Competition of calcified calmodulin N lobe and PIP₂ to an LQT mutation site in Kv7.1 channel

Significance

Voltage-gated potassium 7.1 (Kv7.1) channel and KCNE1 protein coassembly forms the I_KS K⁺ current that repolarizes the cardiac action potential, and mutations in Kv7.1 and KCNE1 genes cause cardiac arrhythmias. The proximal Kv7.1 C terminus binds calmodulin and the phospholipid phosphatidylinositol-4,5-bisphosphate (PIP₂); however, it is unknown whether their binding sites overlap physically and functionally. Here, we reveal the competition of PIP₂ and the calcified form of the calmodulin N lobe to a previously unidentified site in helix B of the proximal Kv7.1 C terminus. Notably, this site bears a mutation causing a cardiac arrhythmia called the long-QT syndrome. Our results suggest that, after receptor-mediated PIP₂ depletion and increased cytosolic Ca²⁺, calcified calmodulin N lobe interacts with helix B in place of PIP₂ to limit excessive I_KS current depression.

Abstract

Voltage-gated potassium 7.1 (Kv7.1) channel and KCNE1 protein coassembly forms the slow potassium current I_KS that repolarizes the cardiac action potential. The physiological importance of the I_KS channel is underscored by the existence of mutations in human Kv7.1 and KCNE1 genes, which cause cardiac arrhythmias, such as the long-QT syndrome (LQT) and atrial fibrillation. The proximal Kv7.1 C terminus (CT) binds calmodulin (CaM) and phosphatidylinositol-4,5-bisphosphate (PIP₂), but the role of CaM in channel function is still unclear, and its possible interaction with PIP₂ is unknown. Our recent crystallographic study showed that CaM embraces helices A and B with the apo C lobe and calcified N lobe, respectively. Here, we reveal the competition of PIP₂ and the calcified CaM N lobe to a previously unidentified site in Kv7.1 helix B, also known to harbor an LQT mutation. Protein pulldown, molecular docking, molecular dynamics simulations, and patch-clamp recordings indicate that residues K526 and K527 in Kv7.1 helix B form a critical site where CaM competes with PIP₂ to stabilize the channel open state. Data indicate that both PIP₂ and Ca²⁺-CaM perform the same function on I_KS channel gating by producing a left shift in the voltage dependence of activation. The LQT mutant K526E revealed a severely impaired channel function with a right shift in the voltage dependence of activation, a reduced current density, and insensitivity to gating modulation by Ca²⁺-CaM. The results suggest that, after receptor-mediated PIP₂ depletion and increased cytosolic Ca²⁺, calcified CaM N lobe interacts with helix B in place of PIP₂ to limit excessive I_KS current inhibition.

Five voltage-gated potassium 7 (Kv7) channel (or KCNQ) channel members form a subfamily of Kv channels that plays important functions in various tissues, including brain, heart, kidney, stomach, pancreas, or inner ear (1). Kv7.1 α-subunits can interact with each of five KCNE β-subunits, displaying distinct current characteristics (2–5). Coassembly of Kv7.1 with KCNE1 produces the I_KS current, which together with I_Kr [human ether-à-go-go–related gene (hERG) channel], forms the main repolarizing currents of the cardiac action potential (6–8). Mutations in either Kv7.1 or KCNE1 genes lead to life-threatening cardiac arrhythmias, causing long-QT (LQT) or short-QT syndromes and atrial fibrillation (9, 10).

Similar to all Kv channels, the Kv7.1 structure features six transmembrane segments (S1–S6) containing a voltage-sensing module (S1–S4) and a pore domain (S5 and S6). In contrast to Shaker-like Kv channels, Kv7 does not harbor an N-terminal T1 tetramerization domain but does possess a large C terminus (CT), which was shown to be important for channel gating, assembly, and trafficking (11–15). The Kv7 CT comprises amphipathic α-helices that form three coiled coils. The proximal helices A and B, adjacent to the membrane, form a coiled coil that binds calmodulin (CaM) (16–19), whereas the distal helices C and D form tandem coiled coils that serve as a tetramerization domain (13, 15). CaM seems to be an essential auxiliary subunit of all Kv7 channels (11, 14, 17, 19–23). Although it is clear that proper CaM binding to the proximal Kv7 CT is required to produce functional channels, the role of CaM in Kv7 channel function is not well-understood and remains controversial. We and others have shown that LQT mutations impairing CaM binding to Kv7.1 proximal CT affect channel gating, folding, and trafficking (11, 14). Overexpression of CaM in CHO cells was found to robustly reduce currents of Kv7.2, Kv7.4, and Kv7.5 but not reduce those of Kv7.1 and Kv7.3 (21). We showed that Kv7.1 and I_KS currents are stimulated by increases in intracellular Ca²⁺ and markedly inhibited by CaM antagonists (14). Our recent structural study showed that CaM embraces the antiparallel coiled coil helices A and B with an apo C lobe and a calcified N lobe, respectively (16), begging additional elucidation of the mechanism of calcium sensing.

Phosphatidylinositol-4,5-bisphosphate (PIP₂) is required for proper Kv7 channel function (24, 25). Earlier work mapped the site of PIP₂ binding on Kv7.2–4 channels to the long intervening linker connecting helices A and B (26). However, a more recent study indicated that this linker is not required for PIP₂ regulation of Kv7.2 (27, 28). PIP₂ is also necessary for maintaining Kv7.1 channel activity (29). The spontaneous rundown of Kv7.1 or I_KS channels observed in excised patches is markedly reduced by replenishing PIP₂ via metabolic manipulation or exogenous application (29–34). PIP₂ regulates Kv7.1 channel function by increasing the coupling between the voltage sensor domain and the pore region, thereby stabilizing channel open conformation and leading to increased current amplitude, slower deactivation kinetics, and a negative shift in the voltage dependence of activation (29, 32, 33, 35, 36). Recent studies identified clusters of basic residues in the Kv7.1 membrane domain, specifically at the S2–S3 and S4–S5 intracellular linkers and in prehelix A, to be involved in PIP₂ binding (33, 37, 38). Another cluster of basic residues in helix C was also found to be involved in Kv7.1 current rundown after PIP₂ depletion (32, 39). In addition, KCNE1 was found to increase PIP₂ sensitivity 100-fold over that of the Kv7.1 α-subunit alone (31).

Because the Kv7 proximal CT contains sites for modulation by both PIP₂ and CaM, it has been speculated that the CaM and PIP₂ binding modules may overlap physically and functionally (12, 35). However, a direct PIP₂ and CaM interaction in Kv7 channels remains unknown. Here, we reveal the competitive binding of PIP₂ and calcified CaM N lobe to a previously unidentified site in the Kv7.1 helix B. Our data suggest that, after receptor-mediated PIP₂ hydrolysis and subsequent increase in cytosolic Ca²⁺, calcified CaM N lobe binds helix B instead of PIP₂ to prevent excessive I_KS channel depression.

Results

WT CaM and PIP₂ Compete for Binding to Purified Kv7.1 CT Only in the Presence of Ca²⁺.

We first investigated the competition between WT CaM and PIP₂ binding to the Kv7.1 CT. The purified Kv7.1 CT protein was pulled down using PIP₂-coated agarose beads in an incubation medium containing either 0.1 mM Ca²⁺ or 1 mM EGTA (Fig. 1, labeled as +Ca²⁺ or −Ca²⁺, respectively) in the presence of increasing concentrations of WT CaM. To obtain folded recombinant protein, the Kv7.1 CT protein was always coexpressed and copurified with WT CaM (14). The purified recombinant Kv7.1 CT protein (input tetramer ∼ 0.068 µM; 352–622, ∆406–504) included helices A–D, with a deletion of the intervening linker connecting helices A and B (Fig. 1A). We previously showed that this linker deletion impaired neither CaM binding nor channel function (40). Increasing amounts of WT CaM displaced PIP₂ binding to the Kv7.1 CT protein only in the presence of Ca²⁺, with an apparent IC₅₀ of 1.4 µM (Fig. 1 B and C). In contrast, no competition was observed by increasing CaM concentrations in the absence of Ca²⁺ (EGTA) (Fig. 1B). Detectable amounts of CaM were also pulled down in the complex in both the presence and absence of Ca²⁺, likely because of the interactions of calcified CaM N lobe with helix B and apo CaM C lobe with helix A, respectively. To evaluate the reciprocal competition of PIP₂ for CaM binding to Kv7.1 CT, we pulled down purified Kv7.1 CT using CaM-coated agarose beads with increasing concentrations of 1,2-dioctanoyl-sn-glycero-3-phospho-(1′-myo-inositol-4′,5′-bisphosphate) (diC8)-PIP₂ in the presence of either 1 mM Ca²⁺ or 1 mM EGTA. Data indicate that diC8-PIP₂ dose-dependently displaced the binding of CaM to Kv7.1 CT in the presence of Ca²⁺, with an apparent IC₅₀ of 39 μM (Fig. 1 D and E). Much lower amounts of Kv7.1 CT were pulled down in the absence of Ca²⁺. Strikingly, under Ca²⁺-free conditions, no competition was observed, and instead, the presence of PIP₂ significantly increased the binding of CaM to Kv7.1 CT (three- to eightfold) (Fig. 1 D and E). Control experiments showed that WT CaM and CaM mutants do not bind directly to PIP₂ in PIP₂ agarose pulldown. Taken together, the data suggest that Ca²⁺-CaM reduces the binding affinity of PIP₂ to Kv7.1 CT and vice versa.

Fig. 1.

CaM and PIP₂ compete for Kv7.1 CT binding in a Ca²⁺-dependent manner. (A) Schematic depiction of the recombinant His-tagged WT Kv7.1 CT 352–622∆406–504 (Kv7.1 CT) construct used in the indicated pulldown (PD) experiments. This construct was coexpressed and copurified with WT CaM. (B) Representative immunoblots of PIP₂ PD by PIP₂-coated beads of His-tagged Kv7.1 CT in the presence of increasing concentrations of CaM with (Left) 0.1 mM Ca²⁺ (+Ca²⁺) or (Right) 1 mM EGTA (−Ca²⁺). Inputs are shown in the lower two rows. Blots were probed with HRP-conjugated anti-His and anti-CaM antibodies. (C) Dose–response curve with apparent IC₅₀, the concentration of Ca²⁺-CaM necessary to inhibit one-half of maximum PIP₂ binding to Kv7.1 CT (IC₅₀ = 1.4 ± 0.1 µM; n = 4). (D) Representative immunoblots of CaM PD by CaM-coated beads of His-tagged Kv7.1 CT in the presence of increasing concentrations of diC8-PIP₂, with (Left) 1 mM Ca²⁺ (+Ca²⁺) or (Right) 1 mM EGTA (−Ca²⁺). (E, Left) Dose–response curve with apparent IC₅₀, the concentration of PIP₂ necessary to inhibit one-half of maximum Ca²⁺-CaM binding to Kv7.1 CT (IC₅₀ = 39 ± 3 µM; n = 3). (E, Right) Quantification of CaM PD immunoblots signals that were corrected for Kv7.1 CT input normalized to the value without PIP₂ and expressed as ratios (n = 3). *P < 0.05 (unpaired two-tailed t test compared with control without PIP₂).

Calcified CaM N Lobe Competes with PIP₂ Binding to Kv7.1 Helix B.

To determine which calcified lobe of CaM is involved in the competition with PIP₂, we pulled down the Kv7.1 CT protein with PIP₂-coated beads in the presence of Ca²⁺ and increasing concentrations of CaM₁₂, CaM₃₄, and CaM_1,234 mutants that are unable to ligate Ca²⁺ in the N lobe, C lobe, or both lobes, respectively (Fig. 2). Results indicate that only CaM₃₄, which possesses an intact N lobe, was able to potently displace PIP₂ binding (IC₅₀ = 1 µM) (Fig. 2C). CaM₁₂ and CaM_1,234 were unable to compete with PIP₂ binding, suggesting that only the calcified CaM N lobe competes with PIP₂ binding to the Kv7.1 CT. The anti-CaM antibodies exhibited higher affinity for apo-CaM mutants (CaM_12, CaM₃₄, and CaM_1,234) than WT CaM (41) and clearly detected CaM₃₄ and CaM_1,234 but not CaM₁₂ in the PIP₂ pulldown complex in the presence of Ca²⁺ (Fig. 2 B–D). In contrast, CaM₁₂ was identified in the pulldown complex in the absence of Ca²⁺ (Fig. 2D). The detection in the complex of CaM₁₂ in the absence of Ca²⁺ and the absence of CaM₃₄ and CaM_1,234 (but not CaM₁₂) in the presence of Ca²⁺ suggests that only the apo form of CaM C lobe can interact with Kv7.1 helix A and could be identified in the pulldown. The data also suggest that, in the presence of high Ca²⁺, the calcified CaM C lobe (CaM₁₂) dissociates from helix A, being thereby barely detected in the pulldown complex. To examine whether an isolated calcified CaM N lobe is necessary and sufficient to compete with PIP₂ binding to the Kv7.1 CT, we purified the isolated CaM N lobe (amino acids 1–78) and CaM C lobe (amino acids 79–148) and performed PIP₂ pulldown experiments in the presence or absence of Ca²⁺. No competition was observed with the isolated CaM N lobe or the isolated CaM C lobe with or without Ca²⁺ (Fig. S1). This data suggests that the two lobes of CaM and the R74–K75 linker residues are necessary to produce the conformational change that allows the calcified CaM N lobe to compete with PIP₂ binding to the Kv7.1 CT (see below).

Fig. 2.

The calcified N lobe of CaM competes with PIP₂ for Kv7.1 CT binding. (A) Schematic depiction of the recombinant His-tagged WT Kv7.1 CT construct used in the indicated pulldown (PD) experiments. (B) Representative immunoblots of PIP₂ PD of His-tagged Kv7.1 CT in the presence of increasing concentrations of CaM_1,234 with 0.1 mM Ca²⁺. (C, Left) Representative immunoblots of PIP₂ PD of His-tagged Kv7.1 CT in the presence of increasing concentrations of CaM₃₄ with 0.1 mM Ca²⁺. (C, Right) Dose–response curve with apparent IC₅₀, the concentration of CaM₃₄ necessary to inhibit one-half of maximum PIP₂ binding to Kv7.1 CT (IC₅₀ = 1.0 ± 0.2 µM; n = 4). (D) Representative immunoblots of PIP₂ PD of His-tagged Kv7.1 CT in the presence of increasing concentrations of CaM₁₂ with (Left) 1 mM Ca²⁺ (+Ca²⁺) or (Right) 1 mM EGTA (−Ca²⁺). Blots were probed with HRP-conjugated anti-His and anti-CaM antibodies.

Fig. S1.

Isolated CaM N lobe or CaM C lobe does not compete with PIP₂ binding to Kv7.1 CT. PIP₂ pulldown (PD) experiments were performed with purified CaM N lobe (amino acids 1–78) and CaM C lobe (amino acids 79–148) in the presence or absence of Ca²⁺. No competition was observed with the isolated CaM N lobe or the isolated CaM C lobe with or without Ca²⁺.

Several clusters of basic residues have been identified in the Kv7.1 CT as potential PIP₂ binding sites, including sites in prehelix A and helix C (32, 34, 37, 39). Therefore, we purified His-tagged Kv7.1 CT that lacks helices C and D (352–539, ∆406–504) and checked whether the competition of Ca²⁺-CaM to PIP₂ binding still occurred in PIP₂ pulldown assays. Indeed, a robust competition was observed (Fig. 3A). Next, we pulled down Kv7.1 CT lacking prehelix A and helices C and D (361–539, ∆406–504) with PIP₂-coated beads in the presence of increasing concentrations of Ca²⁺-CaM (Fig. 3B). Strong competition was still observed, suggesting that CaM competes with PIP₂ binding only in the presence of calcium and to a site likely localized to helix B, because no clusters of basic residues are found in helix A and consistent with our structural data (16). Then, we used a complementary approach to characterize the interaction of CaM with Kv7.1 proximal CT containing only helices A and B (352–539, ∆406–504). We monitored the fluorescence changes of the dansyl group covalently bound to CaM [dansyl-calmodulin (D-CaM)]. D-CaM reports the binding to target peptides or Ca²⁺ based on a fluorescence increase when the environment of the dansyl groups becomes hydrophobic (42, 43). Dose–response curves were constructed with increasing concentrations of Kv7.1 proximal CT and fixed concentrations of D-CaM (100 nM) in the presence (100 µM free Ca²⁺) or absence of Ca²⁺ (5 mM EGTA) without and with 50 µM diC8-PIP₂ (Fig. 3C and Fig. S2). In the presence of Ca²⁺ and without diC8-PIP₂, an apparent K_d value of 4.69 ± 1.70 µM was obtained, whereas with diC8-PIP₂, a 4.25-fold lower affinity was measured with a K_d value of 19.97 ± 2.24 µM. In the absence of Ca²⁺ (5 mM EGTA) and without diC8-PIP₂, an apparent K_d value of 16.55 ± 0.94 µM was measured, whereas with diC8-PIP₂, a similar affinity was obtained with K_d = 20.50 ± 1.21 µM (Fig. 3C). These results are in line with the pulldown data and suggest that, in the presence of calcium, PIP₂ lowers the binding affinity of CaM for Kv7.1 proximal CT. In the absence of Ca²⁺, the affinity of CaM for Kv7.1 proximal CT is lower, and PIP₂ does not modify it.

Fig. 3.

Calcified CaM and PIP₂ compete for binding to Kv7.1 helix AB module. Representative immunoblots of PIP₂ pulldown (PD) of (A) His-tagged Kv7.1 CT (352–539∆406–504; n = 3) or (B) His-tagged Kv7.1 CT (361–539∆406–504; n = 3) in the presence of increasing concentrations of WT CaM with 0.1 mM Ca²⁺. (C) Fluorescence binding curves of Kv7.1-AB and D-CaM (100 nM) with 0.1 mM Ca²⁺ in the absence (dark blue circles) and presence (light blue circles) of diC8-PIP₂ (50 µM) and without Ca²⁺ (5 mM EGTA) in the absence (red squares) and presence (pink squares) of diC8-PIP₂ (50 µM). D-CaM apparent binding affinity to Kv7.1-AB: with 0.1 mM Ca²⁺ in the absence or presence of 50 µM PIP₂, apparent K_d = 4.69 ± 1.70 µM and K_d = 19.97 ± 2.24 µM (n = 4; P < 0.05); without Ca²⁺ (5 mM EGTA) in the absence or presence of 50 µM PIP₂, apparent K_d = 16.55 ± 0.94 µM and K_d = 20.50 ± 1.21 µM (n = 4).

Fig. S2.

D-CaM fluorescence emission in the presence and absence of Kv7.1 CT proximal CT. Representative experiment showing the D-CaM fluorescence (100 nM D-CaM) emission spectrum in the absence and presence of increasing concentration of Kv7.1 CT proximal CT (helices A and B; 0–60 μM Kv7.1 CT proximal CT).

Functional Impact of the Calcified CaM N Lobe Competition with PIP₂ Binding to Kv7.1 Helix B.

To investigate the functional significance of the calcified CaM N-lobe interaction with the Kv7.1 proximal CT, we recorded the I_KS current from transfected CHO cells after introduction into the pipette solution of purified WT CaM in the presence of Ca²⁺ (3 μM CaM and 5 μM free Ca²⁺) or its absence using 5 mM BAPTA [1,2-bis(o-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid], a fast chelator of free Ca²⁺ (3 μM CaM and 0 μM free Ca²⁺). The I_KS current density was dramatically decreased when the pipette solution contained CaM/BAPTA (Fig. 4A and Table S1). There was also a significant right shift (+21 mV) of the I_KS voltage dependence of activation (Fig. 4A and Table S1). To examine the contribution of BAPTA to this dramatic effect, the same experiment was performed with 5 mM BAPTA alone in the pipette solution. Similar sharp decreases in I_KS current density and right shift (+23.7 mV) in the voltage dependence of activation were observed (Fig. 4A and Table S1). Thus, BAPTA was responsible for the effect, or in other words, the removal of Ca²⁺ was deleterious for I_KS channel gating, although endogenous apo-CaM can be tethered to Kv7.1, a feature in line with our previous work (14, 16). Next, we determined the extent of the shift in the voltage dependence of activation as a function of increasing free Ca²⁺ concentration titrated with BAPTA in the pipette solution. An apparent EC₅₀ value of 96 ± 21 nM free Ca²⁺ produced one-half of the maximal left shift (ΔV₅₀ = −18.6 mV) in the voltage dependence of activation (Fig. 4B) (V₅₀ = 25.0 ± 0.7 mV and V₅₀ = 6.4 ± 1.0 mV for 0 and 5 µM free Ca²⁺, respectively; n = 4–5; P < 0.05). Along the same line, we recently showed that adding to the patch pipette 5 μM free Ca²⁺ with WT CaM or CaM₃₄ but neither CaM₁₂ nor CaM_1,234 noticeably maintained a high I_KS current density (16). These results suggest that preservation of high I_KS current density and channel gating involves the calcified N lobe of CaM, consistent with our structural studies (16).

Fig. 4.

Ca²⁺-CaM and PIP₂ are required for I_KS channel function. (A, Upper) Representative WT I_KS currents recorded from transfected CHO cells with pipette solution containing 3 µM purified recombinant WT CaM with either 5 µM free Ca²⁺ or 5 mM BAPTA. From a holding potential of −90 mV, cells were stepped for 3 s from −60 to +60 mV in 10-mV increments and repolarized for 1.5 s to −60 mV. (A, Lower) Shown are current density–voltage relations and normalized conductance–voltage relations. Activation curves were fitted by a single Boltzmann distribution. (B, Left) Normalized conductance–voltage relations of WT I_KS currents expressed in CHO cells and recorded in the presence of different concentration of internal free Ca²⁺ (0, 30, 100, 300, and 5,000 nM) in the pipette solution titrated with BAPTA. (B, Right) ΔV₅₀ values were normalized and plotted as a function of the log of free Ca²⁺ concentrations (ΔV₅₀ = 1; corresponds to the left shift obtained for free Ca²⁺ = 5,000 nM vs. 0 nM free Ca²⁺). The apparent EC₅₀ determined from a Hill fit was 96 nM, with a hill slope of 1.88. (n = 5). (C) I_KS channels (Kv7.1 and KCNE1) and Dr-VSP were expressed in CHO cells, and currents were recorded with pipette solution containing one of the following conditions: 5 mM BAPTA alone, 5 mM EGTA alone, or 5 mM EGTA and 3 µM purified recombinant WT CaM and then, in the presence of 5 µM free Ca²⁺ (titrated with EGTA), no CaM or 3 µM WT CaM, CaM₁₂, or CaM₃₄. The triple-pulse protocol is shown in Top. (Middle) Representative current traces of WT I_KS before and after activation of Dr-VSP under conditions of pipette solution containing 5 µM free Ca²⁺ without and with 3 µM WT CaM. After pipette content dialysis and rupture of the patch (∼2 min), membrane potential was first stepped for 2 s from −90 to +10 mV to open the I_KS channel (black traces). Then, PIP₂ depletion was induced by activating Dr-VSP using a 200-ms step depolarization from −90 to +90 mV 100 times. To measure the consequences of PIP₂ depletion induced by Dr-VSP activation on I_KS current amplitude, a 2-s step depolarization from −90 to +10 mV was reapplied (red traces). (Bottom) For the various experimental conditions, I_KS current amplitudes before and after Dr-VSP activation were determined and expressed as percentages of current inhibition (n = 7–33). Asterisks indicated significance level by one-way ANOVA and Bonferroni's posttest. **P < 0.01; ***P < 0.001. ns, not significant.

Table S1.

Electrophysiological parameters of WT and mutant I_KS channels under different internal recording solutions

Channel	Patch pipette solution	Current density (pA/pF) at +60 mV	V50 (mV)	Slope (mV)	n
WT IKS	5 μM Free Ca2+	299 ± 35	10.0 ± 2.8	12.8 ± 1.0	13
WT IKS	3 μM CaM + 5 μM free Ca2+	328 ± 42	0.7 ± 3.0	10.8 ± 1.0	7
WT IKS	5 mM BAPTA	128 ± 9*†	25.2 ± 2.4*†	14.9 ± 0.3	4
WT IKS	3 μM CaM + 5 mM BAPTA	103 ± 23***‡	22.4 ± 3.5***‡	12.0 ± 0.8	8
K526N and KCNE1	5 μM Free Ca2+	148 ± 29**†	37.6 ± 5.0***†	17.7 ± 1.7*†	5
K526E and KCNE1	5 μM Free Ca2+	114 ± 24***†	43.6 ± 4.2***†	22.3 ± 2.2***†	6
K527N and KCNE1	5 μM Free Ca2+	175 ± 25*†	20.1 ± 2.8*†	18.1 ± 2.2*†	6
K526N–K527N and KCNE1	5 μM Free Ca2+	35 ± 6***†	36.8 ± 2.5***†	17.4 ± 1.5	6

Significance level (*P < 0.05; **P < 0.01; ***P < 0.001) by unpaired two-tailed t test compared with control (WT I_KS recorded with 3 μM CaM and 5 μM free Ca²⁺) or WT I_KS recorded with 5 μM free Ca²⁺.

^†WT I_KS recorded with 5 μM free Ca²⁺.

^‡Control WT I_KS recorded with 3 μM CaM and 5 μM free Ca²⁺.

To evaluate the functional consequences of the reciprocal competition of CaM and PIP₂, we probed the PIP₂ sensitivity of I_KS currents using the Dr-VSP (Danio rerio voltage-sensitive phosphatase) (44). We challenged I_KS PIP₂ sensitivity by introducing in the pipette solution purified CaM (3 µM) in the absence or presence of Ca²⁺ (5 µM free Ca²⁺). After pipette content dialysis after rupture of the patch (∼2 min), we first stepped CHO cells from −90 to +10 mV for 2 s to open I_KS channels (Fig. 4C, black traces). Then, we induced PIP₂ depletion by activating Dr-VSP using a 200-ms step depolarization from −90 to +90 mV 100 times. To measure the consequences of Dr-VSP–induced PIP₂ depletion, we reapplied a 2-s step depolarization from −90 to +10 mV and compared I_KS current amplitude before and after Dr-VSP activation (Fig. 4C, red traces). In agreement with pulldown assays, the electrophysiological findings show that only the calcified CaM N lobe lessened the decrease in I_KS current amplitude resulting from PIP₂ depletion by Dr-VSP. Indeed, in the presence of Ca²⁺, only purified WT CaM or CaM₃₄ added to the patch pipette significantly attenuated the I_KS inhibition to 46 or 43%, respectively, compared with CaM₁₂ (61%), absence of CaM (65%), BAPTA (73%), EGTA (70%), and EGTA and CaM (72%; n = 9–28; P < 0.001 and P < 0.01) (Fig. 4C). Similar results were obtained when PIP₂ depletion was induced by 10 μM wortmannin, which at this concentration, prevents PIP₂ synthesis by inhibiting phosphatidylinositol 4 (PI4) kinase (34, 45). Treatment of CHO cells with 10 μM wortmannin produced 65% inhibition of the I_KS current (Fig. S3) (n = 6–8; P < 0.001), whereas adding CaM₃₄ to the patch pipette significantly reduced I_KS inhibition to 28%. Addition of CaM₁₂ to the pipette solution produced a 70% inhibition of I_KS current on wortmannin treatment. Altogether, the data indicate that both PIP₂ and Ca²⁺-CaM perform the same function on I_KS channel gating by producing a left shift in the voltage dependence of activation (14, 16, 29). Therefore, when PIP₂ is depleted from the helix B site by Dr-VSP or wortmannin treatments, the current rundown can be significantly dampened by the calcified CaM N lobe that interacts with Kv7.1 helix B in place of PIP₂ and substitute for its function. The rundown attenuation is partial, because PIP₂ is also depleted from other sites (S2–S3 and S4–S5 linkers and prehelix A), in which CaM does not interact and cannot rescue. Reflecting the functional convergence of WT Ca²⁺-CaM and PIP₂, we found that diC8-PIP₂ (100 µM; 5 μM free Ca²⁺) or diC8-PIP₂ (100 µM; 5 μM free Ca²⁺) and WT CaM (3 μM CaM) added to the pipette solution produced a similar left shift in the voltage dependence of activation of WT I_KS currents compared with currents recorded with only 5 μM free Ca²⁺ (Fig. S4). This lack of additivity between Ca²⁺-CaM and PIP₂ in ΔV₅₀ suggests that they interact with a common site to stabilize the channel open state.

Fig. S3.

CaM₃₄ but not CaM₁₂ attenuates the I_KS current inhibition induced by wortmannin. Transfected CHO cells were recorded with pipette solution without or with 10 μM wortmannin and containing 5 μM Ca²⁺ in the absence or presence of 3 μM CaM₁₂ or CaM₃₄. After membrane patch rupture, current rundown was recorded at +30 mV until it reached steady state (∼7–10 min). Shown are the current densities of the different experimental conditions (n = 6–8). *P < 0.05; ***P < 0.001.

Fig. S4.

Lack of additivity between Ca²⁺-CaM and PIP₂. WT I_KS currents were recorded with a pipette solution containing 5 μM free Ca²⁺ (dark blue), 100 µM diC8-PIP₂ and 5 μM free Ca²⁺ (light blue), or 100 µM diC8-PIP₂, 5 μM free Ca²⁺, and 3 μM CaM (violet) and yielded V₅₀ values of +9.1 ± 2.2, −7.0 ± 1.0, and −7.2 ± 1.7 mV, respectively (n = 4–8).

Docking of PIP₂ on Kv7.1 Helix B and Molecular Dynamics Simulation.

To determine exactly where calcified CaM N lobe and PIP₂ compete in the Kv7.1 proximal CT, we docked PIP₂ onto the putative interacting helix B and performed molecular dynamic simulations (MDSs) in the presence of Ca²⁺ using a model based on the crystal structure (16). PIP₂ molecules bear a net negative charge at neutral pH that allows them to engage in electrostatic interactions with positively charged regions of proteins. The structural model, which includes CaM embracing the Kv7.1 helices A and B, has a surface electrostatic potential distribution, which is depicted in Fig. 5A. Although the two lobes of CaM interacting with the Kv7.1 proximal helices exhibited negative surface charge, a gorge region in between the two CaM lobes and along the Kv7.1 helix B bears surface-accessible positive charges (Fig. 5A). Molecular docking revealed that PIP₂ could fit into this gorge, interacting with previously unidentified residues: notably positively charged amino acids from Kv7.1 helix B and CaM that interact with the PIP₂ phosphate head groups. These helix B residues include K526, which interacts with P5 of PIP₂ (distance_O52-HZ3 = 2.0 Å), and K527, which interacts with P5 (distance_O53-HZ3 = 1.7 Å) and a PIP₂ phosphodiester moiety (distance_O11-HZ2 = 1.8 Å). Also, CaM K75, located in the linker that bridges the lobes (distance_O4P-HZ3 = 1.9 Å) and R74 (distance_O43-2HH2 = 1.8 Å), interacts with P4 (Fig. 5A). We next performed MDS (10 ns) using the PIP₂-docked structural model to explore potential changes in the dynamic interactions among residues locally. Results indicate that P5 interacted frequently via H bonding with K526 and K527 in helix B and K75 in the CaM linker. P4 also interacted with CaM R74, and the PIP₂ phosphodiester moiety interacted with K527 (Fig. 5 B and C). This strategic location of PIP₂ at the interface of CaM and Kv7.1 helix B suggests that changes in CaM conformation, such as those occurring on calcification of its N lobe, could disrupt this ternary complex and dislodge PIP₂ from its binding pocket.

Fig. 5.

Molecular docking of PIP₂ on Kv7.1 helix B and MDS. (A) Molecular docking of PIP₂ on the structural model of the crystal structure complex of CaM and Kv7.1 helices A and B. PIP₂ docking is shown with its interactions to helix B and CaM on the background of a surface electrostatic potential distribution using the Pymol 1.3 software. (B) PIP₂ interacts with the residues Lys526 and Lys527 in Kv7.1 and Arg74 and Lys75 in CaM. (C) A schematic of detailed diC1-PIP₂ atom interaction with the CaM–Kv7.1 residues. Interactions that occur for more than 30% of the simulation time in the selected trajectory (5–10 ns) are shown (interactions with >100% indicate that the residue of CaM–Kv7.1 has multiple interactions with the same diC1-PIP₂ atom). (D) Simulation interactions diagram. CaM–Kv7.1 interactions with the diC1-PIP₂ monitored throughout the simulation, including hydrogen bonds and hydrophobic and ionic interactions. The stacked bar charts in C are normalized over the course of the trajectory (values over 1.0 indicate that the residue of CaM–Kv7.1 makes multiple contacts of the same subtype with the diC1-PIP₂).

K75 in CaM and K526 and K527 in Kv7.1 Helix B Form a Site for PIP₂ Interaction That Bears an LQT Mutation.

To validate biochemically the PIP₂ docking and the MDS results, we purposely used the Kv7.1 CT construct that includes all helices A–D and prehelix A (352–622, ∆406–504). Our goal was to examine whether the identified PIP₂–CaM interaction site has a significant impact in the context of the other putative PIP₂ binding sites previously described in the Kv7.1 CT, notably prehelix A and helix C (32, 37–39). Thus, we mutated a cluster of three conserved lysines in helix B (K526, K527, and K528) into asparagine and expressed, purified, and probed them individually for their ability to bind to PIP₂ and CaM using pulldown assays in the presence or absence of Ca²⁺. The single (K526N and K527N) and double (K526N–K527N) mutants showed significantly weaker PIP₂ binding compared with the WT in the presence of Ca²⁺ (Fig. 6A and Fig. S5). Notably, the single K528N mutant, which was not anticipated to interact with PIP₂ by the docking and MDS predictions, bound normally to PIP₂, confirming the specificity of the putative binding site (Fig. S5). In the absence of Ca²⁺, mutants K526N and K527N interacted more avidly with PIP₂ than the WT (Fig. 6A). Similar results were obtained when helix B mutants were probed for binding to CaM-coated beads. In the presence of Ca²⁺, the single and double mutants K526N, K527N, and K526–K527 showed significantly lower binding to CaM than the WT, whereas in the absence of Ca²⁺, they exhibited a more avid interaction with CaM than the WT (Fig. 6B). These data suggest that residues K526 and K527 are important for both PIP₂ and CaM interactions. We then tested the CaM linker mutant K75N on PIP₂ binding to Kv7.1 CT in pulldown assays. In line with PIP₂ docking and MDS results, the CaM mutant K75N lost its ability to compete with PIP₂ binding to the Kv7.1 helix B in the presence of Ca²⁺ (Fig. 6C).

Fig. 6.

K526 and K527 in Kv7.1 helix B and K75 in CaM form a site for PIP₂ and CaM interaction. (A, Upper) Representative immunoblots of PIP₂ pulldown (PD) by PIP₂-coated beads of His-tagged WT Kv7.1 CT 352–622∆406–504 (Kv7.1 CT) and helix B single mutants (K526N and K527N) and double mutants (K526N–K527N) in the presence of (Center) 0.1 mM Ca²⁺ (+Ca2⁺) or (Right) 1 mM EGTA (−Ca²⁺). (A, Lower) PIP₂ PD immunoblot intensities were corrected to input signals normalized to WT Kv7.1 CT PD signal and are expressed as ratios. (B, Upper) Representative immunoblots of CaM PD by CaM-coated beads of His-tagged WT Kv7.1 CT and helix B single mutants (K526N and K527N) and double mutants (K526N–K527N) in the presence of (Center) 0.1 mM Ca²⁺ (+Ca²⁺) or (Right) 1 mM EGTA (−Ca²⁺). (B, Lower) CaM PD immunoblots signals were corrected to inputs, normalized to WT Kv7.1 CT PD signal, and expressed as ratios. For A and B, asterisks indicate significance level (n = 3–6) by unpaired two-tailed t test compared with control (WT Kv7.1 CT). *P < 0.05; **P < 0.01; ***P < 0.001. (C) Representative immunoblots of PIP₂ PD by PIP₂-coated beads of His-tagged WT Kv7.1 CT in the presence of increasing concentrations of mutant CaM K75N done with 0.1 mM Ca²⁺ in the PD buffer (n = 3). (D) I_KS channels (Kv7.1 and KCNE1) and Dr-VSP were expressed in CHO cells, and currents were recorded with pipette solution containing 5 µM free Ca²⁺ in one of the following conditions: no CaM, WT CaM (3 µM), or CaM K75N (3 µM). I_KS current amplitudes before and after Dr-VSP activation were expressed as percentages of current inhibition (n = 5–33). ***P < 0.001.

Fig. S5.

Helix B mutant K528N normally interacts with PIP₂ in contrast to K526N and K527N. (Left) Representative immunoblots of PIP₂ pulldown (PD) by PIP₂-coated agarose beads of His-tagged WT Kv7.1 CT, K526N, K527N, and K528N Kv7.1 mutants in the presence of 0.1 mM Ca²⁺. (Right) PIP₂ PD immunoblot signals were corrected to inputs and normalized to WT Kv7.1 CT PD signal and are expressed as ratios. Asterisks indicate significance level (n = 3–4) by unpaired two-tailed t test compared with control (WT Kv7.1 CT). **P < 0.01; ***P < 0.001.

To validate functionally the PIP₂ docking and the MDS, we transfected CHO cells with KCNE1 and Kv7.1 bearing the single and double mutants of helix B and compared their electrophysiological properties with WT I_KS currents using whole-cell patch-clamp recording. Fig. 7A shows that, compared with WT I_KS, each individual lysine mutant (Kv7.1 K526N and Kv7.1 K527N) as well as the double mutant exhibited a significant right shift of the voltage dependence of activation (Table S1). In addition, Kv7.1 lysine mutants exhibited significantly slower activation and faster deactivation kinetics compared with WT I_KS (T_1/2 act = 0.85 ± 0.10 s and T_1/2 act = 1.26 ± 0.05 s for the WT and K526N, respectively; n = 8–13; P < 0.01; τ_deact = 472 ± 28 ms and τ_deact = 321 ± 19 ms for the WT and K526N, respectively; n = 8–13; P < 0.001).

Fig. 7.

Functional properties of helix B mutants K526N and K527N and the LQT mutation K526E. (A, Upper) Representative current traces of KCNE1 coexpressed with WT Kv7.1; mutants K526N, K527, and K526N–K527N; and the LQT1 mutant K526E. The voltage-clamp protocol is the same as described in Fig. 4, except that higher membrane voltages of up to +100 mV were applied for the mutants because of their depolarizing shift. (A, Lower) Current density–voltage relations and normalized conductance–voltage relations of WT Kv7.1 and mutants expressed with WT KCNE1 (n = 5–13). Activation curves were fitted by a single Boltzmann distribution. (B, Upper) Representative currents traces of the Kv7.1 mutant K526N expressed with WT KCNE1. Currents were recorded with pipette solutions containing 3 µM purified recombinant WT CaM with either 5 µM free Ca²⁺ or 5 mM BAPTA. No PIP₂ was added in the patch pipette. (B, Lower) Current density–voltage relations and normalized conductance–voltage relations of K526N and LQT1 K526E (n = 6). The current densities at +60 mV were as follows: 127 ± 34, 149 ± 37, 95 ± 17, and 100 ± 37 pA/pF for K526E CaM/Ca²⁺, K526E CaM/BAPTA, K526E CaM/Ca²⁺, and K526E CaM/BAPTA, respectively. The V₅₀ values were as follows: 30.3 ± 1.2, 37.2 ± 1.5, 33.2 ± 1.4, and 42.7 ± 3.8 mV for K526N CaM/Ca²⁺, K526N CaM/BAPTA, K526E CaM/Ca²⁺, and K526E CaM/BAPTA, respectively.

Remarkably, the functional importance of the helix B residue K526 is underscored by the existence of the LQT mutation K526E (46). Therefore, we probed the electrophysiological properties of CHO cells transfected with KCNE1 and Kv7.1 K526E. The LQT mutant produced a marked right shift (ΔV₅₀ = +33.7 mV) in the voltage dependence of activation and a significantly lower current density (Fig. 7A and Table S1). To explore further the functional importance of residue K526 in the Kv7.1 helix B, we recorded CHO cells transfected with KCNE1 and Kv7.1 K526N or Kv7.1 K526E. Similar to the experimental paradigm presented in Fig. 4A for WT I_KS, we introduced into the pipette solution purified WT CaM in the presence of Ca²⁺ (3 μM CaM and 5 μM free Ca²⁺) or 5 mM BAPTA (0 μM free Ca²⁺). In contrast to WT I_KS, the currents generated by I_KS K526N and I_KS LQT mutant K526E exhibited features seen in BAPTA, which cannot be recovered by Ca²⁺-CaM, with no significant changes in current densities and voltage dependence of activation (compare Fig. 4A with Fig. 7B). These mutants are insensitive to gating modulation by Ca²⁺-CaM.

By cotransfecting Dr-VSP and WT I_KS (Kv7.1 and KCNE1) and using the protocol described in Fig. 4B, we examined the functional importance of the CaM mutant K75N, which was unable to compete with PIP₂ binding to the Kv7.1 helix B in the presence of Ca²⁺ (Fig. 6 C and D). Dr-VSP activation led to a 65% inhibition of I_KS currents (Ca²⁺ in pipette solution), whereas the presence of purified WT Ca²⁺-CaM added to the patch pipette significantly attenuated I_KS current inhibition to 46%. In line with the docking, MDS, and PIP₂ pulldown data, when the pipette solution contained the Ca²⁺-CaM mutant K75N, Dr-VSP produced a 68% inhibition of I_KS currents (Fig. 6D), suggesting that CaM K75N does not properly interact with Kv7.1 helix B to rescue the current rundown arising from PIP₂ depletion.

To evaluate functionally the weaker PIP₂ binding of the helix B mutant K526N assessed in pulldown assays in the presence of Ca²⁺, we determined in transfected CHO cells the sensitivity of WT I_KS and K526N I_KS to PIP₂ by dialyzing cells with increasing concentrations of diC8-PIP₂ in the pipette solution. Two minutes after membrane patch rupture, current run up was recorded at +30 mV until it reached steady state. To construct the dose–response curves, the average current stimulation for each diC8-PIP₂ concentration was normalized to the maximum value obtained at 100 µM diC8-PIP₂ for WT I_KS and K526N (Fig. 8A). Although WT I_KS showed an apparent EC₅₀ of about 13 µM, in agreement with previous work (31), the I_KS mutant K526N exhibited a much lower affinity, with an underestimated apparent EC₅₀ of more than 200 µM (Fig. 8A) (n = 8–10). For comparison, the apparent EC₅₀ values of diC8-PIP₂ for Kv7.2, Kv7.3, and Kv7.4 homomeric channels were previously found to be 111, 6, and 154 μM, respectively (26). No Hill fit could be obtained, because above 100 µM diC8-PIP₂ added to the pipette solution, the patch became unstable. Nevertheless, introduction of the highest diC8-PIP₂ concentration (100 µM) into the patch pipette significantly left-shifted the voltage dependence of activation of WT I_KS and mutant I_KS K526N (Fig. 8B) (ΔV₅₀ = −16.1 mV and ΔV₅₀ = −13.3 mV, respectively). To further probe the lower PIP₂ sensitivity of the I_KS mutant K526N compared with the WT, the kinetics of current decline were analyzed in the absence or presence of Dr-VSP. As implemented in previous studies (39), the rate of current decline is expected to reflect channel affinity for PIP₂. Thus, assuming that the rate of PIP₂ resynthesis is roughly similar, the faster the rate of current decrease, the lower the channel affinity for PIP₂. Indeed, results indicate that the mutant I_KS K526N exhibited significantly faster current decline and larger maximal inhibition compared with the WT (Fig. 8C) (at 5 s, 8 and 50% inhibition for WT I_KS and mutant K526N, respectively; n = 7–10; P < 0.01). These results suggest that mutation of the positively charged residue K526 severely compromised PIP₂ and calcified CaM N-lobe interaction with helix B and thereby, markedly impaired I_KS channel gating (Figs. 4A, 6 A and B, 7, and 8).

Fig. 8.

PIP₂ sensitivity of WT I_KS and K526N mutant currents and model of CaM–PIP₂ interaction. (A) The sensitivity of WT I_KS and K526N I_KS to PIP₂ was determined by dialyzing cells via the pipette solution with various concentrations of diC8-PIP₂. Two minutes after the membrane patch rupture, the current run up was recorded at +30 mV until it reached steady state. To construct the dose–response curves, the average current stimulation for each diC8-PIP₂ concentration was normalized to the maximum value obtained at 100 µM diC8-PIP₂ for WT I_KS and K526N. The curve was fitted to log (agonist) vs. response with variable slope. WT I_KS showed an apparent EC₅₀ of 13 µM, whereas the I_KS mutant K526N exhibited a much lower affinity, and no fit was possible (n = 8–10). (B) Normalized conductance–voltage relations of WT I_KS and K526N mutant were obtained using the same voltage-clamp protocol as described in Fig. 4 in the absence or presence of 100 μM diC8-PIP₂ in the pipette solution (n = 4–13). Activation curves were fitted by a single Boltzmann distribution. The V₅₀ values were as follows: 9.1 ± 2.2 and 26.8 ± 1.6 mV for WT I_KS and K526N I_KS mutant in the absence of PIP₂, respectively, and −7.0 ± 1.0 and 13.3 ± 2.2 mV for WT I_KS and K526N I_KS mutant in the presence of 100 µM diC8-PIP₂ in the pipette solution, respectively. (C) WT I_KS and K526N I_KS mutant sensitivity to voltage-sensitive phosphatase Dr-VSP recorded using a triple-pulse protocol, where membrane potential is first stepped for 1.5 s from −90 to +10 mV to open the channel followed by a 2-s +100-mV voltage step to activate Dr-VSP and then, return for 1.5 s to +10 mV. This protocol was repeatedly applied every 5 s for 1 min (n = 5–10). Kinetics of current decline for WT and mutant channels are shown in the absence or presence of Dr-VSP. Current decline was quantified as the fractional current change from the amplitude obtained +10 mV at time t divided by that measured at time 0. The K526N I_KS mutant exhibited a significantly faster current decline and larger maximal inhibition compared with the WT (at 5 and 10 s, 8 and 23% inhibition for WT I_KS vs. 50 and 48% inhibition for mutant K526N, respectively; n = 7–10; P < 0.01). (D) Cells were dialyzed with (5 µM free Ca²⁺) or without Ca²⁺ (5 mM EGTA) in the absence or presence of PIP₂ (30 µM). After membrane patch rupture, current run up was recorded at +30 mV until it reached steady state, and current density was determined for each condition. *Significance calculated by unpaired two-tailed Student's t test with P < 0.01. (E) Model: at resting cytosolic Ca²⁺, a ternary complex exists, where the Kv7.1 helix B can interact with and accommodate PIP₂ and the calcified CaM N lobe, whereas the Kv7.1 helix A binds the CaM apo C lobe. After receptor-mediated PIP₂ depletion and increased cytosolic Ca²⁺, we suggest that the calcified CaM C lobe unbinds helix A and that the calcified CaM N lobe displaces PIP₂ from its binding site in helix B to limit the decrease in I_KS channel activity arising from PIP₂ hydrolysis.

Finally, we examined the functional relevance of the stronger PIP₂ binding of the helix B mutant K526N in pulldown assays in the absence of Ca²⁺ by dialyzing cells with 30 μM PIP₂. As described above, after membrane patch rupture, current run up was recorded at +30 mV until it reached steady state. In the presence of Ca²⁺, 30 μM PIP₂ expectedly increased by 2.36-fold the WT I_KS current density and barely affected that of mutant K526N. Strikingly, in the absence of Ca²⁺, PIP₂ stimulated by 2.43-fold the current density of K526N without altering that of WT I_KS (Fig. 8D). These data nicely matched the PIP₂ pulldown described in Fig. 6A, Right and suggest that, in the absence of Ca²⁺, the helix B mutants (N526 and N527) undergo a conformational change leading to better CaM binding (Fig. 6B), which also increases their avidity for PIP₂ binding (Fig. 6A). We suggest that the CaM linker residues R74 and K75 can provide the positive charges necessary for strong PIP₂ interaction. Another nonmutually exclusive possibility is that, under zero Ca²⁺ conditions, the positively charged residues K358 and R360 in prehelix A and/or residues R555, K557, and R562 in helix C can provide the positive charges that tighten the PIP₂ binding to Kv7.1 mutants K526N/K527N.

Discussion

The Kv7 proximal CT contains sites for modulation by PIP₂ and CaM, which raises the question of whether the CaM and PIP₂ binding modules overlap physically and functionally (12, 24, 35). Although Kwon et al. (47) found that PIP₂ reduced Ca²⁺-CaM binding to several ion channels, including TRPC1, TRPC5–7, TRPV1, Kv7.1, and Cav1.2, the mechanisms by which they may compete with each other remain unclear, and a direct PIP₂–CaM interaction has not been previously explored in Kv7 channels. In this work, we reveal the existence of a previously unidentified site in Kv7.1 helix B, where competitive binding of PIP₂ and calcified CaM N lobe takes place. Reciprocal CaM and PIP₂ pulldown experiments showed Ca²⁺-dependent competitive binding of PIP₂ and CaM to purified Kv7.1 CT, with apparent IC₅₀ values of 39 and 1.1 µM, respectively (Fig. 1).

The apparent PIP₂ affinity determined here in vitro is slightly lower than but of the same order of magnitude as the one that we measured by electrophysiology for the WT I_KS current (13 µM) (Fig. 8), which is also in line with previous values obtained from inside-out patch recording (∼5–8 µM) (31, 32). PIP₂ affinity is a crucial determinant of ion channel sensitivity to cellular PIP₂ dynamics. If calculated as global cellular concentration, PIP₂ is about 10 μM (48); however, superresolution microscopy studies showed that PIP₂ molecules can cluster and segregate into distinct nanoscale lipid domains of the plasma membrane (49). Thus, the moderate affinity of WT I_KS channels for PIP₂ is within the dynamic range of PIP₂ availability in the cell.

Although the resting concentration of free calcified CaM is likely to be limiting (∼50 nM), studies suggest that free apo-CaM is accessible at the micromolar to submicromolar range (50) and could potentially become calcified after receptor-mediated increase in cytosolic Ca²⁺. Consequently, the apparent micromolar affinity of calcified CaM necessary to displace PIP₂ binding from purified Kv7.1 CT fits well within the dynamic range of Ca²⁺-CaM concentration generated from free apo-CaM on increased intracellular Ca²⁺. In line with our earlier structural findings (16), the potent reduction of I_KS current density and the right shift of the voltage dependence of activation after introduction of BAPTA-CaM or BAPTA alone into the pipette solution show the strict requirement of calcified CaM for normal channel gating (Fig. 4).

The PIP₂ pulldown performed with various purified Kv7.1 CT constructs indicates that the competition of Ca²⁺-CaM with PIP₂ binding occurs in helix B and strictly depends on calcium (Figs. 1, 2, and 3). Prehelix A, helix C, and helix D are not involved in the PIP₂–Ca²⁺–CaM competitive interaction. As shown in our previous work (16), helix B is well-suited to be a site for PIP₂ binding because of its presumed proximity to the inner leaflet of the plasma membrane. Nevertheless, it is likely that, in addition to helix B, several PIP₂ molecules bind to Kv7.1 channels at multiple contact sites or migrate to different places, including those located at the S2–S3 and S4–S5 intracellular linkers, prehelix A, and helix C, to achieve a specific function (33, 37, 38). For example, it was shown that PIP₂ preferentially interacts with the S4–S5 linker in the Kv7.2 open state, whereas it contacts the S2–S3 loop in the closed state (51, 52).

In this work, several lines of evidence indicate that the calcified CaM N lobe competes with PIP₂ binding to the Kv7.1 helix B. (i) Despite the presence of Ca²⁺, CaM₁₂ and CaM_1,234, which are unable to bind Ca²⁺ in the N lobe or both lobes, respectively, do not compete with PIP₂ binding. In contrast, CaM₃₄, which cannot bind Ca²⁺ in the C lobe but can calcify its N lobe, potently displaced PIP₂ binding (Fig. 2). (ii) PIP₂ lowers the binding affinity of Kv7.1 proximal CT for CaM in the presence of Ca²⁺ but does not in its absence as measured by D-CaM fluorescence (Fig. 3C). (iii) In the presence of Ca²⁺, addition to the pipette solution of purified WT CaM or CaM₃₄ but not CaM₁₂ attenuated the decrease in I_KS current amplitude resulting from PIP₂ depletion by Dr-VSP. (iv) There is a lack of additivity between Ca²⁺-CaM and PIP₂ in producing a left shift in the voltage dependence of I_KS channel activation when introduced in the pipette solution (Fig. S4).

These structural and physiological features of I_KS channels are reminiscent to those described for Ca²⁺-activated small conductance SK2 K⁺ channels (53–56). In SK2 channels, the interactions of the CaM binding module at the proximal CT with the apo C lobe and the linker of CaM are thought to be responsible for the constitutive association between CaM and the channels, whereas the interaction with the calcified N lobe accounts for channel gating (53–56).

The retrospective validation of PIP₂ docking and MDS points to the pivotal importance of helix B residues K526 and K527 and confirms experimentally their importance for both PIP₂ and CaM binding interactions (Fig. 5). (i) In the presence of Ca²⁺, PIP₂ and CaM bound significantly weaker to K526N, K527N, and K526N–K527N than the WT in pulldown experiments (Fig. 6). (ii) K526 and K527 had a substantial impact on PIP₂ binding, even in the context of other potential PIP₂ binding sites (Fig. 6) (32, 37–39). (iii) In contrast to WT CaM, the CaM mutant K75N was unable to compete with PIP₂ binding to Kv7.1 CT in the presence of Ca²⁺ and rescue the current rundown arising from PIP₂ depletion, suggesting that it does not properly interact with Kv7.1 helix B (Fig. 6 C and D). (iv) K526N, K527N, and K526N–K527N displayed a right shift in the voltage dependence of activation and lower current densities (Fig. 7A). (v) The mutant K526N showed a lower affinity for diC8-PIP₂ (>200 μM) and a faster rate of current decline compared with WT after Dr-VSP–induced PIP₂ depletion (Fig. 8 A and C) and in BAPTA experiments, was insensitive to gating modulation by Ca²⁺-CaM in contrast to the WT (Fig. 7B). (vi) The LQT mutant K526E revealed a severely impaired channel function with a right shift in the voltage dependence of activation, reduced current density, and insensitivity to gating modulation by Ca²⁺-CaM (Fig. 7B).

How does calcified CaM N lobe displace PIP₂ from its binding site in helix B after an increase in cytosolic Ca²⁺? In our structural models, we see that a rich network of interactions exists between several CaM residues and PIP₂ and that the second EF hand of CaM N lobe (EF2) is closer to PIP₂ than the first one (EF1) (Fig. S6). In the CaM N lobe, residues N53 and E54 can engage into hydrogen bonding interactions with an oxygen of the PIP₂ fatty moiety. In the CaM linker, the guanidinium side chain of R74 can interact with the P4 PIP₂ head group, whereas the ε-amine side chain of K75 can contact the P5 PIP₂ head groups. In fact, the PIP₂ binding site resides in a strategic location at the interface of CaM and Kv7.1 helix B, and it is plausible to assume that binding of Ca²⁺ to CaM produces a conformational change that disrupts the ternary complex. Under these conditions, the calcified CaM C lobe dissociates from helix A, and the calcified N lobe dislocates PIP₂ from its helix B site (Fig. S6). Such a critical location of PIP₂ is reminiscent of that described for Ca²⁺-activated SK2 K⁺ channels, where the PIP₂ binding site is located at the interface of CaM and the SK2 CT (57).

Fig. S6.

Network of interactions between CaM residues and PIP₂. The structure of CaM and Kv7.1 AB is pictured using the Pymol 1.3 software. CaM is shown in deep purple, whereas Kv7.1 helices A and B are colored in cyan and green, respectively. The CaM N lobe is calcified, and the residues of EF2 that chelate calcium are shown as spheres. In the CaM N lobe, residues N53 and E54 main and side chains can engage in hydrogen bonding interactions (d = 3.2 and 2.1 Å, respectively) with an oxygen of the PIP₂ fatty moiety. In the CaM linker, the guanidinium side chain of R74 can interact with the P4 PIP₂ head group (d = 1.8 Å), whereas the ε-amine side chain of K75 can contact the P5 PIP₂ head groups (d = 2.8 Å).

What is the physiological significance of this CaM–PIP₂ competitive interaction for I_KS channel function? Recent studies suggest an interplay between CaM and PIP₂, notably in Kv7.2 channels. A rise in intracellular Ca²⁺ was hypothesized to induce a change in the mode of CaM binding to Kv7.2, leading to the reduction of channel affinity for PIP₂ and subsequent current suppression (58). Similarly, remote coiled coil formation at helix D was proposed to induce different CaM interaction modes, each conferring different PIP₂ dependency to Kv7.2 channels (59). In addition, casein kinase 2-mediated phosphorylation of CaM was shown to strengthen its binding to Kv7.2 channel, causing resistance to PIP₂ depletion, and increase in Kv7.2 current amplitude (60). However, the mechanisms underlying the above CaM–PIP₂ interplay for Kv7.2 remain unclear. Does it result from a direct competition? If so, at which site? Here, we show that the calcified CaM N lobe and PIP₂ compete for the same binding site molded by K526 and K527 at the Kv7.1 helix B to stabilize the channel open state. Our data show that both PIP₂ and Ca²⁺-CaM perform the same function on I_KS channel gating by producing a left shift in the voltage dependence of activation. We assume that, at resting cytosolic Ca²⁺ (Fig. 8D), a ternary complex exists, where the Kv7.1 helix B can interact with and accommodate PIP₂ and the calcified CaM N lobe, whereas the Kv7.1 helix A binds the CaM apo C lobe. A similar mode of CaM interaction was suggested for Nav1.2 sodium channels (61). However, PIP₂ levels can significantly change in cardiomyocytes by activation of phospholipase C via Gq protein-coupled receptors, such as M1-muscarinic receptors (62). After receptor-mediated PIP₂ depletion and increased cytosolic Ca²⁺ (Fig. 8D), we suggest that the calcified CaM C lobe unbinds helix A, whereas the calcified CaM N lobe interacts with helix B in place of PIP₂ to limit the decrease in I_KS channel activity that arises from PIP₂ hydrolysis. Failure to achieve this vital function, as with the LQT mutation K526E, leads to compromised I_KS channel gating.

Experimental Procedures

Human Kv7.1 and KCNE1 were cloned into pCDNA3 vector to allow eukaryotic expression. The mutations in Kv7.1 (K526N, K526N, K526E, K527N, and K526N–K527N) were introduced using the PCR-based Quikchange Site-Directed Mutagenesis (Stratagene). Details of expression and purification of Kv7.1 CT–CaM complexes, D-CaM fluorescence assay, pulldown experiments, cell culture and transfection, electrophysiology, molecular docking, MDSs, and data analyses are in SI Experimental Procedures.

SI Experimental Procedures

Expression and Purification of Kv7.1 CT–CaM Complexes.

The various Kv7.1 CTs, including 352–622, ∆406–504; 352–539, ∆406–504; and 361–539, ∆406–504 and bearing deletion of the intervening loop between helix A and B (residues 406–504), were cloned into the multiple cloning site I (MCS I) of the pET-Duet vector (Novagen) located downstream of an His tag and a Tobacco etch virus protease site. CaM was in MCS II. Kv7.1 CT constructs were copurified with WT CaM as previously described (15, 40). The mutations K526N, K527N, K526N–K527N, and K528N on Kv7.1 CT helix B were introduced in the construct 352–622∆406–504. WT CaM and CaM mutants were cloned into the MCS II of the pET-Duet vector (Novagen). MCS I was empty, and CaM was expressed without any tags. WT and CaM mutants were purified by lysis, heating of the soluble fraction to 70 °C for 5 min, and then, centrifugation for 45 min. Protein was diluted twofold with buffer M [50 mM Tris⋅HCl, pH 7.5, 10% (wt/vol) glycerol] and loaded onto a Q-Sepharose Column (GE Healthcare) preequilibrated with buffer Q. The column was washed with a linear gradient of buffer Q containing 50–800 mM NaCl. Fractions were pooled and applied to a hydroxyapatite column preequilibrated with 200 mM NaCl. The column was washed with a linear gradient of buffer H containing 100 mM NaCl and 0–400 mM potassium phosphate. Fractions were pooled and applied to a Superdex 75 Hi-Prep Gel Filtration Column (GE Healthcare) preequilibrated with buffer F (20 mM Tris⋅HCl, pH 7.5, 200 mM NaCl).

D-CaM Fluorescence Assay.

Primary amines of the purified WT CaM were labeled with 5-(dimethylamino)naphthalene-1-sulfonyl chloride (Molecular Probes) as described (43). D-CaM at 100 nM was incubated with different protein Kv7.1-AB concentrations in the absence or presence of 50 μM diC8-PIP₂ with (0.1 mM Ca²⁺) or without Ca²⁺ (5 mM EGTA) at room temperature (RT) for 10 min in fluorescence buffer (10 mM Hepes, pH 7.4, 150 mM NaCl, 0.1 Ca²⁺). Fluorescent measurements were done at RT using a Horiba Fluorolog-3 Spectrofluorimeter. Spectral data (400–600 nm) were collected using a 340-nm excitation wavelength (5-nm slit width throughout). Normalized Kv7.1-AB titration curves were generated by measuring peak fluorescence signals (average of ±5 nm for noise reduction) after normalization using Fnorm = (F − F0)/(Fmax − F0), where F is the measured fluorescence value for each concentration, Fmax is the maximum measured fluorescence, and F0 is the measured fluorescence without adding the Kv7.1-AB protein. Data were fitted to one site-specific binding with Hill slope function of the form

$$\text{Y}=\frac{\text{Bmax}\times {\text{X}}^{\text{h}}}{\left(K{\text{d}}^{\text{h}}+{\text{X}}^{\text{h}}\right)},$$

where Y is the normalized fluorescence Fnorm, Bmax is the maximal normalized fluorescence, X is the concentration of the purified Kv7.1 proximal CT, K_d is the Kv7.1 proximal CT concentration needed to achieve a half-maximum binding exchange at equilibrium expressed in the same units as X, and h is the Hill slope.

Pulldown Experiments.

Pulldown experiments were performed as previously published (39). In brief, for PIP₂ pulldown, 5 μg purified WT or mutant His-tagged Kv7.1 CT proteins were incubated with equal amounts of PIP₂-coated agarose beads (Echelon Biosciences) for 2 h at RT in binding buffer containing 10 mM Hepes, pH 7.4, 150 mM NaCl, and 0.25% Igepal with either 0.1 mM Ca²⁺ or 1 mM EGTA. After sample centrifugation at 600 × g, three washes in binding buffer were performed and followed by boiling with 4× Laemmli sample buffer at 95 °C. Samples were subjected to SDS/PAGE followed by Western blotting. Input samples were taken for quantification purposes. Detection was performed using HRP-conjugated anti-His antibody (Roche) or anti-CaM antibodies (Millipore) and ECL (Millipore). For CaM pulldown, 5 µg purified WT or mutant His-tagged Kv7.1 CT proteins were incubated overnight with equal amounts of CaM-agarose beads (Sigma) in pulldown buffer containing 150 mM NaCl, 50 mM Tris, pH 7.5, protease mixture inhibitor (Sigma), and 1% Triton with either 1 mM Ca²⁺ or 1 mM EGTA. Beads were washed and eluted with SDS sample buffer. Equal amounts of lysate proteins (input) and pulldown samples were resolved by 8% (wt/vol) SDS/PAGE and subjected to Western blotting as described for PIP₂ pulldown.

Cell Culture and Transfection.

CHO cells were grown in DMEM supplemented with 2 mM glutamine, 10% (wt/vol) FCS, and antibiotics. In brief, 40,000 cells seeded on poly-d-lysine–coated glass coverslips (13 mm in diameter) in a 24-multiwell plate were transfected with pIRES-CD8 (0.3 µg) as a marker for transfection with 0.5 µg WT or mutant Kv7.1 and 1 µg KCNE1. Transfection was performed using 3.6 µL X-tremeGENE 9 (Roche) according to the manufacturer’s protocol. For electrophysiology, transfected cells were visualized ∼40 h after transfection using the anti-CD8 antibody-coated beads method as described previously (39).

Electrophysiology.

Recordings were performed using the whole-cell configuration of the patch-clamp technique. Signals were amplified using an Axopatch 200B Patch-Clamp Amplifier (Axon Instruments), sampled at 5 kHz, and filtered at 2.4 kHz via a four-pole Bessel Low-Pass Filter. Data were acquired using pClamp 10.5 software in conjunction with a DigiData 1440A Interface. The patch pipettes were pulled from borosilicate glass (Harvard Apparatus), with a resistance of 3–7 Mohm. The intracellular pipette solution contained 130 mM KCl, 5 mM K2-ATP, 5 mM EGTA (or 5 mM BAPTA when indicated), 10 mM Hepes, pH 7.3 (adjusted with KOH), and CaCl₂ (as needed for different values of free Ca²⁺ concentration according to MAXCHELATOR software), with sucrose added to adjust osmolarity to 290 mosM. The external solution contained 140 mM NaCl, 4 mM KCl, 1.8 mM CaCl₂, 1.2 mM MgCl₂, 11 mM glucose, and 5.5 mM Hepes adjusted with NaOH to pH 7.3 (310 mosM). Series resistances were compensated (75–90%) and periodically monitored.

Molecular Docking.

An automatic molecular docking program, AUTODOCK4.2 (63), was used for the docking of PIP₂ into the structure of the CaM–Kv7.1 complex (Protein Data Bank ID code 4V0C). The partial charges of PIP₂ were obtained from ab initio quantum chemistry calculations at the HF/6–31+G* level using the CHELPG charge-fitting scheme (64) of the GAUSSIAN 98 program as described previously (65). We used the PIP₂ analog diC1, which replaces the two long acyl tails of PIP₂ by two methyl groups, to reduce the size and flexibility of the molecule for docking simulations. The grid potential maps were generated for the CaM–Kv7.1 complex using carbon, hydrogen, nitrogen, oxygen, and phosphor elements sampled on a uniform grid containing 80 × 80 × 80 points 0.375 Å apart. The center of the grid box was set to a cluster of basic residues: K526, K527, K528, R74, K75, and K77. The Lamarckian Genetic Algorithm was selected to identify the binding conformations of the ligands. The side chains for residues K526, K527, R74, and K75 were allowed to be flexible; 100 docking simulations were performed, and the final docked diC1-PIP₂ configurations were selected on the basis of docked binding energies and relative orientation of the diC1-PIP₂ to the CaM–Kv7.1 complex.

MDSs.

Complexes of CaM–Kv7.1 and diC1-PIP₂ were immersed in 0.15 M KCl solvation. The optimized potentials for liquid simulations (OPLS) force field was used for CaM–Kv7.1 and diC1-PIP₂, and the simple point charge (SPC) model was used for water molecules. All of the MDSs were accomplished by Desmond 3.1 (66). After relaxed by Desmond standard NPT [substance (N), pressure (P), and temperature (T)] relaxation protocol, complexes were subjected to 10-ns MDSs without any restraint at a constant temperature of 298 K. In total, 400 structures were generated during the MDSs. The CaM–Kv7.1 and diC1-PIP₂ interaction was analyzed using the simulation interactions diagram module of the Maestro program (Schrödinger, LLC) based on the last 5 ns of the MDS trajectory.

Data Analyses.

Data analysis was performed using the Clampfit program (pClamp 10.5; Axon Instruments), Microsoft Excel (Microsoft), and Prism 5.0 (GraphPad Software, Inc.). Leak subtraction was performed offline using the Clampfit program of the pClamp 10.5 software. Chord conductance (G) was calculated by using the following equation: G = I/(V − Vrev), where I corresponds to the current amplitude measured at the end of the pulse, and Vrev is the calculated reversal potential assumed to be −90 mV in CHO cells. G was estimated at various test voltages (V) and then, normalized to a maximal conductance value, Gmax. Activation curves were fitted by one Boltzmann distribution: G/Gmax = 1/{1 + exp[(V₅₀ − V)/s]}, where V₅₀ is the voltage at which the current is half-activated, and s is the slope factor. All data were expressed as mean ± SEM. For electrophysiology, if not indicated otherwise, statistically significant differences were assessed by unpaired two-tailed Student’s t test.