KCNQ-like Potassium Channels in Caenorhabditis elegans

The human KCNQ gene family encodes potassium channels linked to several genetic syndromes including neonatal epilepsy, cardiac arrhythmia, and progressive deafness. KCNQ channels form M-type potassium channels, which are critical regulators of neuronal excitability that mediate autonomic responses, pain, and higher brain function. Fundamental mechanisms of the normal and abnormal cellular roles for these channels may be gained from their study in simple model organisms. Here we report that a multigene family of KCNQ-like channels is present in the nematode, Caenorhabditis elegans. We show that many aspects of the functional properties, tissue expression pattern, and modulation of these C. elegans channels are conserved, including suppression by the M1 muscarinic receptor. We also describe a conserved mechanism of modulation by diacylglycerol for a subset of C. elegans and vertebrate KCNQ/KQT channels, which is dependent upon the carboxyl-terminal domains of channel subunits and activated protein kinase C.

The human genome contains five KCNQ genes that encode a family of K ϩ channel ␣-subunits possessing six transmembrane domains and a single pore loop. Functional channels are assembled from four ␣ subunits, and may be either homo-or heterotetramers, depending on cell type. These channels serve a wide range of physiological roles. In the heart, KCNQ1 (originally designated KvLTQ1) is co-assembled with the product of the KCNE1 gene (variously designated minK, IsK, and MiRP) to form the cardiac I Ks delayed rectifier-like K ϩ current (1,2). Mutations in either KCNQ1 or KCNE1 cause inherited long QT syndrome (LQT1 or LQT5, respectively), a condition leading to arrhythmia (Romano-Ward syndrome), as well as an associated form of deafness (Jervell and Lange-Nielsen syndrome) (3)(4)(5)(6)(7). KCNQ1 can also co-assemble with KCNE3, and may form the channel carrying the basolateral cAMP-regulated K ϩ current present in colonic crypt cells (8,9). KCNQ2/KCNQ3 heteromultimers are thought to underlie the prototypic M-current. Mutations in either of these genes cause an inherited neonatal epilepsy (Benign Familial Neonatal Convulsion, BFNC) (10 -14). The KCNQ4 gene may encode the molecular correlate of I(K,n) in outer hair cells of the cochlea, and I(K,L) in Type I hair cells of the vestibular apparatus, mutations that lead to a form of inherited progressive adult deafness (Autosomal Dominant Non-syndromic Deafness, DFNA2) (15). The recently identified KCNQ5 gene is expressed in brain and skeletal muscle and can co-assemble with KCNQ3, suggesting it may contribute to M-current heterogeneity (16,17), although no linkage to a hereditary disease has yet been reported. Thus, mutations in four of the five human KCNQ genes are associated with hereditary diseases, suggesting a uniquely important role for this class of channels in a variety of physiological functions.
In this article we show that a three-gene family of KCNQlike channels is also present in the nematode worm, Caenorhabditis elegans, which we call KQT channels. Many aspects of their functional properties, tissue distribution, and modulation have striking parallels with their mammalian orthologues. For example, C. elegans pharyngeal muscles possess many cardiac-like properties including electrical coupling and rhythmic myogenic contractions generated by prolonged action potentials (18). We show that kqt-1 is expressed in pharyngeal muscles, where it may serve a role analogous to I Ks in mammalian cardiac muscle. Additional kqt expression was found in nematode mechanosensory neurons, chemosensory neurons, and intestinal cells that may mediate cellular functions analogous to those in vertebrate species.
We also show, that like all mammalian KCNQ channels, C. elegans KQT channels are modulated by the M1 muscarinic receptor that utilizes a G␣ q signaling pathway. We address one potential signaling pathway of receptor-stimulated inhibition of M-currents using kqt and KCNQ genes from both C. elegans and vertebrates. We find that a subset of these kqt/KCNQ genes encode channels that are potently suppressed by submicromolar concentrations of the water soluble diacylglycerol analog, 1-oleoyl-2-acetyl-sn-glycerol (OAG), implicating DAG as a signaling intermediary. By analysis of chimeric subunit constructs, we find that this OAG 1 suppression is mediated through the carboxyl-terminal "tail" domains of OAG-sensitive channel subunits. We observe that OAG-mediated suppression * This work was supported by National Science Foundation Grant IBN-0117341 (to A. W.), National Institutes of Health Grants R01GM067154-02 and R24RR017342-02 (to L. S.), and a pilot grant from the Digestive Diseases Research Core Center and Clinical Nutrition Research Unit at Washington University School of Medicine (to A. W.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence reported in this paper has been submitted to the DDBJ/GenBank TM /EBI Data Bank with accession numbers AY572974 and AY572975. is mimicked by phorbol 12-myristate 13-acetate (PMA), a potent pharmacological activator of protein kinase C (PKC). Pretreatment with staurosporine, a PKC-specific inhibitor effectively blocks OAG inhibition. These results suggest that DAGstimulated PKC may mediate receptor-coupled inhibition of a subset of M-currents through a mechanism involving the carboxyl-terminal tails of OAG-sensitive channel subunits. This mechanism is conserved for a subset of KCNQ/KQT channels in both C. elegans and vertebrates.

EXPERIMENTAL PROCEDURES
Molecular Biology and C. elegans Transformed Strains-Full-length cDNAs of kqt-1, kqt-2, and kqt-3 were generated by a combination of hydridization screens with 32 P-labeled DNA probes, PCR from a commercial C. elegans cDNA library (number 937007, Stratagene, San Diego, CA), and reverse transcriptase-PCR from mRNA extracted from mixed staged wild-type animals, using oligonucleotide primer sequences based on predicted cDNAs from cosmid sequences C25B8 (kqt-1; GenBank TM accession number U41556), M60 (kqt-2; GenBank accession number U39995), and YAC sequence Y54G9a (kqt-3; Gen-Bank accession number AL032648). PCR-generated cDNAs were sequenced and compared with GenBank entries, as well as independently sequenced wild-type genomic fragments in some instances, to resolve sequence discrepancies with GenBank predictions. These cDNA sequences can be accessed with GenBank accession numbers AY572974 (kqt-1) and AY572975 (kqt-3). A full-length kqt-2 cDNA was contained within an EST (yk25b9) kindly provided by Yuji Kohara (National Institutes of Genetics, Mishima, Japan). C. elegans kqt cDNAs were modified by introduction of a Kozak initiation consensus sequence to each initiation methionine, and deletion of all 5Ј non-coding sequences. Full-length KCNQ4 was reconstructed from a partial cDNA (AK074957, obtained from the National Institute of Technology and Evaluation, Department of Biotechnology, Biological Resource Center, Chiba, Japan), with additional 5Ј fragments generated by PCR from cDNAs derived from human brain RNA and genomic DNA (Clontech, Palo Alto, CA), using primers based on GenBank accession number AF105202. Additional cDNAs used in this study were generously provided by Jacques Barhanin (19). All cDNAs were subcloned into pOX (20) for Xenopus oocyte expression studies. To promote efficient heterologous expression in Xenopus oocytes, pOX incorporates 5Ј and 3Ј untranslated sequences from the Xenopus ␤-globin gene into transcribed cRNAs. All constructs were completely verified by sequencing.
C. elegans translational GFP fusion constructs for each kqt gene were generated using subclones of genomic cosmid (C25B8, M60) and YAC (Y54G9a) clones provided by the C. elegans Sequencing Consortium (Genome Sequencing Center, Washington University School of Medicine, and Sanger Institute, Hinxton, UK). Subclones were designed to encompass most or all exons for each kqt gene, and substantial lengths of 5Ј and 3Ј non-coding sequences. The kqt-1::GFP construct was created by first subcloning an 11.6-kb NruI fragment of C25B8, which includes the entire kqt-1 gene and ϳ9.0 kb of 5Ј untranslated and ϳ2.1 kb of 3Ј untranslated sequences, into pBluescriptII KSϩ (Stratagene, San Diego). This genomic subclone was then modified by the insertion of a GFP cDNA sequence in-frame at a unique SpeI site within exon 16 (immediately 3Ј of the codon corresponding to T672), preserving the native 3Ј splice site of exon 16. The kqt-2::GFP construct was created by subcloning a 11.9-kb NsiI fragment from Cosmid M-60, encompassing the first 12 exons of kqt-2 and ϳ9.0 kb of 5Ј non-coding sequence, into pPD95.81 (gift of Andy Fire, Carnegie Institution of Washington), linearized with PstI. This construct thus produces a 3Ј translational fusion of KQT-2 at Ala-605 with GFP, lacking the last 70 predicted carboxylterminal amino acids encoded in exon 13. The kqt-3::GFP construct was created by subcloning a ϳ5.0-kb NsiI/NheI fragment amplified from Y54G9a by PCR, into pPD95.75 (gift of Andy Fire, Carnegie Institution of Washington). This construct encompasses all kqt-3 exons and ϳ1.5 kb of the non-coding sequence 5Ј of the initiation methionine, and creates a translational fusion of GFP with the C terminus of KQT-3. Two additional overlapping PCR-generated genomic fragments were made, encompassing sequences ϳ6.2 kb further 5Ј of the initial subcloned kqt-3 fragment. Coinjection of these PCR-generated fragments with the 5Ј linearized kqt-3::GFP construct yielded lines with stable labeling of chemosensory neurons, because of transgenes presumably generated by efficient in vivo recombination (21). Transformed C. elegans strains were created by the standard germline injection technique (22), using ϳ20 -50 ng/l of each construct and rol-6(ϩ) as a selectable marker. GFP-positive chemosensory neurons were identified in L1 staged larvae, assisted by co-labeling amphid sensory neurons (ASK, ADL, ASI, AWB, ASH, and ASJ) with DiI (Molecular Probes, Eugene, OR), as landmark cells (23).
Electrophysiology-Two-electrode voltage-clamp and patch clamp recordings were made from Xenopus oocytes injected with cRNAs, as previously described (20). Dose-response series were obtained with a recording chamber with a volume of ϳ150 l, and solution changes with ϳ3 times the recording chamber volume, applied manually without plastic tubing to minimize potential error because of retention of lipophilic reagents to tubing. Two-electrode voltage-clamp measurements were made at steady-state, typically 2-3 min after drug application. Between drug series, all recording surfaces were flushed with 70% ethanol to remove residual drug samples, then with ND96 recording solution. Stock solutions of drugs were dissolved in Me 2 SO, stored at Ϫ20°C, and diluted in ND96 for each experimental series. In all instances, final Me 2 SO concentrations never exceeded 1% (v/v). Control experiments showed that 1% Me 2 SO had no effect on either endogenous or heterologously expressed currents in Xenopus oocytes. All measurements were made with 1.0 mM DIDS (Sigma) to block endogenous Ca 2ϩ -activated Cl Ϫ currents present in Xenopus oocytes. Composition of ND96 (in mM) was: 96 NaCl, 2.0 KCl, 1.8 CaCl 2 , 1.0 MgCl 2 , 5.0 HEPES (pH 7.2).
Excised inside-out patch-clamp recordings were made under symmetric 160 mM K ϩ recording conditions, with zero Ca 2ϩ and nearly Cl Ϫ free solutions. Recordings were obtained and low-pass filtered at 2. Linopirdine was a gift from Barry Brown (DuPont Pharmaceuticals, Wilmington, DE). OAG, phorbol 12-mystristate 13-acetate (PMA), and poly-lysine HCl (15-30,000 M r ) were purchased from Sigma. Staurosporine and oxotremorine were purchased from Tocris Cookson (Ellisville, MO).

RESULTS
A KCNQ-like Multigene Family in C. elegans-As in humans, KCNQ-like genes are represented in C. elegans as a multigene family. The C. elegans genome encodes 71 potassium channel genes (24), of which three have similarity to KCNQ genes, forming a family of genes we call kqt for "K ϩ channel related to QT interval." The three members of this family are kqt-1 (C25B8.1, Wormbase notation), kqt-2 (M60.2), and kqt-3 (Y54G9a.3). Primary sequence alignments revealed two domains of high conservation (ϳ70% identity), defining the "core" transmembrane segments S1-S6, and an additional region of ϳ115 residues within the putative cytoplasmic carboxyl-terminal tail, implicated in subunit multimerization (25, 26) (Supplemental Materials Fig. S1). Phylogenetic comparison of all C. elegans KQT and human KCNQ primary sequences identified two conserved subfamilies: C. elegans KQT-1 defining one subfamily with KCNQ2-5, whereas C. elegans KQT-3 defined a second subfamily with KCNQ1. KQT-2 did not group with either subfamily, nor did it possess a human ortholog. However, BLAST similarity searches clearly showed higher KQT-2 similarity in the vertebrate KCNQ gene family than to any other vertebrate potassium channel gene family (Fig. 1A).
All vertebrate KCNQ channels can be modulated through the M1 muscarinic receptor (17,27). Significantly, we observed that this mode of regulation was conserved with C. elegans subunits, by coexpressing KQT-1 or KQT-3 with the human M1 muscarinic receptor in Xenopus oocytes. KQT-1 or KQT-3 currents expressed in these experiments were rapidly suppressed by bath application of the muscarinic agonist, oxotremorine (10 M) (Fig. 2, E and F). M1 mediated suppression was more effective for KQT-1 (ϳ90% inhibition) than for KQT-3 (ϳ16% inhibition) (Fig. 2G). These experiments demonstrate that C. elegans KQT channels encode M-currents, which are functionally similar to M-currents encoded by vertebrate KCNQ channels.
A Subset of C. elegans KQT and Vertebrate KCNQ Channels Are Suppressed by Submicromolar OAG-The lipid metabolite diacylglycerol generated by phospholipase C-mediated hydrolysis of phosphatidylinositol 4,5-bisphosphate (PIP 2 ) has been described as a potential signaling intermediary for receptorcoupled inhibition of native M-currents in gastric smooth muscles (29) and perhaps other tissues (30). To test if DAG may mediate inhibition of cloned KQT channels, we applied the water soluble DAG analog, OAG, to oocytes expressing KQT-1 or KQT-3 channels. Potent suppression of KQT-1 currents was observed with submicromolar concentrations of OAG (IC 50 ϭ 0.2 M), resulting in ϳ80% maximal inhibition at saturating concentrations, within 2 min (Fig. 3, A and B). KQT-3 currents were also suppressed by OAG, but with ϳ1000-fold lower sensitivity (IC 50 ϭ 201 M) (Fig. 3B).
To test whether OAG-mediated inhibition observed with C. elegans KQT channels is conserved with vertebrate KCNQ channels, we examined the OAG sensitivity of currents carried by all the vertebrate KCNQ channels (KCNQ1-5) expressed as homomeric channels in Xenopus oocytes. Of these vertebrate channels, KCNQ5 currents were uniquely sensitive to inhibition by OAG at submicromolar concentrations (IC 50 ϭ 0.1 M), similar to C. elegans KQT-1 (Fig. 3C). Other KCNQ currents examined were far less sensitive to OAG inhibition, similar to C. elegans KQT-3, requiring ϳ1000-fold greater OAG concentrations for inhibition (KCNQ1, IC 50 ϭ ϳ717 M; KCNQ3, IC 50 ϭ ϳ1.1 mM; (Fig. 3C) (KCNQ2 and KCNQ4 were similarly uninhibited by 125 M OAG (data not shown)). Because KCNQ5 forms heteromeric channels with KCNQ3 in heterologous expression systems (16,17), and most likely in vivo (31), we examined the OAG sensitivity of currents produced by co-expression of KCNQ5 and KCNQ3 cRNAs. In agreement with these previous reports, we observed that KCNQ3/5 coinjected oocytes produced currents with amplitudes 5-8 times greater than that predicted by the linear sum of individually injected KCNQ3 and KCNQ5 oocytes, consistent with efficient heteromeric channel formation. Moreover, heteromeric KCNQ3/5 currents, like homomeric KCNQ5 currents, were effectively suppressed by OAG at submicromolar concentrations (IC 50 ϭ 0.6 M) (Fig. 3D). Vertebrate KCNQ5 and C. elegans KQT-1 subunits thus define a conserved molecular subclass of M-channels with high sensitivity to OAG. Furthermore, KCNQ5 confers high OAG sensitivity to heteromeric channels formed with KCNQ3 subunits.
Suppression by OAG Is Mediated Through the COOH-terminal Tail of OAG-sensitive KQT/KCNQ Subunits-To investigate the structural regions necessary for OAG-mediated inhibition of channels formed by KCNQ and KQT subunits, we generated a series of chimeric constructs between subunits with high and low sensitivity. Other functional classes of potassium and non-selective cation channels gated by cytosolic  4). E, suppression of KQT-1 currents in oocytes coexpressing KQT-1 and human M1 muscarinic receptor, before (left) and 2 min after application of muscarinic agonist (10 M oxotremorine) (right). Currents evoked by voltage steps from Ϫ80 to 10 mV, in 10-mV increments. F, partial suppression of KQT-3 currents in oocytes coexpressing KQT-3 and human M1 muscarinic receptor, before (left) and 2 min after application of 10 M oxotremorine (right). Currents evoked by voltage steps from Ϫ80 to 20 mV, in 20-mV increments. G, summary of data for E and F. Activated M1-coupled suppression of KQT-1 (ϳ10% of maximal current; n ϭ 8) was more effective than KQT-3 suppression (ϳ85% of maximal current; n ϭ 16). Conductance data (C) fitted by the single Boltzmann function, G/G max ϭ (1 ϩ exp(Ϫs(V Ϫ V 50 )/kT)) Ϫ1 , where G/G max ϭ fractional maximal conductance, s ϭ slope factor reflecting intrinsic voltage sensitivity, V 50 ϭ voltage of half-maximal activation, T ϭ absolute temperature, and k ϭ Boltzmann constant. Conductance was calculated based on a potassium reversal potential of Ϫ90 mV. Dose-response data (D) fitted by the Hill function, I ϭ 1 Ϫ (I max /1 ϩ (IC 50 /[ligand]) s ), where I ϭ fraction maximal current, I max ϭ maximal current inhibition, IC 50 ϭ concentration of half-maximal current inhibition, and s ϭ Hill coefficient, with I max fixed at 1.0. All data were plotted with S.E. ligands (Slo, SK, and CNG) are encoded by subunits that appear to be modularly composed of a core voltage-sensing and pore-forming domain (S1-S6), linked to unique carboxyl-terminal tail structures that modify gating (20,32,33). By analogy, we reasoned that OAG may exert its effects through carboxylterminal domains conserved in OAG-sensitive KCNQ and KQT subunits.
We first examined chimeric channels formed by exchanging carboxyl-terminal tail domains between C. elegans KQT-1 and KQT-3 subunits. Attaching the tail from the OAG-sensitive KQT-1 subunit to the core of KQT-3 created a functional chimeric subunit (K3 CR /K1 TL ) with high OAG sensitivity, essentially identical to wild-type KQT-1 channels (Fig. 3E). To address if this structural requirement for inhibition by OAG may employ an evolutionarily conserved mechanism, an analogous chimeric subunit was generated attaching the tail from the vertebrate OAG-sensitive subunit, KCNQ5, to the core of C. elegans KQT-3 (K3 CR /Q5 TL ). As with K3 CR /K1 TL , the interspecies K3 CR /Q5 TL chimeric subunit also produced channels with high OAG sensitivity, and an OAG dose-response profile essentially identical to wild-type KCNQ5 (Fig. 3F). High OAG sensitivity could thus be conferred to C. elegans KQT-3 by substituting the tail from the OAG-sensitive channel subunits from either vertebrate or C. elegans species.
Conversely, low OAG sensitivity was also conferred by attaching the tail from vertebrate KCNQ1 to the core of C. el-egans KQT-1 (K1 CR /Q1 TL ) (Fig. 3G), although the analogous intraspecies chimera K1 CR /K3 TL failed to express (see construct 1.1, Fig. 4). Results from additional interspecies core and tail chimeric combinations were consistent with a determinative role of the tail in OAG sensitivity. Thus, chimeric subunits with the core of C. elegans KQT-1 and tail of vertebrate KCNQ5 (K1 CR /Q5 TL ) exhibited high OAG sensitivity (IC 50 ϭ 0.032 M), whereas chimera with the core of C. elegans KQT-3 and tail of vertebrate KCNQ1 (K3 CR /Q1 TL ) exhibited low OAG sensitivity (IC 50 ϭ 200 M) (data not shown). Taken together, these results are consistent with OAG sensitivity being determined by the carboxyl-terminal tails of KCNQ/KQT subunits.
Although we determined that the full-length carboxyl-terminal tails fully confer either high or low OAG sensitivities, attempts to localize OAG sensitivities to a smaller domain within the tails were unsuccessful (Fig. 4). No single tail subregion was able to confer complete OAG sensitivity. Our results suggest that the structural requirements for OAG sensitivity are possibly complex and not mediated through a single subregion of the carboxyl-terminal tail, despite the apparent functional modularity of the entire tail domain.
Suppression Is Not Mediated by Direct Binding of OAG to the Channel Protein-Because DAG directly interacts to alter the gating of cyclic-GMP gated cation channels (34 -36) and a subset of TRPC cation channels (37, 38), we tested the possible ability of OAG to directly inhibit KQT/KCNQ channels in ex- cised inside-out patches. We reasoned that OAG may exert its inhibitory effect on OAG-sensitive KCNQ/KQT channels either by a direct interaction with carboxyl-terminal tail domains, or indirectly through the stimulation of PKC-dependent phosphorylation of the channel protein or other accessory proteins. To test the possibility of a direct action of OAG on the channel protein itself, we perfused OAG onto the cytoplasmic face of excised inside-out patches containing KQT-1 channels, in the absence of ATP to preclude any kinase activity possibly retained on the patch. Inhibition of KQT-1 activity was not observed after application of 100 M OAG, when patches were subjected to either steady membrane depolarization, or a set of voltage step pulses (Fig. 5, A and B).

FIG. 3. Inhibition of wild-type C. elegans KQT and vertebrate KCNQ channels by OAG, and inhibition of chimeric constructs made by exchanging cores (CR) and tails (TL). Inhibition is mediated through the COOH-terminal tail of OAG-sensitive subunits. A, dose-dependent inhibition of KQT-1 currents in response to bath applied OAG concentration series (in
Under our recording conditions, KQT-1 channels in excised patches also exhibited no evidence of "run-down." Similarly, excised patches containing KCNQ5 channels failed to exhibit run-down (Fig. 5E). These results are in contrast to the rapid run-down, which has been described for mammalian KCNQ channels (39,40), which we also observed in excised patches containing KCNQ2/3 or KCNQ5/3 heteromeric channels under our recording conditions (Fig. 5, C and D). Because run-down for many channels is attributed to rapid PIP 2 turnover by active lipid phosphatases present in Xenopus oocyte membranes (41), run-down has been interpreted as evidence supporting the hypothesis that the loss of endogenous PIP 2 may mediate receptor-coupled inhibition of KCNQ channels (39,40,42). Our results suggest that KQT-1 and KCNQ5 channels may be particularly insensitive to a decrease of PIP 2 from endogenous levels, and that additional signaling mechanisms may be utilized by these channels. Consistent with this interpretation, KQT-1 channels in inside-out patches exposed to poly-lysine (30 g/ml) at a concentration reported to block PIP 2 effects on KCNQ channels (40), showed no inhibitory effect (data not shown).
Thus, OAG inhibits KQT-1 channels through a mechanism that does not involve direct binding of OAG to the channel protein. Furthermore, our observations suggest that fluctuations of PIP 2 levels near the endogenous membrane concentrations in Xenopus oocytes may not contribute significantly to modulating either KQT-1 or KCNQ5 channel activity, under our recording conditions. OAG Suppression Acts through Activated PKC-To test the possible involvement of PKC in mediating suppression of KQT-1, a pharmacological activator and inhibitor of PKC were applied to oocytes expressing KQT-1, assayed by two-electrode voltage clamp. Bath application of the potent PKC activator PMA at 20 nM rapidly suppressed KQT-1 currents with kinetics that mimic those observed with 1.25 M OAG (Fig. 6, A-D) (for OAG, 82% suppression of maximal current, ϭ 54 s; for PMA, 86% suppression of maximal current, ϭ 56 s). OAG-mediated suppression of KQT-1 current was blocked nearly completely by pretreating oocytes with the PKC-specific inhibitor, staurosporine (1 M, 8-h soak) (Fig. 6, A and D). These results suggest that nearly all of the suppression of KQT-1 currents that we observe with OAG can be explained through the action of activated PKC. channels linked to diverse hereditary diseases (28). However, many aspects of their role in cellular functions and their characteristic modulation by G-protein receptor-coupled signaling remain to be clarified. Conserved mechanisms underlying these processes may be revealed by the study of KCNQ orthologs in model organisms. We describe here the characterization of a small family of KCNQ-like genes in C. elegans (kqt-1, -2, -3) with high functional conservation, and tissue expression profiles in many cell types that may be analogous to those expressing vertebrate KCNQ genes. Heterologous expression of kqt-1 and kqt-3 revealed currents similar to vertebrate KCNQ currents with respect to kinetics of gating, voltage dependence, pharmacology, and the ability to be suppressed by activation of the M1 muscarinic receptor. Sequence comparison of the 3 C. elegans and 5 human members of this gene family revealed two distinctly conserved subfamilies: one defined by C. elegans KQT-3 and human KCNQ1, and a second defined by C. elegans KQT-1 and vertebrate KCNQ2-5. The functional correlates of this ancient subdivision of the KCNQ/KQT gene family are unclear. However, it is interesting to note that among the vertebrate subunits, functional heteromeric channels form only between KCNQ3 and other subunits of the second subfamily (KCNQ2-5) (28). By contrast, KCNE accessory subunits assemble only with KCNQ1 and not KCNQ2-5 (26,28). It remains to be fully tested whether these functional distinctions among vertebrate KCNQ genes are conserved with the C. elegans channels. We have coexpressed KQT-1 with vertebrate KCNE1, and failed to detect any obvious changes in gating kinetics (data not shown).

Conservation of KCNQ Channels between
A distantly similar C. elegans KCNE-like gene (mps-1) has been reported (43) that modifies the gating kinetics of a C. elegans voltage-gated potassium channel (KVS-1), but its ability to modify KQT channel activity is unknown.
Receptor-coupled Inhibition of KCNQ/KQT (M-current) Channels-The ability of M-channels to act as robust effectors for a variety of G-protein-coupled receptors provides a linkage by which diverse extracellular ligands may modulate sustained changes in membrane excitability. A common requirement for G-protein-coupled receptor-coupled inhibition of M-channels appears to be receptor specificity for G␣ q/11 , mediating receptorstimulated phospholipase C activity (45)(46)(47). Phosphatidylinositide metabolites have thus been implicated as important second messenger molecules required for further downstream signaling steps. PIP 2 is one likely second messenger (39 -42), however, other observations suggest that additional molecules may be employed. Other candidate second messengers include diacylgycerol (29,30), Ca 2ϩ (48), Ca 2ϩ -calmodulin (49), calcineurin (50), Src tyrosine kinase (51), and cyclic ADP-ribose (52).
We provided evidence that diacylgycerol can act as a potent second messenger for a molecular subset of M-channels encoded by kqt-1 in C. elegans and KCNQ5 in vertebrates. Using the water-soluble DAG analog, OAG, we found that among the channels formed by KQT/KCNQ subunits, only KQT-1, KCNQ5, and KCNQ5/3 heteromeric channels are uniquely sensitive to inhibition by submicromolar concentrations of OAG. Through an analysis of chimeric constructs, we demonstrated that high OAG sensitivity is dependent upon the carboxylterminal tail domains of OAG-sensitive subunit. The mechanism mediating inhibition through the tail is likely conserved because OAG sensitivity can be transferred in cross-species chimeric subunits. This mechanism acts in a dominant fashion, because OAG-sensitive KCNQ5 subunits conferred high OAG sensitivity to heteromeric channels composed of KCNQ5 and OAG-insensitive KCNQ3 subunits. However, we were unable to identify a more restricted subdomain of the carboxyl-terminal tail that could fully confer either high or low OAG sensitivity. We interpret from these results that OAG inhibition acts through a conserved mechanism requiring the carboxyl-terminal tail domain, and that the structural requirements for inhibition may be complex, unlikely transferable by any single subregion of the tail.
We also concluded from pharmacological experiments that the inhibitory action of OAG on KQT-1 acts through activated PKC. This interpretation is consistent with our inability to observe a direct inhibitory effect of OAG (100 M) applied onto excised inside-out patches containing KQT-1 channels. The mechanism of PKC action is unaddressed by our findings, but we hypothesize that activated PKC may phosphorylate either the carboxyl tail domain of OAG-sensitive channel subunits, or carboxyl tail-associated accessory proteins to inhibit KQT-1 channel activity, and that a similar mechanism may underlie OAG-mediated inhibition for KCNQ5 channels. Sensitivity of KCNQ channels to activated kinases may also depend upon assembly into signaling complexes by scaffold molecules such as Yotiao and AKAP150 (53,54). Although scaffolding molecules that bind either KQT-1 or KCNQ5 have not been identified, our interpretations based on expression in Xenopus oocytes could thus be modified in the context of other cell types.
Our results suggest that DAG mediates receptor-coupled inhibition for a molecular subset of M channels. This signaling mechanism may act in parallel to that proposed for PIP 2 depletion (39 -42). This OAG signaling mechanism may be particularly relevant for membrane environments with low endogenous levels of PIP 2 (55), given the submicromolar potency that we observe for OAG. Our specific failure to observe run-down with excised patches containing KQT-1 or KCNQ5 suggests that fluctuations near endogenous levels of PIP 2 may not be a prominent mechanism for inhibition of these channels, under our recording conditions. Multiple signaling intermediaries may thus mediate receptor-coupled inhibition of M-channels to allow flexibility for the modulation of this important class of potassium channels in a variety of cellular contexts. Future analysis of native KQT-1 channels in C. elegans cells (56) may allow an examination of these issues in the context of specifically engineered genetic backgrounds.