of Na_v1.2_IQp on Ca²⁺-binding shown in Figure 2. These data indicated that, when bound to Na_v1.2_IQp, each domain of CaM_1-148 binds 2 Ca²⁺, and sites I and II have a slightly more favorable Ca²⁺-binding affinity than sites III and IV.

Effect of Ca²⁺ on CaM_1-148:Na_v1.2_IQp

As shown in Figure 2C, apo CaM_1-148 bound to Fl-Na_v1.2_IQp stoichiometrically at a 1:1 ratio, with binding mediated by the C-domain of CaM. If Ca²⁺ binding to CaM_1-148 induced association of the N-domain with Na_v1.2_IQp, then the structure would become more compact. To test this, Ca²⁺ was added to apo CaM_1-148:Fl-Na_v1.2_IQp (Figure S5A). The hydrodynamic behavior of the complex was monitored by fluorescence anisotropy, but no change was observed. Evaluated with the NMR spectra in Figures 3 and 7, this suggests that the Ca²⁺-saturated N-domain is free to bind elsewhere on Na_v1.2 or to another target protein.

Effect of Na⁺ upon Na_v1.2_IQp Binding to CaM

Proteins in the plasma membrane experience large fluctuations in Na⁺ concentration during Na_v1.2 gating (Venosa, 1974) which might affect the affinity of Na_v1.2_IQp for CaM. To test this, Fl-Na_v1.2_IQp was saturated with apo CaM_76-148 or apo CaM_1-148 and then titrated with NaCl in a matching buffer over a range from 0 to 650 mM (Figure S5B). The observed anisotropy of Na_v1.2_IQp was constant, indicating that NaCl has a negligible effect on Na_v1.2_IQp dissociation from CaM.

DISCUSSION

Na_v1.2 contains a CaM-binding IQ motif involved in channel inactivation which is necessary for controlling Na⁺ balance. Thermodynamic studies and the structure of apo semi-open CaM_76-148 bound to Na_v1.2_IQp reported here provide insights regarding the roles of residues that are conserved in IQ motifs in ion channels, and differences between the two domains of CaM. They provides a foundation for further studies of Ca²⁺-mediated switching of channel activity.

Domain-Specific Binding of CaM to Na_v1.2_IQp

The energetics of CaM binding to Na_v1.2_IQp reflect large differences between the two domains of CaM that would permit major regulatory differences. In the complex, apo CaM_76-148 did not contact R1902, a residue implicated in familial autism (R1902C is known to interfere with binding of (Ca²⁺)₄-CaM (Weiss et al., 2003)). This leaves open the question of how (Ca²⁺)₂-CaM_76-148 interacts with R1902, and whether the mutation may exert its influence via effects on the tertiary structure of the channel. The CaM C-domain is necessary and sufficient for binding Na_v1.2_IQp. Its high affinity (nM K_d) for CaM would promote constitutive binding to Na_v1.2 independent of intracellular [Ca²⁺]. Strong, Ca²⁺-independent binding is similar to behavior observed for Na_v1.4 in electrophysiological and FRET studies by Tomaselli and coworkers who showed that CaM remained bound to the IQ motif region as Na_v1.4 adopted multiple conformations in response to voltage changes and Ca²⁺ (Biswas et al., 2008). The weak (μM) affinity of CaM_1-80 for Na_v1.2_IQp and residue-specific NMR studies showing that Na_v1.2_IQp caused barely detectable perturbations of the N-domain (either as a fragment or within full-length CaM) both argue that the N-domain of CaM does not bind Na_v1.2_IQp in a manner that is physiologically significant.

Effect of Na_v1.2_IQp on Ca²⁺ Binding

Unlike kinases that increase the Ca²⁺-binding affinity of both domains of CaM by interacting favorably with the open conformation of each domain, binding of Na_v1.2_IQp selectively lowered the overall Ca²⁺-binding affinity of CaM by 4.03 kcal/mol (Figure 2D). This decrease was attributable almost entirely to interactions of the C-domain of CaM (••G₂ of 3.89 kcal/mol; Table S1). Binding of Na_v1.2_IQp imposed tertiary constraints that lowered the probability of Ca²⁺ binding to sites III and IV. The effects of Na_v1.2_IQp on •G₂ for Ca²⁺ binding to the individual domains of CaM was additive (i.e., similar that observed in CaM_1-148), indicating that interdomain interactions in CaM_1-148 were minimal (Table S1). This was consistent with domain-specific effects seen with fluorescence in Figure 2A, B, C and with NMR in Figures 3, S1, S2, and S3.

Semi-Open Apo C-domain Binds IQ Motifs

Figure 8A shows a structural comparison of (a) apo CaM_76-148 bound to Na_v1.2_IQp (determined in this study) to (b) the C-domains of two apo CaM_1-148 molecules bound to two neighboring IQ motifs found in myosin V (2IX7), (c) apo CaM-like proteins bound to IQ motifs of MYO2P (1M46, 1M45, 1N2D) and (d) ELC bound to myosin heavy chain (3JVT). In all of these, each set of paired EF-hands of the C-domain adopts a semi-open conformation when bound to its respective IQ motif. Moreover, the residues located at positions “0”/Ile and “1”/Gln of the IQ motif interact with a similar subset of residues in the apo C-domain.

Figure 8. Conserved Features of Apo Semi-Open EF-Hand Domains and Model for CaM:Na_V1.2 Binding to Recognition Motifs.

A: Structural comparison of apo CaM_76-148 (red):Na_v1.2_IQp (green rod) complex (2KXW) with the C-domains of apo CaM and apo CaM-like proteins (gray) bound to canonical IQ motifs (2IX7, 1M45, 1M46, 1ND2, and 3JVT). I and Q residues of all IQ motifs are sticks; C_α of residue 113 of CaM is a red sphere. B: Structural comparison of apo CaM_76-148 (red):Na_v1.2_IQp (green rod) complex with the apo C-domain of (Ca²⁺)₂-CaM_1-148 (orange) bound to SK_p (cyan). I1912 and Q1913 (Na_v1.2_IQp) and L428 and N426 (SK_p) are ball-and-stick; CaM residue 113 shown as red or orange sphere. Sequences of Na_v1.2_IQp and SK_p are aligned structurally based on the hydrophobic interaction of I1912 and L428 and carboxamide-containing Q1913 and N426. C: Two alternative polarities for peptides binding to the C-domain of CaM; _NF-G_C depicts the N-terminus (blue) closest to helix F (red), while _CF-G_N depicts the C-terminus (magenta) closest to helix F and farthest from helix G (orange). D: Relative positions of the hydrophobic (cyan spheres) and carboxamide (green spheres) residues. The blue-to-magenta color gradient indicates _NF-G_C polarity. E: Proposed model of CaM_1-148 interacting with Na_v1.2. The IQ motif interacts solely with the semi-open C-domain of CaM. The N-domain is in a closed conformation that may interact with an unidentified site X elsewhere on Na_v1.2. Under Ca²⁺-saturating conditions, the C-domain remains bound to Na_v1.2_IQp; the open N-domain may bind to site Y (potentially distinct from site X). Supported by Figure S6.

At position “0”, a hydrophobic (typically branched) residue contacts hydrophobic sidechains of the semi-open C-domain, burying what would otherwise be solvent-exposed hydrophobic surface. Conservation of Gln at position “1” of canonical IQ motifs preserves hydrogen bonding interactions between its carboxamide and the backbone of CaM residues located in the turn connecting α-helices F and G (Figure 6E). As observed in the ¹⁵N/¹H-HSQC data (Figure 4A and 4D), residues located in this turn exhibited the largest chemical shift perturbation upon peptide binding. Examination of amide hydrogen exchange protection factors mapped onto the structure (Figure 4F) suggests that this turn is solvent-exposed. Therefore, the H-bonds involved in this turn may be different from those observed in classic α-helices and β-sheets.

Tomaselli and coworkers, studying Na_V1.4, showed that residues I and Q are required for gating, and that substitution by alanine eliminated the proximity of CaM to the C-terminus of Na_v1.4 under all conditions (e.g., resting and high Ca²⁺) (Biswas et al., 2008). The interactions observed between the IQ residues of Na_v1.2_IQp and apo CaM could not be satisfied by alanine substitutions. With Ala at position “0”, there would be less of a penalty for solvent exposure relative to Ile, Leu, Met, or Phe at that position. With Ala at position “1”, there would be no carboxamide interaction between the sodium channel and atoms in the backbone of the loop between α-helices F and G of CaM.

Semi-Open Apo C-Domain Binding to a Non-IQ motif

An apo domain of CaM can bind to a sequence without an IQ motif, as seen in the 1G4Y structure of partially apo CaM bound to a peptide (SK_p, gating domain residues 413-488) derived from the small conductance potassium channel (SK Channel) (Schumacher et al., 2001). A superposition of apo CaM_76-148:Na_v1.2_IQp and the apo C-domain of 1G4Y showed that both adopt the semi-open conformation (Figure 8B).

SK_p shares a key feature seen in the IQ motif-containing peptides bound to apo CaM, CaM-like proteins, and ELC. Figures 8A and 8B show that a carboxamide-containing side chain (Asn in SK_p vs. Gln in IQ motifs) forms hydrogen bonds with backbone residues between the F and G helices of CaM. This conserved hydrogen-bonding network influences the position of the IQ motif or SK_p helix relative to the apo C-domain. Another similarity was apparent in the conserved hydrophobic interaction between I1912 (Na_v1.2_IQp) or L428 (SK_p) and the core of the apo C-domain (Figure 8B), even though these residues differ in their position in the primary sequence relative to the carboxamide side chain (Q1913 in Na_v1.2_IQp or N426 in SK_p).

As depicted in Figure 8C, both SK_p and Na_v1.2_IQp adopted the _NF-G_C orientation. where F and G refer to the helices of CaM between sites III and IV, and N and C refer to the termini of the CaM-binding sequence. In Na_v1.2_IQp, the conserved hydrophobic residue (I1912) that inserts into the core of apo CaM_76-148 precedes the conserved carboxamide-containing residue (Q1913). Although in SK_p, L428 (the residue homologous to I1912) is located 2 residues distal to the carboxamide-containing residue (N426), the relative angular positions of IQ and N-L are the same (Figure 8D). This illustrates how peptides can recognize the semi-open conformation of apo C-domain regardless of whether they contain an IQ motif.

Orientations of IQ motif Binding to CaM

If the contacts seen in Figure 8A were maintained by all apo CaM-binding motifs, then all canonical IQ motifs would bind to the semi-open C-domain of apo CaM in the _NF-G_C orientation. At this time, there are too few deposited structures of apo CaM bound to IQ motifs to test this prediction. Figure 9 (upper panel) depicts available structures of apo CaM and (Ca²⁺)₄-CaM bound to peptides containing canonical IQ motifs derived from indicated target proteins. In these cases, position “0” is occupied by Ile in all structures except that of myosin V^B (which has Val), and position “1” is only occupied by Gln. The orientation of peptide binding to each of this is _NF-G_C (N- and C- termini of peptide blue and magenta respectively), as depicted by rightward arrows.

Figure 9. Orientation of Canonical and Non-Canonical IQ Motifs Bound to CaM.

CaM is bound to peptides containing canonical (upper panel) or non-canonical (lower panel) IQ motifs. Each ribbon diagram is labeled with PDB ID and degree of Ca²⁺ binding (yellow spheres). In CaM, helix F is red, and G is orange. Helical polarity of each IQ motif is indicated by a color gradient from N- (blue) to C-terminus (magenta) and corresponding arrow. Residues at positions 0 (cyan) and 1 (green) are indicated by spheres. In the lower panel, sequences of binding motifs are aligned according to positions 0 and 1.

We infer that apo CaM binds to Na_v1.2_IQp and other canonical IQ motifs in the _NF-G_C orientation because of the Gln located at position “1”. The simplest model for Ca²⁺-mediated regulation of Na_V1.2 is that the _NF-G_C orientation is also used by (Ca²⁺)₄-CaM when binding Na_v1.2, because _NF-G_C has been observed in all structures of (Ca²⁺)₄-CaM bound to canonical IQ-motifs from other channels. Adoption of the opposite (_CF-G_N) orientation by (Ca²⁺)₄-CaM would require the C-domain of apo CaM to release from the IQ motif upon Ca²⁺ influx, and re-associate when Ca²⁺-saturated.

Both orientations of (Ca²⁺)₄-CaM have been observed (Figure 9, lower panel) in complexes with peptides that contain non-canonical IQ motifs (e.g., sequences having Met or Phe at position “1”, rather than Gln). Structures of (Ca²⁺)₄-CaM bound to fragments with these sites from Ca_v2.1, Ca_v2.2, and Ca_v2.3 adopted alternative orientations in different studies. Different peptide lengths of Ca_v2.1 and Ca_v2.3, and distinct crystallization conditions were used in these studies, which may account for the structural differences (Kim et al., 2008; Mori et al., 2008). At this time, there are no available structures of apo CaM bound to a Ca_V IQ-motif.

Role of the N-domain of CaM_1-148 in Na_v1.2 Recognition

The N-domains of both apo and (Ca²⁺)₄-CaM_1-148 do not interact with Na_v1.2_IQp as shown by titrations monitored by steady-state fluorescence (Figures 2A and 2B), fluorescence anisotropy (Figure 2C), and additive ¹⁵N/¹H-HSQC spectra (Figures 3A, S2). However, studies presented here do not rule out that it may interact elsewhere on Na_V1.2, as suggested schematically in Figure 8E by site X (for the apo closed N-domain) and site Y (for the Ca²⁺-saturated open N-domain). The interaction of apo N-domain might mimic that seen in a complex of CaM bound to a fragment of the anthrax edema factor (Drum et al., 2002) which involves a helix-helix interaction rather than an interface with the cleft of CaM.

Previous studies (Mori et al., 2000) showed that Na_v1.2 contains a second CaM-binding motif comprised of residues 1913-1938 (Na_v1.2_{BAA-1913-1938}) that binds to (Ca²⁺)₄-CaM, but not apo CaM. This sequence has characteristics of a classic BAA-motif that preferentially binds the open conformation of Ca²⁺-saturated CaM. Thus, the N-domain could bind to a site within Na_v1.2_1913-1938 while the C-domain remains tethered to Na_v1.2_IQp (Figure 8E). To test this model, Fl-Na_v1.2_IQp was titrated with apo CaM_1-148 and monitored by fluorescence anisotropy (Figure S6A) to assure saturation. When Na_v1.2_1913-1938 was added in 6-fold excess, it competed for Na_v1.2_IQp; there was no evidence for simultaneous binding of both Na_v1.2 2_IQp and Na_v1.2_1913-1938 to (Ca²⁺)₄-CaM. This suggests that a binding site for Ca²⁺-saturated N-domain may be located elsewhere in Na_v1.2 or another protein (shown schematically as site Y in Figure 8E). For example, the intracellular linker between domains III and IV of other sodium channels binds (Ca²⁺)₄-CaM (Sarhan et al., 2009; Shah et al., 2006). Alternatively, the conformation of these residues within Na_v1.2 may not be fully represented by a peptide.

Conclusion

The studies reported here provide the first structure of apo CaM_76-148:Na_v1.2_IQp. The structural interface and the binding energy for association of Na_v1.2_IQp and apo CaM are mediated by the C-domain alone. This implies that a preferred binding site for the N-domain of CaM must lie elsewhere. In the apo CaM_76-148:Na_v1.2_IQp structure, CaM adopts a semi-open conformation similar to the C-domain of apo CaM alone (Swindells and Ikura, 1996), CaM-like proteins (Terrak et al., 2003), essential light chain (ELC) (Swindells and Ikura, 1996), apo CaM when bound to IQ motifs in myosin (Houdusse et al., 2006) and the apo C-domain of CaM bound to the SK channel (Schumacher et al., 2001). At the CaM-peptide interface in apo CaM_76-148:Na_v1.2_IQp, the highest number of noncovalent interactions was observed between apo CaM_76-148 and 4 residues (I1912, Q1913, Y1916, and Y1919) of Na_v1.2_IQp. These sidechains are strongly conserved in IQ motifs found in Na_V isoforms, with the exception of the Y1919 which is a histidine in four of the ten isoforms (Figure 1A). There are now multiple structures of (Ca²⁺)₄-CaM bound to peptides representing IQ motifs of ion channels, but there are no cases in which the corresponding complex of apo CaM bound to the same peptide is available. Discovering how Ca²⁺-mediated switching of CaM regulates Na_V1.2 will require future analysis of calcium-saturated CaM_76-148:Na_v1.2_IQp, studies of apo CaM binding to Na_v1.2_IQp having substitutions that mimic mutations known to alter gating properties, and dissection of interactions of CaM with larger fragments of the intracellular C-terminal sequence of the channel.

EXPERIMENTAL PROCEDURES

Calmodulin

Isotopes were obtained from Cambridge Isotope Laboratories (Andover, MA). CaM over-expression was IPTG-induced in E. coli BL21(DE3) cells containing a recombinant pT7-7 vector expressing full-length (residues 1-148), N-domain (1-80), or C-domain (76-148) of Paramecium CaM (C. Kung, University of Wisconsin, Madison, WI) which is 88% identical to mammalian CaM. A genetic screen of Paramecia demonstrated that CaM mutations directly affected Na⁺ channel function and chemotactic behavior (Kung et al., 1992). ¹⁵N-labeled proteins were over-expressed in minimal media (2 g/L unlabeled glucose as carbon source, 1 g/L ¹⁵NH₄Cl as sole nitrogen source). ¹³C,¹⁵N-labeled proteins were produced using 2 g/L ¹³C-glucose as sole carbon source and 1 g/L ¹⁵NH₄Cl as sole nitrogen source. CaM was purified as described previously (Putkey et al., 1985). Proteins were 97-99% pure as judged by silver-stained SDS-PAGE gels; concentration was determined by UV spectroscopy of CaM native at pH 7.4 or denatured with NaOH.

Peptides

GenScript Corporation (Scotch Plains, NJ) synthesized peptides representing residues 1901-1927 of the Na_v1.2 α-subunit (KRKQEEVSAIVIQRAYRRYLLKQKVKK, 3.36 kDa) with (Fl-Nav1.2_IQp) or without (Nav1.2_IQp) fluorescein at the N-terminus. The W.M. Keck Biotechnology Resource Center (New Haven, CT) synthesized Na_v1.2_{BAA-1913-1938} representing residues 1913 to 1938 of the Na_v1.2 α-subunit (QRAYRRYLLKQKVKKVSSIYKKDKGK).

Emission Spectra

Emission spectra (λ_ex 430 nm, bandpasses 2 nm (excitation), 10 nm (emission)) of Fl-Na_v1.2_IQp (1 μM) ± 3 μM CaM_1-148, CaM_1-80 or CaM_76-148 in 50 mM HEPES, 100 mM KCl, 5 mM NTA, 50 μM EGTA, and 1 mM MgCl₂ (pH 7.4) were collected at 22 °C with a PTI-QM4 Fluorimeter (PTI, Birmingham, NJ). Apo samples were Ca²⁺-saturated by addition of CaCl₂ in matching buffer to a final concentration of 10 mM. Spectra were normalized to λ_max of Fl-Na_v1.2_IQp alone and corrected for dilution.

Equilibrium Calcium Titrations

Ca²⁺ titrations of sites I and II (N-domain) and sites III and IV (C-domain) of CaM alone (6 μM) or with 9 μM Na_v1.2_IQp (50% excess) were monitored at 22 °C with a PTI-QM4 Fluorimeter (Photon Technology International, Birmingham, NJ) and analyzed as described previously (Theoharis et al., 2008).

Na_v1.2_IQp Titration of CaM_1-148

¹⁵N/¹H-HSQC spectroscopy monitored a titration of Na_v1.2_IQp into apo ¹⁵N-CaM_1-148 (400 μM) in 10 mM D₄-imidazole, 100 mM KCl, 50 μM D₁₆-EDTA, 0.01% NaN₃, pH 6.8. Upon saturation of apo CaM_1-148 with Na_v1.2_IQp, CaCl₂ was added to a final concentration of 5 mM. Na_v1.2_IQp and Ca²⁺ saturation of CaM_1-148 were confirmed by plateaus in the intensity of peaks corresponding to either the Na_v1.2_IQp:apo CaM_1-148 or Na_v1.2_IQp:(Ca²⁺)₄-CaM_1-148.

Preparation of ¹³C,¹⁵N-CaM_76-148:Na_v1.2_IQp Complex

A 1.5 mM 1:1 complex of ¹²C,¹⁴N-Na 1.2_IQp:¹³C, ¹⁵N-apo CaM_76-148 was prepared in 10 mM D₄-imidazole, 100 mM KCl, 50 μM D₁₆-EDTA, 0.01% NaN₃, pH 6.8. Nitrogen-based experiments were conducted in 90% H₂0 / 10% D₂0, while carbon-based experiments were conducted in 100% D₂O. All spectra, with the exception of amide exchange, were collected on samples in Shigemi (Allison Park, PA) microscale NMR tubes whose magnetic susceptibility was matched to D₂O.

¹⁵N/¹H-HSQC Monitored Amide Exchange of ¹⁵N-CaM_76-148:Nav1.2_IQp Complex

The ¹²C,¹⁴N-Na_v1.2_IQp:¹³C, ¹⁵N-apo CaM_76-148 complex (described above) was lyophilized in a Speed-Vac Model VG-5 (Savant). The complex was re-suspended in 99.96% D₂O (Cambridge Isotope Laboratories, Andover, MA). HMQC spectra prior to lyophilization in H₂O were identical to those taken in D₂O. The pH of the sample was not adjusted after addition of D₂O. The sample was placed immediately in a Bruker 500 MHz Avance II spectrometer; data acquisition began within 5 min. ¹⁵N/¹H-HSQC spectra were acquired at intervals of 30 min for 23 h. Spectra were processed with NMRPipe (Delaglio et al., 1995). Individual peak intensities were determined with Sparky (Goddard and Kneller). Using the Solver function in Microsoft Excel, peak intensities were fit to Equation 1.

$$I=I_0{e}^{-(t-t_0)∕\tau }+b$$ (1)

where I is intensity at time t, I₀ is initial intensity, t₀ is time zero, τ is the apparent decay lifetime, and b represents intensity at the asymptote.

Structure Determination

NMR spectra were collected at 25 °C on either a Bruker Avance II 500 MHz or cryoprobe-equipped 800 MHz spectrometer. The ¹H, ¹⁵N, and ¹³C resonances of the apo CaM_76-148 backbone were assigned using triple resonance experiments (HNCA, HN(CO)CA, HNCACB, HN(CO)CACB, HNCO, and HN(CA)CO) (Yamazaki et al., 1994) with uniformly ¹³C,¹⁵N-labeled apo CaM_76-148 bound to unlabeled Na_v1.2_IQp. ¹H_α resonances were assigned from an ¹⁵N-edited TOCSY spectrum using a uniformly ¹⁵N-labeled protein (Clore and Gronenborn, 1994) and from HA(CACO)NH experiments using a uniformly ¹⁵N and ¹³C-labeled sample. Side chains were assigned from 3D H(CCO)NH-TOCSY, C(CO)NH-TOCSY, HCCH-TOCSY, ¹⁵N-edited TOCSY, and ¹⁵N or ¹³C edited NOESY spectra (Clore and Gronenborn, 1994; Fesik and Zuiderweg, 1988).

CaM was ¹³C, ¹⁵N-labeled but Na_v1.2_IQp had natural abundance isotopes. We used isotope-filtered 2D TOCSY and NOESY experiments with mixing times of 26 to 46 ms (Gemmecker et al., 1992; Ikura and Bax, 1992; Otting and Wuthrich, 1989) to assign the unlabeled Na_v1.2_IQp. Isotope-filters for ¹³C and ¹⁵N were employed twice (once preceding t₁ and once preceding t₂) to effectively suppress the signals derived from ¹³C-attached protons and ¹⁵N-attached protons. As a result, 2D spectra for protons located on the unlabeled peptide were obtained; thus, only intrapeptide signals were observed in these doubly isotope-filtered spectra. To assign intermolecular NOEs between the unlabeled peptide and ¹³C,¹⁵N-CaM, we used ¹³C-edited and ¹⁴N,¹²C-filtered 3D NOESY experiments with mixing times of 80 to 120 ms as described previously (Burgering et al., 1993). Representative intrapeptide and intermolecular NOEs of the CaM/peptide complex were shown in Figure 6. All spectra were processed with NMRPipe (Delaglio et al., 1995) and analyzed with Sparky (Goddard and Kneller).

Apo CaM_76-148 :Na_v1.2_IQp Structure Calculations

Structures of the complex were generated using a torsion-angle molecular dynamics protocol (Stein et al., 1997) with CNS (Brunger et al., 1998). Structure calculations employed 1819 NMR-derived distance restraints from the analysis of 3D ¹⁵N- and ¹³C-resolved NOESY spectra acquired with a mixing time of 120 ms (Fesik and Zuiderweg, 1988). The NOE-derived distance restraints were given upper bounds of 3.0, 4.0, 5.0, and 6.0 Å based upon measured NOE intensities. From an analysis of amide exchange rates measured from a series of ¹⁵N/¹H-HSQC spectra recorded after the addition of D₂O, 46 hydrogen bonds for the α-helices were included in the structural calculations. In addition, 109 Φ and angular restraints derived from an analysis of C, N, C_α, H_α, and C_β chemical shifts using the TALOS program (Cornilescu et al., 1999) were included in the structural calculations. A square-well potential was employed to constrain the NMR-derived distance restraints with F_NOE set to 150 and 50 kcal mol⁻¹ Å⁻² during stages of high temperature and slow-cooling torsion angle dynamics and final stage of conjugate gradient minimization, respectively. Force constants of 100 and 200 kcal mol⁻¹ rad⁻² were applied to all torsional restraints during the stage of high temperature torsion angle dynamics and the rest stages of structural calculations, respectively.

Quantification of Chemical Shifts

To determine the change in chemical shift upon Na_v1.2_IQp binding to apo CaM_76-148, or Ca²⁺ binding to apo CaM_1-148, net perturbations in both the ¹H and ¹⁵N dimensions were quantified using the modified Pythagorean theorem in Equation 2.

$$\Delta \delta \left(\mathrm{ppm}\right)={[{\left({\Delta }^{1}H\right(800.224\mathit{Hz}∕\mathit{ppm}\left)\right)}^2+{\left({\Delta }^{15}N\right(81.095\mathit{Hz}∕\mathit{ppm}\left)\right)}^2]}^{1∕2}$$ (2)

Here, Δδ refers to the linear displacement of a specific resonance peak from its initial position in the reference spectrum.

CaM Binding to Na_v1.2_IQp

Binding of CaM to Fl-Na_v1.2_IQp (50 mM HEPES, 1 mM MgCl₂, 5 mM NTA, 50 μM EGTA, pH 7.4) was monitored by fluorescence anisotropy at 22 °C using a Fluorolog 3 (Jobin Yvon, Horiba) fluorimeter. A slight excess of CaM was added (reaching a 1.2:1 ratio of CaM:Na_v1.2_IQp). The complex was then titrated with a matching buffer that contained 5 M NaCl. Changes in the anisotropy of Fl-Na_v1.2_IQp were monitored and analyzed numerically as described previously (Theoharis et al., 2008).

Accession Numbers

The coordinates for this structure were deposited in May, 2010 with the PDB code of 2KXW. A preliminary report of this structure was presented at the 2010 Annual Biophysical Meeting (Feldkamp, Yu, and Shea, 2010).

Note Added in Proof

During revision of this article, Chagot and Chazin deposited and reported a structure (2L53) of apo CaM_1-148 bound to a peptide representing the IQ motif of NaV1.5 (E1901-L1927) with a short N-terminal tag (GPGS) (Chagot and Chazin, 2011). Superposition of 2KXW and 2L53 using pairs of 68 Cα atoms located in the well defined secondary structure elements showed a RMSD of 1.2 Å. In both 2KXW and 2L53, the apo C-domain adopts the semi-open tertiary conformation. The peptides bound in the same orientation and location despite differences in their sequences. The IQ residues in both peptides interact with almost identical sets of residues on CaM. Slight differences are observed in the two calcium binding loops which are very mobile in the absence of bound calcium. The orientation of the sidechain of Y1916 in our structure is different from that of the corresponding residue F1912 in 2L53, probably reflecting the slight differences in the sequences of both CaM and peptides used in these studies.

Abbreviations

NOE

Nuclear Overhauser effect