Kv7.1 (KCNQ1) properties and channelopathies

Abstract

KCNQ1 is the pore-forming subunit of a channel complex whose expression and function have been rather well characterized in the heart. Almost 300 mutations of KCNQ1 have been identified in patients and a vast majority of the described mutations are linked to the long QT syndrome. Only a few mutations are linked to other pathologies such as atrial fibrillation and the short QT syndrome. However, a considerable amount of work remains to be done to get a clear picture of the molecular mechanisms responsible for the pathogenesis related to each mutation. The present review gives three examples of recent studies towards this goal and illustrates the diversity of the molecular mechanisms involved.

KCNQ1 is the α-subunit of a voltage-dependent potassium channel expressed in various cell types including cardiac myocytes and epithelial cells. In the heart, KCNQ1 assembles with KCNE1 to form a channel complex constituting the slow component of the delayed rectifier current I_Ks (Barhanin et al. 1996; Sanguinetti et al. 1996). Other KCNE subunits (such as KCNE4) may be present in the complex and play a significant role (Bendahhou et al. 2005). To date, nearly three hundred mutations in KCNQ1 have been identified. The majority of these mutations cause a loss of function of KCNQ1, resulting in the long QT syndrome (LQTS), an inherited disorder characterized by a delayed ventricular repolarization, syncope and sudden death (Vohra, 2007). Only a few gain-of-function mutations have been identified and are linked to atrial fibrillation (Chen et al. 2003) or the short QT syndrome (Bellocq et al. 2004).

For most of these three hundred KCNQ1 mutations, the molecular mechanisms leading to the pathologies were poorly understood and frequently limited to one of the following categories: impaired trafficking, impaired voltage dependency, impaired selectivity and impaired tetramerization. Impaired tetramerization was observed in one deletion–insertion mutation at residue 544 (Schmitt et al. 2000), causing the autosomal-recessive form of the LQTS (Jervell and Lange-Nielsen, JLN), as opposed to the more frequent autosomal-dominant form (Romano-Ward, RW). The impaired tetramerization provides a molecular basis for the autosomal-recessive form of the syndrome. Indeed, a KCNQ1 mutant may not have a dominant negative effect if it does not interact with a wild-type (WT) monomer, that is if it does not make a tetramer.

This review describes how recent studies gave further insights into some molecular mechanisms involved in the pathogenesis of LQTS. This is illustrated by three examples: the first example shows that classical methods (microscopy, patch clamp) are useful to investigate new hypotheses. However, these methods are limited to shedding light on the mechanisms at the amino-acid level, and require other methods such as structure prediction; the second and third examples show how structural and biochemical, but also patch-clamp data, can give further insights into the molecular mechanisms leading to the pathology.

Example 1: A hypothetical mechanism for different effects of mutations located in the same region of KCNQ1

We have recently studied three RW mutations (Y111C, L114P and P117L) located in the KCNQ1 N-terminus, just before the first transmembrane domain (Dahimene et al. 2006). Combining classical methodologies and structural model prediction, we characterized a new structural motif and described how these mutations within this motif lead to loss of function of the channel. Using immunofluorescence microscopy and patch-clamp experiments we observed that these three mutants did not reach the plasma membrane but remained trapped in the endoplasmic reticulum (ER) when expressed as homotetramers. Surprisingly, when mutated proteins were co-expressed with WT subunits, they exerted different effects on the KCNQ1 current. Y111C and L114P mutants produced a strong decrease of current density whereas the P117L mutant did not. Fluorescence microscopy analysis provided a clue to the mechanism. Indeed, in the presence of WT-KCNQ1, Y111C and L114P remained trapped in the ER whereas the P117L mutant was expressed at the cell surface. This indicated that the WT subunit could somehow rescue P117L trafficking but not that of Y111C and L114P. This raised an intriguing question: how can spatially close mutations have such different effects? Structure prediction analysis, performed in collaboration with Annick Thomas and Robert Brasseur (University of Gembloux, Belgium) unveiled potential molecular mechanisms. Indeed, structure prediction showed that this region consists of a short helix and a loop connected to the first transmembrane domain. This structural motif is stabilized by hydrophobic interactions between Y111 in the short helix and an aromatic cluster of amino acids in the first transmembrane domain including H126 (Fig. 1). The model predicted that both Y111C and L114P substitutions ultimately hinder the interaction of Y111 with the aromatic cluster and thus destabilize the structure. P117L increases the length of the helix and hinders the formation of the loop, but glycine at position 119 may still confer sufficient flexibility to the helix to bend on the membrane and allow Y111 to participate in the native interactions. Therefore, one can speculate that WT subunits could help the P117L mutant, but not Y111C and L114P, to adopt the native conformation and to pass through the ER control quality.

Figure 1. Ribbon structure diagram of the KCNQ1 structure in the N-terminal juxtamembranous domain, illustrating the potential effects of the L114P mutation on the structure of the trafficking determinant.

The L114P mutation pulls apart Y111C from H126 located in the first transmembrane domain. The effect is highlighted in the cartoon on the right where WT KCNQ1 is shown in grey and L114P mutant in gold (see Material and Methods in Dahimene et al. 2006).

As a conclusion, thanks to classical methods, we were able to identify a new trafficking motif in the N-terminal juxtamembranous domain and envisage molecular mechanisms backed up by structural prediction analysis.

Example 2: Mutations in the assembly domain prevent channel membrane targeting, but without impeding tetramerization

While, in the preceding example, we began our study with biological and pathophysiological observations to develop a potential structural model, in this second example structural studies have led to new insights into the molecular mechanisms of LQTS (Howard et al. 2007). In previous studies, it has been proposed that many LQTS-related mutations in the KCNQ1 C-terminus hindered channel subunit association. Howard and coworkers have revisited those mechanisms based on the crystallographic structure analysis of a KCNQ C-terminal assembly domain called the A-domain. In fact, the KCNQ channel family includes five members that can form homo and heterotetramers, with distinct assembly preferences controlled by the A-domain. Using X-ray diffraction analysis, circular dichroism, size exclusion chromatography, analytical ultracentrifugation, sedimentation equilibrium and gel filtration, they have proposed a common mechanism for KCNQ1 mutations located in this domain. In a first set of experiments, the authors have determined the crystallographic structure of the A-domain tail (a.a. 610–640) of KCNQ4: this is a tightly twisted left-handed four-stranded coiled coil with a predominantly polar surface. Interestingly, the KCNQ3 A-domain tail, contrary to all other KCNQs, was not able to form homotetramers in gel filtration chromatography experiments. Structure comparisons pointed to F622, D631 and G633 as the key residues that hampered this homo association in KCNQ3. This explained why KCNQ3 only forms heterotetramers with other KCNQ subtypes. Interestingly, and in contrast to what was previously thought, none of the LQTS mutations (T587M, R589D, A590T, R591H, R594H) disrupted the KCNQ1 A-domain assembly, ruling out the previous hypothesis that these mutations affected tetramerization and, as a consequence, channel trafficking. Thus, and thanks to crystallographic structure of the KCNQ4 A-domain tail, the authors proposed a new explanation. They were able to map these LQTS mutations onto equivalent positions in the KCNQ4 A-domain tail and observed that they formed a ‘hotspot’ at the surface of the A-domain tail, opposite to the association surface. As the G589D mutation was located in the middle of this ‘hotspot’ and was known to impair association with the regulatory protein kinase yotiao, the authors proposed that all these mutations act by a similar mechanism: they impair regulatory proteins to bind to the channel, suggesting that binding of accessory proteins in this region may be a mechanism of quality control for correctly assembled subunits.

In conclusion, these new data lead us to redefine the role of the A-domain in KCNQ1: its presence is necessary for tetramerization and for the channel to reach the plasma membrane; missense LQTS mutations in this domain do not prevent tetramerization but impair regulatory proteins in binding to the channel.

Example 3: Some mutations alter the channel voltage dependency through a reduced interaction with membrane PIP₂

Numerous mutations of KCNQ1 affecting cytosol facing arginine residues are involved in the LQTS. To date five of them have been functionally studied: R190Q, R243H, R533W, R539W and R555C (Chouabe et al. 1997, 2000) but the molecular mechanism was missing. We attempted to determine the molecular mechanism behind the alteration of the gating properties of KCNQ1 by mutated cytosolic arginines.

Previous studies have shown that several cytosolic basic residues are crucial for the regulation by phosphatidylinositol-4,5-biphosphate (PIP₂) of many Kir channels (Lopes et al. 2002; Logothetis et al. 2007) and several Kv channels such as HERG (Bian & McDonald, 2007) and KCNQ2–3 (Zhang et al. 2003). PIP₂ is a phospholipid present at the inner leaflet of the plasma membrane that is notably the precursor of the second messengers DAG and IP₃. Its effects on inwardly rectifying potassium channels (Kir) are now well documented (Logothetis et al. 2007). Our hypothesis was thus that the loss of function of R243H, R539W and R555C can be explained by a modification of their regulation by PIP₂.

First, we characterized the PIP₂ regulation of KCNQ1–KCNE1 channel complexes contained in excised patches of COS-7 cells (Loussouarn et al. 2003). In that configuration, the PIP₂ concentration in the membrane decreases with time presumably because of an enhanced activity of phosphatases or phospholipases. This decrease is accompanied by a rundown of the maximal current that can be markedly slowed by cytosolic application of PIP₂ (and completely prevented by concomitant MgATP application, Fig. 2). These results emphasize the up-regulation of KCNQ1–KCNE1 by PIP₂. Minute analysis of the biophysical modifications induced by PIP₂ on KCNQ1–KCNE1 reveals a slowing of the deactivation kinetics and a shift toward negative potentials of the activation curve besides the increase in the maximal current. It is noteworthy that a decrease in the PIP₂ level has the same consequences on the channel biophysical properties as R243H, R539W and R555C mutations.

Figure 2. The effects of PIP₂ on KCNQ1–KCNE1 currents.

Average time-dependent currents (I_rel) measured at the end of a 1 s depolarizing step to +40 mV relative to their maximum value measured after patch excision. Patches were excised in control solution (○, n = 9), control solution plus 1.4 mm MgATP (0.6 mm free Mg²⁺; •, n = 3), control solution plus 5 mg ml⁻¹ PIP₂ (dashed line, n = 5), control solution plus 1.4 mm MgATP plus 5 mg ml⁻¹ PIP₂ (▪, n = 9). (Figure adapted from Loussouarn et al. 2003; with permission.)

To establish that regulation by PIP₂ of the three KCNQ1 mutants is modified, we have compared the affinities for PIP₂ of WT and mutants (Park et al. 2005). It is impossible to control the membrane concentration of externally applied non-soluble PIP₂ and thus measure the dose–response of the channel but this can be overcome with the use of a short-chain soluble analogue: diC8-PIP₂ (Fig. 3). The measured response is the current amplitude after a 1 s depolarizing step to +80 mV in the presence of various concentrations of diC8-PIP₂ on an excised patch of transfected COS-7. The EC₅₀ value of diC8-PIP₂ for WT KCNQ1–KCNE1 is 10 μm. Mutant channels have a reduced affinity for diC8-PIP₂: 14 μm (R243H), 22 μm (R539W) and 23 μm (R555C). Hill coefficients are 1.3 for WT and close to 1 for mutant channels. Analysis of KCNQ2–KCNQ4 in the cell-attached mode indicates that the physiological membrane PIP₂ concentration would have equivalent effects as 23 μm diC8-PIP₂ intracellular solution in the inside-out configuration (Li et al. 2005), i.e. in the range of WT KCNQ1–KCNE1 sensitivity. KCNQ1–KCNE1 affinity is thus such that the channel seems likely to respond significantly to moderate physiological PIP₂ variations and that the activity of the mutants in vivo may be notably reduced compared with WT. However, firmly establishing these assumptions requires overcoming the difficulty of a dynamic quantification of the PIP₂ concentration in living cells (with fluorescent probes), and possibly to deal with the question of compartmentalization of both PIP₂ and channels in the plasma membrane (Gamper & Shapiro, 2007).

Figure 3. Dose–response relationship of the PIP₂ effects on WT and mutant KCNQ1–KCNE1 concatemer channels.

Currents measured at the end of a 1 s depolarazing step to +80 mV relative to their maximum values extrapolated from a Hill fit. R243H, R539W and R555C mutants present a lower diC8-PIP₂ affinity than WT channels. (Figure adapted from Park et al. 2005; with permission.)

Arginine mutations were suspected to modify regulation by PIP₂ because cytosolic positive charges can bind PIP₂. Nevertheless, our experiments cannot distinguish allosteric effects from direct alterations of the PIP₂ binding site. In the case of R555C, a cytosolic application of methanethiosulphonate ethylammonium brings a positive charge on the residue and restores the channel activity as expected from a disrupted electrostatic interaction.

Insights into molecular mechanisms are also obtained by kinetic models. In the case of WT KCNQ1–KCNE1, the biophysical modifications due to PIP₂ loss and recovery can be fully taken into account by a kinetic model of the channel in which voltage-dependent transitions precede a final transition to the open state that is facilitated by PIP₂ (Loussouarn et al. 2003). Similarly to Kir channels (Enkvetchakul et al. 2000), PIP₂ would act on KCNQ1–KCNE1 by stabilizing the open state. Based on this model, the effects of mutations R539W and R555C can also be fully explained by a weaker stabilization of the open state (Park et al. 2005). In the case of R243H, the kinetic model suggests an additional change of its voltage-dependent transitions. This mutation may alter both the voltage sensing and the stabilization of the channel open state. R243 belongs to the linker between the 4th and 5th transmembrane domains whose interaction with the pore lining 6th transmembrane domain is supposed to be crucial for the voltage transduction (Long et al. 2005).

Weakening of the PIP₂ up-regulation of KCNQ1–KCNE1 seems likely to explain the modifications of the voltage-dependent gating of some LQT mutants. Already identified for Kir channels (Lopes et al. 2002), the regulation of channels by PIP₂ is thus revealed by our experiments as a relevant molecular explanation for KCNQ1 channelopathies.