An unconventional calmodulin-anchoring site within the AB module of Kv7.2 channels

Kv7.2 (also known as KCNQ2) is the main channel mediating the non-inactivating K⁺ M-current, which opens at subthreshold membrane potentials, providing a brake for membrane excitation (Soldovieri et al., 2011). Kv7 channels have the canonical six-transmembrane architecture found throughout the voltage ion channel family. The cytoplasmic C-terminal tail is composed of four conserved elements, believed to adopt an α-helix configuration, known as helices A–D (Yus-Nájera et al., 2002). This region forms an important multifunctional hub that is central for channel assembly, gating and the formation of complexes with regulatory factors including calmodulin (CaM) (Haitin and Attali, 2008; Yus-Nájera et al., 2002). CaM binds to the module containing helices A and B and regulates M-channel plasma membrane density, and sensitivity to phosphatidylinositol 4,5-bisphosphate (PIP₂) and Ca²⁺ (Etxeberria et al., 2004,, 2008; Etzioni et al., 2011; Kang et al., 2014; Kosenko and Hoshi, 2013; Gamper and Shapiro, 2003; Xu et al., 2007; Yus-Nájera et al., 2002). This small acidic protein is formed from two highly homologous lobes joined by a flexible linker, referred to as the N- and C-lobe. Each lobe engages with a different hydrophobic residue. Frequently, there is a secondary hydrophobic anchoring residue separated from the primary one by an α-helical turn. In many targets, the N- and C-target recognition sites (TRSs) overlap, but in others, they do not. For instance, in SK2 K⁺ channels (KCa2.2, also known as KCNN2), the N- and C-TRSs are located in different subunits of the tetrameric protein (Schumacher et al., 2001), whereas in Ca_V1 Ca²⁺ channels the second TRS is located in a distant part of the same polypeptide (Dick et al., 2008).

The transduction mechanism for Ca_V1 Ca²⁺ and SK2 K⁺ channels involves two steps. The first step is docking of the C-lobe to the C-TRS under resting intracellular Ca²⁺ levels (Ben Johny et al., 2013; Dick et al., 2008; Schumacher et al., 2001; Zhang et al., 2013). Docking to the C-lobe places the N-lobe closer to the N-TRS and, therefore, increases the effective N-lobe local concentration by orders of magnitude. The second step is the engagement of the N-lobe with its correspondent N-TRS. After completion of the second step, which requires Ca²⁺ binding to the N-lobe, Ca²⁺ channels inactivate or SK2 K⁺ channels open the pore. Quantitative binding data support a similar model for Kv7.2 channels in which the initial docking step takes place between the N-lobe and helix B under low or resting Ca²⁺ levels (Alaimo et al., 2014). The current view is that CaM engagement with helix A is essential for Kv7.2 channel function, and that anchoring helix B to the N-lobe facilitates C-lobe binding to helix A, which is located distantly from helix B in the same or in another subunit of the tetrameric channel (Alaimo et al., 2009,, 2013a,, 2014).

Several studies lend support to the hypothesis that CaM binding to Kv7.2 is essential for channel function, yet a mutation in helix B that prevents CaM binding, S511D, yields functional channels in cells with elevated CaM expression (Alaimo et al., 2013a; Etxeberria et al., 2008; Gómez-Posada et al., 2011). This suggests that there is an alternative site for N-lobe engagement. Identification of such a binding site is crucial to understanding how CaM regulates Kv7 channels. Despite a number of biochemical and structural studies (Aivar et al., 2012; Alaimo et al., 2013a,, 2014; Etzioni et al., 2011; Gómez-Posada et al., 2011; Sachyani et al., 2014; Xu et al., 2013), the TRSs that control Kv7.2 function are uncertain. Close inspection of the sequence does not reveal a potential site upstream or downstream of the IQ motif. If the Kv7.2 segment A sequence adopts an α-helical configuration, as seen in the CaM–Kv7.1 complex (Sachyani et al., 2014), the first bulky hydrophobic residue downstream of the IQ motif (L351) will not be positioned properly to be anchored by the N-lobe. Thus, it is unlikely that the N-lobe adopts a configuration resembling that seen in complexes with different myosins or Na_V channels that harbor similar IQ motifs (see Fig. 1). The N-lobe is engaged upstream the IQ motif in Ca_V1 channels, but this kind of arrangement seems unlikely for Kv7.2 because the equivalent position is occupied by an asparagine, a residue that is not found occupying the N-lobe anchoring pocket of any CaM–protein complex solved so far (Villarroel et al., 2014).

Here, we employed flow cytometry, in vitro binding and electrophysiology for a systematic delineation of the post-helix A CaM-binding region to unravel the requirements for CaM-dependent Kv7.2 function. We show that this region contains a TW CaM-binding motif reminiscent of the TRC for the C-lobe found in SK2 K⁺ channels (Schumacher et al., 2001; see Fig. 1). This site is not essential for Kv7.2 function, but appears to assist in the recognition and/or transduction of the CaM signal. In agreement with structural information (Sachyani et al., 2014), the data is consistent with the idea that the dispensable TW site acts as a staple that bridges the N-Lobe–helix B and C-lobe–helix A, a function that becomes essential when either the helix A or helix B interaction with CaM is weakened.

We have previously shown that when the AB module from Kv7.2 channels is transferred to the Tac membrane protein, trafficking to the plasma membrane becomes CaM-dependent. This assay was employed to assist in the identification of sites involved in CaM-dependent Kv7.2 regulation (Etxeberria et al., 2008; Gómez-Posada et al., 2011). The flow cytometry surface expression histogram from a population of cells expressing Tac protein displayed a well-defined peak to the right, which corresponds to cells expressing high levels of the protein at the plasma membrane. No significant effect on the surface expression flow cytometry profile was revealed after removal of the A–B linker from Y372 to K493 (data not shown). In contrast, the histogram changed dramatically when the region from T359 to T501 of the A–B linker was removed (ΔT359–T501), suggesting the existence of crucial CaM TRSs within this region (Fig. 2). In addition, the surface expression histogram from cells expressing ΔT359–T501 was indistinguishable from those from cells expressing ΔT359–Y372 (n=3, data not shown); therefore, we focused on the N-terminal boundary between T359 and Y372.

Segment A was extended by nine, six or three residues, generating deletions starting at T368, R365 and Y362, and all ending at T501. Fig. 2 shows that the surface expression histogram profile for the construct resulting of just adding three residues (T359, W360 and Q361) presented a well-defined peak, similar to that from cells expressing the reporter carrying the complete wild-type (WT) CaM-binding domain (denoted Q2AB). Therefore, the influence of region between T359 and Y362 was studied by extending the A region by one and two residues. Whereas the area under the peak was not altered significantly, the peaks of the surface expression histograms were progressively taller and narrower as more residues were incorporated, and no further difference was evident when the sequence was extended six or nine residues (Fig. 2). These results reveal the existence of an important site for CaM-dependent Kv7.2 regulation that roughly extends from T359 up to Y362.

The impact of altering this post-IQ region was assessed in vitro using dansylated CaM (D-CaM) and purified monodisperse recombinant proteins (Alaimo et al., 2013b), both in the absence and presence of a physiologically relevant Ca²⁺ concentration. The ionic conditions (Mg²⁺, ionic strength and pH) were based on the intracellular solution used for whole-cell recording, in an effort to obtain binding parameters under near-physiological ionic conditions. This test reveals binding to Ca²⁺ as a blue-shift on the peak emission, which represents a convenient internal control to confirm the absence of Ca²⁺ contamination (Alaimo et al., 2013b). The results were notable (Fig. 3), removal of the T359–T501 linker resulted in a negligible increase in D-CaM fluorescence. Maximal fluorescence was restored both in the absence and in the presence of Ca²⁺ by incorporating W360. Further extension of the N-terminal sequence had relatively minor consequences in maximal fluorescence emission. The impact on apparent affinity was small in both the presence and absence of Ca²⁺, and the binding parameters tended to improve as more residues were incorporated into the protein (Fig. 3).

This assay reports CaM conformational changes upon binding to targets, Ca²⁺ or both, as an increase in fluorescence emission (Alaimo et al., 2009). CaM adopts at least three configurations that are in equilibrium (Fig. 4). In this model, the transition rate k₁ from free-CaM to CaM–Q2AB depends on the concentration of Q2AB, whereas the complex CaM–Q2AB undergoes an intramolecular rearrangement to adopt a configuration, CaM*–Q2AB, that displays augmented fluorescence emission. In this simple model, the intramolecular transition k₂ from CaM–Q2AB to CaM*–Q2AB does not depend on the concentration of the Q2AB protein.

To further delineate the post-IQ region and to collect data to decipher the relationship (Eqn 1) between maximal fluorescence and the apparent binding affinity, a series of single or double point mutations were introduced in Q2AB, and recombinant monodisperse proteins were purified and tested for binding to D-CaM (Fig. 3B). If a mutation affects binding (i.e. k₁ or b₁), the dose-dependent relationship should be displaced and reach the same maximal effect, resulting in a relationship as depicted using dashed lines in Fig. 3C. Fig. 3C shows the result of fitting Eqn 1 to the relationship between maximal fluorescence increase and affiliated apparent affinity, which resulted in an extrapolated binding dissociation constant for the first transition of 226±35 nM and 292±34 nM (mean±s.e.m.) in the absence or the presence of Ca²⁺. Within the frame of a three-state model, these data are consistent with the proposal that the mutations at the post-IQ region affect the second transition to the activated state rather than binding itself (the first transition).

The single or double point mutations were introduced into the full-length Kv7.2 channel for functional evaluation in cells recorded under whole-cell conditions. As shown in Fig. 5, none of the mutations had a significant impact on current density. We returned to the Tac reporter for further evaluation of the functional consequences of alterations at the TW site. Tac is a monomeric protein that undergoes multiple glycosylation events, resulting in the appearance of two bands in western blots due to differential glycosylation occurring at the endoplasmic reticulum (ER) or at the Golgi complex (faster and slower migrating bands, respectively). The ratio between the intensity of both bands results in an index for trafficking from the ER to the Golgi complex (Etxeberria et al., 2008). The samples from cells expressing WT Q2AB or a construct carrying the T359–T501 deletion, which eliminates the TW site, showed similar glycosylation ratios. In addition, overexpressing CaM led to the appearance of the mature band in similar proportion for both proteins (Fig. 6). For further assessment of the role of the TW site, the consequences of overexpressing CaM on the function of the channels were evaluated under whole-cell recording conditions. The current density in cells expressing either the full-length or a mutant channel carrying an AB linker deletion ΔT359–T501 were indistinguishable. Furthermore, current density increased to similar levels for both channels upon CaM overexpression (Fig. 6). The voltage-dependence was not altered because of the linker deletion or elevation of CaM levels (Fig. 6E). In contrast, the current kinetics were faster for the ΔT359–T501 channel without the link (Fig. 6F), which is in line with the behavior of C-terminal alternative splicing variants (Pan et al., 2001). However, this influence was not evident in a subunit carrying the more limited deletion ΔT359-Y372 (see Fig. 7) or in the mutant channels described on Fig. 5 (data not shown). Thus, the TW site is not essential for CaM-mediated regulation of Kv7.2 channels.

CaM binds with higher affinity to helix B than to helix A (Alaimo et al., 2013a), and helix B displays a marked preference for the N-lobe (Alaimo et al., 2014). Binding to helix B in Kv7.2 channels is dispensable, as suggested by the observation that when CaM binding to the helix B becomes undetectable by introducing the S511D mutation, function can be recovered by overexpressing CaM (Alaimo et al., 2013a). Where does CaM N-lobe bind when helix B is not available? Could it be that binding to the TW site sustains the functional rescue of the S511D mutant? We explored this idea by introducing the S511D mutation in combination with the ΔT359–T501 deletion. Consistent with the alternate TW–helix-B model, the resulting channels were non-functional, even after CaM overexpression (n=15, data not shown). As a further test, the activity of a channel carrying a more limited deletion, expanding from T359 to Y372, was evaluated. Similar to the longer deletion, this mutant channel was functional and sensitive to CaM, but when combined with the S511D mutation, creating a double mutant with both the TW and helix B sites disabled, function was lost irrespective of CaM overexpression (Fig. 7).

We addressed the possibility that the TW site can assist in CaM-dependent function when the engagement to helix A is weakened. Given that binding to helix A is essential for Kv7.2 function (Alaimo et al., 2013a), we tested the impact on a channel carrying the I340A mutation that partially interferes with CaM binding to helix A (Etxeberria et al., 2008). Similar to channels with the helix B S511D mutation, the function of channels carrying the helix A I340A mutation can be rescued by overexpressing CaM (Alaimo et al., 2013a). However, the function of doubly mutated channels, without the TW site and carrying the helix A I340A mutation, were not functional even after co-expressing CaM (Fig. 7).

The data of this report reveal the existence of a site located in the inter-helical region between segments A and B of the Kv7.2 terminus that influences CaM-dependent regulation. Analysis of the relationship between the maximal D-CaM fluorescence and apparent binding affinity suggests that mutations at this site affect intramolecular rearrangements of the CaM–Kv7.2 complex. Functional examination by whole-cell recording reveals that this site is not essential for function by itself. Rather, this site assists in maintaining the stability of an active CaM–Kv7.2 configuration, and becomes crucial when the interactions of CaM with helix B or helix A are impaired.

Mutations at this new site influenced trafficking of a Tac–Kv7.2 chimera. The reference flow cytometry histograms presented a sharp peak that reflects the existence of a homogenous population of cells with similar levels of the Tac reporter carrying the intact CaM-binding site at the plasma membrane, but with the surface expression was spread over a wide range when the TW site was removed. However, the area under the peak remained unaltered and the global extent of glycosylation at the Golgi and at the ER, which is an indication of trafficking, was not affected by these modifications. The simplest interpretation is that removal of the site leads to increased heterogeneity, with more cells with lower and higher surface expression than in the control group, such that similar levels of surface expression are achieved on average. Nonetheless, this increase in heterogeneity for Tac–AB trafficking was not reflected in a variability of current density for the different mutants analyzed by whole-cell recording. Keeping in mind that Tac is a monomeric protein and Kv7.2 channels are tetramers, the possibility that the assembly of four subunits tends to increase the homogeneity of the population of Kv7.2 channels should be considered. By contrast, CaM-dependent potentiation of Kv7.2 function occurred both in the presence and absence of the TW motif, suggesting that the channels retain the ability to interact with CaM regardless of the absence of the TW site. This seems to contradict the binding data, but, as discussed below, lack of signal does not necessarily means absence of interaction (see Alaimo et al., 2013a).

The impact of different mutations on the dose-dependent fluorescence enhancement of D-CaM is consistent with the idea that D-CaM reports conformational changes that take place after docking, and it is in line with the concept that the mutations at the TW site do not primarily affect binding itself. The hyperbolic relationship between apparent binding affinity and maximal effect is an indication that mutations at the TW site are disturbing mainly a downstream transition. According to theory, this relationship extrapolates to the ‘true’ affinity, yielding values of 226 and 292 nM in the absence and presence of Ca²⁺, respectively. These ‘true’ affinities are weaker than those for the reference AB protein. Thus, the initial binding step is relatively feeble, but the subsequent conformational changes stabilize the complex, resulting in one order of magnitude more favorable apparent affinities (11 and 27 nM in the absence and presence of Ca²⁺, respectively). In comparative terms, the difference in affinities in presence and absence of Ca²⁺ are much larger when, according to the three-state model, the second transition can take place. Within the framework of the three-state model, the data indicates that Ca²⁺ is preferentially affecting the second transition, and the true binding affinities derived from this analysis are of similar magnitude. This suggests that the structural changes that CaM undergoes when loaded with Ca²⁺ have relatively minor consequences for the initial docking steps during the Kv7.2 target recognition process.

Our study suggests that the S511D mutant, which is nonfunctional in HEK293T cells, becomes functional upon overexpression of CaM because of the existence of the TW site, as upon deletion of the site the functional rescue by CaM is eliminated (Fig. 7A). Taking into account that the corresponding helix B mutation in Kv7.3 channels (A518D) renders non-functional channels (Gómez-Posada et al., 2011), it might seem illogical that in the absence of any binding to helix B there is functional rescue by overexpressed CaM. We cannot discard the possibility that CaM binding to S511D escapes detection even in the sensitive competition assay that we have previously performed (Alaimo et al., 2013a). Alternatively, the interaction of CaM with the TW site of Kv7.3 might be weaker than for Kv7.2 subunits and unable to compensate for the missing interaction with helix B in the A518D mutant.

The TW site presents a remarkable similarity with the C-lobe TRS of SK2 channels (Fig. 1). In the SK2–CaM complex the C-lobe is found embracing the TW motif, in a disposition very similar to that seen in other complexes with an IQ motif (Villarroel et al., 2014). Interestingly, the structure of the SK2 CaM-binding domain on its own in solution (without CaM) contains a distorted helical region that includes the core sequence TWLIY, whereas the rest of the molecule lacks stable overall folding (Wissmann et al., 2002). This TW region is crucial for the interaction in the absence of Ca²⁺, and it is very plausible that it is the site for initial docking during the process of SK2 target recognition. Based on this similarity, it is tempting to propose that the Kv7.2 TW site also adopts an α-helical configuration spontaneously, even in the absence of CaM. However, given that the TW region of Kv7.2 channels is dispensable, and considering the arguments regarding binding to D-CaM presented above, it seems unlikely that it plays an important role in the initial steps of Kv7.2 target recognition. Furthermore, the recently solved atomic structure of a dimeric Kv7.1 complex shows that the TW helix is not located into the grooves formed by the CaM lobes. Instead, it is simultaneously touching the tips of both lobes. As shown in Fig. 8, the Kv7.1 TW site adopts an α-helical configuration, but in contrast to that of SK2, the TW helix is not encapsulated by the CaM C-lobe (nor by the N-lobe). The Kv7.1–CaM structure presents a stereotypical IQ-motif–apoCaM engagement with the C-lobe, whereas the N-lobe is found anchoring helix B (Sachyani et al., 2014). The atomic details of helix B–CaM binding for the Kv7.1 and Kv7.4 complexes show that the positions equivalent to Kv7.2 S511 in Kv7.1 and Kv7.4 are the main anchoring site for engagement with the N-lobe, providing a straightforward explanation for the disrupting effects of the S511D mutation on CaM binding to Kv7.2 channels (Sachyani et al., 2014; Xu et al., 2013). Besides contacting both CaM lobes, the TW site creates a contact network with helix B (Fig. 8B). The structural details suggest that, after the initial CaM docking, there is a critical reorganization, during which the TW site helps pull helix B towards helix A. This hypothetical scenario is consistent with the lack of function of the double mutants analyzed in this study.

Deletions, point mutations, and epitope insertions in the KCNQ subunits were constructed by PCR-based mutagenesis in the human KCNQ2 GenBank code, from PubMed Nucleotide (Y15065) sequence. The cDNA encoding rat CaM was provided by the group of John P. Adelman (Vollum Institute, Oregon Health Sciences University, Portland, OR, USA). The Tac–CFP construct was generated using the Tac receptor provided by Steve Standley (Laboratory of Neurochemistry, NIDCD, NIH, Bethesda, MD), which was cloned into a modified version of the expression vector pEGFP (Clontech), where eGFP has been exchanged with mCFP. Single mutations were introduced by using QuikChange mutagenesis.

HEK293T cells were maintained in Dulbecco's modified Eagle's medium (DMEM, Sigma-Aldrich) supplemented with non-essential amino acids (Sigma) and 10% fetal bovine serum (FBS) at 37°C in 5% CO₂. HEK293T cells grown in T25 Flasks were transiently transfected with Tac-Kv7.2-mCFP by the calcium phosphate method. After 36 h, cells were detached and disaggregated using a buffered solution of trypsin-EDTA and dissociation buffer (1:2, Gibco), transferred to 1.5 ml Eppendorf tubes, washed and resuspended in 700 ml PBS. Cells were fixed for 20 min at room temperature with 3% paraformaldehyde (Fluka) and washed with PBS. After pre-incubation with 5% BSA (Sigma) for 30 min, cells were labeled for 1 h at room temperature with an antibody recognizing a Tac extracellular epitope, washed three times with PBS, blocked for 15 min with 5% BSA and incubated for 1 h at room temperature with an Alexa-Fluor-488-conjugated anti-mouse-IgG secondary antibody. Cells were washed four times and resuspended in PBS before analysis on a Gallios flow cytometer (775014, Beckman Coulter). Data were collected from at least 10,000 healthy cells with emission intensities above the background level determined using untransfected HEK293T cells. Histograms normalized to 10,000 events were visualized using WinMDI 2.9 software, analyzed in Excel and represented using SigmaPlot.

HEK293T cells were transfected by the calcium phosphate precipitation method with 1 µg of DNA per T25 flask, and analyzed after 36 h. Cells were removed from the flasks, pelleted by centrifugation, lysed using 15 µl SDS-PAGE sample buffer and boiled before SDS-PAGE analysis. Proteins were resolved on a 10% SDS-PAGE gel and transferred onto polyvinylidene difluoride membranes (PVDF) Inmobilon-P (EMD Millipore Corp.). PVDF membranes were blocked with 5% (w/v) non-fat dried milk powder in PBS containing 0.05% Tween 20 for 1 h at room temperature, incubated for 1 h at room temperature with an anti-GFP primary antibody diluted 1:1000 that was visualized using a secondary anti-mouse-IgG antibody conjugated to peroxidase diluted 1:5000 in PBS containing 0.05% Tween 20.

The following primary monoclonal antibodies were used: mouse anti-CaM (1:2000; 05-173; Millipore); mouse anti-GFP (1:2000; 1814460; clones 7.1 and 13.1; Roche Applied Science) and mouse anti-Tac (1:1000; BD Bioscience) antibodies. The secondary antibodies used were a peroxidase-coupled anti-mouse-IgG (1:5000; 1706516, Bio-Rad Laboratories), and a fluorescent secondary goat anti-mouse-IgG conjugated to Alexa Fluor 488 (1:1000; A11001, Invitrogen).

Proteins were visualized using SuperSignal West Pico Chemiluminescent Substrate (34078, Pierce) and SuperSignal ELISA Femto (37075, Pierce). At least 10 cumulative images (30 s exposition) were acquired using the Versadoc Imaging System (Bio-Rad Laboratories). Protein bands were analyzed with ImageJ software v. 1.45.

Whole-cell patch recordings of HEK293T cells were obtained at room temperature (21–25°C) 48 h after transfection using Lipofectamine 2000 (Invitrogen). Cells were bath-perfused with the following solution (in mM): 140 NaCl, 4 KCl, 2 CaCl₂, 2 MgCl₂, 10 HEPES (Na), 5 D-glucose, adjusted to pH 7.4 with NaOH. The osmolarity was adjusted with mannitol to ∼315 mOsm. Pipettes were pulled from borosilicate glass capillaries (Sutter Instruments) using a Narishige micro-pipette puller (PC-10; Narishige Instrument Co.). Membrane currents were measured using a VE-2 Alembic amplifier (Alembic Instruments) with pipette and membrane capacitance cancellation and >95% series resistance compensation. Pipettes were filled with an internal solution containing (in mM): 125 KCl, 10 HEPES (K), 5 MgCl₂, 5 EGTA, 5 Na₂ATP, adjusted to pH 7.2 with KOH and the osmolarity adjusted to ∼300 mOsm with mannitol. The pipette resistance was typically 2–3 MΩ. The amplitude of the Kv7 current was defined as the peak difference in current relaxation measured at −30 mV after 1500 ms pulses to −120 mV (all channels closed) and to +50 mV (all channels opened). Current–voltage relationships were fitted using the Boltzmann equation I=I_max/(1+e^{((V−V1/2)/S)}). To measure channel kinetics, current relaxations were fitted with an exponential function of the form: y=A×e^(−t/τ), where t is time, A is the amplitude and τ is the time constant.

Data were acquired and analyzed using pCLAMP software (version 8.2), normalized in Excel (Microsoft) and plotted using SigmaPlot (SPSS Corp.). Data are shown as the mean±s.e.m. (n indicates the number of samples). Student's t-test was used to compare means of two groups from data with a normal distribution. Statistical analysis was performed using SigmaPlot software. *P>0.05, **P<0.001, and ***P<0.001.

Recombinant proteins encoding the Kv7.2 CaM-binding domain fused to GST (GST-tagged Q2AB WT, mutants and deletions), were produced in E. coli strain BL21-DE3 and obtained principally as inclusion bodies. Refolding and purification of these proteins, and purification of rat brain CaM was performed as described (Alaimo et al., 2009). Fluorescent dansylated CaM [D-CaM, 5-(dimethylamino) naphtalene-1-sulfonyl-calmodulin] was prepared using recombinant CaM and dansyl chloride as described previously (Alaimo et al., 2013a). Dispersion of the samples was evaluated by dynamic light scattering (DLS) using a Zetasizer Nano instrument (Malvern Instruments) in order to exclude the presence of aggregates.

Prior to data acquisition, D-CaM and other proteins were dialyzed against 2 l of fluorescence buffer containing 25 mM Tris-HCl pH 7.4, 120 mM KCl, 5 mM NaCl, 2 mM MgCl₂, EGTA 10 mM for 48 h, changing the buffer every 12 h. Steady-state fluorescence measurements were performed at room temperature with an Aminco Bowman series 2 (SLM Aminco) fluorescence spectrophotometer in a final volume of 100 μl using a quartz cuvette at room temperature. The excitation wavelength was 340 nm and emissions were recorded from 400 to 660 nm. Slit widths were set at 4 nm for excitation and 4 nm for emission. Titration experiments were performed recording emission spectra at 1 min after adding increasing concentrations of each fusion protein to D-CaM (12.5 nM) in fluorescence buffer. Experiments were also performed in the presence of an excess of free Ca²⁺ (3.9 µM), adding 9.63 mM Ca²⁺. Free Ca²⁺ concentration was determined using Fura-2 (Invitrogen) following the manufacturer's instructions.

Fluorescence enhancement was plotted against the protein concentration to generate the concentration–response curves. The parameters of the Hill equation (fluorescence increase=A×[Ligand]^h/(EC₅₀^h+[Ligand]^h); where A is the maximal fluorescence increase and h is the Hill coefficient) were fitted to the data by curvilinear regression. Data are shown as average of three or more independent experiments.