Long QT syndrome‐associated calmodulin variants disrupt the activity of the slowly activating delayed rectifier potassium channel

Abstract Calmodulin (CaM) is a highly conserved mediator of calcium (Ca2+)‐dependent signalling and modulates various cardiac ion channels. Genotyping has revealed several CaM mutations associated with long QT syndrome (LQTS). LQTS patients display prolonged ventricular recovery times (QT interval), increasing their risk of incurring life‐threatening arrhythmic events. Loss‐of‐function mutations to Kv7.1 (which drives the slow delayed rectifier potassium current, IKs, a key ventricular repolarising current) are the largest contributor to congenital LQTS (>50% of cases). CaM modulates Kv7.1 to produce a Ca2+‐sensitive IKs, but little is known about the consequences of LQTS‐associated CaM mutations on Kv7.1 function. Here, we present novel data characterising the biophysical and modulatory properties of three LQTS‐associated CaM variants (D95V, N97I and D131H). We showed that mutations induced structural alterations in CaM and reduced affinity for Kv7.1, when compared with wild‐type (WT). Using HEK293T cells expressing Kv7.1 channel subunits (KCNQ1/KCNE1) and patch‐clamp electrophysiology, we demonstrated that LQTS‐associated CaM variants reduced current density at systolic Ca2+ concentrations (1 μm), revealing a direct QT‐prolonging modulatory effect. Our data highlight for the first time that LQTS‐associated perturbations to CaM's structure impede complex formation with Kv7.1 and subsequently result in reduced IKs. This provides a novel mechanistic insight into how the perturbed structure–function relationship of CaM variants contributes to the LQTS phenotype. Key points Calmodulin (CaM) is a ubiquitous, highly conserved calcium (Ca2+) sensor playing a key role in cardiac muscle contraction. Genotyping has revealed several CaM mutations associated with long QT syndrome (LQTS), a life‐threatening cardiac arrhythmia syndrome. LQTS‐associated CaM variants (D95V, N97I and D131H) induced structural alterations, altered binding to Kv7.1 and reduced IKs. Our data provide a novel mechanistic insight into how the perturbed structure–function relationship of CaM variants contributes to the LQTS phenotype.

CaM is a highly conserved, ubiquitous Ca 2+ -sensing protein which serves as a central mediator of Ca 2+ -dependent signalling. It is encoded by three independent genes (CALM1, CALM2 and CALM3) which all translate an identical 148 amino acid long protein (Fischer et al., 1988;Friedberg & Rhoads, 2001a, b;Halling et al., 2016). CaM contains four EF-hand motifs which can bind up to four Ca 2+ ions, EF-hands 1 and 2 are found in the N-lobe while EF-hands 3 and 4 are in the C-lobe. Both lobes are tethered to each other in a 'dumbbell-like' manner by a flexible α-helix linker, this allows CaM to conformationally adapt to embrace many target proteins (Ikura & Ames, 2006;Shimoyama & Takeda-Shitaka, 2017;Zhang et al., 2012). Furthermore, the isolated, yet physically coupled globular domains of CaM permit a dynamic interplay with target binding where target affinity of one lobe can be modulated depending on the complex formed at the neighbouring lobe (Sondergaard et al., 2017). Upon binding to Ca 2+ , CaM undergoes significant conformational change and adopts a more open conformation, exposing otherwise buried hydrophobic patches. This reversible conformational transition permits Ca 2+ -dependent interaction with, and modulation of, a wide range of targets. CaM is therefore able to dynamically translate changes in intracellular [Ca 2+ ] to mediate a myriad of cellular processes, including cardiac muscle contraction. In cardiomyocytes, CaM regulates many of the central proteins involved in excitation-contraction coupling such as the L-type voltage-gated Ca 2+ channel (Cav1.2), the cardiac ryanodine receptor (RyR2), the voltage-gated Na + channel (Nav1.5) and the voltage-gated K + channel (Kv7.1) (Balshaw et al., 2002;Brohus et al., 2019;Gabelli et al., 2014;Ghosh et al., 2006;Kang et al., 2020;Kang et al., 2021;Xu & Meissner, 2004;Yamaguchi et al., 2003;Zuhlke et al., 1999).
In this paper, we investigate three LQTS-associated CaM variants: D95V, N97I and D131H. These variants have missense mutations within the C-lobe of CaM, specifically at residues which directly coordinate Ca 2+ binding (Fig. 1). D95V and N97I are located in EF-hand III, while D131H is in EF-hand IV. The consequences of these variants have previously been clinically described in patients presenting with de novo CaM (CALM2) mutations (Crotti et al., 2013;Makita et al., 2014;Pipilas et al., 2016). The clinical presentations of all mutations were highly pathogenic and resulted in severe and recurrent episodes of life-threatening arrythmias. The perturbed Ca 2+ -binding ability of D95V, N97I and D131H CaM variants is a likely driver of disease pathogenesis J Physiol 601.17 (Crotti et al., 2013;Makita et al., 2014;Pipilas et al., 2016). Previous studies highlight the difficulty in predicting how perturbed Ca 2+ binding in arrhythmia-associated CaM variants translates to defective target modulation (Chazin & Johnson, 2020;Hussey et al., 2023;Jensen et al., 2018). We employed a multidisciplinary approach to gain novel insight into the molecular aetiology of how LQTS-associated CaM mutants contribute to arrythmia through impaired Kv7.1 function. We demonstrate that LQTS-associated CaM mutations reduce IKs, most likely through altered CaM structure and impaired interaction with Kv7.1. The presented data aid in elucidating the contributions of LQTS-associated CaM mutants in electrical disease of the heart and how they may exacerbate the most common mechanism of LQTS, perturbed IKs generation.

Molecular biology
For biophysical and structural biology experiments. The sequence of human wild-type (WT) CaM was subcloned into the pE-SUMOPro-Kan vector (LifeSensors, USA) as previously described (Prakash et al., 2021(Prakash et al., , 2023. A series of site-directed mutagenesis (SDM) reactions were performed using the QuikChangeII kit (Agilent Technologies, USA), according to the manufacturer's recommendation, in order to generate LQTS-associated CaM mutants D95V, N97I and D131H. Primers used are presented in Table 1 (SDM primers).
For electrophysiology experiments. DNA corresponding to CaM variants (from the pE-SUMOPro-Kan constructs), KCNQ1 and KCNE1 (Addgene plasmid 53 048, 53 050, gifts from Michael Sanguinetti) (Sanguinetti et al., 1996) was subcloned into pHIV-IRES-dTomato or pHIV-IRES-EGFP (Addgene plasmid 21 374, 21 373, gifts from Bryan Welm) using Gibson Assembly (NEBuilder HiFi, New England Biolabs), according to the manufacturer's guidelines. The primers used are presented in Table 1 (IRES primers). In these constructs, proteins of interest and a fluorescent marker (dTomato or EGFP) were co-expressed under the control of the same promoter, as two distinct proteins and not as fusion proteins.
For flow cytometry experiments. The trafficking assay requires KCNQ1-BBS, a modified KCNQ1 construct whereby a 13-residue bungarotoxin-binding site (BBS) was introduced into the extracellular S1-S2 loop to allow channel detection at the cell surface through a cell-impermeable, fluorescent α-bungarotoxin (BTX) conjugate. KCNQ1-BBS (gift from Henry M. Colecraft) (Aromolaran et al., 2014), ECFP (Addgene plasmid 70 104, gift from Harald Sitte) (Sucic et al., 2010) and the pHIV backbone (Addgene plasmid 21 373, gift from Bryan Welm) were assembled in a single reaction via Gibson Assembly (NEBuilder HiFi, New England Biolabs) according to the manufacturer's guidelines. The primers were designed to remove the IRES-EGFP sequence, allowing the generation of pHIV-ECFP-KCNQ1-BBS (Table 1). The regulatory subunit KCNE1/minK was a gift from Michael Sanguinetti (Addgene plasmid 53 050) (Sanguinetti et al., 1996). For CaM constructs, the CaM variants were subcloned from the pGEX-6P-1 vector (Lian et al., 2014) into pEGFP-N1 by restriction-ligation (KpnI/BamHI) using primers displayed in Table 1. A series of SDM reactions were performed following the QuikChangeII protocol (Agilent Technologies) as . Right illustrates their location within the C-lobe, all of which occur within residues which directly coordinate Ca 2+ , as depicted by black lines. Mutants D95V and N97I are located in the third Ca 2+ -binding site of CaM (EF-hand III), whereas D131H is found in the fourth EF-hand (EF-hand IV).
HEK293T cells transiently transfected with CaM-IRES-EGFP, KCNQ1-IRES-dTomato and KCNE1-IRES-dTomato were added dropwise into the perfusion bath and allowed to settle onto the glass-bottomed chamber. CaM-expressing cells were identified by EGFP fluorescence (Nikon Eclipse TE200 inverted microscope with epifluorescence attachment). The IKs activation protocol consisted of 4 s voltage steps ranging from −60 to +100 mV (20 mV increments) from a holding potential of −80 mV, followed by J Physiol 601.17 a 3 s repolarisation step at −40 mV. Half maximal activation (V 1/2 ) was calculated by fitting normalised peak conductance using the Boltzmann sigmoid equation. All data analysis was performed using the Axon pClamp software package (version 10.7.0.3, Molecular Devices) and GraphPad Prism.

Secondary structure content and thermal stability
Secondary structure content. Far-UV spectra (180-260 nm) were collected at 20°C in a 0.1 cm path length quartz cell using a Jasco J-1100 circular dichroism spectrometer equipped with a Jasco MCB-100 mini circulation bath for temperature control. CaM (10 μM) spectra were measured in 2 mm HEPES, pH 7.5 supplemented with either 1 mm EGTA (pH 7.5) or 5 mm CaCl 2 for Ca 2+ -free (apo) and Ca 2+ -bound experiments, respectively. For each sample, three scans were averaged (scan rate of 100 nm/min). After buffer subtraction, data were normalised to mean residual ellipticity and secondary structure content was calculated using the CDSSTR prediction programme (Dichroweb, reference dataset 7) (Whitmore & Wallace, 2004. Thermal stability. Sensitivity of apo-CaM (10 μM) to temperature was assessed by decrease in α-helical content measured by circular dichroism at 222 nm. Temperature ranged from 15°C to 90°C in 1°C increments, with a ramp increase rate of 1°C/min and a 180 s equilibration period between recordings. Data were normalised and fitted to the Boltzmann sigmoid equation (GraphPad Prism) to derive the melting temperature of CaM (T m ).

Susceptibility to trypsin digestion
Sensitivity of CaM to trypsin hydrolysis was assessed by SDS-PAGE and densitometry analysis. Purified CaM proteins (5 μM) were incubated with trypsin for 30 min at 37°C in 25 mM HEPES, 100 mM NaCl, pH 7.5 supplemented with trypsin at 0-10 mg/ml in apo conditions (10 mM EGTA) and 0-30 mg/ml in Ca 2+ -bound conditions (5 mM CaCl 2 ). Reactions were rapidly terminated by the addition of SDS-containing sample buffer and heating at 95°C for 10 min. Proteins were separated by SDS-PAGE (NuPAGE 4-12% Bis-Tris, Life Technologies) and stained with InstantBlue (Abcam). Images were obtained on a ChemiDoc XRS+ transilluminator (Bio-Rad) and the fraction of intact CaM was quantified by densitometry using Fiji software (Schindelin et al., 2012).

Isothermal titration calorimetry (ITC)
Experiments were performed in 50 mM Na + -HEPES, 100 mM KCl, 2 mM MgCl 2 (pH 7.5) supplemented with either 1 mM EGTA or 5 mM CaCl 2 for Ca 2+ -independent or -dependent interactions, respectively. Kv7.1-HA 370-389 or Kv7.1-HB 507-536 peptides were titrated against CaM proteins across 20 injections (2 μl each) lasting 4 s with a 180 s grace period between each injection. Peptide was typically titrated into the cell at a concentration 10-20 times higher than CaM ([CaM] ∼ 50 μM). All titrations were performed using MicroCal iTC200 and automated PEAQ-ITC systems (Malvern Panalytical) at 25°C with continuous stirring at 800 rpm. Data were processed using MicroCal PEAQ-ITC software and fitted according to a one-site or two-site binding model to determine binding characteristics (affinity, stoichiometry, enthalpy change, entropy change and Gibbs free energy).

Flow cytometry
Kv7.1 cell surface density was determined by flow cytometry in live HEK293T cells expressing ECFP-KCNQ1-BBS, KCNE1 and CaM-EGFP variants, following previously described protocols (Aromolaran et al., 2014). Briefly, cells (0.3 × 10 6 ) were plated in standard six-well plates and co-transfected with ECFP-KCNQ1-BBS, KCNE1 and CaM-EGFP variants using Lipofectamine2000 (1:1:1 molar ratio), following the manufacturer's recommendations. Cells were gently washed with ice cold PBS containing 1 mm CaCl 2 and 0.5 mm MgCl 2 (pH 7.4). Post-transfection (48 h), cells were blocked in DMEM/3% bovine serum albumin on ice for 30 min and then incubated with 1 μM Alexa Fluor 647 conjugated α-bungarotoxin (ThermoFisher) on ice for 1 h in the dark. Cells were washed three times with PBS (containing Ca 2+ and Mg 2+ ) and harvested with 0.05% Trypsin-EDTA (ThermoFisher). Cells were resuspended in normal PBS and assayed using a BD FACSCanto II flow cytometer (BD Biosciences). ECFP-and EGFP-tagged proteins were excited with violet (405 nm) and blue (488 nm) lasers, respectively, and Alexa Fluor 647 was excited with a red (633 nm) laser. Fluorescence signals from flow cytometry were analysed using the BD FACSDiva 9.0 software. For each group of experiments, live cells were discriminated from heterogenous populations by size through comparison of cell area via side and forward scatter (SSC-A and FSC-A). Single cells were gated from doublets through comparison of forward scatter height and area (FSC-H and FSC-A). Isochronal untransfected and single colour controls were used to manually set the threshold and gain setting for each fluorophore.
After selecting for EGFP-positive cells (CaM), Kv7.1 surface density was calculated as:  (Lloyd et al., 2008). The model is available at https://models.cellml. org/e/5a0/ohara_rudy_cipa_v1_2017.cellml/view. We made a simple addition to the O'Hara-Rudy model in terms of a direct CaM-IKs interaction where CaM concentration blocked IKs according to the Hill equation (eqn 2): is the CaM concentration specified elsewhere in the model and Kb is an arbitrary constant leading to the experimentally observed block in peak IKs density.

Statistical analysis
Statistical analyses were performed using GraphPad Prism version 9. Number of replicates and type of statistical tests are indicated in the figure legends. P value <0.05 was considered statistically significant. P values are represented by asterisks with * P < 0.05, * * P < 0.01, * * * P < 0.001, * * * * P < 0.0001. J Physiol 601.17

Arrhythmogenic CaM variants reduce IKs densities and impair voltage-dependent activation at resting (100 nM) and high (1 μM) Ca 2+ levels
To investigate the effect of the LQTS-associated CaM variants on IKs, whole-cell patch-clamp electrophysiology was performed using HEK-293T cells co-transfected with CaM, KCNQ1 and KCNE1. At resting intracellular Ca 2+ levels (100 nm), we observed a significant decrease in IKs densities ( Fig. 2A, B) and disrupted activation kinetics when channels were modulated by LQTS-CaM variants (Fig. 2C). Current densities at +100 mV were significantly reduced for CaM-D95V, CaM-N97I and CaM-D131H, when compared with CaM-WT ( Fig. 2A, B and Table 2). The voltage at which 50% of the channels were active (V 1/2 ) was reduced by up to threefold for CaM-D95V and CaM-D131H, when compared with CaM-WT (Fig. 2C, Table 2).
At systolic levels of intracellular Ca 2+ (1 μm), reduced current densities and impaired activation kinetics were observed for CaM-D95V and CaM-N97I (Fig. 3). Current densities at +100 mV were reduced for CaM-D95V and CaM-N97I, when compared with CaM-WT (Fig. 3A, B and Table 2). CaM-D131H remained unchanged. The voltage at which 50% of the channels were active (V 1/2 ) were reduced for CaM-D95V and CaM-D131H, while modulation by CaM-N97I remained similar to CaM-WT (Fig. 3C, Table 2). Together, this indicates that LQTS-associated CaM variants affect Kv7.1 activity at both resting and high Ca 2+ concentrations.

Trafficking of KCNQ1 channels to the plasma membrane is not altered by the LQTS-CaM variants
A reduction in whole-cell currents can be caused by changes in channel behaviour or by a change in the number of active channels at the cell surface. Using flow Currents were obtained in whole-cell voltage-clamp configuration by holding cells at −80 mV and stepping for 4 s from −60 mV to +100 mV in 20 mV increments, followed by a repolarising step at −40 mV. B, current-voltage (I/V) relationships of IKs currents modulated by CaM. Differences between groups were determined using a two-way ANOVA with Dunnett's multiple comparisons tests. C, activation kinetics. (Left panel) mean ± s.e.m. Channel conductance, G, normalised to peak conductance, Gmax, to give mean activation/activation curves. (Right panel) mean ± s.e.m. Half maximal activation voltages, V 1/2 , calculated from individual curves fitted using the Boltzmann equation. Differences between groups were determined using a one-way ANOVA with Dunnett's multiple comparisons tests. cytometry, we quantitatively assessed the Kv7.1 (KCNQ1) relative surface density in live cells to determine whether trafficking of the channel was impaired in the presence of the CaM variants. KCNQ1 was expressed as an ECFP fusion and contained a 13-residue high-affinity α-BBS in its extracellular S1-S2 loop to allow labelling with extracellularly applied α-bungarotoxin-Alexa Fluor 647. Therefore, ECFP levels represented the total Kv7.1 expressed in cells, while ECFP+Alexa Fluor 647 levels indicated Kv7.1 channels at the surface (Fig. 4A). Based on this method developed by Colecraft's laboratory (Aromolaran et al., 2014), we observed that KCNQ1 surface density (ECFP+Alexa Fluor 647 labelled) was 35.0 ± 1.2% (of total KCNQ1 channels, ECFP only, n = 5) and that LQTS-associated CaM variants did not significantly alter the percentage of channels at the plasma membrane. Kv7.1 surface density was 30.8 ± 2.6% for D95V (n = 5, P = 0.3140), 30.4 ± 1.7% (n = 5, P = 0.2502) for N97I and 35.4 ± 1.9% for D131H (n = 5, P = 0.9976) (Fig. 4).
Using 1 H-15 N HSQC NMR, we investigated structural differences between the CaM variants in Ca 2+ -bound conditions (Fig. 6). Spectra showed that LQTS-associated variants exhibited distinct spectra when compared with CaM-WT (Fig. 6A). Spectra for D95V and N97I (EF-hand III variants) showed high degrees of homology with each other, while D131H (EF-hand IV variant) was distinct. Chemical shift perturbation analysis showed that for D95V and N97I, the structure of CaM was perturbed locally, near the site of the mutation (Fig. 6B). Due to difficulties in transferring assignments from CaM-WT to the LQTS variants, peak shift analysis could not be performed for many residues (Fig. 6B, unassigned residues shown as negative values). For D131H, due to extensive structural changes spanning beyond the region immediate to the site of mutation, the majority of resonances within the C-lobe could not be transferred from the wild-type (∼7% of residues assigned).

LQTS-associated variants increase CaM susceptibility to protease digestion while temperature sensitivity remains unchanged
To determine the effect of the mutations on the 3D structure of CaM, we investigated the variants' susceptibility to protease digestion. CaM proteins were incubated with various concentrations of trypsin, a serine protease which cleaves peptide chains at the carboxyl-side of lysine and arginine residues. In the absence of Ca 2+ , CaM variants D95V and N97I displayed an increased susceptibility to trypsin digestion, whereas D131H showed a decreased susceptibility, when compared with CaM-WT (Fig. 7A, left panel). In the presence of Ca 2+ , CaM proteins required higher concentrations of trypsin to reach near-complete proteolysis, with all mutants showing an increased susceptibility to trypsin proteolysis, when compared with CaM-WT (Fig. 7A, right panel).
Using circular dichroism, we monitored the unfolding of apo-CaM as a function of temperature (at 222 nm, J Physiol 601.17 characteristic of α-helices). Melting curves were fitted to the Boltzmann sigmoidal equation to determine the midpoint of the unfolding transition (melting temperature, T m ) (Fig. 7B, left panel). From the melting curve, we determined the T m for CaM-WT as 42 ± 1°C (n = 3) and no significant differences were observed for LQTS-associated CaM variants. The T m was 43 ± 1°C for D95V (n = 3, P = 0.0899), 42 ± 1°C for N97I (n = 3, P = 0.3446) and 41 ± 1°C for D131H (n = 3, P = 0.1290).

CaM interacts with the helix A domain of Kv7.1 in a Ca 2+ -dependent manner
Helix A is a CaM binding domain within the C-terminus of KCNQ1 (Kv7.1-HA 370-389 ) (Yus-Najera et al., 2002). Using ITC, we showed that in the absence of Ca 2+ , no measurable interaction between CaM-WT and Kv7.1-HA 370-389 was observed (Fig. 8A).
At saturating Ca 2+ concentrations, all CaM proteins showed two distinct binding events, at one and two molar excess of Kv7.1-HB 507-536 to CaM (Fig. 10, Table 4). We observed a tight first interaction (K d1 = 0.56 ± 0.03 nm for CaM-WT) followed by a lower affinity second interaction (K d2 = 538 ± 30 nm for CaM-WT) (Fig. 10A, B). All LQTS-associated mutations were found to significantly reduce affinities of both first and second interactions. The affinity was reduced up to 17-fold (K d1 = 9.70 ± 0.8 nm for D131H) and 1.7-fold (K d2 = 908 ± 29 nm for D131H), for the first and second binding events, respectively, when compared with CaM-WT (Fig. 10B,  D). Thermodynamic signatures revealed that interactions between Ca 2+ /CaM and Kv7.1-HB 507-536 are energetically favourable, exothermic and enthalpy-driven (Fig. 10C ,E). For the first binding event, all LQTS-associated variants showed a reduced G and H as well as altered S, suggesting less energetically favourable interactions and unfavourable conformational changes. For the second binding event, G was reduced for all variants, while H and S were significantly altered only for the N97I and D131H variants, when compared with CaM-WT (Table 4). In the absence of Ca 2+ , 1 H-15 N HSQC NMR spectral overlays revealed good general consistency between apo-CaM:Kv7.1-HB 507-536 complexes (Fig. 11A). EF-hand III mutant spectra (D95V and N97I) showed near-complete consistency with that of CaM-WT, in contrast to EF-hand IV mutant D131H.

Predicted effect of IKs inhibition by CaM variants on action potential duration (APD)
First, based on our experimental data, we mathematically altered IKs to match the effect of the LQTS-associated CaM variants (Fig. 1A). Then, using the adapted version of the O'Hara-Rudy model, we simulated ventricular action potentials and measured the duration at 50% of the amplitude (APD50). APD50 increased from 216.8 ms (CaM-WT) to 224.7 ms (N97I), 226.7 ms (D131H) and 232.5 ms (D95V). This represents a relative increase of APD of 3.6% (N97I), 4.6% (D131H) and 7.2% (D95V), when compared with CaM-WT ( Fig. 12B-D). Data represent averages of five replicates ± s.e.m. Differences between groups were determined using a two-way ANOVA with Dunnett's multiple comparisons tests.

Discussion
CaM is a highly conserved, Ca 2+ -sensing protein which has emerged as a modulator of many key proteins and ion channels which govern both excitation-contraction coupling and the spatio-temporal topology of the ventricular action potential. CaM mutations which perturb modulation of these targets promote LQTS (Chazin & Johnson, 2020;Hussey et al., 2023;Jensen et al., 2018), the most common genetic aetiology of which arises from loss-of-function mutations in Kv7.1 (∼50% of cases) (Crotti et al., 2008;Schwartz et al., 2012). CaM facilitates the Ca 2+ -dependent potentiation of IKs (Bai et al., 2005;Bartos et al., 2017;Nitta et al., 1994;Shamgar et al., 2006;Tobelaim et al., 2017;Tohse, 1990) and contributes to proper channel folding, tetramer assembly, membrane trafficking and post-translational modification of Kv7.1 (Asada et al., 2009;Ghosh et al., 2006;Shamgar et al., 2006). Disruption of this calmodulation appears to be a prominent driver of arrhythmia, as observed through LQTS-associated Kv7.1 mutations which disrupt CaM interactions at the channel C-terminus (Ghosh et al., 2006;Gonzalez-Garrido et al., 2021;Mousavi Nik et al., 2015;Sachyani et al., 2014;Schmitt et al., 2007;Shamgar et al., 2006;Tobelaim et al., 2017;Yang et al., 2009;Zhou et al., 2016). The consequences of CaM mutants on Kv7.1 function, however, remain elusive. Through adopting a multidisciplinary approach to better understand three LQTS-associated CaM mutants (D95V, N97I and D131H), we reveal the molecular mechanisms which explain the perturbed structure-function relationships of these CaM variants, how their altered structures hinder complex formation with Kv7.1, and how this modulation results in a LQTS-compatible IKs. While the modulatory consequences of arrythmia-associated CaM variants have been well characterised in other key cardiac targets such as Cav1.2 and RyR2 (Chazin & Johnson, 2020;Hussey et al., 2023;Jensen et al., 2018), little is known concerning their effects on Kv7.1 (Kato et al., 2022;Rocchetti et al., 2017). Using a HEK293T cellular model, we demonstrated that, at elevated [Ca 2+ ] cyt , CaM-WT hastened activation kinetics of Kv7.1/KCNE1 and facilitated channel opening at lower membrane potentials when compared with resting Ca 2+ concentrations (100 nm). These findings agree with previously published work whereby CaM was shown to increase IKs current amplitude and left-shift the voltage dependence of activation in a Ca 2+ -dependent manner (Adam et al., 1993;Nitta et al., 1994;Tobelaim et al., 2017). Such regulation of IKs by CaM is vital in coordinating an appropriate repolarising response to depolarising stimuli (increased [Ca 2+ ] cyt ) and prevents arrhythmogenic prolongation of the ventricular APD. With regards to the modulatory effects of LQTS-associated CaM variants, we revealed that IKs current density was reduced across a range of membrane potentials at resting [Ca 2+ ] cyt when modulated by all CaM variants, whereas only EF-hand III mutants (D95V and N97I) significantly reduced IKs generated at high [Ca 2+ ] cyt . Interestingly, only D95V and D131H CaM reduced voltage sensitivity of activation at both high and low [Ca 2+ ] cyt , suggesting perturbation of the voltage-sensitive domain of Kv7.1. [Ca 2+ ] free of 100 nm and 1 μm were chosen to mimic the cytosolic environment of a myocyte during diastole and systole, respectively. Because CaM has been described to regulate trafficking of Kv7.1 to the plasma membrane (Ghosh et al., 2006;Shamgar et al., 2006), the reduced current densities for CaM variants could be associated with a trafficking defect. However, we did not observe any significant impairment of Kv7.1 trafficking to the plasma membrane for any of the LQTS-associated CaM variants assayed, consistent with other studies of mutant CaM-Kv7.1 trafficking (Kato et al., 2022). These data suggest diverse mechanisms by which CaM variants modulate IKs, with reduced Ca 2+ -sensitivity, reduced current amplitude and depolarised voltage dependence of activation (positive shift in V 1/2 ) observed, all of which contribute to a LQTS-compatible IKs current.
To decipher the mechanisms behind this pathogenic modulation, the conformational consequences of structural perturbations to CaM were investigated. Circular dichroism experiments revealed that LQTS-CaM variants were less able to undergo Ca 2+ -induced conformational change, a structural transition essential to CaM's function as a Ca 2+ -signalling protein. This was most apparent in EF-hand IV variant D131H and agrees with other studies where differences in structural distributions for CaM-mutant compared with CaM-WT were observed (Dal Cortivo et al., 2022;Hennessey et al., 1987). Using chemical shift analysis of HSQC NMR spectra to more precisely identify areas of structural change, we revealed that in their Ca 2+ -saturated states, D95V and N97I presented with localised perturbation within the C-lobe, while perturbations to D131H were global, affecting both the N-lobe and the C-lobe. These findings agree with other studies which have demonstrated a site-dependent, ranging degree of chemical shift perturbation across LQTS-CaM variants, which present more apparently when variants are compared with WT in their calcified states Pipilas et al., 2016;Wang et al., 2020;Wren et al., 2019). Additionally, we demonstrated that LQTS-associated CaM variants displayed altered susceptibility to protease hydrolysis, consistent with other works (Crotti et al., 2013;Dal Cortivo et al., 2022;Prakash et al., 2023). This was more apparent when mutants were compared in Ca 2+ -saturating conditions, reflecting the more distinct Ca 2+ -bound conformations of mutants, rather than their more subtly dissimilar apo-conformations. A specific order of chelation exists within the four EF-hands of CaM. Precise allosteric Experiments were performed at 25°C in the presence of 1 mM EGTA. DP, differential power. Differences between groups were determined using a one-way ANOVA with Dunnett's multiple comparisons tests. regulation and positive cooperativity between Ca 2+ binding sites facilitate a system whereby EF-hand IV binds Ca 2+ first, then EF-hand III, followed by EF-hand II and finally EF-hand I (Liu et al., 2019). Therefore, mutations which reduce the Ca 2+ binding of EF-hand IV are likely to disrupt downstream calcification of all other EF-hands, whereas disruptions to EF-hand III would still benefit from the positive cooperativity from calcification of EF-hand IV. Mutations within EF-hand IV typically reduce Ca 2+ binding affinity more so than mutations in EF-hand III (Jensen et al., 2018) and so help explain how D131H appears to more significantly perturb the structure of CaM compared with D95V and N97I. The data outline the significant effects which single missense, C-lobe mutations infer on the conformation of CaM.
To better understand how LQTS-CaM mutants infer altered modulation of IKs, the interactions between CaM proteins and their isolated binding domains of Kv7.1 were characterised. We showed that CaM interacts with Kv7.1-HA 370-389 exclusively in the presence of Ca 2+ , similar to works performed on Kv7.4 (Archer et al., 2019). The affinity of this interaction was low and could not be accurately determined for the LQTS-associated CaM variants. While the apo-CaM:Kv7.1-HA 370-389 could not be fully characterised, crystal structures of the CaM-KCNQ1 complex at HA and HB consistently reveal the apo C-lobe of CaM bound to HA, even in high molar excesses of Ca 2+ (Sachyani et al., 2014). This has also been suggested in pull down assays between the Cterminus of KCNQ1 and CaM (Tobelaim et al., 2017). For Kv7.1-HB 507-536 , we showed that CaM can interact in both Ca 2+ -independent and dependent manners. In apo conditions, HSQC NMR spectral overlays revealed good general consistency between apo-CaM:Kv7.1-HB 507-536 complexes. EF-hand III mutant spectra (D95V and N97I) showed near-complete overlap with CaM-WT, contrasting with EF-hand IV mutant D131H and further supporting the more apparent conformational divergence of the EF-hand IV variant. At saturating Ca 2+ concentrations, the HSQC NMR spectral overlays revealed incomplete homology with the distribution of CaM-WT signals, indicating that LQTS-associated variant proteins adopt alternative conformations when bound to Kv7.1-HB 507-536 . Using ITC, we showed that the apo-CaM:Kv7.1-HB 507-536 interaction was driven by hydrophobic interactions, with all apo-mutants revealing less favourable conformational changes (reflected by their significantly increased -T S contributions). Variants D95V and D131H exhibited significant increases in measured K d . In the presence of Ca 2+ , interaction of CaM variants with Kv7.1-HB 507-536 was driven by hydrogen bonding and Van der Waals forces (reflected in negative H values). For all LQTS-associated mutants, we observed a reduction in affinity. The trends across data appear consistent, whereby EF-hand IV mutant (D131H) J Physiol 601.17 confers the greatest reduction in affinity for Kv7.1, most notably in Ca 2+ -saturating conditions. Compared with EF-hand III variants (D95V and N97I), this mutant was most structurally distinct from WT-CaM, particularly when compared in the presence of Ca 2+ .
Together, the findings presented here reveal that single residue substitutions to CaM's highly conserved structure have substantial effects on global protein structure. In the case of the variants studied (D95V, N97I and D131H), substitutions occur at residues which Figure 10. Ca 2+ -CaM binding to Kv7.1-HB 507-536 is decreased for LQTS-associated variants A, representative ITC titration curves (upper panel) and binding isotherms (lower panel) for the interaction between apo-CaM and helix B of the Kv7.1 C-terminus (Kv7.1-HB 507-536 ). B, D, affinity and (C, E) thermodynamic profile of the binding of apo-CaM to Kv7.1-HB 507-536 obtained by fitting to a two-site binding model. B, C, binding parameters for the first interaction and (D, E) for the second interaction. Data are means ± s.e.m. N, stoichiometry; n, number of experimental replicates. The sum of the change in enthalpy ( H) and the change in entropy ( S) multiplied by the absolute temperature (T) gives the change in free energy ( G). Experiments were performed at 25°C in the presence of 5 mM CaCl 2 . DP, differential power. Differences between groups were determined using a one-way ANOVA with Dunnett's multiple comparisons tests. directly coordinate Ca 2+ at the C-lobe of the protein, resulting in reduced Ca 2+ -binding (Crotti et al., 2013;Makita et al., 2014;Pipilas et al., 2016;Sondergaard et al., 2015;Vassilakopoulou et al., 2015) and reduced conformational plasticity in response to increases in [Ca 2+ ]. The perturbed structures of CaM mutants were found to reduce their affinity of interaction at the C-terminus of Kv7.1, both in the absence and presence of calcium. Ca 2+ -dependent binding was significantly weakened compared with Ca 2+ -independent interactions of mutants, reflected by their more divergent calcified conformations than apo ones. The conformations adopted by CaM mutants with their binding sites were significantly distinct from those adopted by WT CaM, with exacerbations to perturbation found more so in the presence of Ca 2+ than in Ca 2+ -free conditions. The alternative structures which CaM mutants adopt with Kv7.1 were found to result in aberrant channel modulation, while trafficking remained unaffected. Ca 2+ -sensitivity of LQTS-CaM modulated Kv7.1 channels were reduced and generated smaller IKs when elicited in conditions which mimicked those of physiological cardiac contraction (1 μm [Ca 2+ ] free ). Such modulation would reduce the repolarisation capacity of IKs, extending the APD and contributing to the LQTS phenotype which presented in patients harbouring said CaM mutations. LQTS-associated CaM variants have been shown to perturb a range of other cardiac ion channels, including reducing the Ca 2+ -dependent inactivation of Cav1.2 Limpitikul et al., 2014;Prakash et al., 2023;Yin et al., 2014), altered inhibition of RyR2 (Nomikos et al., 2014;Vassilakopoulou et al., 2015) and activation of CaMKIIδ (Berchtold et al., 2016;Prakash et al., 2023). The large repertoire of targets which CaM interacts with, combined with the limited clinical data from CaM-driven LQTS patients, make it difficult to establish a direct correlation between CaM mutation and disease severity. Using mathematical modelling, we predicted that IKs reduction for the LQTS-CaM variants would cause an APD prolongation of 3.6% (N97I), 4.6% (D131H) and 7.2% (D95V). Considering that QTc interval is prolonged by 26% (N97I, from 440 to 555 ms) (Crotti et al., 2013), 48% (D131H, from 440 to 651 ms) (Makita et al., 2014) and 57% (D95V, from 440 to 690 ms) (Pipilas et al., 2016), data suggest a substantial IKs contribution to the LQTS phenotype across all variants: 14% (N97I), 10% (D131H) and 13% (D95V). It should be appreciated that the functional effects of CaM mutants described here would span beyond Kv7.1 modulation and would simultaneously exacerbate multiple modalities of LQTS (enhanced I Ca,L, I Na ). We demonstrate here that LQTS-CaM mutants have the capacity to prolong membrane repolarisation, CaM mutants have also been demonstrated to perturb Cav1.2 inactivation and could modulate a range of  A, different values of Kb (eqn 2, see Methods) were used until the resulting block of peak IKs matched that of the experimental data for CaM-WT, D95V, N97I and D131H variants. B, output simulated ventricular action potentials for each of these situations. The dotted line indicates the action potential 50% level (APD50) where action potential duration has been measured. C, expanded view of the simulated action potentials shown in panel B. D, plot of APD duration increase for the given reduction in IKs current density. In each of A,B, the waveforms are colour coded to the simulation condition: black is CaM-WT, blue is D95V, green is N97I and red is D131H.
arrhythmogenic substrates (Benitah et al., 2010;January et al., 1988;Madhvani et al., 2015;Weiss et al., 2010). The most common mechanism of arrythmia involves the generation of early-after depolarisations (EADs) via self-amplification of I Ca,L (Weiss et al., 2010). EADs promote discordant refractory periods across the myocardium and create functional obstacles through which the depolarising wave of activity must navigate (Sato et al., 2009;Weiss et al., 2010). The results in a multifocal, self-sustaining arrhythmia where the site of origin continuously shifts throughout the tissue, producing the characteristic 'Torsade de Pointes' , which can lead to ventricular fibrillation.
In summary, our study demonstrates that LQTS-associated single residue substitutions to CaM significantly disrupt its structure-function relationship as a Ca 2+ -sensing protein. Pathogenic CaM variants; D95V, N97I and D131H contribute to aberrant IKs generation through adopting alternative conformations with Kv7.1. This work aids in elucidating the site-dependent effects of CaM mutations and highlights the significant disease processes which CaM mutants underpin. Pharmacologically, LQTS patients are usually treated with β-blockers (Farzam & Jan, 2022). While broadly effective, many patients still suffer from recurrent arrhythmic episodes while treated with β-blockers. The multiple genotypes which contribute to LQTS therefore likely reduce the effectiveness of one blanket pharmacological agent (Ahn et al., 2017). This results in almost one third (32%) of LQTS patients experiencing recurrent cardiac events despite being on β-blocker therapies (Moss et al., 2000). Specifically, LQTS patients with CaM mutations respond varyingly to these therapies (Ahn et al., 2017). Understanding the molecular mechanisms by which CaM can aberrantly modulate targets has proven valuable and allows clinicians to better guide management of patients harbouring CaM mutations (Webster et al., 2017).