The IQ domains in neuromodulin and PEP19 represent two major functional classes

Abstract

The affinities of Ca²⁺-saturated and Ca²⁺-free calmodulin for a fluorescent reporter construct containing the PEP19 IQ domain differ by a factor of ~100, with K_d values of 11.0±1.2 and 1128.4±176.5 µM, while the affinities of a reporter containing the neuromodulin IQ domain are essentially identical, with respective K_d values of 2.9±0.3 and 2.4±0.3 µM. When Ca²⁺ is bound only to the C-ter pair of Ca²⁺-binding sites in calmodulin the K_d value for the PEP19 reporter complex is decreased ~5-fold, while the value for the the neuromodulin reporter complex is increased by the same factor. When Ca²⁺ is bound only to the N-ter pair of Ca²⁺-binding sites the K_d value for the PEP19 reporter complex is unaffected, but the value for the complex with the neuromodulin reporter is increased ~12-fold. These functional differences are largely accounted for by three differences in the CaM-binding sequences of the two reporters. Replacement of a central Gly in the neuromodulin IQ domain with the Lys at this position in the PEP19 sequence almost entirely accounts for the distinctive patterns of Ca²⁺-dependent stability changes exhibited by the two complexes. Replacement of a Lys immediately before the “IQ” amino acid pair in the neuromodulin sequence with an Ala accounts for the remaining Ca²⁺-dependent differences. Replacement of an Ala in the N-ter half of the IQ domain with a Gln accounts for about half the Ca²⁺-independent difference in the stabilities of the two reporter complexes, with the Ca²⁺-independent effect of the Lys replacement accounting for most of the remainder. Since the central Gly in the neuromodulin IQ domain is conserved in half of all known IQ domain sequences, these results suggest that the presence or absence of this residue defines two major functional classes of IQ domains.

The Ca²⁺-binding protein calmodulin (CaM¹) participates at multiple levels in essentially all cellular processes. It interacts with an estimated 100 intracellular proteins, whose activities it modulates to varying degrees with varying dependencies on the intracellular free Ca²⁺ concentration. CaM contains two pairs of EF hand Ca²⁺-binding domains, each comprising a globular structure which we term a “lobe”. The two lobes are joined by a solvent-exposed flexible helix (1). Due to cooperativity within each pair of Ca²⁺-binding sites (Ca²⁺)₂CaM, with either the N-ter or C-ter EF hand pair occupied, and (Ca²⁺)₄CaM are the major Ca²⁺ liganded species produced (2). A number of proteins, including neuromodulin, neurogranin and PEP19, the unconventional myosins, ion channels (3–5), and modulators of small G-proteins, bind CaM through so called “IQ domains”. Depending upon their amino acid sequences, these domains can bind Ca²⁺-free and Ca²⁺-bound forms of CaM with quite different degrees of preference, allowing for a variety of Ca²⁺-dependent switching behaviors (4).

IQ domains were first identified as light chain subunit binding sites in conventional myosins, and the consensus sequence was defined as IQxxxRGxxxR (4, 6). As additional homologous domains have been discovered in other proteins a less restricted consensus sequence has emerged: [I,L,V]QxxxR[G,x]xxx[R,K] (4, 6). Structural studies demonstrate that a narrow hydrophobic cleft in the Ca²⁺-free C-ter CaM lobe and in the homologous regions in the myosin light chain subunits is involved in recognition of the “IQ” amino acid pair in the IQ domain (7–9). The Ca²⁺-free N-ter lobes in CaM and the light chains contain no such cleft, and their interactions with the IQ domain appear to be variable (7–9).

Structural and biophysical studies indicate that in the absence of Ca²⁺ there are two types of light chain or CaM complexes with IQ domains: a compact structure in which the N-ter lobe interacts closely with the IQ domain, and an extended form in which the N-ter lobe is only weakly associated (7, 8). The compact structure is formed if the semi-conserved central Gly residue in the IQ consensus sequence is present, and the extended structure is formed if there is a bulky amino acid at this position (7, 8). Extensive hydrophobic surfaces are exposed on both CaM lobes when they are replete with Ca²⁺, and these appear to enfold IQ domains to form structures similar to those observed with other types of Ca²⁺-saturated CaM complexes (10, 11). Structural and biophysical studies further suggest that IQ domains adopt an α-helical conformations when they are bound to Ca²⁺-free or Ca²⁺-saturated CaM (7–11).

The semi-conserved Gly in the IQ domain consensus sequence is present in neuromodulin, but has been replaced by a Lys in PEP19. The complexes between CaM and the IQ domains in these proteins are therefore expected to belong to different structural classes. In order to investigate the possible functional correlates of these classes, we compared the affinities of the Ca²⁺-free, Ca²⁺-saturated and intermediate Ca²⁺-bound CaM complexes with the neuromodulin and PEP19 IQ domains. Our results suggest that the IQ domains in neuromodulin and PEP19 represent two major functional classes whose complexes with CaM each exhibit a distinctive pattern of Ca²⁺-dependent stability changes.

Materials and Methods

The cDNA encoding B_IQ, a fluorescent protein reporter that binds CaM via an IQ domain insert sequence derived from neuromodulin, has been described in detail elsewhere (12–14). The CaM-binding sequences in these reporter constructs are inserted between cyan and yellow variants of green fluorescence protein. When CaM is bound fluorescence resonance transfer from the cyan donor to the yellow acceptor is decreased (12–14). The B_IQ reporter construct, which contains the neuromodulin IQ domain, and intact neuromodulin have previously been shown to bind CaM with similar affinities (12). The insert sequences in the reporter constructs we have used in these studies are listed in Fig. 1. Fluorescent protein reporter constructs were expressed in E. coli BL21(DE3) and purified as described previously (15). Native and mutant vertebrate CaMs were expressed in E. coli and purified as described in detail elsewhere(16, 17). Two mutant CaMs were employed to determine the effects of Ca²⁺-binding solely to the N-ter and C-ter EF hand pair: N_xCCaM (N_xC), in which Ca²⁺ ligands at positions 31 and 67 in the N-ter EF hand pair have been replaced with alanines, and NC_xCaM (NC_x), in which the homologous ligands at positions 104 and 140 in the C-ter EF hand have been replaced (12). We have previously determined that the mutant EF hand pairs in these proteins do not bind Ca²⁺ under the experimental conditions used for these investigations, and that the mutant CaM lobes mimic the properties of Ca²⁺-free native lobes (12).

FIG 1. Native and mutant IQ domain sequences that were investigated.

The sequences listed are inserts in a previously described reporter construct that responds to CaM binding with decreases in fluorescence resonance energy transfer between two fluorescent protein variants (12–14). These decreases were monitored based on fluorescence emission at 525 nm (430 nm excitation). The aligned insert sequences are numbered in relation to the first residue (0) in the 11 residue consensus IQ motif sequence: [I,L,V]QxxxR[G,x]xxx[R,K], which is underscored by a heavy line. Positions -1, 3 and 6 (boxed) were the focus of these investigations. The insert sequences in B_IQ and B_IQPEP are from neuromodulin and PEP19. The others were derived by replacing amino acids in the B_IQ sequence with those in the B_IQPEP sequence at one or all three (B_IQTr) of the indicated positions.

Fluorescence Measurements

A Photon Technologies International (Monmouth Junction, NJ) QM-1 fluorometer operated in photon-counting mode was used for all equilibrium fluorescence measurements. Monochromator excitation and emission slit widths were set to produce bandwidths of ~2.5 nm. All experiments were performed at 23 °C. The standard experimental buffer contained 25 mM Tris (pH 7.5), 100 mM KCl, 100 µg/mL BSA, 25 µM TPEN, and other components as specified in the text or captions. Nominally Ca²⁺-free conditions were produced by including 3 mM BAPTA. Saturation of functional Ca²⁺ binding sites in native or mutant CaMs was achieved by adding CaCl₂ to the standard buffer and CaM stock solutions to produce a final free Ca²⁺ concentration of ~250 µM.

Analysis of Fluorescence Data

Decreases in yellow acceptor fluorescence protein emission at 525 nm (cyan donor excited at 430 nm) due to CaM binding were used to determine fractional binding to reporter IQ domains as described in detail elsewhere (12). Apparent K_d values were derived from these measurements using a standard hyperbolic binding equation: FR = [CaM]/([CaM] + K_d). Fractional response (FR) is defined as (F_max - F)/(F_max - F_min), where F corresponds to the 525 nm fluorescence emission measured after each addition of CaM, and F_max and F_min correspond to the fluorescence of the CaM-free and CaM-saturated reporter. In cases of very low affinity interactions where F_min could not be approached, its value was allowed to float during fitting calculations. The F_min values thus obtained were not found to differ significantly from the experimental values determined for higher affinity interactions. A hyperbolic binding equation is applicable to these data because the 15–100 nM reporter concentrations used are at least 5-fold below the lowest CaM concentrations used. Thus, the free and total CaM concentrations can be considered to be the same. This equation describes a single-site dependence of the fractional response on [CaM], so the excellent fits obtained support a 1:1 binding stoichiometry. We have previously confirmed this by performing stoichiometric titrations of the reference B_IQ reporter (data not shown). Reported K_d values are the means of 3–5 independent determinations. Errors are expressed as the standard error of the mean (SEM). The difference between two values is considered statistically significantly if the p value derived from an unpaired t-test is less than 0.05.

Results

CaM binding to B_IQ and B_IQPEP

The first step in these investigations was to evaluate binding by B_IQ and B_IQPEP of CaM in its four principal Ca²⁺-liganded states, namely Ca²⁺-free, fully Ca²⁺-saturated, and partially saturated; with Ca²⁺ bound only to the N-ter (N₂CCaM) or C-ter EF (NC₂CaM) hand pair. Representative data are presented in Fig 2 and Fig 3, and the K_d values derived from these and similar data are listed in Table 1. Determinations performed with B_IQ confirm results obtained previously with this construct (12). Thus, Ca²⁺-free and Ca²⁺-saturated CaM bind this reporter with similar K_d values of 2.4±0.3 and 2.9±0.3 µM. NC₂CaM and NC₂CaM are both bound more weakly, with respective K_d values of 12.7±1.1 and 31.3±2.7 µM. In contrast, B_IQPEP binds Ca²⁺-free and Ca²⁺-saturated CaM with significantly different K_d values of 1128.4±176.5 and 11.0±1.2 µM, and it binds NC₂CaM and N₂CCaM with respective K_d values of 207.6±16.5 and 988.4±146.8 µM. Thus, Ca²⁺ binding to only one EF hand pair either increases the affinity of the B_IQPEP complex or has little effect, dependent upon which pair of sites is occupied.

FIG 2. Comparison of the CaM-binding properties of reporters containing the neuromodulin and PEP19 IQ domains.

Binding was evaluated based on the fractional changes in the fluorescence emission of the B_IQ (■, □) or B_IQPEP (●,○) reporter constructs. Binding experiments were performed in the presence of ~250 µM free Ca²⁺ (filled symbols) or 3 mM BAPTA (empty symbols). Apparent K_d values derived from fits of these (solid lines) and similar data to a hyperbolic binding equation are listed in Table 1. Fluorescent reporter concentrations were 25–100 nM. The standard experimental buffer contained 25 mM Tris (pH 7.5), 100 mM KCl, 100 µg/mL BSA, 25 µM TPEN, and other components as specified. The fitted F_min value for binding of Ca²⁺-free CaM to B_IQPEP is 0.55±0.07. This is not different from the value of 0.56±0.07 derived for the Ca²⁺-free CaM complex with B_IQ.

FIG 3. The effects of Ca²⁺ binding to the N-ter or C-ter EF hand pair in CaM on its affinities for reporters containing the neuromodulin and PEP19 IQ domains.

Data are presented for binding of B_IQ to N_xC (■) or NC_x (▲) CaM and for binding of B_IQPEP to N_xC (●) or NC_x (◆) CaM. At the ~250 µM free Ca²⁺ concentration used for these experiments the native EF hand pairs in N_xCCaM and NC_xCaM are Ca²⁺ saturated. Apparent K_d values derived from fits of a hyperbolic binding equation to these (solid lines) and similar data are listed in Table 1. Additional experimental details are given in the caption to Fig. 2. The fitted F_min values for binding of NC₂CaM and N₂CCaM CaM to B_IQPEP are 0.72±0.03 and 0.70±0.04. These are not different from the respective values of 0.74±0.03 and 0.71±0.01 derived for the complexes with B_IQ.

Table 1. K_d and ΔΔG_B values for binding of Ca²⁺-free and Ca²⁺-bound CaM.

N and C refer to the N-terminal and C-terminal EF hand pairs in CaM, respectively, with subscripts indicating their Ca²⁺-liganded states (2 or no Ca²⁺ bound). K_d values are given as means ± standard error of the mean calculated from 3 – 5 individual determinations. The row labeled “SUM” contains the arithmetic sums of the ΔΔG_B values for the B_IQ variants containing single amino acid replacements. The nomenclature for IQ domain reporter constructs is defined in Fig. 1. Differences in binding energy (ΔΔG_B) were calculated according to the relation: ΔΔG_B = RTln(K_dB_IQN/K_dB_IQ), where K_dB_IQ is the K_d value for a Ca²⁺-free or Ca²⁺-bound CaM-B_IQ complex and K_dB_IQN is the K_d value for the corresponding complex with B_IQPEP or a substituted B_IQ variant, as specified in the table.

	NC		NC2*		N2C*		N2C2
Construct	Kd (µM)	ΔΔGB (kJ/mol)	Kd (µM)	ΔΔGB (kJ/mol)	Kd (µM)	ΔΔGB (kJ/mol)	Kd (µM)	ΔΔGB (kJ/mol)
BIQ	2.4±0.3	-	12.7±1.1	-	31.3±2.7	-	2.9±0.3	-
BIQK−1A	21.1±2.8	5.4±0.9	46.6±3.9	3.2±0.4	118.1±8.4	3.3±0.4	5.0±0.4	1.3±0.2
BIQA3Q	18.6±1.8	5.0±0.7	83.6±14.0	4.6±0.9	562.7±60.0	7.1±0.9	12.3±1.1	3.6±0.5
BIQG6K	5.8±0.8	2.2±0.4	1.4±0.2	−5.4±0.8	12.7±1.3	−2.2±0.3	0.3±0.03	−5.6±0.8
SUM	-	12.6±1.8	-	2.4±1.3	-	8.2±1.0	-	−0.7±1.0
BIQTr	413.5±23.4	12.7±1.7	47.2±4.2	3.2±0.4	344.2±46.1	5.9±0.9	7.9±0.7	2.5±0.4
BIQPEP	1128.4±176.5	15.1±2.3	207.6±16.5	6.9±0.7	988.4±146.8	8.5±1.5	11.0±1.2	3.3±0.5

^*The affinities of complexes with these Ca²⁺-liganded states of CaM were determined in the presence of 250 µM free Ca²⁺ using N_xCCaM or NC_xCaM, which contain defective N-ter and C-ter EF hand pairs, as indicated by the subscript “x”.

For comparative purposes it is preferable to express differences in the affinities of reporter complexes in terms of their stabilities instead of their K_d values, which do not scale linearly with stability. We have calculated the stability differences (ΔΔG_B) between the CaM complexes with B_IQPEP or B_IQ variants and the corresponding B_IQ complexes (Table 1). A negative ΔΔG_B value means that B_IQPEP or B_IQ variant complex is more stable than the B_IQ complex, while a positive value means it is less stable.

The ΔΔG_B values for the B_IQPEP complexes with Ca²⁺-free and Ca²⁺-saturated CaM are 15.1±2.3 and 3.3±0.5 kJ/mol, and the values for the complexes with NC₂CaM and N₂CCaM are 6.9±0.7 and 8.5±1.5 kJ/mol. Although the B_IQPEP complex is less stable overall than the B_IQ complex, this difference is significantly reduced when Ca²⁺ is bound to either EF hand pair in CaM, and is reduced still further when Ca²⁺ is bound to both EF hand pairs. The two complexes therefore differ both in terms of their overall stabilities and the changes in stability produced when Ca²⁺ is bound to calmodulin.

CaM binding to B_IQ variants

As seen in Fig 1, the inserts in B_IQ and B_IQPEP differ in length and in amino sequence at several positions. For these initial investigations the functional significance of amino acid sequence differences at positions -1, 3 and 6 was examined. The Gly at position 6 in the neuromodulin IQ domain has been replaced by a Lys in the PEP19 sequence. Since a Gly is present at this position in about half of all IQ domains, we were particularly interested in the functional consequences of this replacement. The other two replacements are of interest because structural data suggest that they are likely to interact with the C-ter CaM lobe, which plays a key role in Ca²⁺-free CaM-IQ domain complexes (7–9). Three variants of B_IQ with single amino acid replacements were generated: B_IQK⁻¹A, B_IQA³Q and B_IQG⁶K, together with one containing all three replacements: B_IQTr. The complete insert sequences for these reporter constructs are listed in Fig 1. Representative data for binding of Ca²⁺-free and Ca²⁺-saturated CaM, NC₂CaM, and N₂CCaM by all of these reporters are presented in Fig 4 through Fig 7. The K_d and ΔΔG_B values derived from these and similar data are presented in Table 1.

FIG 4. Binding of Ca²⁺-free CaM to reporters containing mutant neuromodulin IQ domains altered to match the PEP19 sequence at positions -1, 3 and 6.

Data for the following reporter constructs are presented: B_IQK⁻¹A (□), B_IQA³Q (○), B_IQG⁶K (△), and B_IQTr (◇), which contains all three amino acid changes The complete amino acid sequences of the IQ domains in these reporters are listed in Fig. 1. Apparent K_d values derived from fits of a hyperbolic binding equation to these (solid lines) and similar data are listed in Table 1. Nominally Ca²⁺-free conditions were produced by addition of 3 mM BAPTA. Additional experimental details are given in “Experimental Procedures” or in the caption to Fig. 2. Fitted curves for binding of Ca²⁺-free CaM to B_IQ (-.-) and B_IQPEP (…) are presented for comparative purposes (see Fig. 2.). The fitted F_min value for B_IQTr is 0.60±0.01. This is not different from the values derived for the other reporter constructs, which range from 0.57±0.01 to 0.59±0.01.

FIG 7. The effects of Ca²⁺ binding to the N-terminal EF hand pair in CaM on its affinities for reporters containing mutant neuromodulin IQ domains altered to match the PEP19 sequence at positions -1, 3 and 6.

Data for binding of NC_xCaM to the following reporter constructs are presented: B_IQK⁻¹A (■), B_IQA³Q (●), B_IQG⁶K (▲), and B_IQTr (◆), which contains all three amino acid changes. The complete amino acid sequences of the IQ domains in these reporters are listed in Fig. 1. At the ~250 µM free Ca²⁺ concentration used for these experiments the native EF hand pair in NC_xCaM is Ca²⁺ saturated. Apparent K_d values derived from fits of a hyperbolic binding equation to these (solid lines) and similar data are listed in Table 1. Additional experimental details are given in “Experimental Procedures” or in the caption to Fig. 2. Fitted curves for binding of Ca²⁺-saturated NC_xCaM to B_IQ (-.-) and B_IQPEP (…) are presented for comparative purposes (see Fig. 2). The fitted F_min values for B_IQA³Q and B_IQTr are 0.70±0.03 and 0.70±0.04. These are not different from the values derived for B_IQK⁻¹A and B_IQG⁶K, which are 0.69±0.01 and 0.71±0.01.

The respective ΔΔG_B values for the Ca²⁺-free CaM complexes with B_IQK⁻¹A, B_IQA³Q and B_IQG⁶K are 5.4±0.9, 5.0±0.7 kJ/mol and 2.2±0.4 kJ/mol. Thus, the K⁻¹A and A³Q replacements significantly decrease the stability of the complex, while the G⁶K replacement has a relatively small effect, corresponding with only a ~2-fold increase in the K_d value (Table 1). The ΔΔG_B values for the various Ca²⁺-bound forms of the B_IQA³Q complex are similar to the value for the Ca²⁺-free form of this complex (Table 1). The effects of the A³Q replacement on stability thus appear to be Ca²⁺-independent. The ΔΔG_B values for the intermediate Ca²⁺-bound forms of the B_IQK⁻¹A complex both appear to be less than the 5.4±0.9 kJ/mol value for the Ca²⁺-free complex, but the differences are not quite statistically significant. The 1.3±0.2 kJ/mol ΔΔG_B value for the Ca²⁺-saturated form of this complex is significantly lower (Table 1). The K⁻¹A replacement therefore appears to have significant Ca²⁺-dependent and independent effects on stability. The ΔΔG_B values for the complexes between B_IQG⁶K and the Ca²⁺-bound forms of CaM all are negative, with values ranging from −5.6±0.8 to −2.2±0.3 kJ/mol (Table 1), and the effects of the G⁶K replacement on stability are primarily Ca²⁺-dependent.

The ΔΔG_B values for the CaM complex with B_IQTr, which contains all three amino replacements, are similar to the corresponding values for the B_IQPEP complex (Table 1). The only statistically significant difference is between the values for the two NC₂CaM complexes. Thus, the combination of Ca²⁺-independent and Ca²⁺-dependent changes in stability produced by the three amino acid replacements appear to largely explain the functional differences between the B_IQ and B_IQPEP complexes. The arithmetic sums of the ΔΔG_B values for the CaM complexes with B_IQ variants containing single replacements are similar to the ΔΔG_B values for the B_IQTr reporter complex (Table 1). The only statistically significant difference is between the two values for the Ca²⁺-saturated complex. This additivity suggests that there is little interaction among positions -1, 3 and 6.

Energy coupling in the CaM complexes with B_IQ and B_IQPEP

It is evident that Ca²⁺-binding has quite distinct effects on the stabilities of the B_IQ and B_IQPEP complexes. To examine these differences we have derived energy coupling (ΔΔG_C) values for the various reporter complexes from ratios of the K_d values for the different Ca²⁺-free and Ca²⁺ liganded forms of CaM, as indicated in Table 2. These values define the changes in stability produced when Ca²⁺ is initially bound to the C-ter or N-ter EF hand pair in CaM (NC₂/NC or N₂C/NC ratio), and the changes produced when the remaining EF hand pair is occupied (N₂C₂/NC₂ or N₂C₂/N₂C ratio). The overall effect of Ca²⁺ binding on stability is defined by the ΔΔG_C value derived from the N₂C₂/N₂C K_d ratio (Table 2). This value is, by definition, independent of whether the C-ter (C-ter→N-ter binding order) or N-ter (N-ter→C-ter binding order) EF hand pair is occupied first. Although ΔΔG_C values could in principle be calculated from ΔΔG_B values, calculating them from the K_d values for a single reporter complex is preferable because it reduces the number of experimentally determined values in the calculations.

Table 2. ΔΔG_C values for CaM-IQ domain complexes.

The indicated K_d ratios were calculated for each reporter construct from the K_d values in Table 1 for binding of the different Ca²⁺⁻ liganded forms of CaM (NC, NC₂, N₂C or N₂C₂). Energy coupling (ΔΔG_C) values were calculated from these K_d ratios according to the relation: ΔΔG_C = RTln(K_d ratio). A positive ΔΔG_C value means that Ca²⁺ binding to the indicated EF hand pair(s) increases the K_d for the complex; a negative value means that Ca²⁺ binding decreases the K_d for the complex.

	NC2/NC		N2C2/NC2		N2C/NC		N2C2/N2C		N2C2/NC
Construct	Kd Ratio	ΔΔGC (kJ/mol)	Kd Ratio	ΔΔGC (kJ/mol)	Kd Ratio	ΔΔGC (kJ/mol)	Kd Ratio	ΔΔGC (kJ/mol)	Kd Ratio	ΔΔGC (kJ/mol)
BIQ	5.4±0.8	4.1±0.6	0.2±0.03	−3.5±0.5	14.1±1.9	6.5±0.9	0.09±0.012	−5.8±0.8	1.3±0.2	0.7±0.1
BIQK−1A	2.2±0.4	1.9±0.3	0.1±0.01	−5.5±0.6	5.6±0.9	4.2±0.7	0.04±0.005	−7.8±0.8	0.2±0.09	−3.5±0.6
BIQA3Q	4.8±1.0	3.8±0.8	0.1±0.03	−4.8±1.1	33.3±4.9	8.6±1.4	0.02±0.003	−9.2±1.7	0.7±0.08	−0.7±0.2
BIQG6K	0.2±0.04	−3.4±0.6	0.2±0.04	−3.6±0.6	1.8±0.3	1.5±0.3	0.03±0.004	−8.5±1.2	0.06±0.008	−7.0±1.1
BIQTr	0.1±0.02	−5.3±0.7	0.2±0.02	−4.4±0.7	0.8±0.2	−0.4±0.1	0.02±0.004	−9.3±1.5	0.02±0.002	−9.7±1.2
BIQPEP	0.2±0.03	−4.2±0.5	0.1±0.02	−6.3±1.1	0.8±0.2	−0.5±0.1	0.01±0.003	−10.9±2.1	0.01±0.002	−11.4±1.8

The ΔΔG_C values derived for the B_IQ and B_IQPEP complexes are listed in Table 2. As we have previously demonstrated (12), Ca²⁺ binding only to the C-ter or N-ter EF hand pair in CaM lowers the stability of the B_IQ complex, with respective ΔΔG_C values of 4.1±0.6 and 6.5±0.9 kJ/mol (Table 2). These are balanced by the ΔΔG_C values for Ca²⁺ binding to the remaining N-ter or C-ter EF hand pair, which are respectively −3.5±0.5 and −5.8±0.8 kJ/mol. As a result, the stabilities of the Ca²⁺-free and Ca²⁺-saturated complexes are closely similar, with an overall ΔΔG_C value of 0.7±0.1 kJ/mol. As might be inferred from its ΔΔG_B values, the situation with the B_IQPEP complex is quite different. When Ca²⁺ is bound only to the C-ter or N-ter EF hand pair in CaM the stability of the complex is increased or remains approximately the same, with respective ΔΔG_C values of −4.2±0.5 and −0.5±0.1 kJ/mol. In both cases binding to the unoccupied N-ter or C-ter EF hand pair increases stability, with respective ΔΔG_C values of −6.3±1.1 or −10.9±2.1 kJ/mol. The overall ΔΔG_C value for the B_IQPEP complex is −11.4±1.8 kJ/mol, corresponding with the ~100-fold decrease in the K_d value noted earlier (Table 2). A key aspect of this behavior is its asymmetry with respect to the order of Ca²⁺-binding: in the C-ter→N-ter order the overall increase in stability is distributed over both steps, while in the in N-ter→C-ter order it occurs entirely at the second step. The ΔΔG_C values for the B_IQPEP complex are consistent with the equilibrium Ca²⁺-binding constants reported for the CaM complex with a peptide containing the PEP19 IQ domain (18).

Energy coupling in the CaM complexes with B_IQ variants

The overall ΔΔG_C value for the B_IQK⁻¹A complex is −3.5±0.6 kJ/mol, which is ~4 kJ/mol less than the overall ΔΔG_C value for the B_IQ complex and ~8 kJ/mol larger than the value for the B_IQPEP complex (Table 2). The ΔΔG_C values for the individual steps in both the C-ter→N-ter and N-ter→C-ter Ca²⁺-binding orders appear to be ~2 kJ/mol less than the values for the B_IQ complex. However, most of these differences are not quite statistically significant. It is therefore unclear how the K⁻¹A substitution affects stability at the level of the individual Ca²⁺-binding steps, although the significant decrease in the overall ΔΔG_C value is apparently not produced at a single step.

The overall ΔΔG_C value for the B_IQG⁶K complex is −7.0±1.1 kJ/mol, which is ~8 kJ/mol less than the overall value for the B_IQ complex, and ~4 kJ/mol larger than the value for the B_IQPEP complex. The steps in the C-ter→N-ter Ca²⁺-binding order have respective ΔΔG_C values of −3.4±0.6 and −3.6±0.6 kJ/mol, while the respective values for the steps in the N-ter→C-ter order are 1.8±0.3 and −8.5±1.2 kJ/mol. This asymmetric pattern of Ca²⁺-dependent stability changes closely resembles the one exhibited by the B_IQPEP complex (Table 2).

The overall ΔΔG_C value for B_IQA³Q complex is identical to the value for the B_IQ complex (Table 2), and ΔΔG_C values for the individual steps in the C-ter→N-ter and N-ter→C-ter Ca²⁺-binding orders do not differ to a statistically significant degree from the values for the B_IQ complex (Table 2). Thus, as suggested by the ΔΔG_B values for this complex, the ~5 kJ/mol decrease in stability produced by the A³Q replacement appears to be Ca²⁺-independent.

The ΔΔG_C values for the B_IQTr and B_IQPEP complexes appear to be similar, and none of the apparent differences between them are statistically significant (Table 2). Since the effect of the A³Q replacement on stability is Ca²⁺-independent, the sum of the Ca²⁺-dependent effects of the K⁻¹A and G⁶K replacements must account for the distinctive patterns of Ca²⁺-dependent stability changes exhibited by the B_IQ and B_IQPEP complexes. Indeed, the G⁶K replacement alone appears to account for most of the Ca²⁺-dependent differences between the two reporter complexes.

Discussion

As reported previously and confirmed here, the stability of the B_IQ reporter complex decreases when Ca²⁺ is bound to a single EF hand pair, but Ca²⁺ binding has little or no overall effect on stability (12). This produces a biphasic change in the stability of the complex as the free Ca²⁺-concentration is increased (12). These characteristics are shared by the CaM complex with full-length neuromodulin (12). In contrast, when Ca²⁺ is bound to a single EF hand pair in the B_IQPEP reporter complex it either increases stability or has little effect, depending upon whether the C-ter or N-ter EF hand pair is occupied, and occupancy of both EF hand pairs substantially increases the stability of this complex, producing a ~100-fold decrease in the K_d value. Our results indicate that a G⁶K replacement largely accounts for the distinctive patterns of intermediate and overall Ca²⁺-dependent stability changes exhibited by the B_IQ and B_IQPEP complexes. The K⁻¹A replacement accounts for the remaining Ca²⁺-dependent differences. This and the A³Q replacement account for most of the Ca²⁺-independent differences between the two complexes. An interesting aspect of these results is that under equilibrium conditions, unless there is a Gly at position 6, the distinction between IQ domains and other types of CaM-binding domains does not appear to be clear-cut.

A Gly residue is found at position 6 in about half all known IQ domain sequences (Fig. 8), so our results suggest that the presence or absence of this residue defines two major functional classes of IQ domains. Consistent with this hypothesis, a variety of published data demonstrate that the identity of the amino acid at this position significantly affects the structure of a CaM-IQ domain complex.

FIG 8. Logo representation of the aligned sequences of 3000 IQ domain regions.

The sequence is numbered in relation to the first residue (0) in the 11 residue IQ motif, which has been underlined. Each position is represented by a stack of one-letter amino acid symbols. The total height of a stack is an indication of how well conserved that position is; the height of each amino acid symbol in a stack is an indication of its relative frequency. The arrow indicates a semi-conserved Gly residue that appears to be a major determinant of the pattern of Ca²⁺-dependent stability changes exhibited by a CaM-IQ domain complex.

Two major structural classes of CaM-IQ domain complexes

Structural and biophysical studies indicate that myosin light chain subunits and Ca²⁺-free CaM can form two different types of complexes with IQ domains. One is compact, with the N-ter lobe interacting closely with the IQ domain, and the other extended, with the N-ter lobe largely free of the IQ domain (7, 8). Although crystal structures supporting an extended conformation have been determined only for light chain subunits bound to myosin IQ domains, fluorescence energy transfer measurements performed with labeled CaM and peptides indicate that CaM forms similar complexes with these domains (7, 8). Crystal structures supporting a compact conformation have been determined for both CaM and light chain subunit complexes with myosin IQ domains containing the semi-conserved Gly (7–9). Structural and biophysical investigations of native and specifically altered IQ domains indicate that for the compact structure to be formed Gly or another small amino acid (Ala, Ser, Thr, Val) must be present at position 6 in the IQ domain, and a basic residue must also be present at position 10, which is well-conserved (see Fig. 8). The presence of a bulky amino acid (Lys, Arg, Met) at position 6 appears to result in the extended conformation (7, 8). Structural studies suggest that the N-ter CaM lobe interacts extensively with the IQ domain in Ca²⁺-saturated complexes, regardless of the identity of the residue at position 6, and this is presumably also the case for the N-ter lobes in the Ca²⁺-saturated B_IQ and B_IQPEP complexes (10, 11). In both Ca²⁺-free and Ca²⁺-saturated CaM-IQ domain complexes the C-ter CaM lobe appears to interact in the same orientation with the N-ter half of the IQ domain (7, 8, 10, 11). This is consistent with the significant Ca²⁺-independent effects on stability produced by the K⁻¹A and A³Q replacements, which are in this region. The semi-conserved nature of the Gly at position 6 (Fig. 8) suggests that its role goes beyond allowing the compact Ca²⁺-free CaM-IQ domain complex, since this does not appear to specifically require a Gly. We are currently investigating additional B_IQ variants in order to determine the specific functional effects of different amino acid replacements at this position.

Relating the functional and probable structural effects of the G⁶K substitution

Based on the above dissussion, the Ca²⁺-free CaM complex with the B_IQ is likely to be in the compact conformation, while the complexes with B_IQPEP and B_IQG⁶K are likely to be in the extended conformation. Although the ~2 kJ/mol drop in stability associated with the G⁶K substitution is consistent with loss of interactions with the Ca²⁺-free N-ter lobe, this relatively small effect may be an underestimate of the contribution of these interactions to stability. This is because replacement of the Gly might be also expected to be stabilizing, given an α-helical conformation for the bound IQ domain (7–9).

When Ca²⁺ is bound only to the N-ter EF hand pair in CaM it has a minimal effect on the stabilities of the B_IQPEP and B_IQG⁶K complexes, suggesting that the the N-ter lobe is not only extended in these complexes in the absence of Ca²⁺, but remains so in this Ca²⁺-bound state. In contrast, Ca²⁺-binding only to the N-ter EF hand pair in the B_IQ complex produces a ~7 kJ/mol decrease in stability. This may reflect a loss of interactions with the N-ter lobe when Ca²⁺ is bound and/or formation of an antagonistic interaction between the Ca²⁺-bound N-ter lobe and the IQ domain.

When Ca²⁺ is bound only to the C-ter EF hand pair in CaM it decreases the stability of the B_IQ complex by ~4 kJ/mol, and has the opposite effect on the B_IQPEP and B_IQG⁶K complexes (Table 2). The essentially identical increases in the stabilities of the all the complexes examined produced when Ca²⁺ is subsequently bound to the N-ter EF hand pair suggests that the N-ter CaM lobe is in a similar state in the intermediate Ca²⁺-bound complexes (Table 2).

Our results demonstrate that the presence or absence of a Gly at position 6 defines two classes of IQ domains whose complexes with CaM exhibit distinct patterns of Ca²⁺-dependent stability changes. This is likely to at least in part be a reflection of differences in interactions with the N-ter CaM lobe in these two classes. Additional structural and biophysical investigations are needed to confirm and extend this hypothesis. Structures of the intermediate Ca²⁺-bound states of CaM-IQ domain complexes are particular important, as major differences between the stabilities these states appears to be a key feature of the two classes.

FIG 5. Binding of Ca²⁺-saturated CaM to reporters containing mutant neuromodulin IQ domains altered to match the PEP19 sequence at positions -1, 3 and 6.

Data for the following reporter constructs are presented: B_IQK⁻¹A (■), B_IQA³Q (●), B_IQG⁶K (▲), and B_IQTr (◆), which contains all three amino acid changes. The complete amino acid sequences of the IQ domains in these reporters are listed in Fig. 1. Apparent K_d values derived from fits of a hyperbolic binding equation to these (solid lines) and similar data are listed in Table 1. Ca²⁺ saturation was effected by addition of ~250 µM free Ca²⁺. Additional experimental details are given in “Experimental Procedures” or in the caption to Fig. 2. Fitted curves for binding of Ca²⁺-saturated CaM to B_IQ (-.-) and B_IQPEP (…) are presented for comparative purposes (see Fig. 2).

FIG 6. The effects of Ca²⁺ binding to the C-terminal EF hand pair in CaM on its affinities for reporters containing mutant neuromodulin IQ domains altered to match the PEP19 sequence at positions -1, 3 and 6.

Data for binding of N_xCCaM to the following reporter constructs are presented: B_IQK⁻¹A (■), B_IQA³Q (●), B_IQG⁶K (▲), and B_IQTr (◆), which contains all three amino acid changes. The complete amino acid sequences of the IQ domains in these reporters are listed in Fig. 1. At the ~250 µM free Ca²⁺ concentration used for these experiments the native EF hand pair in N_xCCaM is Ca²⁺ saturated. Apparent K_d values derived from fits of a hyperbolic binding equation to these (solid lines) and similar data are listed in Table 1. Additional experimental details are given in “Experimental Procedures” or in the caption to Fig. 2. Fitted curves for binding of Ca²⁺-saturated N_xCCaM to B_IQ (-.-) and B_IQPEP (…) are presented for comparative purposes (see Fig. 2).