Regulation of neural KCNQ channels: signalling pathways, structural motifs and functional implications

Abstract

Neural M-type (KCNQ/Kv7) K⁺ channels control somatic excitability, bursting and neurotransmitter release throughout the nervous system. Their activity is regulated by multiple signalling pathways. In superior cervical ganglion sympathetic neurons, muscarinic M₁, angiotensin II AT₁, bradykinin B₂ and purinergic P2Y agonists suppress M current (i_m). Probes of PLC activity show agonists of all four receptors to induce robust PIP₂ hydrolysis. We have grouped these receptors into two related modes of action. One mode involves depletion of phosphatidylinositol 4,5-bisphosphate (PIP₂) in the membrane, whose interaction with the channels is thought necessary for their function. The other involves IP₃-mediated intracellular Ca²⁺ signals that stimulate PIP₂ synthesis, preventing its depletion, and suppress i_m via calmodulin. Carbon-fibre amperometry can evaluate the effect of M channel activity on release of neurotransmitter. Consistent with the dominant role of M current in control of neuronal discharge, M channel openers, or blockers, reduced or augmented the evoked release of noradrenaline neurotransmitter from superior cervical ganglion (SCG) neurons, respectively. We seek to localize the subdomains on the channels critical to their regulation by PIP₂. Based on single-channel recordings from chimeras between high-PIP₂ affinity KCNQ3 and low-PIP₂ affinity KCNQ4 channels, we focus on a 57-residue domain within the carboxy-terminus that is a possible PIP₂ binding site. Homology modelling of this domain using the published structure of IRK1 channels as a template predicts a structure very similar to an analogous region in IRK1 channels, and shows a cluster of basic residues in the KCNQ2 domain to correspond to those implicated in PIP₂ regulation of Kir channels. We discuss some important issues dealing with these topics.

Over two decades have passed since Brown and colleagues described a neuronal K⁺ current in bullfrog sympathetic cells that they named ‘M current’ due to its suppression by stimulation of muscarinic acetylcholine receptors (mAChRs) (Brown & Adams, 1980; Constanti & Brown, 1981). This time- and voltage-dependent K⁺ current has a threshold for activation at typical neuronal resting potentials, with greater activity upon depolarization. This characteristic and its lack of inactivation give M current a major impact on neuronal excitability. Suppression of M current by the activation of appropriate receptors or by pharmacological blockade allows neurons to fire more rapidly due to the reduction in accommodation or spike frequency adaptation (Jones et al. 1995; Wang et al. 1998; Peretz et al. 2005; Gamper et al. 2006; Yue & Yaari, 2006; Zaika et al. 2006, 2007). Underlain by the KCNQ family of genes (Wang et al. 1998), M channels, also known as Kv7 (Gutman et al. 2003), are localized throughout the nervous system (Wang et al. 1998; Cooper et al. 2001; Roche et al. 2002; Shah et al. 2002; Devaux et al. 2004; Pan et al. 2006), where the channels are inhibited by stimulation of a variety of G_q/11-coupled neurotransmitter receptors. The transduction mechanism between the receptors and channels has been the subject of intense investigation, and work over the past five years has shed considerable light on this topic (Delmas & Brown, 2005). This report focuses on our recent findings in the modes of inhibition of M-type channels in sympathetic neurons of the superior cervical ganglion (SCG), as well as the physiological roles these signalling pathways play.

Receptor specificity

Much effort has gone into understanding how specificity is achieved in cellular signalling. One paradigm invokes compartmentalization of signalling proteins within membrane microdomains (Kreienkamp, 2002; Delmas et al. 2004). Another hypothesizes local production or degradation of intracellular messengers that results in their local concentration gradients (reviewed in Gamper & Shapiro, 2007b). We consider PIP₂ hydrolysis by PLC as the first step in signalling by G_q/11-coupled receptors. This process can now be monitored in real-time in individual living cells using probes such as the green fluorescent protein (GFP)-tagged PH domain of PLCδ (PLCδ-PH) (Stauffer et al. 1998; Raucher et al. 2000). This probe binds to both PIP₂ and IP₃. In unstimulated cells (Fig. 1C, left), cytoplasmic [IP₃] is very low and there is much membrane PIP₂, to which the probe localizes. Upon stimulation of PLC-linked receptors, the probe translocates to the IP₃ accumulating in the cytoplasm (Fig. 1C, right), which can be optically monitored. In SCG neurons, the responses of PIP₂ probes to muscarinic and bradykinin stimulation have been found to be equally strong (Gamper et al. 2004; Winks et al. 2005; Hughes et al. 2007). To compare them to those from other G_q/11-coupled receptors that act on M channels, such as the purinergic P2Y and angiotensin AT₁ types, we performed confocal microscopy on cells stimulated with multiple agonists. Figure 1A shows such an experiment on a SCG neuron transfected with PLCδ-PH using the biolistic ‘gene gun’ method (Gamper & Shapiro, 2006). Sequential stimulation of P2Y, bradykinin B₂ and muscarinic M₁ receptors by their respective agonists induces a reversible translocation of the probe from membrane to cytoplasm, indicating strong PIP₂ hydrolysis by all three. It is important to note that such translocations cannot be easily used to report actual depletion of membrane PIP₂ since the probe binds to IP₃ with an affinity > 10-fold greater than that of PIP₂ (Hirose et al. 1999); instead, careful IP₃ titration experiments are required (Xu et al. 2003; Winks et al. 2005). Figure 1B summarizes the increases in cytoplasmic fluorescence from the probe induced by each agonist. Note that the translocation produced by bradykinin is the strongest, although other experiments discussed below indicate that stimulation of B₂ receptors does not appreciably deplete PIP₂ in these cells. What these results do indicate is that stimulation of M₁, AT₁, B₂ and P2Y receptors results in signalling mechanisms that have in common the activation of PLC and vigorous PIP₂ hydrolysis.

Figure 1. The PLCδ-PH probe reports PIP₂ hydrolysis upon stimulation of multiple G_q/11-coupled receptors.

A, confocal images from a SCG neuron sequentially stimulated by UTP, bradykinin (BK), and oxotremorine (Oxo-M). In the unstimulated cell (control), the probe was concentrated mainly in the membrane, bound to PIP₂. Upon stimulation by agonist, activation of PLC hydrolyses PIP₂ and the probe translocates to the cytoplasm, bound to IP₃. B, summarized data of the average fluorescence intensity, F, in a cytoplasmic region, normalized to the average intensity over 30 s before agonist application, F_o, quantified as the percentage increase in F (ΔF/F_o) induced by the four agonists. C, a schematic depiction of movement of the PLCδ-PH probe before (left panel) and after (right panel) activation of PLC via stimulation G_q/11-coupled receptors. D, an example of M current regulation under whole-cell clamp before (a, c) and after (b, d) application of either bradykinin or Oxo-M, and after complete blockade after application of linopirdine (e). GPCR, G protein coupled receptor; PLC, phospholipase C; DAG, diacylglycerol; BK or brady, bradykinin; Oxo-M or oxo, oxotremorine-M; LP, linopirdine. (G. P. Tolstykh, O. Zaika and M.S. Shapiro, unpublished observations.)

Figure 1D shows an example of M current suppression in an SCG cell studied under whole-cell clamp by the muscarinic agonist oxotremorine and by bradykinin. Although PIP₂ hydrolysis by stimulation of both M₁ and B₂ receptors must produce much cytoplasmic IP₃, the former does not induce IP₃-mediated rises in intracellular Ca²⁺ ([Ca²⁺]_i) (Wanke et al. 1987; Beech et al. 1991; Shapiro et al. 1994; Delmas et al. 2002; Zaika et al. 2007). In contrast, stimulation of bradykinin B₂ receptors does induce [Ca²⁺]_i transients, and bradykinin suppression of M current requires IP₃-mediated rises in cytoplasmic Ca²⁺. Thus, clamping of intracellular Ca²⁺, blockade of IP₃ receptors and depletion of internal Ca²⁺ stores all severely attenuate M current suppression by bradykinin, but all have little effect on the muscarinic action (Cruzblanca et al. 1998; Bofill-Cardona et al. 2000). In SCG cells, purinergic P2Y and B₂ receptors have seemed to act similarly (Bofill-Cardona et al. 2000) and we have recently shown over-expression of an IP₃ scavenger or an IP₃ 5-phosphatase, both of which prevent IP₃ accumulation, to selectively block the purinergic or bradykinin inhibitions of M current, but to spare the muscarinic one (Zaika et al. 2007). We have further shown calmodulin (CaM) to be involved in the signal involving IP₃-mediated [Ca²⁺]_i rises, acting as the Ca²⁺ sensor of M-type channels. Thus, expression in SCG neurons of a dominant-negative CaM that cannot bind Ca²⁺ (Geiser et al. 1991) makes native M current mostly insensitive to bradykinin or purinergic stimulation, but again spares the muscarinic action (Gamper & Shapiro, 2003; Gamper et al. 2005; Zaika et al. 2007).

The receptor-specific [Ca²⁺]_i signals have important implications for phosphoinositide (PI) signalling, since not only can hormonal stimulation hydrolyse PIP₂, it can also strongly stimulate PIP₂ synthesis by acceleration of PI kinases (Loew, 2007). One such pathway involves the Ca²⁺-binding protein neuronal Ca²⁺ sensor-1 (NCS-1), a signalling molecule with a wide spectrum of actions (Burgoyne & Weiss, 2001), including up-regulation of PI4-kinase, the first step in PIP₂ synthesis from PI (Zhao et al. 2001; Taverna et al. 2002; Koizumi et al. 2002; Rajebhosale et al. 2003). Here, our thinking on M current signalling is fertilized by our work on modulation of N-type Ca²⁺ channels of SCG neurons by G_q/11-coupled receptors. As for P/Q-type channels (Wu et al. 2002), N-type channels also require membrane PIP₂ to be functional (Gamper et al. 2004). Whereas the suppression of the Ca²⁺ current by muscarinic or angiotensin II stimulation is robust (Hille, 1994), it is little suppressed by stimulation of bradykinin or purinergic receptors (Gamper et al. 2004; Zaika et al. 2007). We hypothesized that this receptor-specific modulation of Ca²⁺ channels is due to receptor-specific depletion of PIP₂. In this scenario, concurrent stimulation of PIP₂ synthesis by bradykinin and purinergic agonists via Ca²⁺/NCS-1-mediated acceleration of PI4-kinase activity compensates for consumption of PIP₂ by PLC. Indeed, over-expression of a DN NCS-1 that cannot bind Ca²⁺ bestowed onto bradykinin and purinergic agonists the effect of Ca²⁺ current suppression (Gamper et al. 2004; Zaika et al. 2007), consistent with PIP₂ levels now falling in the absence of compensatory PIP₂ synthesis (Delmas et al. 2005). Evidence for NCS-1 action on PIP₂ production comes from other over-expression studies in SCG neurons (Winks et al. 2005), neuroendocrine PC12 cells (Koizumi et al. 2002; Rajebhosale et al. 2003), chromaffin cells (Pan et al. 2002) and in brain synaptosomes (Zheng et al. 2005). All of these studies are consistent with hormonal stimulation of PIP₂ synthesis via ${\text{Ca}}_{\text{i}}^{2+}$ signals, and subsequent amplification of PI turnover.

Although this hypothesis is attractive, it raises several crucial issues. As for much in the lipid kinase field, the control of a PI4-kinase by an orthologue of mammalian NCS-1 was discovered in yeast, where the respective proteins are Pik1 and Frq1 (frequenin in Drosophila) (Hendricks et al. 1999). The mammalian orthologue of yeast Pik1 is PI4-kinase IIIβ, one of four isoforms of mammalian PI4-kinases cloned so far, and this is the isoform shown to be stimulated by NCS-1 (Balla & Balla, 2006). Several studies have localized the bulk of wortmannin-sensitive PI4-kinases IIIα and IIIβ to ER and golgi membranes (Wong et al. 1997; Taverna et al. 2002; Balla & Balla, 2006), not to the plasma membrane where replenishment of PIP₂ pools would seem to have most relevance for channel modulation. Moreover, PI4-kinase IIIα has been suggested to be necessary for the replenishment of plasma membrane PIP₂, not the PI4-kinase IIIβ whose activity is known to be augmented by NCS-1 (Balla et al. 2005, 2008). Clearly, the possible regulation of PI4-kinase IIIα as well by NCS-1 needs to be carefully explored. Furthermore, the large literature documenting the control of plasma membrane PIP₂ abundance by wortmannin indicates either greater localization there of type III PI4-kinases than the current literature suggest, or intriguingly, the rapid transfer of PI(4)P from ER to plasma membranes, perhaps at spots of apposition of the two membranes such at those thought intrinsic to I_Crac made by the Stim/Orai complex (Liou et al. 2005; Wu et al. 2006). Likewise, the clear effects of NCS-1 on plasma membrane levels of PIP and PIP₂ are hard to understand without invoking regulation of PI4-kinases. A mechanism may be the rapid translocation of NCS-1/type III PI4-kinases to the membrane upon activation of PLC and rises in [Ca²⁺]_i (Taverna et al. 2002). Or rather, the kinase may translocate to NCS-1 prelocalized to the membrane via its myristoylation, a feature necessary for its association with, and up-regulation of, PI4-kinases (Zhao et al. 2001; Rajebhosale et al. 2003; Zheng et al. 2005).

Our scheme for the dual modulatory pathways acting on M channels is summarized in Fig. 2. It envisions two basic types of G_q/11-mediated signalling mechanisms in SCG neurons. Both involve phospholipase C (PLC) and are underpinned by the need of M channels for membrane phosphatidylinositol 4,5-bisphospate (PIP₂) to function (Suh & Hille, 2002; Ford et al. 2003; Zhang et al. 2003; Li et al. 2005; Suh et al. 2006). The first mechanism is used by M₁ mACh and angiotensin II AT₁ receptors and principally involves depletion of PIP₂ (Zaika et al. 2006; Suh & Hille, 2007). Perhaps due to the lack of spatial colocalization with IP₃ receptors (Delmas & Brown, 2002), stimulation of these receptors does not elicit [Ca²⁺]_i rises, membrane PIP₂ abundance is allowed to fall and the M channels are inhibited. The second mechanism is based on the inositol trisphosphate (IP₃)-mediated rises in intracellular Ca²⁺ induced by bradykinin (BK) B₂ and purinergic P2Y receptors (Cruzblanca et al. 1998; Bofill-Cardona et al. 2000; Delmas et al. 2002; Zaika et al. 2007) and subsequent Ca²⁺ binding to calmodulin (CaM), which acts on the channels (Gamper & Shapiro, 2003; Gamper et al. 2005; Zaika et al. 2007). The ${\text{Ca}}_{\text{i}}^{2+}$ signals produced by these receptors also augment PI4-kinase activity via NCS-1, thereby stabilizing PIP₂ levels. The Ca²⁺/CaM action could involve reduction in the channels' affinity for PIP₂, which would then unbind from the channel proteins. Such competitive or allosteric regulation of the affinity of membrane transport proteins for regulatory PIP₂ molecules as a mechanism for modulation is widespread, and may act as a coincidence-detector motif for spatiotemporal targeting of receptor stimulation to the proper ion channel targets within the cell (Gamper & Shapiro, 2007a,b; Logothetis et al. 2007b; Lopes et al. 2007).

Figure 2. Modes of M channel regulation in sympathetic neurons.

Shown are mechanisms of inhibition of M channels used by two types of G_q/11-coupled receptors in the SCG. Both types activate PLCβ, which in turn hydrolyses PIP₂ to IP₃ and diacylglycerol. The first is used by M₁ muscarinic acetylcholine and AT₁ angiotensin II receptors (left). These agonists are ineffective in producing cytoplasmic Ca²⁺ signals, probably because the receptors are too far away from ER IP₃ receptors, and the IP₃ produced dissipates away (thick red arrow). Thus, much PIP₂ is consumed, PIP₂ unbinds from M channels down the [PIP₂] gradient, and M channels are suppressed. The second is used by bradykinin B₂ and purinergic P2Y₆ receptors (right). Due to the spatial colocalization of these receptors to ER IP₃ receptors, cytoplasmic Ca²⁺ signals are elicited. The released Ca²⁺ binds to neuronal Ca²⁺ sensor-1 (NCS-1) and to calmodulin (CaM). NCS-1 promotes PIP₂ synthesis via acceleration of PI4-kinase, providing PIP₂ to the membrane (purple arrows), and stabilizing PIP₂ levels in the face of PLC activity. CaM binds to carboxy-terminal domains of the channel and likely acts by reducing its affinity for PIP₂, which then unbinds from the channel since tonic [PIP₂] is now insufficient to maintain association with the channel, and M current is suppressed. PLC, phospholipase C; PIP₂, phosphatidylinositol 4,5-bisphosphate; CaM, calmodulin; PI4-kinase, phosphoinositide 4-kinase; NCS-1, neuronal calcium sensor-1 protein; IP₃R, inositol trisphosphate receptor; ER, endoplasmic reticulum.

Neurotransmitter release: presynaptic modulation?

Given the established role for M channels in regulation of action potential firing, it would seem to follow that M current should also control the release of neurotransmitter at release sites. This question can be easily investigated in sympathetic neurons that release noradrenaline (NA) transmitter, which is oxidized at a suitable carbon fibre electrode (CFE) held at a potential greater than the chemical potential of NA (Zhou & Misler, 1995). Figure 3A is a schematic diagram showing the amperometry technique, in which a CFE held at +600 mV is placed proximal to a cluster of SCG neurons (Koh, 2006). A stimulating electrode is placed against the bundle of processes emanating from the cells and a shock train applied sufficient to elicit action potentials at the somas. The spikes in the amperometric record report the exocytosis of vesicles very near the electrode, and the baseline reflects the NA concentration in the solution near the electrode. The area under the amperometric records reflects the total amount of NA released by the shock train. As shown in Fig. 3B, application of the specific M channel opener retigabine (10 μm) blunted the evoked release of NA, whereas Fig. 3C shows that application of the specific M channel blocker XE991 (10 μm) augmented it. Such experiments are summarized in Fig. 3D, showing the significant and reversible effects on evoked release by the M channel specific compounds. These data are consistent with the regulation of neurotransmitter release by M channel activity, likely via the control of action-potential firing in response to excitatory inputs. We predict that neurotransmitters which modulate M channels will likewise regulate presynaptic release, although the issue is made complex by the concurrent modulation of Ca²⁺ channels by stimulation of several of these same G_q/11-coupled receptors. Indeed, stimulation of NA receptors in SCG neurons, which inhibits Ca²⁺ channels, but not M current, has been shown via this same method to strongly reduce the evoked release of NA (Koh & Hille, 1997).

Figure 3. M channels affect the release of NA from cultured sympathetic neurons detected by carbon-fibre amperometry.

A, schematic illustration of experimental conditions. A carbon-fibre electrode (CFE) is placed proximal to a cluster of SCG neurons, and the cells excited by 18 1 ms shocks from an extracellular field electrode (e-stim). The applied shocks induce many action potentials, causing the release of NA, both from the soma, and from varicosities in the neurites that are symbolized as small spots. A multibarrelled solution exchange system for perfusion of experimental solutions completes the set-up. In control conditions (B and C), the train of shocks elicited a total amperometric current that is proportional to the concentration of NA. The effects on NA release of retigabine (B) and XE991 (C) are shown. Consistent with the effects of these compounds on somatic excitability and action potential firing, retigabine caused a strong decrease in released NA, and XE991 caused an increase in released NA, suggesting that the effects of altering M channel activity on neuronal excitability may be paralleled by their effects on neurotransmitter release at nerve terminals. D, bars show the summarized data from these experiments. In the presence of retigabine, the train of shocks evoked a total charge of 68 ± 22% (P < 0.05) of control, and after XE991, the train of shocks evoked a total charge of 148 ± 19% (P < 0.05) of control. The effects of both drugs were fully reversible (O. Zaika and M. S. Shapiro, unpublished observations.)

Much accumulating work shows the role of somatodendritic M channels in control of excitability and neuronal discharge in a variety of peripheral and central neurons (Jones et al. 1997; Passmore et al. 2003; Yue & Yaari, 2004; Gu et al. 2005; Peters et al. 2005; Shen et al. 2005; Lawrence et al. 2006; Otto et al. 2006; Vervaeke et al. 2006; Wladyka & Kunze, 2006; Yue & Yaari, 2006; Zaika et al. 2006, 2007), which must indirectly regulate release of neurotransmitter. However, is the regulation of neurotransmitter release by M channels only through control of action-potential firing, or is there a direct role at the nerve terminal? The experimental evidence on this question is promising. Presynaptic M channels have been suggested to regulate neurotransmitter release of loaded [³H]noradrenaline, [³H]GABA and d-[³H]aspartate from hippocampal synaptosomes (Martire et al. 2004). In hippocampal slices, the M channel opener NH6 blunted spontaneous spiking behaviour and reduced the frequency of spontaneous EPSCs in cultured neurons, as well as evoked discharge, whereas the M channel blocker linopirdine had the opposite effect (Peretz et al. 2007). The simplest explanation for these presynaptic inhibitory actions would be M channel activation that hyperpolarizes the presynaptic terminals, thereby directly or indirectly depressing vesicular transmitter release. This notion is supported by simultaneous ionic current and Ca²⁺ measurements at the calyx of Held that show an exquisite dependence of neurotransmitter release on the resting potential in the nerve terminal, via tuning of tonic [Ca²⁺]_i (Awatramani et al. 2005).

In sympathetic ganglia, preganglionic axons synapse onto the somatodendritic region of the primary ganglionic neurons that we routinely study. It is clear in those cells that suppression of M current alters cell excitability and spike-frequency adaptation (Brown, 1988; Jones et al. 1995; Zaika et al. 2006, 2007). Thus, it would seem that M₁, AT₁, B₂ and P2Y receptors would all increase evoked NA release from sympathetic neurons in vitro (Kubista & Boehm, 2006). The results reported here using CFE amperometry and M channel specific drugs are consistent with the link between M current activity and NA release. However, CFE experiments that assayed the effect of stimulation of receptors that suppress the Ca²⁺ currents in SCG neurons indicated that those acting on Ca²⁺ channels alone, and not M channels, uniformly caused an inhibition of NA release, whereas those that concurrently suppress both types of channels, such as M₁ receptors, had little effect. One explanation may be compensatory excitatory and inhibitory effects on neurotransmitter release by these two types of channels (Koh & Hille, 1997). Indeed, AT₁ receptor stimulation, which also suppresses both types of channels (Shapiro et al. 1994; Zaika et al. 2006), also does not affect NA release (Gobel et al. 2000; our unpublished observations) whereas stimulation of B₂ bradykinin and P2Y₆ purinergic receptors, which inhibit M channels, but not Ca²⁺ channels, results in robust augmentation of NA release (Bofill-Cardona et al. 2000; Edelbauer et al. 2005). Since these effects on NA release are prevented by use of the M channel opener retigabine (Lechner et al. 2003), the actions are mediated via control of M channel activity. It will be interesting to rigorously assay the end-result on neurotransmitter release of receptor stimulation which acts on both types of channels, which may play different roles in regulating synaptic strength.

PIP₂ sensitivity

With the role of PIP₂ in the regulation of M channels established, our lab is currently seeking to determine the site(s) of action of PIP₂ on the channel proteins, and the loci that determine PIP₂ efficacy. Compared to other Kv channels, the C terminus of KCNQ channels is relatively extended, and has been shown to contain the sites of action for several regulatory molecules such as Ca²⁺/CaM, protein kinase C and A-kinase anchoring protein 79/150, and circumstantial evidence suggests PIP₂ acts there as well (Delmas & Brown, 2005). Using recombinant KCNQ subunits studied in single-channel patches, we have found these channels to display divergent open probabilities arising from differential PIP₂ sensitivities (Li et al. 2004, 2005). Based on the existence of highly basic clusters in the C-termini of the channels suggestive of PIP₂-binding domains (Rosenhouse-Dantsker & Logothetis, 2007), as well as the PIP₂ affinity shift caused by neutralization of a histidine residue located very proximal in the C-terminus of KCNQ2 (Zhang et al. 2003), we hypothesize that the C-terminus contains the molecular determinants of PIP₂ regulation. To address this question, we are taking advantage of the highly differential PIP₂ apparent affinities of KCNQ2–4 to construct KCNQ3/4 chimeras containing differing amounts of the C-terminal tail, and are mutating positively charged amino acids in a linker located between the first and second regions of highly conserved sequences in the C-terminus. The chimera data suggest this linker at least partly determines PIP₂ apparent affinity, and is a possible binding site. Point mutants within this linker in KCNQ2 and KCNQ3 further localize the critical region to a cluster of basic residues (unpublished observations).

To gain further insight into possible interactions between PIP₂ and KCNQ channels, a homology model of the C-terminal domain of KCNQ2 was constructed using the solved crystal structure of the cytoplasmic domain of IRK1 (Kir2.1) (Pegan et al. 2005) as a template (Protein Database accession number 1u4f), using SWISS-MODEL (Schwede et al. 2003). Since no crystal structure of the possible modulatory regions of KCNQ C-termini is available, we aligned the 57-residue core (aa 428–484) of our putative PIP₂-binding domain of KCNQ2 to residues 186–245 in IRK, which has been identified as a PIP₂ binding locus in those channels (Zhang et al. 1999). Figure 4A shows a homology model of the linker domain of KCNQ2 (blue) superimposed with the homologous region of IRK1 (red), indicating a marked structural similarity between the two with a CαRMS (carbon α root mean-square) of 0.55 Å. Although such a homology model is highly speculative, it may serve to guide more specific mutagenesis. Figure 4B depicts the two domains individually, with certain basic residues indicated. In the KCNQ2 linker, we are investigating the possible involvement of basic residues K452, R459, R461, R463 and R467 (Fig. 4B, right), which may play similar roles to R189, R218, K219 and R228 (Fig. 4B, left) previously identified in IRK1 as critical for PIP₂ action (Logothetis et al. 2007a). We hypothesize this domain of KCNQ channels to be a locus of PIP₂ binding.

Figure 4. Homology modelling of a putative PIP₂-binding domain within the C terminus of KCNQ2 shows structural similarity to a similar domain in IRK1.

Shown are the results of homology modelling using the program SWISS-MODEL. A, superimposed are the structure of residues 186–245 of IRK1 (Kir2.1) and the predicted structure of residues 428–484 of KCNQ2 using the IRK1 structure as a template. B, shown individually are the homology model of the linker domain of KCNQ2 (blue) and the homologous region of IRK1 (red), with elements of secondary structure indicated. Shown are clusters of basic residues (stick model) located in a putative PIP₂ binding site on IRK1 (Logothetis et al. 2007a), as well as a cluster of basic residues located in the C-terminal domain of KCNQ currently under investigation as critical for PIP₂ regulation. (C. C. Hernandez and M. S. Shapiro, unpublished.)

Conclusions

Given the identification of the KCNQ2 and KCNQ3 gene products that underlie many M channels (Wang et al. 1998) from inherited human epileptic syndromes (Biervert et al. 1998; Charlier et al. 1998; Singh et al. 1998; Lerche et al. 2001), it is unsurprising that these channels are under widespread study for new therapeutic modes for a variety of diseases. Thus, specific M channel openers, such as retigabine, are currently being developed as novel antiepileptic drugs (Cooper & Jan, 2003) and specific blockers, such as XE991 (a linopirdine derivative) (Zaczek et al. 1998) may find use as ‘cognition enhancers’. Since M current plays such a powerful role in tuning synaptic efficacy and neurotransmitter action, we also predict M channels to be potential targets to ameliorate a variety of psychiatric diseases, as well as new treatments for chronic pains, in accord with the identification of M currents in a variety of nociceptive/sensory neurons (Passmore et al. 2003; Wladyka & Kunze, 2006). We look forward to the expanding literature that will indicate the diverse physiological roles of M channels, the molecules that regulate them, and their multifaceted modulatory pathways.