Determinants in CaV1 Channels That Regulate the Ca2+ Sensitivity of Bound Calmodulin*

Calmodulin binds to IQ motifs in the α1 subunit of CaV1.1 and CaV1.2, but the affinities of calmodulin for the motif and for Ca2+ are higher when bound to CaV1.2 IQ. The CaV1.1 IQ and CaV1.2 IQ sequences differ by four amino acids. We determined the structure of calmodulin bound to CaV1.1 IQ and compared it with that of calmodulin bound to CaV1.2 IQ. Four methionines in Ca2+-calmodulin form a hydrophobic binding pocket for the peptide, but only one of the four nonconserved amino acids (His-1532 of CaV1.1 and Tyr-1675 of CaV1.2) contacts this calmodulin pocket. However, Tyr-1675 in CaV1.2 contributes only modestly to the higher affinity of this peptide for calmodulin; the other three amino acids in CaV1.2 contribute significantly to the difference in the Ca2+ affinity of the bound calmodulin despite having no direct contact with calmodulin. Those residues appear to allow an interaction with calmodulin with one lobe Ca2+-bound and one lobe Ca2+-free. Our data also provide evidence for lobe-lobe interactions in calmodulin bound to CaV1.2.

CDF in voltage-dependent Ca 2ϩ channels by competing with CaM for Ca 2ϩ (16).
The conformation of the carboxyl terminus of the ␣ 1 subunit is critical for channel function and has been proposed to regulate the gating machinery of the channel (17,18). Several interactions of this region include intramolecular contacts with the pore inactivation machinery and intermolecular contacts with CaM kinase II and ryanodine receptors (17, 19 -22). Ca 2ϩ regulation of Ca V 1.2 may involve several motifs within this highly conserved region, including an EF hand motif and three contiguous CaM-binding sequences (10,12). ApoCaM and Ca 2ϩ -CaM-binding sites appear to overlap at the site designated as the "IQ motif" (9,12,13), which are critical for channel function at the molecular and cellular level (14,23).
Differences in the rate at which 1,2-bis(o-aminophenoxy)ethane-N,N,NЈ,NЈ-tetraacetic acid affects CDI of Ca V 1.1 and Ca V 1.2 could reflect differences in their interactions with CaM. In this study we describe the differences in CaM interactions with the IQ motifs of the Ca V 1.1 and the Ca V 1.2 channels in terms of crystal structure, CaM affinity, and Ca 2ϩ binding to CaM. We find the structures of Ca 2ϩ -CaM-IQ complexes are similar except for a single amino acid change in the peptide that contributes to its affinity for CaM. We also find that the other three amino acids that differ in Ca V 1.2 and Ca V 1.1 contribute to the ability of Ca V 1.2 to bind a partially Ca 2ϩ -saturated form of CaM.

EXPERIMENTAL PROCEDURES
Materials-All peptides used were either synthesized in the core facility at the Baylor College of Medicine under the direction of Dr. Richard Cook or by GenScript Corp. (Piscataway, NJ). Ca V 1.1 IQB and Ca V 1.2 IQB peptides had a six-carbon biotin linker attached via an additional modified lysine at the carboxyl terminus of the peptide (CPC Scientific Inc., San Jose, CA). Calibrated Ca 2ϩ buffers were ordered through Invitrogen. High grade reagents for crystallization experiments were purchased from Hampton Research (Aliso Viejo, CA) or from Sigma. N-Trimethoxysilylpropyl-N,N,N-trimethylammonium chloride for silylizing glassware was purchased through United Chemical Technologies (Bristol, PA).
Mutant CaM Constructs-Fluorescent recombinant mammalian CaM constructs (F19W, F92W, and F92W/E12Q) were prepared as described previously (1,24). Constructs for the mammalian Ca 2ϩ -binding mutants E12QCaM and E34QCaM were also made previously (13). To prepare F19W/E34Q, we used a DNA construct for F19W in the pET3a vector. The cDNA primers (41 bp) were designed to introduce a glutamine substitution for a glutamate in the Ϫz position of EF hands 3 and 4 (residues Glu-104 and Glu-140) by PCR site-directed mutagenesis. F19W/E12Q and F92W/E34Q were made similarly. Other mutants created using similar procedures included the following: 1) CaM mutants with cysteine mutations in the N-lobe (T34C) and the C-lobe (T110C); 2) E12Q/T34C/T110C; and 3) E34Q/T34C/T110C. The cDNA sequences were verified at the DNA Sequencing Core Facility at the Baylor College of Medicine. Protein expression and purification of these mutants were previously described (25).
Structure Determination-The complex of CaM-Ca V 1.1 IQ peptide complex was purified following the procedure described for the CaM-Ca V 1.2 IQ peptide complex (28). The complex was concentrated to 10 mg/ml in a buffer containing 20 mM MOPS, pH 7.4, 150 mM NaCl, and 10 mM CaCl 2 . Crystals were grown by vapor diffusion by mixing 2 l of complex into a 4-l drop of a milieu from the well containing 32% polyethylene glycol 3500, 50 mM Tris, 50 mM MgCl 2 . Large football-shaped crystals grew to full size in 2 weeks in a Torrey Pines Scientific incubator (San Marcos, CA) at 20°C. Data were collected at the Center for Advanced Microstructures and Devices Gulf Coast Protein Crystallography Consortium beamline at the Louisiana State University Center for Advanced Microstructures and Devices (Baton Rouge, LA). The HKL2000 software package was used for data set reduction (29). The structure of CaM/ Ca V 1.1 IQ was determined by the molecular replacement method (30) using the CaM/Ca V 1.2 IQ structure (PDB code 2f3y) (28) as the search model. The parameters used for solving the crystal structure are presented in Table 1. Structure refinement and analyses were performed using CNS (31) and the CHAIN graphics program (32). The structure was deposited to the Protein Data Bank (PDB) with the PDB code 2vay.
Determination of Affinity of CaM, E12QCaM, and E34QCaM for Peptides-Surface plasmon resonance (SPR) with a Bia-core3000 instrument (GE Biacore, Inc., Piscataway, NJ) was used to assess the affinity of CaM, E12QCaM, and E34QCaM for the biotinylated wild type and mutant IQ peptides (Table 2). Sensor chip SA was conditioned according to the manufacturer's protocol, and biotinylated wild type or mutant IQ peptides were immobilized to a sensor chip SA by loading 100 l of 3 nM  (1). Data were fit by nonlinear regression analyses with either a standard dose-response curve or, if appropriate, a biphasic dose-response curve as modeled previously (1,33).
For FRET measurements, 200 nM labeled CaM was incubated with 1 M peptide for 1 h at room temperature in a 20 M Ca 2ϩ buffer from Molecular Probes (Ca 2ϩ calibration buffer kit 3). Fluorescence was measured at 400 -625 nm with an SLM8000 spectrofluorometer with 350 nm excitation. Settings included 8-nm bandpass excitation and emission slits, 309 nm cut-on excitation filter, 395 nm cut-on emission filter, and 1-s integration times. All spectra had the same spectral maximum near 493 nm, and bar graphs reflect the observations at this wavelength.
Ca 2ϩ Dissociation Kinetics-Stopped-flow experiments were performed as described (1,24) using an Applied Photophysics instrument (model SX.18MV; Leatherhead, UK) to measure rates of Ca 2ϩ dissociation (k off ) at 22°C. Instrument parameters are the same as described (1). Represented data were averages of 5-8 individual traces fit with either a single or double exponential curve after premixing reached equilibrium. Tryptophan fluorescence was measured after rapidly mixing equal volumes where A is the amplitude of the fluorescence change, and k is the rate at which the change is occurring. Ca 2ϩ dissociation rates (1) were also determined with Quin-2.
where A 1 and A 2 are component amplitudes of the fluorescence change, and k 1 and k 2 are the corresponding rates of change. The double exponential reflects the Ca 2ϩ dissociation rates from both the N-lobe (fast) and the C-lobe (slow). The molar quantity of Ca 2ϩ dissociating from CaM was calculated by monitoring the increase in Quin-2 fluorescence with increasing concentrations of Ca 2ϩ standards (10,20,40, and 80 M) (34).

Identification of IQ Residues That Interact Directly with
Ca 2ϩ -CaM-To define the determinants for interaction of Ca V 1.1 IQ and Ca V 1.2 IQ (differing by four amino acids) with CaM, we obtained crystals of Ca 2ϩ -CaM bound to the Ca V 1.1 IQ peptide and determined the structure to 1.94 Å resolution ( Table 1). The crystals are isomorphous to those formed with Ca V 1.2 IQ, and the crystal properties (Table 1) are very similar to those of the complex with Ca V 1.2 IQ peptide, but the CaM/ Ca V 1.2 structure was determined at a higher resolution of 1.45 Å (28).
The N-lobe of CaM binds the amino terminus of the Ca V 1.1 IQ peptide, and the C-lobe binds the carboxyl terminus of Ca V 1.1 IQ in a parallel arrangement similar to that seen with Ca V 1.2 IQ (Fig. 1, A and D) and other IQ peptides (28,35,36). As expected from the identity of the amino-terminal portions of the Ca V 1.1 IQ and Ca V 1.2 IQ peptides, the N-lobe of CaM binds the amino-terminal peptide sequences in nearly identical Calmodulin and Ca V 1 IQ Peptides JULY 24, 2009 • VOLUME 284 • NUMBER 30 conformations ( Fig. 1, B and E). The root mean square deviation (r.m.s.d.) for the backbone atoms of the superimposed N-lobes is 0.34 Å. This value is about 2-fold smaller than the r.m.s.d. of 0.75 Å for the superimposed C-lobes, both of which are smaller than the overall backbone r.m.s.d. of 0.88 Å for the superimposed structures of the two CaM-peptide complexes. This indicates a difference in the relative orientation between the two lobes in the two structures.
Other differences are detected in the C-lobe interactions. There is a salt bridge between Arg-1539 on the Ca V 1.1 IQ peptide and Glu-127 of CaM that was not observed in our Ca V 1.2 IQ-CaM structure (28), but it was seen in the structure by Van Petegem et al. (36). Of the four nonconserved residues, only His-1532 on Ca V 1.1 IQ (Fig. 1, A and C) and Tyr-1675 in Ca V 1.2 IQ (Fig. 1, D and F) contact Ca 2ϩ -CaM. In Ca 2ϩ -CaM/Ca V 1.2 IQ, a water molecule forms hydrogen bonds both with the hydroxyl group side chain of the Tyr-1675 and with the carbonyl oxygen of the Met-124 main chain (28). A hydrophobic pocket formed by four C-lobe methionine side chains is more collapsed around His-1532 (Fig. 1C). This Ca 2ϩ -CaM methionine pocket expands around the bulkier Tyr-1675 on the Ca V 1.2 IQ peptide (Fig. 1F). Although the methionine side chains are slightly farther away from Tyr-1675, the ␣-carbon atoms are actually drawn inward toward the tyrosine. When comparing structures, the main chain ␣-carbon atoms of residues of Met-109, Met-124, Met-144, and Met-145 are displaced by 0.54, 0.18, 0.52, and 0.38 Å, respectively. The ␣-carbon to ␣-carbon distance from Met-109 to Met-145 is 0.41 Å closer in the presence of Ca V 1.2 IQ than in the presence of Ca V 1.1 IQ.
The ␣-carbons are 0.12 Å closer from Met-124 to Met-144. The difference in ␣-carbon positions correlates with small perturbations in backbone structure. As mentioned above, the r.m.s.d. of the C-lobes from both complexes is 0.75 Å, but it is influenced mainly by the difference in the relative positions of the ␣6-␣7 loops; the r.m.s.d. measured without the ␣6-␣7 loop drops to 0.63 Å. Because 0.63 Å is still considerably larger than the r.m.s.d. of the N-lobes, it likely reflects variations in the C-lobe conformations caused by the different peptides.
Because the pH of the crystallization solution is 8.3, His-1532 of Ca V 1.1 IQ is likely in the neutral or nonprotonated form. A charged residue in this position is predicted to be unfavorable. To assess this, we created peptides with an H1532D replacement in Ca V 1.1 IQ. This peptide does not bind Ca 2ϩ -CaM (data not shown), suggesting that the residue in this position (His-1532 of Ca V 1.1 IQ or Tyr-1675 of Ca V 1.2 IQ) is important for the interactions within the methionine pocket. A useful side note is that the H1532D mutation in Ca V 1.1 can be used to abolish CaM binding.
How Nonconserved Amino Acids in IQ Motif Regulate the Affinity for Ca 2ϩ -CaM-To address the question of how the four nonconserved amino acids affect the affinity of the IQ peptides for Ca 2ϩ -CaM, we synthesized the peptides shown in Table 2 and assessed their affinity using SPR. Ca V 1.1 IQ, Ca V 1.2 IQ, Ca V 1.2 Y1675H, Ca V 1.1 H1532Y, and four Ca V 1.1 mutant peptide with the H1532Y substitution and one or more amino acid changes were tested. Briefly, we assessed the affinity of the biotinylated wild type and mutant IQ peptides listed in Table 2 for Ca 2ϩ -CaM in saturating Ca 2ϩ concentrations. We also performed competition experiments for CaM binding to immobilized biotinylated peptides using increasing concentrations of the nonbiotinylated peptides listed in Table 2 (see supporting information). Binding data reflecting the interaction of Ca 2ϩ -CaM with the biotinylated peptides are shown in Fig. 2. The interactions are adequately modeled with a simple bimolecular interaction for the concentration range shown in Fig. 2, with a K D of 7.9 nM for the Ca V 1.1 IQB, 2.5 nM for the Ca V 1.2 IQB, 4 nM for the Ca V 1.1 H1532Y-IQB, and 2.9 nM for the Ca V 1.2 Y1675H-IQB. This finding suggests that at low concentrations (Ͻ50 nM approximately) CaM assumes one binding conformation to the IQ peptides. As concentration increases, the data are fitted better to the two site saturation model (data not shown), suggesting that at higher concentrations CaM binds in two or more different conformations to the IQ peptides. For simplicity, we have chosen to compare only the high affinity interac- tions. The H1532Y mutation in Ca V 1.1 IQ increases the affinity for CaM, but the affinity is still lower than that of Ca V 1.2 IQ. The Y1675H mutation in Ca V 1.2 IQ results in only a decrease in CaM affinity. These findings together with the competition data with the other mutant peptides (see supporting information) suggest that all four amino acids that differ between the peptides contribute to the overall affinity for CaM, despite the finding that only one of them interacts with CaM in the crystal structure, suggesting that the interactions detected in the crystal are not the only interactions that occur in solution.
Amino Acids in the IQ Motif That Regulate the Ca 2ϩ Affinity of Bound CaM-The primary function of CaM is to transduce a Ca 2ϩ signal into a protein response. The apparent affinity of CaM for Ca 2ϩ in the presence of a target peptide is coupled to the affinity of CaM for the peptide. In addition, an interaction of apoCaM (Ca 2ϩ -free CaM) with an IQ peptide can increase the apparent Ca 2ϩ affinity of CaM by altering the conformation of the Ca 2ϩ -binding sites. Tryptophan mutants of CaM have been used to assess apparent Ca 2ϩ affinity to CaM complexed to peptides (24). The apparent Ca 2ϩ affinity of CaM, referred to here as K D,app , is determined from the Ca 2ϩ titration of the tryptophan fluorescence. F19W is a CaM mutant that indirectly measures Ca 2ϩ binding to the N-lobe, and the interaction of Ca V 1.2 IQ with F19W increases its Ca 2ϩ affinity (1). The apparent Ca 2ϩ affinity of the N-lobe of F19W complexed to Ca V 1.1 IQ (254 nM) is less than that of the N-lobe of F19W bound to Ca V 1.2 IQ (49 nM) ( Fig. 3A and Table 3) (1). We used F92WCaM to assess Ca 2ϩ binding to the C-lobe. The Ca 2ϩ titration of the fluorescence of the C-lobe of F92W was fit with a single component for Ca V 1.1 but was distinctly biphasic with Ca V 1.2 (1), as characterized by a Ca 2ϩ -dependent increase followed by a decrease in fluorescence. The decrease in fluorescence with F92W/Ca V 1.2 IQ at higher Ca 2ϩ concentrations is likely to be due to the N-lobe quenching the fluorescence of the C-lobe upon binding Ca 2ϩ . The discrepancy between the N-lobe Ca 2ϩ affinity determined with Ca V 1.2 and F19W (49 nM) and that estimated from the quenching (198 nM) is likely to be due to the difficulty in fitting this complex biphasic curve. In these experiments the peptide is present in a 5-fold molar excess over CaM, and therefore, two CaMs binding to a single peptide is not likely. Overall, the Ca 2ϩ affinity of the C-lobe of F92W complexed to Ca V 1.2 IQ is about 5-fold higher than that of the C-lobe of F92W bound to Ca V 1.1 IQ, a factor that is similar to the change observed at the N-lobe (Fig. 3B and Table 3).
Identification of the Amino Acids Responsible for the Higher Ca 2ϩ Affinity of CaM Bound to Ca V 1.2-As mentioned previously, of the four nonconserved amino acids, only His-1532 in Ca V 1.1 IQ and Tyr-1675 in Ca V 1.2 actually contact Ca 2ϩ -CaM in the crystal structures (Fig. 1). If the residue at this position in the IQ motifs is responsible for the difference in Ca 2ϩ affinity of CaM bound to Ca V 1.2 versus Ca V 1.1, it should be possible to lower the Ca 2ϩ affinity of Ca V 1.2 IQ by converting Tyr-1675 to a His and to increase the Ca 2ϩ affinity of Ca V 1.1 IQ by converting His-1532 to a Tyr. The peptides with amino acids substitutions used in this study are listed in Table 2. The apparent N-lobe Ca 2ϩ affinity of F19W bound to Ca V 1.1 H1532Y is only slightly different from that of F19W bound to Ca V 1.1 IQ, suggesting that this residue alone is not responsible for the higher Ca 2ϩ affinity of the N-lobe of F19W bound to Ca V 1.2 IQ (Fig.  3C). The apparent N-lobe Ca 2ϩ affinity of F19W bound to Ca V 1.2 Y1675H is less than that of F19W bound to Ca V 1.2 IQ, but it is still higher than that of F19W bound to Ca V 1.1 IQ. With F92W to monitor Ca 2ϩ affinity of the C-lobe, the first obvious difference using Ca V 1.2 Y1675H is the absence of the fluorescence quenching seen at higher Ca 2ϩ concentrations with Ca V 1.2 IQ and F92W (Fig. 3D). This suggests that the lobes may not be in close enough proximity when bound to the Ca V 1.2 Y1675H peptide to cause quenching. The Y1675H substitution does reduce the Ca 2ϩ affinity of the C-lobe of the F92W (Fig. 3, B and D, and Table 3). These data suggest that the amino acids at this position influence the Ca 2ϩ affinities of both lobes of CaM but cannot alone account for the difference in Ca 2ϩ affinities of CaM bound to Ca V 1.1 IQ and Ca V 1.2 IQ. The other nonconserved amino acids, despite their lack of contact with Ca 2ϩ -CaM in the crystal structure, must be contributing to these observed differences.
Lys-1680, Lys-1683, and Gln-1685 provide Ca V 1.2 IQ with a ϩ3 net charge compared with Ca V 1.1 IQ. Four peptides, each containing the H1532Y substitution in Ca V 1.1 and one or more of the above amino acid changes (Table 2), were tested for effects on the Ca 2ϩ affinity of bound F19WCaM and F92WCaM. All of these peptides increased the apparent Ca 2ϩ affinity of both F19W and F92W relative to these CaMs complexed to either Ca V 1.1 IQ or Ca V 1.1 H1532Y (highest p value ϭ 0.0017) (Fig. 3, E and F, and Table 3). The most dramatic changes were seen with Ca V 1.1/H1532Y/M1537K and Ca V 1.1/.H1532Y/M1537K/Q1540K. These data suggest that all of the nonconserved residues modulate the Ca 2ϩ affinity of both lobes of CaM.
Partially Ca 2ϩ -saturated CaM Binds with Both Lobes to Ca V 1.2 IQ-One explanation of the observation that amino acids that do not interact with Ca 2ϩ -CaM in the crystal increase Ca 2ϩ affinity is that these amino acids are involved in the binding of CaM in a Ca 2ϩ -free or a partially Ca 2ϩ -saturated state. This type of interaction could increase the apparent Ca 2ϩ affinity of the sites by altering the conformation of the Ca 2ϩ -free sites. We used Ca 2ϩ -binding site mutants of CaM, FRET analyses, and measurement of Ca 2ϩ dissociation rates to determine whether Ca V 1.2 and Ca V 1.1 have different abilities to bind CaM with one lobe Ca 2ϩ -bound and one lobe Ca 2ϩ -free.
To assess the effects of Ca 2ϩ binding at one lobe of CaM on the Ca 2ϩ binding properties of the second lobe, we created a series of CaM mutants (F19W/E34Q, F92W/E12Q, F19W/ E12Q, and F92W/E34Q) that combined the tryptophan substitutions with mutations in either the N-or C-lobe Ca 2ϩ -binding sites. As shown previously with Drosophila CaM, the E12Q mutation (glutamates in the Ϫz positions of EF hands 1 and 2 are mutated to glutamines) abolishes Ca 2ϩ binding to the N-lobe, and the E34Q mutation (glutamates in the Ϫz positions of EF hands 3 and 4 are mutated to glutamines) abolishes Ca 2ϩ binding to the C-lobe (37). The F19W/E34Q and F92W/E12Q mutants are used to detect Ca 2ϩ binding to the N-and C-lobes, respectively. Mutations in the Ca 2ϩ -binding sites in the C-lobe decreased the apparent Ca 2ϩ affinity of the N-lobe, whereas mutations in the N-lobe had lesser effects on the apparent Ca 2ϩ affinity of the C-lobe bound to Ca V 1.2 ( Fig. 4A and Table 4). The apparent Ca 2ϩ affinity of F19W/E34Q when bound to Ca V 1.2 IQ (K D,app for Ca 2ϩ ϭ 208 nM) is greater than when bound to Ca V 1.1 IQ (K D,app for Ca 2ϩ ϭ 690 nM) (Fig. 4A and Table 4). These values demonstrate lower affinities than obtained with F19WCaM complexed to Ca V 1.2 IQ (50 nM) and Ca V 1.1 IQ (190 nM) and are likely to reflect both the contributions of the other lobe and the decreased affinity of the Ca 2ϩ -binding site mutants of CaM for the peptide. The   4B and Table 4). We used two additional mutants as follows: the F19W/E12Q to detect alterations in Ca 2ϩ free N-lobe arising from Ca 2ϩ binding to the C-lobe, and F92W/E34Q to detect alterations in Ca 2ϩ -free C-lobe arising from Ca 2ϩ binding to the N-lobe. We were unable to detect Ca 2ϩ -dependent fluorescence changes with F19W/E12Q and F92W/E34Q either in the absence of peptide or with Ca V 1.1 IQ. However, we were able to detect Ca 2ϩ -dependent changes in fluorescence of both F19W/E12Q and F92W/E34Q in complex with Ca V 1.2 IQ. The K D,app for Ca 2ϩ for F19W/E12QCaM-Ca V 1.2 IQ was 21 nM, whereas that of F19W/E12QCaM-Ca V 1.2 IQ was 120 nM. These findings suggest that Ca 2ϩ binding to either the N-or C-lobe of CaM changes the environment of the tryptophan in the Ca 2ϩ -free lobe, suggesting an interaction between lobes when CaM is bound to Ca V 1.2. In both cases the apparent affinity was higher when the tryptophan was in the Ca 2ϩ -free lobe suggesting that a tryptophan in the Ca 2ϩ -free lobe either facilitates the interaction between the lobes or increases the affinity of the Ca 2ϩfree lobe for Ca V 1.2. Either explanation would support a lobelobe interaction when CaM is bound to Ca V 1.2.
To further support the interaction of partially saturated CaM with Ca V 1.2, we used stopped-flow fluorescence measurements and the F19W and F92W mutants to measure the rate of Ca 2ϩ dissociation from each lobe of CaM in the presence of the peptides. The tryptophan mutations in F19W and F92W have only small effects on Ca 2ϩ dissociation rates compared with unmodified CaM (1). Similar experiments have previously shown that Ca 2ϩ dissociates from the N-lobe faster than from the C-lobe of CaM and that the binding of CaM to peptides slows the rate of Ca 2ϩ dissociation from both lobes (1). The interaction with Ca V 1.1 also slows Ca 2ϩ dissociation from F19W and F92WCaM (Fig. 5). Using F19W we found that the rate of dissociation from the N-lobe was similar when F19W was bound to Ca V 1.1 IQ (6.3 s Ϫ1 ) and Ca V 1.2 IQ (6.4 s Ϫ1 ) (Fig. 5A and Table 5). Initial Ca 2ϩ dissociation is modestly faster from the C-lobe (F92W) when bound to Ca V 1.1 IQ (k d ϭ 1.3 s Ϫ1 ) compared with Ca V 1.2 IQ (0.8 s Ϫ1 ) (Fig. 5B and Table 5). However, F92W complexed to Ca V 1.2 IQ (but not Ca V 1.1 IQ) displays a two component dissociation (1). The fluorescence first increases and then decreases. These data demonstrate that the tryptophan in the F92WCaM-Ca V 1.2 IQ complex can detect a conformational change in the N-lobe as it releases Ca 2ϩ and that a stable intermediate with CaM with its N-lobe Ca 2ϩ free can be detected kinetically. It is possible that CaM also binds Ca V 1.1 IQ with one lobe Ca 2ϩ -free, but the biphasic dissociation is too fast to detect in these experiments.
We also measured Ca 2ϩ dissociation rates using stoppedflow kinetics and the fluorescent Ca 2ϩ chelator Quin-2. With Quin-2, fluorescence increases when it binds Ca 2ϩ . In the absence of peptide, the rate of increase in Quin-2 fluorescence can be fit with a single exponential because the N-lobe Ca 2ϩ dissociation is too fast to be resolved by Quin-2 (represented as a dotted line in Fig. 5C JULY 24, 2009 • VOLUME 284 • NUMBER 30

Calmodulin and Ca V 1 IQ Peptides
peptide. As expected from the tryptophan fluorescence studies, Ca 2ϩ dissociates faster from CaM when bound to Ca V 1.1 IQ than when bound to Ca V 1.2 IQ (Fig. 5C and Table 5). The Quin-2 data also support the existence of a stable intermediate with partially saturated CaM in complex with Ca V 1.2. Peptide Binding Affinity of Partially Ca 2ϩ -saturated CaM-The lobe-lobe interactions of CaM bound to Ca V 1.2 raise the issue of Ca 2ϩ binding affinities of CaM lobes. We address this issue by assessing the affinity of the biotinylated wild type IQ peptides for E12QCaM and E34QCaM in saturating Ca 2ϩ concentrations using SPR (Fig. 6). The interactions of E12QCaM and E34QCaM with the Ca V 1.1 IQB peptide are fit to a one-site saturation model with a K D of 6950 and 4695 nM, respectively, whereas the interactions with Ca V 1.2 IQB are fit to a two-site saturation model with K D values of 4366 and 197 nM for interaction with E12QCaM and K D values of 5745 and 138 nM for the interaction with the E34QCaM. The drastically reduced affinity of Ca V 1 peptides for E12QCaM and E34QCaM indicates reduced Ca 2ϩ binding affinity of each lobe of CaM.
Condensed Versus Extended Conformations of CaM Bound to Peptides-We next used FRET to evaluate proximity of the lobes of CaM to each other when bound to the peptides. We mutated CaM, E12QCaM, and E34QCaM at positions 34 and 110 to place cysteines for labeling with FRET reagents, IAEDNS, and DDPM. DDPM is a nonfluorescent energy-trans-   fer acceptor with IAEDNS as the donor. Residues 34 and 110 of CaM face the solvent in nearly all known structures of CaM and are usually far from the ligand-binding sites. A decrease in the donor fluorescence indicates FRET, as measured relative to CaM labeled with only the donor fluorophore. When fluorescence is decreased in the presence of peptide, the lobes of CaM are closer together. When CaM is labeled with both donor and acceptor (CaM DA ), the FRET in the presence of Ca 2ϩ is greater in the presence of the IQ than in its absence ( Table 6). The peptides alone increase the fluorescence of CaM D , suggesting that the fluorescence of the donor is somewhat influenced by the peptide under the presented conditions. However, as shown in Table 6, the peptide effects on CaM D fluorescence are not responsible for the FRET with CaM DA and therefore do not affect the ability of the acceptor to quench donor fluorescence. The data summarized in Table 6 indicate that fully Ca 2ϩ -bound CaM DA binds to the Cav1.1 IQ and Cav1.2 IQ peptides and assumes a compact conformation with the two lobes in close proximity ( Fig. 7 and Table 6). FRET signals were also obtained with both E12QCaM-Ca V 1.2 IQ and E34QCaM-Cav1.2 IQ (Fig.  7 and Table 6). In contrast, neither the E12QCaM-Ca V 1.  (Fig. 7).

DISCUSSION
One mechanism that would allow CaM to regulate the response of different target proteins to different Ca 2ϩ signals is for the binding sites themselves to regulate the Ca 2ϩ affinity of the bound CaM. We provide evidence for this mechanism by demonstrating that subtle differences in CaM-binding sites of Ca V 1.1 and Ca V 1.2 lead to differences in the Ca 2ϩ binding properties of the lobes of CaM. Assuming these findings accurately reflect Ca 2ϩ sensing by CaM bound to L-type Ca 2ϩ channels, then calmodulin modulation of Ca V 1.2 channels would be expected to be more sensitive to Ca 2ϩ than that of Ca V 1.1 channels. In Ca V 1.2, the tyrosine at residue 1675 has contact with methionines in the C-lobe of CaM and also interacts through a hydrogen bond involving a water molecule and the C-lobe backbone. This difference in conformation compared with CaM in complex with Ca V 1.1 is likely to contribute to the higher affinity of Ca V 1.2 IQ for CaM. In Ca V 1.1 the histidine (His-1532) in place of the tyrosine is likely to increase the flexibility of the 3-4 loop of CaM and Phe-1533 of Ca V 1.1 IQ, resulting in fewer stabilizing interactions and lower affinity. The amino acid in this position (Tyr-1674 in Ca V 1.2 and His-1532 in Ca V 1.1) is obviously important for affinity of CaM for the peptide, but it does not fully account for differences in the Ca 2ϩ affinity of CaM bound to peptides. Our data suggest that the other three amino acids that differ between these two peptides and do not directly contact CaM when CaM is fully Ca 2ϩsaturated participate in the binding of CaM that has one or both lobes Ca 2ϩ -free. This argues that different conformations of CaM bind to Ca V 1.2. Although several studies from other laboratories have shown that the Ca 2ϩ affinity of CaM is influenced by binding to different targets (34, 38 -41), we elucidated the contributions of the nonconserved amino acids within the IQ-binding site responsible for this effect on Ca 2ϩ affinity, and we demonstrated that these amino acids play a role in regulating the apparent Ca 2ϩ affinity by modulating the interactions with CaM with at least one lobe Ca 2ϩ -free. Additional contri-   In the Ca 2ϩ -CaM/Ca V 1.2 structure, Lys-1680 contacts the C-lobe through a water molecule. A question arising from the differences in affinity of Ca V 1.1 and Ca V 1.2 for Ca 2ϩ -CaM is whether the Ca V 1.2 IQ lysines electrostatically attract the negatively charged CaM. CaM is an acidic protein with a net negative charge (Ϫ16e) and a calculated pI of 4.15 (44). The peptide Ca V 1.2 IQ has a ϩ3 greater net charge over Ca V 1.1 IQ, having theoretical pI values of 10.12 and 9.52, respectively. The salinity of the solvent and the positive charges on the peptides may stabilize the compact conformation of the CaM allowing its negatively charged lobes to be close together (45).
Although the parallel arrangement of the N-and C-lobes of CaM bound to Ca V 1.1 IQ peptide presented here is in agreement with the arrangement seen with the Ca V 1.2 IQ and other peptides (28,35,36), it is opposite to the structure of three Ca V 2 IQ-Ca 2ϩ CaM complexes reported by Kim et al. (46). In these structures CaM is bound in an anti-parallel arrangement, with the N-lobe of CaM binding the carboxyl terminus of the Ca V 2-IQ peptide and the C-lobe binding the amino terminus of Ca V 2-IQ.
Our data suggest that CaM has the ability to bind to Ca V 1.2 IQ, in both a fully and a partially Ca 2ϩ -saturated state with higher affinity than to Ca V 1.1. Recently Saucerman and Bers (47) suggested that the affinity of proteins for CaM determines their response to local Ca 2ϩ signals. In the case of the Ca V 1 channels discussed here, the different affinities of skeletal and cardiac isoforms for fully or partially Ca 2ϩsaturated CaM could indicate selective modulation and finetuning of local Ca 2ϩ -dependent pathways or more specifically their CDI and CDF.
In the case of the Ca V 1.2 channel, it has been suggested that the Ca 2ϩ -C-lobe CaM interaction is important for CDI (11), whereas the Ca 2ϩ -N-lobe interaction participates in the CDF (35,36). Given the similarity in the structural arrangement of Ca V 1.1 IQ and Ca V 1.2 IQ complexes with CaM, similar responses are expected for Ca V 1.1. Furthermore, because the nonconserved amino acids between Ca V 1.1 IQ and Ca V 1.2 IQ domains are involved in the interactions with CaM C-lobe, both Ca V 1.1 IQ and Ca V 1.2 IQ domains are expected to have similar response to CDF, which is modulated by the CaM N-lobe.
The Ca 2ϩ titration of the fluorescence of F92WCaM in complex with Ca V 1.2 shows evidence of two components (increase followed by a decrease in fluorescence with increasing Ca 2ϩ ). Substitution of any of the amino acids in Ca V 1.2 for those in Ca V 1.1 eliminates the second phase fluorescence quenching, suggesting that all of these amino acids contribute to this second Ca 2ϩ -dependent event. The biphasic Ca 2ϩ response of F92WCaM bound to Ca V 1.2 IQ is best explained by the binding of Ca 2ϩ at the C-lobe at Ca 2ϩ concentrations less than 100 nM followed by a change in the environment of the tryptophan at amino acid 92 (reflected by a quenching of the fluorescence) when Ca 2ϩ binds to the N-lobe. The ability of the Trp-92 to sense Ca 2ϩ binding to both lobes may reflect its location in the linker helix that connects the lobes. The tryptophan at position 19 in the N-lobe is less likely to sense Ca 2ϩ binding at the C-lobe because it is not directly connected to the linker helix. A twophase Ca 2ϩ equilibrium curve is not apparent when F92W binds to Ca V 1.1 IQ.
Indirect Ca 2ϩ dissociation data with F92W also reveal a stable Ca 2ϩ intermediate state for CaM binding to Ca V 1.2 IQ that is not seen with Ca V 1.1 IQ. During Ca 2ϩ dissociation, the fluorescence of F92W first increases then decreases (1), suggesting that Ca 2ϩ dissociates first from the N-lobe relieving the fluorescence quenching and then dissociates more slowly from the C-lobe. These data again support the existence of a stable intermediate state of CaM with the C-lobe Ca 2ϩ -bound and the N-lobe Ca 2ϩ -free. This state is likely to exist in a cell when the Ca 2ϩ concentration is declining after a transient. The question becomes what is the functional role of this intermediate state?
A Ca 2ϩ -binding site mutant that cannot bind Ca 2ϩ at the N-lobe still supports Ca 2ϩ -dependent inactivation of this channel, and hence the role of this intermediate form may be to help to close or inactivate the channel. The FRET data suggest that CaM can also interact with Ca V 1.2 in a compact conformation with the C-lobe Ca 2ϩ -free and the N-lobe Ca 2ϩ -bound. This would be expected to be the first change in CaM bound to the channel when Ca 2ϩ begins to rise in a cell at the start of the Ca 2ϩ transient. Yue and co-workers (2) suggested that Ca 2ϩ binding to the N-lobe drives Ca 2ϩ -dependent facilitation. We conclude that CaM with intermediate saturation states can assume very different conformations depending on whether the N-or C-lobe is Ca 2ϩ -bound and in doing so may produce very different functional outcomes.
The IQ motif is likely to be at least part of the binding site for apoCaM (Ca 2ϩ -free CaM) (1,9,13,25), but the amino acid residues involved in the binding of apoCaM have not yet been identified. Our data suggest that at least with the partially Ca 2ϩsaturated states the apo-N-lobe and the apo-C-lobe can bind within the IQ sequence of Ca V 1.2, and the residues at positions Tyr-1675, Lys-1680, Lys-1683, and Gln-1685 contribute to the interaction. A more detailed analysis of CaM interactions with larger regions of the C-tail is needed to determine how CaM can move within its binding pocket upon binding Ca 2ϩ . In addition other parts of the channel itself may interact near or with the IQ motif. A recent report by Dick et al. (48) suggests that the N-lobe of a single CaM molecule switches to an amino-terminal site on the Ca V 1 channels. CaM binding could either promote or inhibit protein-protein interactions.
Our previous studies and those of Van Petegem et al. (36) suggest that CaM is most likely binding with both lobes Ca 2ϩ saturated to determinants within the IQ motif during CDF rather than CDI (28,36). The data presented here provide the tools needed to test this hypothesis by providing a means to alter the Ca 2ϩ sensitivity of CaM. Our studies also provide details of how the binding site itself regulates the affinity of EF hands of CaM for Ca 2ϩ and identifies new interactions that contribute to CaM binding to the IQ motif.