Sites on Calmodulin That Interact with the C-terminal Tail of Cav1.2 Channel*

Two fragments of the C-terminal tail of the α1 subunit (CT1, amino acids 1538–1692 and CT2, amino acids 1596–1692) of human cardiac L-type calcium channel (CaV1.2) have been expressed, refolded, and purified. A single Ca2+-calmodulin binds to each fragment, and this interaction with Ca2+-calmodulin is required for proper folding of the fragment. Ca2+-calmodulin, bound to these fragments, is in a more extended conformation than calmodulin bound to a synthetic peptide representing the IQ motif, suggesting that either the conformation of the IQ sequence is different in the context of the longer fragment, or other sequences within CT2 contribute to the binding of calmodulin. NMR amide chemical shift perturbation mapping shows the backbone conformation of calmodulin is nearly identical when bound to CT1 and CT2, suggesting that amino acids 1538–1595 do not contribute to or alter calmodulin binding to amino acids 1596–1692 of CaV1.2. The interaction with CT2 produces the greatest changes in the backbone amides of hydrophobic residues in the N-lobe and hydrophilic residues in the C-lobe of calmodulin and has a greater effect on residues located in Ca2+ binding loops I and II in the N-lobe relative to loops III and IV in the C-lobe. In conclusion, Ca2+-calmodulin assumes a novel conformation when part of a complex with the C-terminal tail of the CaV1.2 α1 subunit that is not duplicated by synthetic peptides corresponding to the putative binding motifs.

We subcloned the cDNA of these sequences into pET23a(ϩ) and pET28a(ϩ) vectors (Novagen, Madison, WI) between NdeI and HindIII sites and used these constructs for the expression of the fragments with and without a His tag. We transformed the subcloned vectors into BL21(DE3) host (Novagen) for expression and grew the cells in LB media containing the suitable antibiotics at 37°C. Cells were induced with 0.5 mM isopropyl-1-thio-␤-D-galactopyranoside at an OD 600 of 0.6 and incubated another 3 h at 37°C before harvesting. The cells were lysed with lysozyme and nuclease and soluble material was removed by sedimentation of the insoluble material. The insoluble pellet was washed with urea and Triton X-100 to obtain an insoluble fraction enriched in inclusion body proteins. Wild-type CaM, N-CaM, and C-CaM were expressed and purified as described previously (25). The Ca 2ϩ binding site mutants of CaM (E12Q CaM, E34Q CaM, and E1234Q CaM) were expressed and purified as described previously (26). Expression and purification of T34C/T110C CaM (cysteine substitution for threonine at positions 34 and 110 of CaM) is the same as that of wild-type CaM.
Refolding of CT1 (Amino Acids 1538 -1692) and CT2 (Amino Acids 1596 -1692)-The inclusion bodies were solubilized with Inclusion Body solubilization reagent (Pierce) and refolded by dialyzing against 90 mM Tris-HCl, pH 7.4, 6 M urea, 6 mM calcium in the presence of CaM (wild-type CaM or CaM mutants). Water was added gradually to the dialysis buffer to lower the urea concentration to 2 M. The remaining urea was removed by dialysis against 30 mM Tris-HCl, pH 7.4 and 2 mM calcium. Soluble and insoluble proteins were then separated by centrifugation at 48,000 ϫ g for 1 h at 4°C. His tags were removed using a Thrombin Cleavage Capture kit (Novagen). The refolding in the presence of CaM was also examined at three different Ca 2ϩ concentrations.
Preparation of CT1 for Analysis of CaM Binding by Nondenaturing Gel Electrophoresis-The untagged, solubilized CT1 was further purified using chelating Sepharose (Amersham Biosciences). To avoid CT1 precipitation, all buffers (equilibration, binding, wash, and elution buffer) included a 50% solution of the Inclusion Body solubilization reagent. CT1 was eluted with 10 mM imidazole. The purified CT1 was refolded by dialysis against 90 mM Tris-HCl, pH 7.4, 6 M urea, 3% Triton X-100, gradually adding water into the dialysis buffer until 2 M urea was reached. At this stage CT1 was dialyzed against 50 mM MOPS, pH 7.4, 1% Triton X-100. Increasing amounts of refolded CT1 were incubated with CaM (2 g in 20 l (6 M) per sample) for 30 min at room temperature in 50 mM MOPS (pH 7.4), 1% Triton X-100, and 2 mM CaCl 2 or 2 mM EGTA. Nondenaturing PAGE (15%) was performed to assess complex formation of CT1 and CaM. Protein concentrations were determined by the method of Lowry (27) using bovine serum albumin as a standard.
Preparation of CT1, CT2, and Donor/Acceptor-labeled CaM for Fluorescence Resonance Energy Transfer (FRET)-The purified complex of CT1 or CT2-CaM was obtained by refolding CT1 or CT2 in the presence of CaM. His tags were cleaved with thrombin. The samples were dialyzed against 30 mM Tris-HCl buffer, pH 7.4, with EGTA to precipitate CT1 or CT2. The pellet was washed with 30 mM Tris-HCl buffer, pH 7.4, to obtain purified CT1 or CT2 (insoluble). Samples were solubilized in the Inclusion Body solubilization reagent. Donor/acceptor-labeled CaM (D/A CaM) was obtained by labeling T34C/T110C CaM with 1,5-IAEDANS (the donor) and DDPM (the acceptor). To obtain T34C/ T110C that was only labeled with the donor, T34C/T110C CaM was labeled with 0.4 mol of donor/mol of protein in 20 mM Tris-HCl, pH 7.5, 100 mM KCl for 2 h at 20°C in the dark. Free donor was removed by a desalting column packed with size exclusion medium, Bio-Gel P-6DG (Bio-Rad). The product had 0.31 mol of the donor per mol of protein. A portion of this partially labeled protein was saved as the donor-alone protein, while the rest was labeled with excess acceptor to give D/A CaM saturated by acceptor at the same reaction conditions. Free acceptor was removed by the same desalting column as above.
FRET-D/A CaM (0.5 M) was incubated with increasing concentrations of CT1, CT2, A-peptide, C-peptide, or IQ-peptide for 5 min at 20°C in 20 mM MOPS pH 7.5, 100 mM KCl, 2 mM CaCl 2 , or 2 mM EGTA. Fluorescence data were collected on a PTI QuantaMaster spectrofluorometer. The excitation wavelength was set at 334 nm and emission spectra were scanned from 400 nm to 600 nm with a 5-nm slit width for excitation and a 10-nm slit width for emission. Spectra were processed by subtracting the background of buffer and/or additives where appropriate and by averaging three sets of scans. To ensure that the fluorescence changes were due to the changes of distance between donor and acceptor and not to the interaction of these fragments with the donor, parallel experiments were performed with donor-alone-labeled CaM. Data were compared with the FRET obtained with a CaM kinase KII peptide (FNARRKLKGAILTTMLATRN, designated FNA-peptide) that is known to bring the two lobes of CaM in close proximity (28).
To calculate the apparent affinity of CT2 and the synthetic peptides for CaM, we titrated D/A CaM in 20 mM MOPS pH 7.5, 100 mM KCl, 2 mM CaCl 2 with increasing CT1-, CT2-, IQ-, A-, or C-peptide. Fluorescence emission data at 490 nm were read on an ISS PC1 Photon Counting Spectrofluorometer (ISS, Champaign, Illinois), with a 1-mm slit for excitation and 2-mm slit for emission, at a 334 nm excitation wavelength. Data were processed by subtracting the background of buffer and/or additives where appropriate. The data were fit with a four parameter-Hill equation in SigmaPlot.
Preparation of CT1 or CT2/ 15 N, 13 C-labeled CaM Complex for Nuclear Magnetic Resonance Spectroscopy (NMR)-15 N, 13 C-labeled CaM was created by growing cells in M9 minimal medium using an expression plasmid coding for CaM. [ 15 N]NH 4 Cl and D-[U-13 C]glucose were the only nitrogen and carbon sources for the cells. Other steps involved in expression and purification of 15 N, 13 C-labeled CaM were the same as for wild-type CaM. The complex of 15 N, 13 C-labeled CaM/CT1 or CT2 for NMR was obtained by refolding the solubilized CT1 or CT2 in the presence of 15 N, 13 C-labeled CaM. The His tag was cleaved with biotinylated thrombin, and the thrombin was removed from solution with streptavidin agarose (Novagen). The cleaved His tag was removed from the complex solution during dialysis against the final buffer (10 mM imidazole pH 6.4, 2 mM CaCl 2 ) with a 10,000 MWCO cassette (Pierce).
NMR Methodology-NMR experiments were collected on a DRX 600 MHz spectrometer instrument using a 5-mm TXI probe or cryoprobe at 47°C. Complexes of Ca 2ϩ -loaded 15 N, 13 C-labeled CaM bound to CT1 or CT2 were prepared as described above and then concentrated to 0.5-1 mM. Peptide-CaM complexes were prepared by adding the appropriate peptide directly to the Ca 2ϩ -CaM NMR sample. Final NMR samples contained 10 mM imidazole, and 2 mM CaCl 2 in 95% H 2 O, 5% D 2 O at pH 6.4. Amide chemical shifts for free Ca 2ϩ -CaM in the absence of salt were assigned by comparison to reported chemical shifts for free Ca 2ϩ -CaM in 100 mM KCl (29) and confirmed by the following triple resonance experiments: CBCANH, CBCA(CO)NH, HNCA, HN(CO)CA, and 15 Nedited NOESYHSQC. CaM complexes were assigned using the above mentioned experiments and in addition three-dimensional HNCO and HN(CA)CO experiments were also collected. All data were processed using Felix 2002 software from Accelrys.

RESULTS
Expression of Fragments of the C-terminal Tail of the Ca V 1.2 ␣ 1 Subunit and Analysis of Their Ability to Bind CaM-To elucidate the determinants of the molecular interactions of CaM with the C-terminal tail of the Ca V 1.2 ␣ 1 subunit, we expressed or synthesized fragments of this region containing all or part of the putative binding motifs (Fig. 1). For these studies we used three synthetic peptides designated A (amino acids 1609 -1628), C (amino acids 1627-1652), and IQ (amino acids 1665-1685). We expressed two fragments of the C-termi- nal tail: CT1 (amino acids 1538 -1692, containing the putative EF hand as well as the A, C, and IQ motifs) and CT2 (amino acids 1596 -1692, containing the A, C, and IQ motifs but missing the putative EF hand). We analyzed the interactions of these fragments and peptides with CaM using nondenaturing gel electrophoresis. When expressed in Escherichia coli, both CT1 and CT2 were found in inclusion bodies. A small amount of soluble CT1 or CT2 can be obtained by extraction of the inclusion bodies and a subsequent refolding step (see "Experimental Procedures"). The interaction of refolded CT1 with CaM as assessed by nondenaturing gel electrophoresis is shown in Fig.  2. CT1 is highly positively charged (isoelectric point 10.5) and does not enter the nondenaturing gel at either high or low Ca 2ϩ concentrations (lane 1 of Fig. 2, A and B, respectively). The interaction of CT1 with CaM was assessed by the appearance of a more slowly migrating complex CT1/CaM or by a decrease in the intensity of CaM band. A stable CT1-CaM complex was formed at high Ca 2ϩ concentrations ( Fig. 2A). The densitometric analysis of the disappearance of the CaM band at different fragment concentrations is shown in Fig. 2C. A very limited and low affinity interaction of this fragment was detected with apoCaM (assessed by the disappearance of the apoCaM band on the gel, Fig. 2, B and C).
Although only a small amount of soluble CT1 or CT2 can be obtained as described above, the presence of CaM during the refolding step greatly enhances the amount of CT1 and CT2 that can be isolated in a soluble form. As shown in the SDSpolyacrylamide gel of the soluble fraction in Fig 1-75, missing the C-lobe, 2) C-CaM, a CaM composed of amino acids 76 -148, missing the N-lobe, 3) E12Q CaM which cannot bind Ca 2ϩ at the N-lobe, 4) E34Q which cannot bind Ca 2ϩ at the C-lobe, and 5) E1234Q which cannot bind Ca 2ϩ at either the N-or C-lobes. These CaM mutants have been described previously (25,26). Although none of the five CaM mutants were as effective as the wild-type CaM for the refolding, N-CaM (lane 4 of Fig. 5, A and B), C-CaM (lane 6 of Fig. 5, A and B), E12Q (lane 8 of Fig. 5, A and B) and E34Q (lane 10 of Fig. 5B) produced some facilitated refolding. In contrast, the E1234Q did not facilitate refolding. This suggests that both lobes of Ca 2ϩ -CaM contribute to the facilitated refolding of the Ca V 1.2 ␣ 1 C-terminal tail fragments.
FRET Analysis to Determine the Effects of CT1, CT2, and Synthetic Peptides on the Conformation of CaM-To determine the effects of binding to the C-terminal tail of the Ca V 1.2 ␣ 1 subunit on the conformation of CaM, we examined the ability of a double-labeled CaM (1,5-IAEDANS and DDPM) bound to the fragments or the peptides to produce FRET at high and low Ca 2ϩ concentrations. Since the efficiency of FRET depends upon the distance between the donor and the acceptor in CaM, the emission fluorescence of the D/A CaM at 490 nm can be used to compare the relative distances between the two lobes of CaM when bound to the fragments or peptides. At high Ca 2ϩ concentrations, CT2 (Fig. 6A), the IQ-peptide (Fig. 6B), and the A-peptide (Fig. 6C) all decreased the fluorescence of D/A CaM. In contrast, the C-peptide (Fig. 6D) increased the fluorescence of D/A CaM. These findings suggest that the lobes of CaM are closer together when bound to CT2, the IQ-peptide, and the A-peptide compared with the C-peptide. The longer CT1 fragment produces similar FRET changes to those determined with CT2 (not shown). As a control we examined the fluorescence at 490 nm with CaM that was labeled only with the donor compound (D/CaM) and found that the fluorescence did not change upon addition of the peptides or fragments (data not shown). With CT1, CT2, and IQ, saturation of the FRET changes was obtained at 1:1 molar ratios of fragment/CaM (Fig. 6, A and B).
A higher molar ratio for saturation of the FRET changes was required for both the A-and C-peptides (Fig. 6, C and D), consistent with our findings that the IQ-peptide has a higher affinity for CaM than either the A-or C-peptide (1, 19). Analysis of the concentration dependence of the FRET changes (Fig.  6E) was used to calculate apparent affinities for CaM of 37 Ϯ 3 nM for CT2 (n ϭ 3), 45 Ϯ 4 nM for IQ (n ϭ 3), and 76 Ϯ 5 nM (n ϭ 3) for A-peptide). The titration curve of CT1 is essentially identical to that of CT2 (data not shown). Our measured K d for CaM binding to CT2 is comparable to the value (K d ϭ 163 nM) reported by Erickson et al. (30) for CaM binding to full-length ␣ 1 C. The maximal fluorescence decrease of the D/A CaM in the presence of either the IQ-peptide or A-peptide is similar to that obtained with the control FNA-peptide (Fig. 6, B and C), which is known to bring the two lobes of CaM into close proximity (28). In contrast, the fragments CT1 and CT2 produced an intermediate decrease of D/A CaM fluorescence, suggesting that the conformation of CaM bound to the fragment is different than when bound to either A-or IQ-peptides. This could reflect a difference in conformation of these sequences (IQ or A) within the larger fragment or a contribution of the C sequence to the CaM interaction. At low Ca 2ϩ concentrations, the fragments and peptides did not produce FRET of D/A CaM (data not shown), indicating that the either conformational change in CaM or its interaction with the fragments and peptides are Ca 2ϩ -dependent.
Effect of CT1 and CT2 on Backbone Amide Chemical Shifts of Ca 2ϩ -CaM-We used amide chemical shift perturbation mapping to compare the interactions of CaM with the expressed fragments of the C-terminal tail of the Ca V 1.2 ␣ 1 subunit (CT1 and CT2) versus the synthetic peptides (C and IQ) corresponding to the putative CaM binding sites in this sequence. Our study was performed in the presence of Ca 2ϩ since both CT1 and CT2 are insoluble in the absence of Ca 2ϩ -CaM. Fig. 7A compares 1 H-15 N HSQC spectra for free (black) and CT2-bound (red) Ca 2ϩ -CaM. The absence of heterogeneity is consistent with a dominant conformation of Ca 2ϩ -CaM bound to CT2 in a 1:1 complex. The boxed region in Fig. 7A is enlarged in Fig. 7, B-D to compare the effects of CT2 and CT1 on chemical shifts for selected amino acids in the N-and C-terminal lobes of CaM. Chemical shifts for the 11 amino acids are identical for Ca 2ϩ -CaM bound to CT2 or CT1 (Fig. 7, C and D). In fact, the cross-peaks for only 5 residues (Ser 17 , Glu 54 , Met 77 , Asp 78 , and Thr 79 ) do not directly overlap in the spectra for CaM bound to CT2 versus CT1. These data indicate that sequences N-terminal to Met 1596 in the cytoplasmic tail of Ca V 1.2 have minor, if any, effect on the conformation of Ca 2ϩ -CaM.
The histogram in Fig. 8A summarizes the weighted chemical shift difference per residue for free CaM versus CaM bound to CT2. Also indicated are the locations of helices A-H and the four Ca 2ϩ binding loops. The N-domain and C-domain include residues 1-76 and 80 -148, respectively. Residues 77-79 form a flexible linker between domains (31). The dashed line at 0.46 ppm corresponds to the mean plus the S.D. of chemical shift differences for all residues, and is used as reference for very significant changes. Bars denoted with closed circles correspond to hydrophobic residues with chemical shift changes Ն0.46 ppm.
CT2 causes large changes in the backbone amide chemical shifts for residues 77 and 78 located in the flexible tether that links helices D and E. Changes in this region, which functions as a hinge to allow the N-and C-domains of CaM to bind to small peptides (32,33), are consistent with Fig. 6, showing that CT2 causes CaM to assume a more compact conformation. Large chemical shift changes in the N-domain are seen for hydrophobic amino acids Phe 19 , Val 55 , Leu 69 , Met 71 , and Met 72 . In contrast, only hydrophobic residue Phe 92 in the C-lobe has a chemical shift perturbation Ն0.46 ppm. The majority of C-lobe amides with chemical shifts perturbation of Ն0.35 are associated with hydrophilic residues. Fig. 9 shows chemical shift differences induced by CT2 mapped on the surface of Ca 2ϩ -CaM bound to a peptide from smooth muscle myosin light chain kinase (34). We selected a peptide-bound model because binding targets to Ca 2ϩ -CaM stabilizes the intrinsically flexible N-and C-domains in more open conformations, as defined by interhelical angles (35). Ca 2ϩ -CaM bound to the myosin light chain kinase peptide was chosen since it is a well studied complex, and since the major difference in backbone tertiary structure of CaM bound to different target peptides is localized to the flexible linker between the N-and C-domains while the structures of the respective domains, especially the C-domain, is similar in the different complexes (36,37). White indicates ⌬␦ of Ͻ0.04 ppm, shades of red indicate increasing ⌬␦ between 0.04 and 0.46 ppm, and dark red indicates ⌬␦ Ն 0.46 ppm. The two views of each domain are rotated roughly 180°degrees to contrast effects on surfaces that include the Ca 2ϩ binding loops and the hydrophobic pockets. The yellow contours encompass contiguous hydrophobic surfaces in the N-and C-lobes.
As anticipated from Fig. 8A, the N-domain hydrophobic pocket shown in the upper panel of Fig. 9 has a well defined dark red surface because of large chemical shift perturbations of hydrophobic residues. The hydrophobic pocket of the C-domain does not show a well grouped cluster of residues that are greatly affected by CT2. Instead, dark red surfaces in the C-domain of Fig. 9 are associated primarily with polar and charged residues that border the hydrophobic pocket. Similar patterns were seen when using structures of Ca 2ϩ -CaM bound to peptides from the Ca 2ϩ -ATPase, CaM-dependent protein kinase II, and anthraces edema factor (see Supplemental Information).
The histogram in Fig. 8A, and the ribbon and surface representations on the right hand side of Fig. 9 (which highlights the Ca 2ϩ binding loops) show more extensive chemical shift perturbations for residues in Ca 2ϩ binding loops I and II in the N-domain relative to loops III and IV of the C-domain. This contrast is especially striking for residues 56 -67 in loop II and residues 129 -140 in loop IV, which have average chemical shift changes of 0.21 and 0.07 ppm, respectively. Loop IV is also distinguished by a relatively small chemical shift perturbation for Ile 128 (0.17 ppm), which immediately precedes the loop. The corresponding residue preceding loops I, II, and III (Phe 19 , Val 55 , and Phe 92 , respectively) have large chemical shift changes (0.51-1.1 ppm).
CaM-target interactions are known to modulate the Ca 2ϩ binding properties of CaM (38 -43). CaM-mediated facilitation and inactivation of channel activity in response to Ca 2ϩ levels and frequency of oscillation may be coupled to channel-induced modulation of Ca 2ϩ binding to CaM. The data in Fig. 8A suggest that this modulation is targeted to sites I-III.
Comparison of CT2 and IQ-Peptide on Chemical Shifts of Ca 2ϩ -CaM-Titration of Ca 2ϩ -CaM with the IQ-peptide caused changes in backbone amide chemical shifts that were characteristic of slow exchange on the NMR time scale. Thus, backbone assignments for 15 N, 13 C-labeled Ca 2ϩ -CaM bound to the IQ-peptide were made using triple resonance experiments described under "Experimental Procedures." Fig. 8B compares backbone amide chemical shifts for free CaM versus the CaM-IQ peptide complexes. Patterns of chemical shift perturbations induced by the peptide are similar to those caused by CT2. The greatest magnitude of change is seen for residues 55-94, which span helices D and E. Hydrophobic residues in the N-domain of CaM are affected to a greater extent than those in the C-domain. Residues in loops I and II are affected to a greater extent than those in loops III and IV. However, as shown in Fig. 8C, 112 out of 148 backbone amide shifts differ by greater than the precision of the measurement when comparing CaM bound to CT2 versus the IQ-peptide. The average chemical shift difference for these residues (0.16 ppm) is 4-fold greater than the error for the calculation. The greatest differences are found in the helices A-D of the N-lobe. Helix E is also different. A comparison of the relative chemical shift differences between free CaM and CaM bound to either CT2 or the IQ-peptide suggests that the IQ-peptide alters the conformation of the N-lobe to a greater extent than CT2.
Comparison of Effects of IQ-and C-Peptides-Both the IQand C-peptides were soluble at the concentrations required for NMR analysis, but the A-peptide was not and was, therefore, not studied further. Peptide C showed relatively weak binding to CaM with fast exchange characteristics on the NMR time scale. This allowed us to assign 118 backbone amides by following the change in chemical shifts during titration of Ca 2ϩ -CaM with the C-peptide. Fig. 8D shows that chemical shift changes induced by the C-peptide are lesser in magnitude than for CT2 or the IQ-peptide, because of weaker binding affinity, but are comparable to the effect of the IQ protein PEP-19 reported previously (44). Several features of Fig. 8D are notable. First, there is a lack of effect of the C-peptide on residues in the central hinge region of CaM. This is particularly evident for residues 77 and 78, which are greatly affected by CT2 and the IQ-peptide, but show little change upon association of the C-peptide. This is consistent with Fig. 6 showing that, in contrast to CT2 and the IQ-peptide, the C-peptide does not induce CaM to adopt a compact conformation. Second, relative to CT2 and the IQ-peptide, the C-peptide causes relatively large changes in chemical shifts for residues in Ca 2ϩ binding loop IV. Third, chemical shift changes are observed in both N-and C-lobes of CaM. Since FRET experiments in Fig. 6 show CaM to be extended when bound to the C-peptide, single peptides may interact with both the N-and C-lobes, or one molecule of the C-peptide may bind to each lobe forming a ternary complex. Such a ternary complex may only be apparent at the high concentrations of CaM and peptide used for NMR experiments. DISCUSSION We have expressed and refolded two different fragments (CT1 and CT2) of the C-terminal tail of the ␣ 1 subunit of the Ca V 1.2 channel. Refolding is greatly facilitated by the presence of CaM. Both CT1 and CT2 contain all three motifs (A, C, and IQ), which have been suggested to contribute to CaM binding (1, 9, 19 -21). Despite the fact that synthetic peptides representing these motifs can each bind CaM, the expressed fragments (CT1 and CT2) bind only a single molecule of CaM, suggesting that either two of these sequences do not bind CaM when part of the larger protein or that the three sequences contribute to a single CaM binding site. A stoichiometry of one CaM binding to each ␣ 1 subunit is consistent with the findings of Mori et al. (45).
Peterson et al. (22) found that a four amino acid cluster (VVTL) within the F helix of the EF hand motif of the Cterminal tail of the ␣ 1 subunit of Ca V 1.2 channel was essential for CDI and suggested that the EF hand motif of ␣ 1 was involved in the transduction of Ca 2ϩ binding to CaM into channel inactivation. A conformational coupling between Ca 2ϩ -CaM binding and the region of the EF hand (or the VVTL sequence identified by Peterson et al.,Ref. 22) seems unlikely since the conformation of CaM, as indicated by amide chemical shifts, is essentially identical when bound to CT1, which has the EF hand and the VVTL sequence, and CT2, which does not. We conclude that the sequences in the cytoplasmic tail N-terminal to Met 1596 , including the EF motif, do not participate in binding to Ca 2ϩ -CaM. A role of this region in association with apoCaM has yet to be determined.
Although Ca 2ϩ binding to the EF hand does not appear to play a role in regulating the conformation of CaM bound to these fragments, Pitt et al. (9) suggested that there was a site within the C-terminal tail of the ␣ 1 subunit outside of the putative EF hand between amino acids 1551-1660 of the mouse sequence (corresponding to amino acids 1599 -1708 of the human sequence) that binds Ca 2ϩ with high affinity, allowing the binding of apoCaM. Romanin et al. (46) also found that Ca 2ϩ binding to a C-terminal sequence (amino acids 1571-1585 of the mouse sequence, corresponding to amino acids 1619 -1633 of the human sequence) was required for apoCaM binding. These sequences are contained within both CT1 and CT2 and may play an important role in the interaction with apoCaM as proposed by these authors. However, a Ca 2ϩ binding site mutant of CaM (E1234Q), even at Ca 2ϩ concentrations expected to saturate these putative Ca 2ϩ binding sites on CT1and CT2, did not produce facilitated refolding of the expressed fragments, suggesting that this is a property of Ca 2ϩ -CaM. CaM mutants unable to bind Ca 2ϩ at either the N or C-lobes could partially support refolding, but binding to both lobes of CaM is required for maximal facilitated refolding of CT1 and CT2.
Previous studies from our laboratory have demonstrated that the IQ-peptide binds Ca 2ϩ -CaM with higher affinity than the A-or C-peptides (19). We have also shown that the C-and IQ-peptides, but not the A-peptide, increase the affinity of the C-terminal lobe of CaM for Ca 2ϩ . Only the IQ-peptide increases the affinity of the N-terminal lobe of CaM for Ca 2ϩ . The differences in Ca 2ϩ affinity of the lobes of CaM when bound to the different sequences and the differences in affinity for these CaM binding motifs for the lobes of CaM may allow CaM to assume different conformations within the binding site. Our FRET data suggest that CaM is in a more extended conformation when bound to CT2 than to the IQ-peptide and is in a very extended conformation when bound to the C-peptide. Amide chemical shift perturbation mapping experiments support the FRET data in that amides in the central linker regions of CaM are greatly affected upon binding to CT2 and IQ, but not to the C-peptide. The differences between CT2 and IQ-peptide bound to CaM could arise, in part, from competition with other sequences within CT2 for binding to the N-lobe of CaM. For example, the IQ motif in CT2 could anchor Ca 2ϩ -CaM via its C-terminal lobe, allowing the N-terminal lobe to interact with other sequences (A and/or C). The N-terminal lobe could be either dominated by interaction with the A or C region or might be in fast exchange between the IQ regions and A and/or C regions, thereby promoting a slightly more extended conformation of CaM bound to CT2 versus IQ. This is consistent with the model of Pitt et al. (9). The relative competitive binding advantage between regions in CT2 for binding to domains of CaM could change significantly depending on the binding of Ca 2ϩ to CaM and/or the functional state of the channel itself. Fig. 8 shows that differences between the conformation of CaM bound to CT2 and IQ extend beyond central linker region and include backbone differences throughout the N-and Cdomains. This is in striking contrast to the results of Kranz et al. (47) who used backbone amide chemical shift perturbation mapping to conclude that the backbone conformation of CaM is virtually identical when bound to either intact CaM kinase I (CKI) or the CaM binding peptide from CKI. Contributions from multiple sequences in CT2 to the CaM binding site may be a general feature of IQ motif proteins that is not shared by proteins such as CKI, which are regulated by autoinhibition involving a restricted CaM binding region.
Amide chemical shifts are extremely sensitive to conformational changes, and can result from allosteric effects, but Figs. 8 and 9 are consistent with interactions between CT2 and regions in the N-and C-lobes of Ca 2ϩ -CaM that include hydrophobic surfaces. This is consistent with the structural paradigm established from NMR and crystal structures of CaMtarget complexes that hydrophobic surfaces in Ca 2ϩ -CaM interact with a variety of ligands (28,32,37,48,49). However, Figs. 8 and 9 reveal an interesting pattern of effects on hydrophobic and hydrophilic residues that suggest a greater contribution of hydrophobic interactions for binding CT2 to the Ndomain of CaM, while binding to the C-lobe may rely more on electrostatic interactions with residues that border the hydrophobic pocket. This is consistent with previous studies demonstrating a role for electrostatic interactions in binding the IQ motif containing protein RC3 and its IQ-peptide to apo and Ca 2ϩ -CaM (50 -53). A different chemical basis for interaction of CT2 with the N-and C-lobes of CaM could be important for the lobe movement on CT2 in response to changes in Ca 2ϩ levels.
A final observation of our NMR studies is that CT fragments and the IQ-peptide have a greater effect on amide chemical shifts of residues located Ca 2ϩ binding loops I and II in the N-lobe, especially loop II, relative to loops III and IV in the C-lobe. This may reflect important differential modulation of the Ca 2ϩ binding properties of the N-and C-domains of CaM. For example, these data are consistent with our previous study (19) showing that the IQ-peptide produces approximately a 70-fold increase in the Ca 2ϩ affinity of the N-lobe of CaM, but only a 16-fold increase in the Ca 2ϩ affinity of the C-lobe. Since the IQ-peptide has a greater overall magnitude of effect than CT2 on the N-lobe of CaM, we predict modulation of Ca 2ϩ to the N-lobe by these ligands.
In summary, these studies suggest that CaM interacts in a unique fashion with the C-terminal tail of the Ca V 1.2 ␣ 1 subunit: 1) the C-tail of Ca V 1.2 binds a single CaM; 2) the conformation or initial folding of this region may be dependent on the presence of CaM; 3) the binding site cannot be adequately mimicked by synthetic peptides, suggesting multiple sites of interaction or a conformation not obtainable in a short peptide; 4) the N-lobe of CaM interacts primarily via hydrophobic interaction while the C-lobe of CaM involves significantly more electrostatic interactions; and 5) the 3rd and 4th EF hand are not greatly affected by the binding of CaM at the binding site.