Calmodulin Is Essential for Cardiac IKS Channel Gating and Assembly
Impaired Function in Long-QT Mutations
The slow IKS K+ channel plays a major role in repolarizing the cardiac action potential and consists of the assembly of KCNQ1 and KCNE1 subunits. Mutations in either KCNQ1 or KCNE1 genes produce the long-QT syndrome, a life-threatening ventricular arrhythmia. Here, we show that long-QT mutations located in the KCNQ1 C terminus impair calmodulin (CaM) binding, which affects both channel gating and assembly. The mutations produce a voltage-dependent macroscopic inactivation and dramatically alter channel assembly. KCNE1 forms a ternary complex with wild-type KCNQ1 and Ca2+-CaM that prevents inactivation, facilitates channel assembly, and mediates a Ca2+-sensitive increase of IKS-current, with a considerable Ca2+-dependent left-shift of the voltage-dependence of activation. Coexpression of KCNQ1 or IKS channels with a Ca2+-insensitive CaM mutant markedly suppresses the currents and produces a right shift in the voltage-dependence of channel activation. KCNE1 association to KCNQ1 long-QT mutants significantly improves mutant channel expression and prevents macroscopic inactivation. However, the marked right shift in channel activation and the subsequent decrease in current amplitude cannot restore normal levels of IKS channel activity. Our data indicate that in healthy individuals, CaM binding to KCNQ1 is essential for correct channel folding and assembly and for conferring Ca2+-sensitive IKS-current stimulation, which increases the cardiac repolarization reserve and hence prevents the risk of ventricular arrhythmias.
KCNQ channels represent a family of voltage-gated K+ channels (Kv7) that plays a major role in brain and cardiac excitability.1,2 Mutations of human KCNQ genes lead to severe cardiovascular and neurological disorders such as the cardiac long-QT syndrome (LQT) and neonatal epilepsy. Coassembly of KCNQ1 with KCNE1 β subunits produces the IKS-current that is crucial for repolarization of the cardiac action potential.3–5
The cytoplasmic KCNQ C-termini were shown to feature 4 α helices.6 We previously identified the last α helix of the C terminus (helix D, aa.589–620) as a region important for the tetrameric assembly of KCNQ1 α subunits.7 This region also binds Yotiao, an A-kinase–anchoring protein that targets PKA on the IKS channel complex.8 The first 2α helices of KCNQ1–5 form a calmodulin-binding domain (CBD), including an IQ motif that mediates Ca2+-free calmodulin (apoCaM) binding.6,9 Although KCNQ channels bind calmodulin (CaM), the role of CaM in channel function remains controversial. Recent studies found a role for CaM as a Ca2+-sensor of KCNQ2/4/5 channels,10,11 whereas others suggested a role in channel assembly.9 So far, no information has been available about the interaction of calmodulin with cardiac IKS channels and its pathophysiological impact to KCNQ1-related LQT channnelopathies. Here, we show that LQT mutations located near the IQ motif of KCNQ1 C terminus impair CaM binding, alter channel assembly, stabilize inactivation, and decrease current density. In healthy individuals, CaM binding to KCNQ1 is necessary for proper channel assembly and for conferring Ca2+-sensitive stimulation of the cardiac IKS-current, which is essential for maintaining the repolarization reserve that prevents excessive action potential prolongation.
Materials and Methods
The molecular interactions between CaM and the various KCNQ1 channel constructs (wild-type [WT] and LQT mutants) were investigated using complementary methods, including the yeast 2-hybrid screen, CaM-agarose pulldown assays, bacterial expression systems, in vitro cotranslations, and coimmunoprecipitations techniques. The steady-state expression levels of the channel proteins were assessed by cell surface biotinylation experiments. The functional analysis of channel activities was studied using the whole-cell patch-clamp technique in transfected Chinese hamster ovarian (CHO) cells and the whole-cell 2-electrode and inside-out macro-patches in micro-injected Xenopus oocytes. The reagents, experimental protocols, and data analyses are described in detail in the online-only data supplement available at http://circres.ahajournals.org.
The KCNQ1 C Terminus Interacts With CaM in the Presence and Absence of Ca2+
Figure 1A shows that KCNQ1 and CaM interact, as demonstrated by immunoprecipitation from lysates of CHO cells stably expressing KCNQ1. To delineate the CaM interacting region in KCNQ1 C terminus, we used the yeast 2-hybrid system (Figure 1B). Deletion of helix A (CT390–676 and CT390–620) showed a weak yeast growth, which indicates that as for KCNQ2–5,6,9,11 this domain is important for CaM interaction. Deletion of the proximal C terminus up to helix A (CT361–676 and CT370–676) led to normal yeast growth. Similarly, CaM interaction was normal when truncating both extremities of the C terminus (CT354–620 and CT370–620). Unexpectedly, deletion of the C-terminal end (CT354–589), including helix D, which corresponds to the putative assembly domain, did not yield yeast growth (Figure 1B). Myc-tagged KCNQ1 C-terminal constructs and CaM were either translated separately or cotranslated. Immunoprecipitation showed that CaM bound directly to the KCNQ1 C-terminus both in the absence and presence of Ca2+ (Figure 1C and 1D). In line with the yeast 2-hybrid data, binding of CaM was abolished when helix D was truncated (CT -589, Figure 1C). CaM binding possibly requires an oligomerized KCNQ1 C terminus, as previously observed for CaM binding to SK-channels.12
LQT Mutations in KCNQ1 C Terminus Impair CaM Binding and Stabilize Inactivation
Various LQT mutations, eg, R366W, R366P, A371T, S373P, and W392R,13–15 as well as a mild polymorphism K393N,16 are located near the KCNQ1 IQ motif (Figure 2). A yeast 2-hybrid assay indicated that A371T and S373P mutants do not interact with CaM, as no yeast cells grew on the selection medium (Figure 2A). R366W and W392R mutants exhibited a weak CaM interaction, whereas the mutant K393N mildly affected yeast growth. The weak interaction obtained with the R366P mutant was not significant in pulldown assays (see below).
We then used CaM-agarose pulldown assays and examined the profile of CaM interaction with WT KCNQ1 and LQT mutants in the presence and absence of Ca2+ in human embryonic kidney (HEK) 293 cells transfected with full length myc-tagged channels. As impaired CaM binding to LQT mutants drastically affected channel assembly (see Figure 6), we incubated CaM-agarose beads with comparable amounts of channel proteins expressed in cell lysates to estimate CaM binding (Figure 2B, row c). Anti-myc antibodies specifically labeled KCNQ1 channels at a molecular weight of ≈75 kDa. Results showed that in the presence of Ca2+, the LQT mutants S373P, W392R, and A371T (not shown) exhibit a much weaker CaM binding compared with WT (27% and 12% of WT, respectively; Figure 2B, row a, and 2C). The mutants K393N, R366W, and R366P bound less CaM, though at levels not significantly different from WT. In the absence of Ca2+ (5 mmol/L EGTA), 5- to 10-fold less CaM bound to WT KCNQ1 (Figure 2B, row b, and 2C). Although K393N and R366P mutants bound apoCaM with the same strength as WT, significantly less apoCaM was bound to the mutants R366W, S373P, and W392R (74%, 19%, and 11% of WT, respectively). These results were confirmed by CaM-agarose pulldown and immunoprecipitation from in vitro translated KCNQ1 C-termini (supplemental Figure I). Immunoprecipitation of in vitro cotranslated CaM and myc-tagged KCNQ1 C-terminal constructs confirmed the lack of interaction of apoCaM with the mutant W392R (Figure 2D).
Next, WT KCNQ1 and LQT mutants were expressed in CHO cells because of their low K+ conductance background compared with HEK 293 cells. Except for K393N, the LQT mutants produced small current densities with a dramatic change in inactivation (Figure 3A through 3G). Compared with WT, the mutants exhibited a voltage-dependent macroscopic inactivation (Figure 3A through 3F and 3I) and produced a significant right shift in the voltage dependence of channel activation, with up to a +44 mV shift for mutant S373P (Figure 3H and supplemental Table II). Similar results were obtained in transfected HEK 293 cells and in Xenopus oocytes (data not shown).
Impact of KCNE1 on CaM Binding and Inactivation Produced by LQT Mutants
As native IKS channels result from the coassembly of KCNQ1 and KCNE1 subunits, we examined the pattern of CaM interaction with WT KCNQ1 and LQT mutants cotransfected with KCNE1 in HEK 293 cells. CaM-agarose pulldown assays showed that in the presence of Ca2+, the mutants W392R, S373P, and, to a lower extent, R366W exhibit a significantly weaker CaM binding than WT IKS (15%, 43%, and 58% of WT, respectively) (Figure 4A, row a, and 4C). In contrast, mutants K393N and R366P bound CaM in amounts comparable to those bound with WT. It is noteworthy that in the presence of Ca2+, CaM-agarose beads could also pulldown KCNE1, which exists in a ternary complex with KCNQ1 and CaM (Figure 4B). Significantly less apoCaM was bound to the mutants R366W, S373P, and W392R (32%, 13%, and 3% of WT, respectively) compared with WT, K393N, and R366P (Figure 4A, row b, and 4C).
Next, we examined the biophysical properties of each mutant cotransfected with KCNE1 in CHO cells. Except mutant R366P, all mutant channels coexpressed with KCNE1 generated noninactivating currents with typical slow IKS activation kinetics (Figure 5). However, most of the LQT mutants produced significant lower current density compared with WT. They also showed a significant right shift of their activation curve (eg, V50=8.2±2.2 mV and V50 46.1±1.6 mV for WT and S373P, respectively; n=6 to 16) (Figure 5G and supplemental Table II). Coexpression of R366W with KCNE1 caused a significant rescue of current density, reaching up to 55% of WT IKS (Figure 5H).
Impact of CaM in Channel Assembly
The marked decrease in current density observed for most mutants suggested to us that the LQT mutations affecting CaM binding might also alter KCNQ1 subunit folding and/or assembly. We tested whether we could isolate a stable complex between CaM and the C terminus of KCNQ1 by use of a bacterial expression system. We first examined whether pET-21 bacterial expression of His-tagged KCNQ1 C terminus (352–622) yielded a soluble protein. Bacterial cell lysates were prepared and centrifuged, and resulting supernatants and pellets were analyzed by SDS-PAGE followed by immunoblotting with anti-His antibodies (Figure 6A). Very little if any of the cell extract was soluble, and the protein was aggregated into inclusion bodies in the pellet fraction. By contrast, coexpression of KCNQ1 C terminus (352–622) with CaM in pET-Duet produced soluble, nonaggregated material. Conversely, coexpression of CaM with KCNQ1 C terminus deleted from CBD (helix A and B) did not express a soluble protein and the material was found in aggregates (Figure 6A). The data suggest a critical role of CaM for folding the KCNQ1 C terminus.
Next, we monitored steady-state levels of WT and mutant KCNQ1 proteins expressed at the cell surface. Cell surface proteins were biotinylated using the membrane impermeable reagent sulfo-NHS-LC-biotin. Biotinylated proteins were pulled down by streptavidin-agarose beads and channel proteins were detected by Western blotting with anti-myc antibodies. Equal amounts of cell lysates were also run to compare the surface versus total cellular channel levels (Figure 6B, lysates). Blots of total cell lysates were probed with anti-Gβ protein antibodies to monitor inputs. No cell surface biotinylation of intracellular Gβ proteins was detected, showing the specificity of surface membrane protein labeling (Figure 6B, left panel, second row). Mutants with impaired CaM binding like R366W, S373P, and W392R were hardly detectable at the cell surface (Figure 6B, left panel, first row). Detection levels of R366W, S373P, and W392R mutants corresponded to ≈1% of WT KCNQ1. In contrast, detection levels of R366P and K393N that bound CaM almost normally were 15±2% and 53±8% of WT, respectively (n=3). We also investigated levels of WT and mutant proteins in total cell lysates. Mutations with defective CaM binding (R366W, S373P, and W392R) dramatically altered channel expression, as reflected by a profound reduction in expressed mutant proteins (Figure 6B, left panel, third row). When normalized to Gβ levels, total cell expression of R366W, S373P, and W392R corresponded to 5±2%, 15±10%, and 3±2% of WT, respectively (n=3, P<0.05). In contrast, expression levels of R366P and K393N mutants were 87±6% and 78±9% of WT, respectively (Figure 6B, left panel, third row; n=3). We performed similar biotinylation experiments on WT KCNQ1 and LQT mutants coexpressed with KCNE1 (Figure 6B). KCNE1 considerably increased both cell surface expression (3.9±0.7-fold, n=3, P<0.01) and total steady-state cellular content (3.9±0.9-fold, n=3, P<0.01) of WT KCNQ1. This observation suggested that KCNE1 significantly improves WT IKS channel expression or assembly. Similar data were obtained for the LQT mutants. The presence of KCNE1 markedly increased cell surface expression of R366W, S373P, and W392R mutants (Figure 6B; n=3). A similar effect of KCNE1 was observed for cell surface expression of R366P and K393N mutants representing 89±11% and 132±28%, respectively, of WT IKS channels (n=3, P<0.01). Total cellular contents of K393N, R366P, R366W, S373P, and W392R were 92±7%, 64±7%, 28±10%, 22±12%, and 19±12%, respectively, of WT IKS channels (Figure 6B; n=3), reflecting the improvement of channel expression by KCNE1.
We asked whether the cellular deficit in LQT mutant expression arose from channel misfolding/misassembly and subsequent degradation along the ubiquitin-proteasome pathway. When proteins fail to fold or assemble properly, this pathway could play a major role in their ER-associated degradation.17 Results shown in the Data Supplement suggest that LQT mutants are targets of proteasomal degradation (supplemental Figure II).
To obtain additional evidence for a role of CaM in KCNQ1 channel assembly, we transfected HEK 293 cells with WT KCNQ1 and the defective CaM-binding mutant S373P with or without CaM overexpression and performed a biotinylation assay. Transfection with CaM increased the levels of CaM by ≈2- to 3-fold over endogenous CaM, as detected with anti-CaM antibodies (Figure 6C, third row). CaM overexpression produced a striking increase in cell surface expression of WT KCNQ1 and S373P by more than 5-fold and 100-fold, respectively (Figure 6C, first row). A comparably drastic stimulation of total cellular channel content was observed in cell lysates, with a 2.8±0.4-fold and 10.2±2.9-fold increase for WT KCNQ1 and S373P, respectively (Figure 6C, second row; n=4, P<0.01).
Signaling of Ca2+-CaM to KCNQ1 and IKS Channels
In addition to its role in KCNQ1 subunit assembly, CaM might also function as a Ca sensor to modulate KCNQ1 and IKS channel gating. We tested this possibility on inside-out patches of Xenopus oocyte membranes. Bath application of the CaM antagonist W7 (50 μmol/L) potently and reversibly inhibited both WT KCNQ1/KCNE1 (IKS) and R366W/KCNE1 currents (Figure 7A). In the absence of KCNE1, R366W current was also reduced by W7 (supplemental Figure III). Increasing free Ca2+ concentrations in the bath from nominally 0 to 290 nmol/L produced a significant left-shift in the voltage dependence of channel activation, with a maximum Ca2+-induced shift ΔV50=22±6 mV (n=4) for WT KCNQ1 and ΔV50=79±8 mV (n=5) for WT KCNQ1/KCNE1 (Figure 7A and 7B; supplemental Table I). The left-shift in V50 was not observed on bath perfusion with Mg2+ at concentrations of up to 5 mmol/L (Figure 7C). Conspicuously, Δz (equivalent gating charge) and ΔV50 varied inversely for WT and mutant channels, but their product remained constant (Figure 7D). In line with inside-out patch data, elevation of internal Ca2+ concentrations by the Ca2+-ionophore ionomycin (5 μmol/L) increased the amplitude of WT KCNQ1 by producing a left-shift of the activation curve (ΔV50=−24 mV; n=7) when measured by 2-electrode voltage-clamp in Cl−--free solutions (supplemental Figures III and IV). In addition, ionomycin accelerated the activation kinetics of KCNQ1. Ionomycin stimulated the R366W mutant current, but the resulting amplitude was far from reaching that of WT KCNQ1 (supplemental Figure III). Though inside-out patch records were done in the absence of ATP and in the presence of phosphatase inhibitors, we checked whether the Ca2+-mediated stimulation of IKS was mediated by CaM kinase II. KN93 (5 μmol/L), an inhibitor of CaM kinase II, did not affect ionomycin-mediated stimulation of IKS currents (supplemental Figure IVE). Altogether, the data indicate that CaM is essential for IKS channel activity and conveys Ca2+ sensitivity. In agreement with this idea, coexpression of KCNQ1 or IKS channels with a Ca2+-insensitive CaM mutant (CaM1234) markedly suppressed the currents (by 81%) and produced a right-shift in the voltage-dependence of channel activation (ΔV50=±25 mV) (supplemental Figure V).
In this study, we showed that KCNQ1 and IKS channels need CaM binding for not only proper channel expression and assembly but also correct gating. CaM binding confers Ca2+-sensitive stimulation of IKS current. LQT mutations located near the IQ motif of KCNQ1 impair CaM binding, confer inactivation, and reduce current amplitude.
The physical interaction of CaM with KCNQ2–5 α subunits has been recently investigated.6,9,11 However, no information has been available so far regarding KCNQ1 and IKS channels. Our in vitro results clearly showed that CaM directly interacts with the C terminus of KCNQ1 in not only the presence of Ca2+ but also its virtual absence. Pulldown experiments indicated that KCNQ1 channels interact more strongly with Ca2+-CaM than with apoCaM.
Mutations in KCNQ1 have been shown to be associated with either the autosomal-dominant or -recessive form of LQT, the Romano-Ward syndrome and the Jervell and Lange-Nielsen syndrome, respectively. We previously showed that mutations in the distal C terminus associated with the recessive Jervell and Lange-Nielsen syndrome phenotype impair KCNQ1 subunit assembly.7 Here, we investigated the impact of various LQT mutations located near the IQ motif of KCNQ1. The mutants A371T, S373P, and W392R showed severely impaired binding of CaM in both its Ca2+-loaded and unloaded form. Although less severe, the mutant R366W had a weaker CaM interaction compared with WT, whereas replacement of arginine by proline at the same location (R366P) bound normally CaM. Probably because of its proximity to the S6 C terminus, the mutant R366P displayed a prominent gating defect. Despite their ability to bind CaM, R366P channels mediated inactivating tiny currents. Except for R366P, decreases in current density of the other LQT mutants tightly correlated with both the severity of the CaM binding defect and the impairment of channel expression. Our data also showed that the LQT mutants analyzed in this study produced a voltage-dependent macroscopic inactivation. However, modulation of macroscopic inactivation is not restricted to the CaM binding domain, as other sites in KCNQ1 are known to confer macroscopic inactivation. KCNE1 association with LQT mutants prevents macroscopic inactivation and significantly improves mutant channel amplitude. However, the marked right-shift in channel activation and the resulting decrease in current amplitude cannot restore normal levels of IKS channel activity. Hence, in these heterozygous Romano-Ward syndrome mutations, even if there is no dominant-negative effect of the mutant channel subunit, the resulting decrease in IKS current density (Figure 5H) will significantly alter the cardiac repolarization reserve.
CaM was suggested to be important for the assembly and trafficking of ion channels like SK4/IK1 Ca2+-activated K+ channels.18 CaM binding to KCNQ1 C terminus is essential for not only proper channel gating but also channel folding and assembly (Figure 8). First, similar to Cav1.2 and Nav1 channels,19,20 recombinant production of a soluble KCNQ1 C terminus required bacterial coexpression of CaM, suggesting that CaM is necessary for proper folding of the C terminus. Second, LQT mutations with impaired CaM binding (R366W, A371T, S373P, and W392R) profoundly disrupted channel expression, as revealed in biotinylation experiments. This disruption may involve not only channel assembly but also processes like degradation pathways (see below). Third, we found that overexpression of WT CaM greatly improved the expression WT KCNQ1 and even more dramatically that of S373P mutant channels. Fourth, deleting helix D compromised CaM interaction with KCNQ1, a feature which suggests a link between KCNQ1 channel assembly and CaM interaction. However, a recent work 21 has shown that KCNQ1 channels lacking helix D are not functional but can coassemble as tetramers, suggesting that helix D is not the sole determinant of subunit assembly. Interestingly, we showed that KCNE1 forms a ternary complex with KCNQ1 and Ca2+-CaM. KCNE1 greatly increased not only cell surface expression but also total cellular content of WT and mutant channels, indicating that it may facilitate IKS channel assembly. We suggest that KCNE1 assembles with CaM-tethered KCNQ1 early in biogenesis of the channel complex, forming a ternary complex that possibly stabilizes the pre-association of CaM to KCNQ1.
Our data with the proteasomal inhibitor MG132 indicate that the deficit in LQT mutant channel expression arises from channel misfolding/misassembly and subsequent degradation along the proteasome pathway. Because the LQT mutations with impaired CaM binding are likely misfolded, blocking the proteasome pathway is ineffective, which explains why the MG132 treatment did not rescue the S373P and R366W mutant expression to the cell surface. Instead, the proteasome blocker probably leads the mutant proteins to accumulate in intracellular aggregates, which was previously shown to be a cellular response to misfolded proteins.22 In contrast, CaM overexpression was effective in retrieving the S373P mutant expression to the cell surface, suggesting that by providing a larger amount of CaM, one could improve mutant channel assembly.
There are prominent differences in the role played by Ca2+-CaM in KCNQ2–5 and KCNQ1 channel signaling. Recently, overexpression of CaM in CHO cells was found to robustly reduce currents of KCNQ2, KCNQ4, and KCNQ5, but not those of KCNQ1 and KCNQ3.10 Here, we showed in Xenopus oocytes that WT KCNQ1, R366W, and IKS currents are stimulated by increases in intracellular Ca2+ and are markedly inhibited by CaM antagonists. Remarkably, intracellular Ca2+ produced a left-shift in the voltage-dependence of activation of WT IKS as determined from inside-out patch recording. Although the LQT mutations are right-shifted in their activation curve, an increase in intracellular Ca2+ still produced a marked left-shift in the activation seen for R366W/KCNE1. Similarly, ionomycin left-shifted the voltage dependence of KCNQ1 activation and accelerated its activation kinetics. In agreement with these data, coexpression of KCNQ1 or IKS channels with a Ca2+-insensitive CaM mutant markedly suppressed the currents and produced a right-shift in the voltage-dependence of channel activation. Altogether, our results suggest that Ca2+-CaM is the Ca2+ sensor that stimulates IKS channels during cardiac activity via a considerable Ca2+ sensitivity. This dual voltage and Ca2+ dependence of IKS channels is somewhat reminiscent of the large-conductance Ca2+-activated BKCa channel behavior.
Interestingly, previous studies showed that elevation of [Ca2+]i in guinea pig ventricular myocytes enhances IKS-currents.23 Noteworthy, IKS-currents increase ≈3-fold on free [Ca2+]i rise from 10−8 M to 10−7 M,23 suggesting that there is substantial Ca2+-CaM bound to IKS channels under resting Ca2+ levels. This stimulation may be accounted for by the Ca2+-induced left-shift of IKS activation. Our preliminary data indicate that the kinetics of Ca2+-induced rise in IKS fit within the time frame of a single action potential (L.M., unpublished data, 2006). Our results are also consistent with a recent study showing that at high stimulation rates, there is a Ca2+-induced increase in cardiac IKS currents that plays a dominant role in shortening action potential duration.24 Interestingly, this Ca2+-induced IKS stimulation was sensitive to the CaM antagonist W7.24
In all, CaM exerts a dual action on IKS channels, being necessary for both correct channel assembly and gating. As such, CaM enhances IKS channel activity after increases in internal Ca2+ (Figure 8). This IKS current boosting is crucial for increasing the repolarization reserve. Given the marked loss of channel function produced by the defective CaM binding LQT mutants, it is not surprising that a decrease in cardiac repolarization reserve would prolong the action potential and increase the risk of ventricular arrhythmias.
We thank Dr J.P. Adelman (Vollum Institute, Portland, Ore) for providing us with the WT and Ca2+-insensitive CaM1234 mutant. This work was supported by the United States–Israel Binational Science Foundation (No: 2001229), Israeli Science Foundation (ISF 672/05), Wolfson Family and Recanati grants to B.A., and the German Research Council support (DFG Po137/36–1) and Leducq Foundation to O.P.
Original received October 30, 2005; revision received March 4, 2006; accepted March 14, 2006.
Yus-Najera E, Santana-Castro I, Villaroel A. The identification and characterization of a noncontinuous calmodulin-binding site in noninactivating voltage-dependent KCNQ potassium channels. J Biol Chem. 2002; 277: 28545–28553.
Schmitt N, Schwarz M, Peretz A, Abitbol I, Attali B, Pongs O. A recessive C-terminal Jervell and Lange-Nielsen mutation of the KCNQ1 channel impairs subunit assembly. EMBO J. 2000; 19: 332–340.
Marx SO, Kurokawa J, Riken S, Motoike H, D’Armiento J, Marks AR, Kass RS. Requirement of a macromolcular signaling complex for beta adrenergic receptor modulation of the KCNQ1-KCNE1 potassium channel. Science. 2002; 295: 496–498.
Wen H, Levitan IB. Calmodulin is an auxiliary subunit of KCNQ2/3 potassium channels. J Neurosci. 2002; 22: 7991–8001.
Gamper N, Li Y, Shapiro MS. Structural requirements for differential sensitivity of KCNQ K+ channels to modulation by Ca2+/calmodulin. Mol Biol Cell. 2005; 16: 3538–3551.
Gamper N, Shapiro MS. Calmodulin mediates Ca2+-dependent modulation of M-type K+ channels. J Gen Physiol. 2003; 122: 17–31.
Donger C, Denjoy I, Berthet M, Neyroud N, Cruaud C, Bennaceur M, Chivoret G, Schwartz K, Coumel P, Guicheney P. KVLQT1 C-terminal missense mutation causes a forme fruste long-QT syndrome. Circulation. 1997; 96: 2778–2781.
Joiner WJ, Khanna R, Schlichter LC, Kaczmarek LK. Calmodulin regulates assembly and trafficking of SK4/IK1 Ca2+-activated K+ channels. J Biol Chem. 2001; 276: 37980–37985.
Kim J, Ghosh S, Liu H, Tateyama M, Kass RS, Pitt GS. Calmodulin mediates Ca2+ sensitivity of sodium channels. J Biol Chem. 2004; 279: 45004–45012.
Kanki H, Kupershmidt S, Yang T, Wells S, Roden DM. A structural requirement for processing the cardiac K+ channel KCNQ1. J Biol Chem. 2004; 279: 33976–33983.
Johnston JA, Ward CL, Kopito RR. Aggresomes: a cellular response to misfolded proteins. J Cell Biol. 1998; 143: 1883–1898.
Bai CX, Namekata I, Kurokawa J, Tanaka H, Shigenobu K, Furukawa T. Role of nitric oxide in Ca2+ sensitivity of the slowly activating delayed rectifier K+ current in cardiac myocytes. Circ Res. 2005; 96: 64–72.