Intra- and Interdomain Effects Due to Mutation of Calcium-binding Sites in Calmodulin

Abstract

The IQ-motif protein PEP-19, binds to the C-domain of calmodulin (CaM) with significantly different k_on and k_off rates in the presence and absence of Ca²⁺, which could play a role in defining the levels of free CaM during Ca²⁺ transients. The initial goal of the current study was to determine whether Ca²⁺ binding to sites III or IV in the C-domain of CaM was responsible for affecting the kinetics of binding PEP-19. EF-hand Ca²⁺-binding sites were selectively inactivated by the common strategy of changing Asp to Ala at the X-coordination position. Although Ca²⁺ binding to both sites III and IV appeared necessary for native-like interactions with PEP-19, the data also indicated that the mutations caused undesirable structural alterations as evidenced by significant changes in amide chemical shifts for apoCaM. Mutations in the C-domain also affected chemical shifts in the unmodified N-domain, and altered the Ca²⁺ binding properties of the N-domain. Conversion of Asp⁹³ to Ala caused the greatest structural perturbations, possibly due to the loss of stabilizing hydrogen bonds between the side chain of Asp⁹³ and backbone amides in apo loop III. Thus, although these mutations inhibit binding of Ca²⁺, the mutated CaM may not be able to support potentially important native-like activity of the apoprotein. This should be taken into account when designing CaM mutants for expression in cell culture.

Introduction

PEP-19 is a small (62 amino acids) protein that binds to calmodulin (CaM)² in the presence or absence of Ca²⁺ (1) via an IQ CaM binding motif. Although PEP-19 has no known intrinsic activity other than binding to CaM, expression of PEP-19 can protect cells from apoptosis (2, 3), and cell death from Ca²⁺ overload or excitotoxicity (3). We showed that PEP-19 binds preferentially to the C-domain of CaM and greatly affects the kinetics of Ca²⁺ binding to the C-domain (4, 5). We also showed that this effect requires the synergistic activities of the IQ motif and an adjacent acidic sequence, and that the rates of association and dissociation of PEP-19 from apoCaM are much slower than for Ca²⁺-CaM (6). These data support a role for PEP-19 in CaM signaling that involves both modulation of Ca²⁺ binding to CaM, and inhibition of other proteins that rely on binding to the C-domain of CaM.

The goal of the current study was to use CaM mutants with modified Ca²⁺-binding loops to determine whether Ca²⁺ binding to sites III or IV has dominant effects on the properties of PEP-19 binding to CaM. Two strategies have been used to inhibit Ca²⁺ binding to EF-hand motifs in CaM. One involves conversion of Glu to Gln, Lys, or Ala at position 12 in the Ca²⁺-binding loops (7, 8) (see Fig. 1 for the consensus Ca²⁺-binding loop for CaM). The other involves conversion of Asp to Ala at position 1 (9). This latter mutation was used previously to inactivate EF-hand Ca²⁺-binding sites in cardiac troponin C (10), skeletal troponin C (11), and myosin regulatory light chain (12). Mutation of position 1 has been used extensively to assess the relative functional contribution of Ca²⁺ binding to the N- and C-domain of CaM by expression of the mutated proteins in cells (13–31). Thus, we elected to convert Asp to Ala in the first ligand position in all CaM Ca²⁺-binding mutants. The proteins CaM1,2, CaM3, CaM4, and CaM3,4 correspond to mutations: D20A/D56A, D93A, D129A, and D93A/D129A, respectively.

FIGURE 1.

Consensus EF-hand Ca²⁺-binding loop and mutation strategy. The consensus Ca²⁺-binding loop for the 4 EF-hand Ca²⁺-binding motifs of CaM consists of 12 amino acids (32). Positions 1 to 6 adopt a coil secondary structure, 7 to 9 form a short β-strand, and 10 to 12 adopt an α-helix. Ca²⁺ is coordinated by side chain oxygens at X, Y, and −Z. The carbonyl oxygens of Glu at the −Z provide bidentate coordination. Water provides a hydrogen bond link to oxygens of side chains at Z and −X. Calcium is coordinated by the backbone amide at −Y. An invariant Gly is at position 6. As described in the text, we elected to inactivate Ca²⁺ binding to CaM by converting Asp to Ala at position Asp⁹³ in site III (CaM3), Asp¹²⁹ in site IV (CaM4), both 93 and 129 (CaM3,4), or both 20 and 56 (CaM1,2).

Our initial experiments indicated that Ca²⁺ binding to both sites III and IV in the C-domain of CaM are necessary for native-like interactions with PEP-19. However, the data also pointed to undesirable structural alterations due to mutation of Asp⁹³ and Asp¹²⁹ to Ala. For example, the affinity of apoCaM and apoCaM3,4 for PEP-19 should be the same if the mutations do not greatly alter the structure of CaM, however, apoCaM3,4 and the other apo mutant CaM proteins had lower affinity for PEP-19 than native apoCaM. Thus, we investigated the structural consequence of these mutations. The data show that, whereas conversion of Asp⁹³ and Asp¹²⁹ to Ala effectively inhibits Ca²⁺ binding to sites III and IV, the resulting mutated C-domain exhibits extensive structural perturbations. This has implications for cell expression studies because the C-domain of CaM3,4 may not support important native-like activities of the apo C-domain of CaM.

EXPERIMENTAL PROCEDURES

Materials

[¹⁵N]NH₄Cl was obtained from Cambridge Isotope Laboratories, Inc. (Andover, MA). Quin-2 (free acid), 1-anilinonaphthalene-8-sulfonic acid (1,8-ANS), and 5-((((2- iodoacetyl)amino)ethyl)amino)naphthalene-1-sulfonic acid (1,5-IAEDANS) were purchased from Invitrogen. N-(4-Dimethylamino-3,5-dinitrophenyl)maleimide (DDPM) was purchased from Alfa Aesar.

Expression and Purification of CaM Proteins

A pET23 expression vector with a mammalian CaM cDNA optimized for bacterial codon usage was kindly provided by Andy Hudmon at the Indiana School of Medicine. QuikChange® II XL Site-directed Mutagenesis Kit (Stratagene) was used to generate all mutant proteins. Calcium binding was inhibited by mutation of the X Ca²⁺ coordination position (see Fig. 1), which is the first of 12 amino acids in a canonical EF-hand Ca²⁺-binding loop (32). In addition, a panel of Ca²⁺-binding mutants with Lys⁷⁵ converted to Cys as described previously (33) was generated for labeling with fluorescent probes. A cDNA encoding human PEP-19 was synthesized with optimal bacterial codon usage (DNA2.0 Inc.) and subcloned into a pET23 expression vector. PEP-19 with a C-terminal Gly-Cys extension was used for labeling with the FRET acceptor, DDPM as described previously (5). All constructs were transformed into BL21 DE3 or BL21 DE3 pLys-S host cells for expression of protein with or without [¹⁵N]NH₄Cl and proteins were purified as described previously (4, 33).

Equilibrium Binding and K_off of CaM Mutants with PEP-19

A FRET-based assay described previously (5) was used to measure equilibrium binding of CaM to PEP-19. Typically, a solution of 1.0 μm 1,5-IAEDANS-labeled CaM(K75C) with 10.0 μm DDPM-labeled PEP-19 was prepared in a base buffer containing 20 mm MOPS, pH 7.5, 100 mm KCl, with either 2 mm CaCl₂ or 1 mm EGTA. A concentrated stock solution of unlabeled native CaM (control) or CaM mutant in the same solution was used to titrate the complex. Apparent IC₅₀ values were derived by plotting the fluorescence response versus log(competitor) and fitting to the following equation.

Where F_min and F_max are fluorescence in the absence or large excess of competitor.

Stopped-flow fluorescence was used to determine the dissociation rate (K_off) of the CaM mutants from PEP-19. Briefly, syringe A contained a solution of 1 μm IAEDANS-labeled CaM mutant and 10 μm DDPM-labeled PEP-19 in base buffer. Syringe B contained buffer only. The final concentration of these reagents in the optical chamber was one-half of these values resulting from the 1:1 mixing ratio.

Measuring Calcium Binding Affinity of CaM Mutants

The relative equilibrium Ca²⁺ binding affinities of the various CaM mutants were determined using Ca²⁺-dependent Tyr fluorescence, and the dye 1,8-ANS, which fluoresces upon binding to hydrophobic surfaces exposed by Ca²⁺ binding to CaM. Briefly, solutions contained 4 μm CaM, 50 mm MOPS, pH 7.5, 100 mm KCl, 1 mm EGTA, 1 mm HEDTA, and 1 mm NTA with or without 4 μm 1,8-ANS. The level of total Ca²⁺ needed to achieve a desired free Ca²⁺ in the presence of the above chelators was calculated using WEBMAXC EXTENDED. Calcium was added from a concentrated stock made in the above buffer such that CaM and 1,8-ANS were not diluted during the titration. Temperature was set at 23 °C. Fluorescence data were collected with a QuantaMaster fluorimeter (Photon Technology International) using excitation wavelengths of 276 for Tyr and 370 for 1,8-ANS. Emission was integrated from 290 to 320 nm for Tyr, and from 430 to 520 nm for 1,8-ANS. Data were fit to a Hill equation with one or two binding components as follows.

Where F₀ is the fluorescence without Ca²⁺, Ca_F is the free Ca²⁺ level, A is the amplitude of change due to a given binding event, n is the Hill coefficient, and K is the K_Ca50.

Stopped-flow Measurements

Experiments were performed at 23 °C using an Applied Photophysics Ltd. (Leatherhead, UK) model SX20 MV sequential stopped-flow spectrofluorimeter with a 150 watt Xe/Hg lamp. The system was configured with two 3-ml syringes (A and B) and has a dead time of 1.7 ms, all solutions contained a base buffer of 20 mm MOPS, pH 7.5, 100 mm KCl. Three independent fluorescence indicators were used to measure the rate of dissociation (k_off) of Ca²⁺ from CaM or CaM mutants; intrinsic tyrosine, 1,8-ANS, and Quin-2. For Tyr fluorescence, syringe A contained 4 μm CaM and 0.1 mm CaCl₂, whereas syringe B contained either 10 mm EGTA or 0.1 mm CaCl₂ as a negative control for each CaM. The optical/mixing chamber was excited at 276 nm and fluorescence was collected using Oriel filter 51662. Changes in fluorescence obtained with Ca²⁺ in syringe B were subtracted from data obtained with EGTA in the syringe. Experiments using 1,8-ANS were performed as for Tyr, but with 4 μm 1,8-ANS included in syringe A. 1,8-ANS was excited at 366 nm and fluorescence was collected using a 430-nm cutoff filter (Oriel 51282). Measurements using Quin-2 included 4 μm CaM and 30–60 μm CaCl₂ in syringe A, and 200–400 μm Quin-2 in syringe B. The samples were excited at 334.5 nm and fluorescence was collected using a 430-nm cutoff filter (Oriel 51282). Data collected using all three fluorescent markers fit well to an equation with a single exponential rate constant. Fluorescence from Quin-2 was calibrated by replacing CaM in syringe A with 2, 4, 6, 8, or 10 μm EGTA. The amplitude of the slow rate of release of Ca²⁺ from EGTA (k_off = 0.7 s⁻¹) could be readily used to calibrate the fluorescence response.

NMR Methodology

All NMR experiments were collected on a DRX 600 MHz spectrometer instrument using a 5-mm TXI Cryoprobe at 298 K for apo samples and 310 K for Ca²⁺ samples. The data were processed using FELIX 2007 software. Resonance assignments were made by comparison to wild-type CaM. For experiments requiring decalcified proteins, proteins were decalcified by adding 1 to 5 mm 1,2-bis(o-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid followed by desalting on a Bio-Gel P-10 column equilibrated in buffers that were decalcified by passage over a Calcium Sponge column (Molecular Probes). The decalcified proteins were lyophilized and resuspended in decalcified buffer containing 10 mm imidazole, 100 mm KCl, 5% D₂O at pH 6.3. Calcium was added to 20 mm to the [¹⁵N]apoCaM mutants and the pH adjusted accordingly. Chemical shift perturbation was calculated using the following equation.

Where ΔδH is the change in ¹H chemical shift and ΔδN is the change in ¹⁵N chemical shift.

RESULTS

Mutation at Ca²⁺-binding Sites Alters Binding of PEP-19 to CaM

Fig. 1 shows the consensus sequence of a canonical EF-hand Ca²⁺-binding loop (32), which consists of 12 amino acids that provide 7 coordination positions for Ca²⁺. All mutated CaMs used in this study have this Asp converted to Ala at position 1 as indicated by the arrowhead. Nomenclature for mutant CaM proteins was adopted from the literature (34). For example, CaM3,4 indicates that Asp⁹³ and Asp¹²⁹ were changed to Ala at the X coordination position of Ca²⁺-binding loops III and IV, respectively.

We first determined the relative affinity of PEP-19 for native CaM and the CaM mutant proteins, with a focus on CaM3, CaM4, and CaM3,4 because PEP-19 binds selectively to the C-domain of CaM. CaM with Lys⁷⁵ changed to Cys was labeled with the FRET donor 1,5-IAEDANS (CaM_D), and PEP-19 with a C-terminal Gly-Cys extension was labeled with the FRET acceptor DDPM (PEP-19_A). The relative affinity of the CaM mutants for PEP-19 in the presence or absence of Ca²⁺ was determined by their ability to compete with CaM_D for binding to PEP-19_A. Because PEP-19 binds to CaM with similar affinity in the presence or absence of Ca²⁺ (6), we anticipated that inactivation of Ca²⁺-binding sites would have little effect on binding affinity. However, Fig. 2 shows that only Ca²⁺-CaM3 had similar affinity for PEP-19 as native Ca²⁺-CaM (see Fig. 2A), all other mutants had lower affinity in either the presence or absence of Ca²⁺ as indicated by a rightward shift in the binding curves. Results shown in Fig. 2B in the absence of Ca²⁺ are particularly interesting. If the mutations inhibit or prevent Ca²⁺ binding to a given EF-hand in CaM, but do not greatly alter the structure of CaM in the absence of Ca²⁺, then all mutant proteins should have similar affinity for PEP-19 in the apo forms. In contrast, the apo forms of all mutant proteins have lower affinity for PEP-19 relative to native apoCaM.

FIGURE 2.

Relative affinity of PEP-19 for native CaM and Ca²⁺-binding mutants. The relative affinity of CaMs for PEP-19 was determined using a competition FRET assay in which IAEDANS-labeled native CaM (donor) bound to DDPM-labeled PEP-19 (acceptor) was displaced by increasing concentrations of the indicated unlabeled CaM or the indicated mutant proteins. Under these conditions, effective competition with IAEDANS-labeled CaM results in an increase in fluorescence. In the presence of Ca²⁺ (panel A), only CaM3 has an affinity for PEP-19 that is comparable with that of native CaM. In the absence of Ca²⁺ (panel B), CaM3, CaM4, and CaM3,4 all show lower affinity for PEP-19 than native CaM. Sample conditions for all assays included 1 μm IAEDANS-labeled CaM, 10 μm DDPM-labeled PEP-19, in 20 mm MOPS, pH 7.5, 100 mm KCl with 1 mm EGTA or 2 mm CaCl₂. The lines indicate fits to the equations described under “Experimental Procedures” to derive IC₅₀ values.

We showed previously that the k_on and k_off rates for binding PEP-19 to Ca²⁺-CaM are too rapid to measure using a stopped-flow fluorimeter (6). Fig. 3A shows that mutation of sites III, IV, or both III and IV greatly decrease the rate of dissociation of PEP-19 from CaM in the presence of Ca²⁺. This indicates that Ca²⁺ binding to both sites III and IV are necessary to allow a fast rate of dissociation of PEP-19 from CaM.

FIGURE 3.

Rate of dissociation of PEP-19 from native CaM versus Ca²⁺-binding mutants. Dissociation of CaM from PEP-19 was determined using a stopped-flow fluorimeter as described under “Experimental Procedures.” Essentially, donor-labeled CaM (1 μm) bound to acceptor-labeled PEP-19 (10 μm) was rapidly diluted to achieve partial dissociation of the complex, and an increase in fluorescence. Syringe A contained labeled CaM (1 μm) and labeled PEP-19 (10 μm) with 1 mm EGTA or 2 mm CaCl₂. Syringe B contained 1 mm EGTA or 2 mm CaCl₂, but no proteins. Panel A shows the dissociation in the presence of Ca²⁺ (2 mm CaCl₂). Dissociation from native Ca²⁺-CaM is too fast to detect using a stopped flow with a 1.7-ms dead time. Panel B shows dissociation in the absence of Ca²⁺ (1 mm +EGTA).

In contrast to Ca²⁺-CaM, both k_on and k_off rates for binding PEP-19 to apoCaM are slow and can be easily measured. Rates for mutant proteins should be identical to CaM if mutation of the Ca²⁺-binding loops does not alter the structure of CaM, however, Fig. 3B shows this is true only for CaM4. The k_off rates for CaM3 and CaM3,4 are 3- to 4-fold greater than native CaM. Based on the data in Figs. 2B and 3B, the k_on rates (μm⁻¹ s⁻¹) for binding PEP-19 are: CaM = 0.36; CaM3 = 0.45; CaM4 = 0.15; CaM3,4 = 0.59. Thus, differences in affinity of binding PEP-19 to the mutant proteins can result from changes in k_on or k_off.

Mutations in the C-domain Affect Ca²⁺ Binding to the N-domain

The data in Figs. 2 and 3 suggest that conversion of Asp⁹³ and/or Asp¹²⁹ to Ala causes undesirable structural perturbations in CaM. Thus, we performed a series of experiments to characterize the mutant proteins, especially CaM3,4. We first compared the Ca²⁺ dependence of fluorescence from Tyr⁹⁹ and Tyr¹³⁸ in the C-domain of CaM. Fig. 4A shows that CaM and CaM1,2 respond similarly to Ca²⁺, indicating that mutations in the N-domain do not greatly affect Ca²⁺-dependent Tyr fluorescence from the C-domain. Interestingly, Tyr fluorescence from both CaM3,4 and CaM3 showed little response to Ca²⁺, but had increased basal Tyr fluorescence in the absence of Ca²⁺. This suggests a more hydrophobic environment for Tyr⁹⁸ and Tyr¹³⁸ in the apo forms of CaM3 and CaM3,4 relative to native apoCaM.

FIGURE 4.

Calcium-binding mutations alter intrinsic Tyr fluorescence and 1,8-ANS binding. Panel A shows Tyr fluorescence in the presence and absence of Ca²⁺. Solutions contained 4 μm CaM, 50 mm MOPS, pH 7.5, 100 mm KCl, 1 mm EGTA or 1 mm CaCl₂. Panel B shows fluorescence from 1,8-ANS, which fluoresces upon binding to hydrophobic surfaces. Conditions were the same as in panel A but with 4 μm 1,8-ANS.

We next compared exposure of hydrophobic surfaces in CaM and mutant proteins using 1,8-ANS, which fluoresces upon binding to hydrophobic surfaces. Fig. 4B shows much greater fluorescence from 1,8-ANS in the presence of Ca²⁺-CaM versus apoCaM. In contrast, there is little change in 1,8-ANS fluorescence when Ca²⁺ is added to CaM1,2, indicating that the increase in 1,8-ANS fluorescence in the presence of native CaM is primarily due to Ca²⁺ binding to sites I and II in the N-domain. This is confirmed by CaM3,4, which does not bind Ca²⁺ to sites III and IV based on Tyr fluorescence (see Fig. 4A), but has a robust Ca²⁺-dependent increase in ANS fluorescence upon binding Ca²⁺. The increase is much greater than observed with native CaM, suggesting that mutation of sites III and IV in the C-domain alters exposure of hydrophobic surfaces in response to Ca²⁺ binding to sites I and II in the N-domain. This is also observed for CaM3, but to a much lesser extent with CaM4.

Relative Ca²⁺ binding affinities were determined by monitoring Tyr or 1,8-ANS fluorescence during titration of CaM and mutant derivatives with Ca²⁺. The results shown in Fig. 5 and Table 1 can be interpreted with respect to binding Ca²⁺ to the C- and N-lobes with K_Ca50 values of around 2 and 13 μm, respectively (5, 35, 36). Tyr fluorescence detects a single class of Ca²⁺-binding sites in native CaM with K_Ca50 of 2.5 μm, which corresponds with Ca²⁺ binding to the C-domain. Fluorescence data for native CaM collected using 1,8-ANS fit well to a single class of binding sites with a K_Ca50 of 5.2 μm. We interpret this K_Ca50 to reflect binding to sites I and II because Fig. 4 shows that 1,8-ANS is most sensitive to Ca²⁺ binding to the N-domain. The lower K_Ca50 relative to the native CaM may be due to a contribution to fluorescence from the C-domain, or an increase in Ca²⁺ binding affinity of the N-domain due to 1,8-ANS binding.

FIGURE 5.

Equilibrium Ca²⁺ binding to native and mutant CaM. Fluorescence from Tyr (panel A) and 1,8-ANS (panel B) was used to measure equilibrium Ca²⁺ binding to CaM. Solutions contained 4 μm CaM, 50 mm MOPS, pH 7.5, 100 mm KCl, 1 mm EGTA, 1 mm HEDTA, 1 mm NTA with or without 4 μm 1,8-ANS. The level of total Ca²⁺ needed to achieve a desired free Ca²⁺ in the presence of the above chelators was calculated using WEBMAXC EXTENDED. Data are expressed as a percent of maximal fluorescence for each protein to normalize for differences in absolute maximal fluorescence shown in Fig. 4. Lines represent fits of the data to a Hill equation with one or two binding components as described under “Experimental Procedures.”

TABLE 1.

Equilibrium calcium binding

Data such as that shown in Fig. 5 were fit to Hill equations with one or two Ca²⁺-binding components. All values are the average ± S.D. of at least three independent titrations. S.D. values of less than 0.1 are reported as 0.1.

	Method	KCa50 (μm) and Hill coefficient (n)
		K1	n1	K2	n2
CaM	Tyr	2.5 ± 0.7	2.1 ± 0.1	—a	—
	ANS	5.2 ± 0.5	1.5 ± 0.1	—	—
CaM1,2	Tyr	2.4 ± 0.5	2.1 ± 0.1	—	—
	ANS	1.9 ± 0.1	1.9 ± 0.1	—	—
CaM3,4	Tyr	—	—	—	—
	ANS	2.1 ± 0.1	2.3 ± 0.1	—	—
CaM3	Tyr	1.0 ± 0.1	2.0 ± 0.1	33 ± 7	1.3 ± 0.1
	ANS	2.2 ± 0.1	2.0 ± 0.1	30 ± 5	1.4 ± 0.1
CaM4	Tyr	10 ± 2	2.2 ± 0.4	44 ± 10	1.3 ± 0.2
	ANS	4.1 ± 0.1	2.2 ± 0.1	53 ± 3	1.5 ± 0.1

^a Indicates either no detectable signal or that the data fit well to an equation with a single binding component.

Both Tyr and 1,8-ANS fluorescence detect a single class of Ca²⁺-binding sites in CaM1,2 with a K_Ca50 of about 2 μm, which is consistent with Ca²⁺ binding to the C-domain. Although CaM3,4 shows no Ca²⁺-dependent change in Tyr fluorescence, 1,8-ANS detects a single binding component with a K_Ca50 of 2.1 μm. This is due to Ca²⁺ binding to the N-domain because NMR data show that Ca²⁺ does not bind to the C-domain of CaM3,4. The K_Ca50 of 2.1 μm is less than for the N-domain of native CaM by about 11 μm (5, 35, 36). Thus, mutation of Ca²⁺ binding sites III and IV increases the affinity of Ca²⁺-binding to sites I and II in the N-domain.

Fig. 5B and Table 1 show that CaM3 and CaM4 have biphasic Ca²⁺ binding curves. Although values derived using Tyr and 1,8-ANS fluorescence are not identical, the trends are clear. Barring some unusual cooperative interactions between N- and C-domain EF-hands, the simplest interpretation of the data is that the higher affinity site(s) associated with a large increase in 1,8-ANS fluorescence is due to binding Ca²⁺ at sites I and II in the N-domain of CaM3 and CaM4 because the Hill coefficient is close to 2. The lower affinity class of binding sites has a K_Ca50 of 30–50 μm, and Hill coefficient of less than 2. This is consistent with Ca²⁺ binding to the remaining active site in the C-domain of CaM3 and CaM4, which would have lower affinity due to the absence of cooperativity (6).

Regardless of the microscopic assignment of Ca²⁺-binding sites, Fig. 5 and Table 1 demonstrate interdomain effects of mutations. First, fluorescence from Tyr⁹⁹ and Tyr¹³⁸ in the C-domain, which is insensitive to Ca²⁺ binding to the N-domain of native CaM, responds to Ca²⁺ binding to the N-domain of CaM3 and CaM4 because biphasic Ca²⁺ binding curves are observed. Second, the apparent affinity of Ca²⁺ binding to sites I and II is increased for CaM3 and CaM3,4 relative to native CaM.

Effect of Ca²⁺-binding Mutations on Rates of Ca²⁺ Dissociation

We next determined the effect of mutations on the rates of Ca²⁺ dissociation (k_off) using stopped-flow techniques to monitor fluorescence from 1,8-ANS, Tyr, and Quin-2. Data obtained with 1,8-ANS are shown in Fig. 6, and rates obtained with all methods are shown in Table 2.

FIGURE 6.

Effect of mutations on the rates of Ca²⁺ dissociation. Dissociation of Ca²⁺ (k_off) from the various CaMs was measured using fluorescence from 1,8-ANS, Tyr, and Quin-2 as described under “Experimental Procedures.” All fluorescent markers gave very similar k_off values as shown in Table 2. The arrow in panel A indicates very rapid release of Ca²⁺ from CaM, which occurs in the dead-time of the stop flow, indicating a k_off of >500.

TABLE 2.

Calcium dissociation rate constants

All values are the average ± S.D. of at least three independent data sets. S.D. values of less than 0.1 are reported at 0.1. Moles of Ca²⁺ released per mol of protein determined using Quin-2 were based on calibrating the fluorescence response with Ca²⁺ release from EGTA.

	Method	koff	Ca2+/protein
		s−1
CaM	Quin-2	9.5 ± 0.2	2.0 ± 0.1
	1,8-ANS	10.5 ± 0.1
	Tyr	10.3 ± 0.4
CaM1,2	Quin-2	9.7 ± 0.1	2.0 ± 0.1
	1,8-ANS	12.5 ± 0.2
	Tyr	9.8 ± 1.1
CaM3,4	Quin-2	296 ± 1	1.9 ± 0.1
	1,8-ANS	270 ± 9
	Tyr	NDa
CaM3	Quin-2	282 ± 11	1.9 ± 0.1
	1,8-ANS	217 ± 6
	Tyr	ND
CaM4	Quin-2	59 ± 3	1.6 ± 0.1
	1,8-ANS	52 ± 2
	Tyr	40 ± 2

^a ND indicates no detectable change in Tyr fluorescence.

The vertical arrow in Fig. 6A shows a rapid decrease of fluorescence from 1,8-ANS during the dead time of the stopped-flow fluorimeter when Ca²⁺-CaM is rapidly mixed with EGTA. This is consistent with rapid dissociation of Ca²⁺ from the N-domain of CaM, which is too fast to be detected by stopped-flow methods. This fast phase is not seen for CaM1,2, because both N-domain sites are inactivated. Both CaM and CaM1,2 show a slower dissociation of 2 Ca²⁺/mol of protein with similar rates of 10–12 s⁻¹, which is consistent with release of Ca²⁺ from sites III and IV in the C-domain. These data, together with Table 1, show that mutation of sites I and II do not affect Ca²⁺ binding to sites III and IV.

Very similar data were obtained using CaM3,4 and CaM3. Neither protein showed a detectable change in Tyr fluorescence, indicating an insensitivity to Ca²⁺ release, or that the change occurs too rapidly to measure using stopped-flow techniques. Data collected with Quin-2 and 1,8-ANS fit single exponential decays with rates of between 200 and 300 s⁻¹, and stoichiometry of 2 Ca²⁺/mol of protein. The simplest explanation for these data is that Quin-2 and 1,8-ANS detect dissociation of Ca²⁺ from the N-domain of CaM. This would mean that mutation of C-domain Ca²⁺-binding sites decreases the rate of Ca²⁺ dissociation from the N-domain, and would account for the increase in affinity of Ca²⁺ binding to the N-domain in CaM3 and CaM3,4 shown in Table 1.

Dissociation of Ca²⁺ from CaM4 shows a single rate of 40 to 60 s⁻¹. There are two possibilities for this result. One is that the observed rate is associated with Ca²⁺ release from the active site III in CaM4, and that release of Ca²⁺ from sites I and II in the N-lobe is too rapid to detect. The other possibility is that release of Ca²⁺ from the active site III is too fast to be observed, and that the observed rate is associated with dissociation of Ca²⁺ from the N-domain, which is sensed by Tyr residues due to mutation of site IV in CaM4. The latter seems more plausible because: 1) low affinity binding of Ca²⁺ to site III would likely be associated with a rapid rate of dissociation; 2) rapid release of Ca²⁺ from the N-lobe would be detected as a large decrease of 1,8-ANS fluorescence during the dead time of the stopped-flow fluorimeter, but this is not observed; and 3) the stoichiometry of Ca²⁺ release is consistent with release of 2 Ca²⁺ from the N-lobe rather than 1 Ca²⁺ from site III.

Together, the data in Figs. 5 and 6 and Tables 1 and 2 indicate that mutation of Ca²⁺-binding sites III and IV in the C-domain can affect Ca²⁺ binding to sites I and II in the N-domain. This is most evident for CaM3,4 because it does not bind Ca²⁺ to the C-domain, and exhibits Ca²⁺ binding to the N-domain with higher affinity and slower k_off rate than native CaM.

NMR Spectra Indicate Altered Apo Structure in CaM3,4 and Interdomain Effects of Mutations

NMR was used to assess potential structural changes due to conversion of Asp⁹³ and Asp¹²⁹ to Ala. Fig. 7A compares a region of the ¹H-¹⁵N HSQC spectra for CaM (gray) and CaM3,4 (black) collected in the absence of Ca²⁺ (see supplemental Fig. S1 for complete spectra). Cross-peaks for amides in the N-domain of CaM3,4 overlap with those of CaM, but amides in the C-domain (residues 76–148) of CaM3,4 do not overlap with those of native CaM due to severe exchange broadening or large differences in chemical shifts. Fig. 7B shows that major differences are also seen in the crowded, central region of the spectra. Almost all C-domain resonances are significantly affected by changing Asp⁹³ and Asp¹²⁹ to Ala, and could not be assigned by comparison to amide chemical shifts for native CaM.

FIGURE 7.

Mutation of Asp⁹³ and Asp¹²⁹ destabilizes the C-domain of CaM. Fig. 7 compares selected regions of ¹H-¹⁵N HSQC spectra collected for native CaM (gray) and CaM3,4 (black) in the absence of Ca²⁺. The upper panel shows that amide resonances for residues in the N-domain of CaM3,4 (residues 1–75) are very similar to those of CaM, but amides for residues in the C-domain (residues 76–148) are greatly affected by the mutations. The lower panel shows that major differences are also seen in the crowded central region of the spectra.

Overall, C-terminal amide resonances in apoCaM3,4 show a loss of dispersion that is consistent with a loss of tertiary structure, or increased disorder in the C-domain due to the mutations. An immediate question is whether conversion of Asp⁹³ or Asp¹²⁹ to Ala is most responsible for this structural change? A quantitative answer would require extensive heteronuclear experiments for backbone assignments. Instead, we compared ¹⁵N HSQC spectra for apo-native CaM, CaM3, and CaM4, and focused on C-domain amides that are in well resolved regions of the spectra. This type of analysis shows that mutation of Asp⁹³ in site III causes the greatest structural change because amide cross-peaks for all of the following residues overlap in the spectra for apoCaM and apoCaM4, but are significantly shifted for apoCaM3: 91, 92, 95, 96, 97, 100, 101, 102, 109, 110, 113, 117, 118, 119, 127, 139, 146, 147, and 148. The “Discussion” will provide a plausible explanation for the greater structural perturbation due to mutation of Asp⁹³ in site III.

The majority of N-lobe amide assignments for the mutant proteins could be made by comparison with native CaM, however, resonances in the N-domain were affected by mutations in the C-domain. Fig. 8 shows chemical shift differences between native apoCaM and the apo forms of CaM3, CaM4, and CaM3,4. Chemical shifts for amides in site II are most affected by mutations in sites III and IV, especially Asn⁶⁰ and Gly⁶¹ in CaM3, CaM4, and CaM3,4, and Asp⁶⁴ in CaM3,4 shown by the gray bars. These data are consistent with Figs. 4–6, which show that Ca²⁺ binding to the N-domain is altered as a consequence of mutations in the C-domain.

FIGURE 8.

Mutation of Asp⁹³ and Asp¹²⁹ affects the conformation of the apo N-domain. The majority resonances for backbone amides in the N-domain of the mutant CaM proteins could be assigned by comparison of ¹H-¹⁵N HSQC spectra of mutant proteins to that of native CaM. Panels A–C compare chemical shift differences between native apoCaM versus apoCaM3, -CaM4, and -CaM3,4. The differences are small but measurable, and the greatest differences are localized to Ca²⁺-binding loops I and II. The insets indicate positions of residues in the solution structure of the apo N-domain that have amide chemical shift differences greater than 0.02.

Fig. 6 suggests that mutation of sites III and IV in the C-domain of CaM affects the structure of the Ca²⁺-bound N-domain because the rate of dissociation of Ca²⁺ from the N- domain of CaM3,4 is significantly slower than for native CaM. To further investigate this, the amide chemical shifts of CaM3,4 were followed during titration with Ca²⁺ and compared with native CaM. Fig. 9 shows chemical shift differences for residues that could be confidently assigned in the Ca²⁺-bound N-domain of CaM3,4 versus native CaM. Observed differences are greater in magnitude than for the apoproteins shown in Fig. 8, and the greatest changes are clustered in the hydrophobic pocket as shown by the inset. A change in exposure of hydrophobic surfaces in the Ca²⁺-bound N-domain due to mutations in the C-domain is consistent with the increased ANS fluorescence from CaM3,4 shown in Fig. 4B.

FIGURE 9.

Interdomain effects of Ca²⁺ binding to CaM3,4. Amide chemical shifts differences are shown for the N-domain of CaM3,4 versus native CaM in the presence of Ca²⁺. The insets indicate positions of residues in the solution structure of the N-domain that have amide chemical shift differences greater than 0.08 in CaM3,4 relative to native Ca²⁺-CaM. The greatest differences are localized to the hydrophobic pocket.

DISCUSSION

The original goal of this study was to determine whether Ca²⁺ binding to sites III or IV in the C-domain of CaM has dominant effects on the characteristics of binding PEP-19. The two most common strategies to inhibit Ca²⁺ binding to EF-hand motifs are: 1) convert Glu at loop position 12 to Gln, Lys, or Ala (7, 8) (see Fig. 1); and 2) convert Asp at loop position 1 to Ala (9). The latter mutation was used previously to inactivate Ca²⁺ binding to cardiac troponin C (10), skeletal troponin C (11), myosin regulatory light chain (12), and CaM (9). Mutation at loop position 1 in CaM has been used extensively to assess the relative functional contribution of Ca²⁺ binding to the N- and C-domains of CaM when expressed in cell culture (13–31). Thus, we elected to convert Asp to Ala in the first ligand position in all Ca²⁺-binding mutants. Although the data in Figs. 2 and 3 indicate that Ca²⁺ binding to both sites III and IV of CaM is important for native-like interactions with PEP-19, apparent structural perturbations induced by the mutations complicated interpretation of the data. This led us to characterize the effect of these mutations on the Ca²⁺ binding properties and structure of CaM.

Data in Figs. 5 and 6 and Tables 1 and 2 show that mutation of Ca²⁺-binding sites in the C-domain increases the Ca²⁺ binding affinity of the N-domain by decreasing the rate of Ca²⁺ dissociation, but that mutation of sites in the N-domain has no apparent effect on the C-domain. This interdomain effect is consistent with other studies showing that mutation of the 1st or 12th ligand positions in sites III and IV decreased the rate of Ca²⁺ dissociation from the N-domain (37, 38), and that mutation of Val¹³⁶ to Gly in site IV increases the Ca²⁺ binding affinity of the N-domain (39).

NMR analyses in Figs. 7–9 show that mutation of Ca²⁺-binding sites in the C-domain causes significant changes in amide chemical shifts that are consistent with a loss of structure. In addition, mutations in the C-domain affect amide chemical shifts in the N-domain. There are at least two general mechanisms for this observation. One involves a specific interaction between domains that are disrupted by mutations. A recent crystal structure of Ca²⁺-CaM shows a collapsed structure in which helix A in the N-domain makes significant H-bonding interactions with helices G and H in the C-domain (40). Such a conformation would undoubtedly be a minor species in solution. Moreover, the apparent K_Ca50 for Ca²⁺ binding to isolated N and C domains of CaM are very similar (<1.5-fold different) to those determined for the intact protein (35, 36, 41, 42). This would not be expected if the Ca²⁺ binding properties of the N- and C-domains of intact CaM experience an intrinsic effect of domain-domain interactions.

A second mechanism for interdomain effects of mutations involves relatively nonspecific interactions between the N-domain and the mutated C-domain. Even though a highly flexible tether links these domains, it is reasonable that random arcing motions would allow nonspecific collisions. Changes in the surface chemistry due to mutations could lead to enhanced interactions between domains. Fig. 9 shows that residues in the hydrophobic pocket of the Ca²⁺-bound N-domain are most affected by mutation of Asp⁹³ and Asp¹²⁹ in the C-domain. This supports a model in which exposed surfaces in the mutated C-domain associate with the N-terminal hydrophobic pocket in the presence of Ca²⁺ to stabilize the Ca²⁺ bound form, and increase Ca²⁺ binding affinity by decreasing the k_off.

Interestingly, inactivation of sites I and II in CaM1,2 does not affect Ca²⁺ binding to the C-domain. Because the overall affinity of Ca²⁺ binding to the C-domain (k_IIIk_IV) is about 30-fold stronger than the N-domain (k_Ik_II) (5, 35, 36, 42, 43), Ca²⁺ binds sequentially first to the C-domain, whereas the N-domain is predominately in the apo state. Thus, inhibition of Ca²⁺ binding to the N-domain would not be expected to affect binding to the C-domain. The relative structural stability of the N- and C-domains may also contribute to interdomain effects of mutations. The apo N-domain has a defined hydrophobic core and a single conformation, but the apo C-domain has an ill-defined aromatic hydrophobic core (44, 45), multiple thermal melting transitions (46), with regions of intrinsic disorder (45, 47), and a high degree of backbone conformational entropy that allows global conformation exchange between at least two conformations (45, 48, 49). Thus, mutation in the C-domain may have a greater effect on structural stability than similar mutations in the N-domain.

Comparison of ¹⁵N HSQC spectra for native apoCaM, CaM3, and CaM4 shows that conversion of Asp⁹³ to Ala in CaM3 has the greatest effect on structure of the C-domain. Structures of apo Ca²⁺-binding loops in Fig. 10 provide a mechanistic basis for this observation. The side chains of Asp¹²⁹ in loop IV, and the corresponding Asp⁵⁶ in loop II (not shown) are both solvent exposed. Thus, significant structural effects would not be predicted from mutation of these residues. In contrast, Fig. 10 shows that the side chain carboxyl group of Asp⁹³ points toward the backbone of other residues in the Ca²⁺-binding loop of EF-hand III and is 2–3 Å from backbone amides of Asp⁹⁵, Gly⁹⁶, Asn⁹⁷, and Gly⁹⁸ as shown by the yellow lines. A similar conformation is seen for Asp²⁰ of loop I in the N-domain. This suggests that the apo forms of loops I and III are stabilized by hydrogen bonds between side chains of Asp at position 1, and backbone amides of residues at positions 3 to 6. Conversion of Asp⁹³ or Asp²⁰ to Ala would eliminate these hydrogen bonds. The extent of structural perturbation in the N- and C-domains may depend on differences in stability in the apo state as summarized above. Specifically, the apo C-domain is less stable than the N-domain and may be more sensitive to mutations.

FIGURE 10.

NMR solution structures for apoCa²⁺-binding loops III and VI. Solution structure coordinates were taken from Kuboniwa et al. (45). Asp⁹³ and Asp¹²⁹ are the first amino acids in loops III and IV, respectively, and provide the X Ca²⁺ coordination ligand. Glu¹⁰⁴ and Glu¹⁴⁰ are the 12 amino acids in the loops and provide the −Z coordination position. Yellow lines indicate backbone amide hydrogens that are within 2 to 3 Å of carboxyl side chain of Asp⁹³.

This raises the question of what is the best way to mutate Ca²⁺-binding sites of EF-hand proteins for in vivo and in vitro studies? Mutation strategies must inhibit Ca²⁺ binding, but should also maintain structural integrity of the apo form to allow productive interactions with target proteins that require the apo state. The data presented here argue that mutation of the first ligand is not the best choice for inactivation of EF-hands III and IV in the C-domain of CaM. However, similar mutations may be suitable for the N-domain, which has greater intrinsic stability in the apo form.

Mutation of Glu at loop position 12 (see Figs. 1 and 10) has been used to inhibit Ca²⁺ binding to CaM (7, 8), because its side chain carboxyl group provides bidentate coordination of Ca²⁺ (32) (see Fig. 1) and it is solvent exposed in the apo form (44, 45). Mutation of Glu¹⁰⁴ or Glu¹⁴⁰ to Gln in sites III and IV, respectively, does not abolish Ca²⁺ binding, but greatly reduces affinity (50, 51). This is consistent with an analogous mutation in site I of skeletal troponin C, which allows weak Ca²⁺ binding via the first 5 coordination positions, but the EF-hand does not undergo a conformational change (52). Relative to CaM3 or CaM4, published NMR spectra of CaM with E104Q or E140Q mutations show fewer and less dramatic changes in amide chemical shifts in the C-domain (50, 51). Although this is desirable, it was shown that CaM E140Q regained Ca²⁺ binding activity in the presence of a high affinity CaM binding peptide (53). We see no evidence for recovery of Ca²⁺ binding to mutated sites in CaM1,2 or CaM3,4 upon association with a high affinity CaM binding peptide from CaM kinase II, CaMKII-(293–312) (data not shown).

The potential to recover Ca²⁺ binding at mutated EF-hands is an important consideration, and this potential will likely depend on the nature of the mutation, the EF-hand protein being studied and its putative target proteins. For example, conversion of Asp to Ala at the Y coordination position in sites II, III, or IV of cardiac troponin C inhibited Ca²⁺ binding to the free protein, however, Ca²⁺ binding activity was recovered at sites III and IV, but not site II, when the mutant proteins were incorporated into a troponin complex (54).

In summary, mutation strategies for inhibition of Ca²⁺ binding to EF-hands must balance retention of structural integrity against the potential for recovery of Ca²⁺ binding activity upon interaction with native protein binding partners. This is especially problematic if the mutated proteins are to be expressed in cell culture where there is the potential for interaction with multiple binding proteins.