Identification of channeis underiying the M-like potassium

current in NG108-15 neurobiastoma-giioma ceiis

A thesis submitted for the degree of Doctor of Philosophy

University of London

by

Jennifer Kathleen Hadley

Department of Pharmacology

University College London

Gower Street

London WC1E 6BT ProQuest Number: U644085

All rights reserved

INFORMATION TO ALL USERS The quality of this reproduction is dependent upon the quality of the copy submitted.

In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if material had to be removed, a note will indicate the deletion. uest.

ProQuest U644085

Published by ProQuest LLC(2016). Copyright of the Dissertation is held by the Author.

All rights reserved. This work is protected against unauthorized copying under Title 17, United States Code. Microform Edition © ProQuest LLC.

ProQuest LLC 789 East Eisenhower Parkway P.O. Box 1346 Ann Arbor, Ml 48106-1346 Abstract

NG108-15 cells express a potassium current resembling the IVI-current found in sympathetic ganglia. I contributed to the identification of the channels underlying this NG108-15 current. I used patch-clamp methodology to characterise the kinetics and pharmacology of the M-like current and of three candidate channel genes, all capable of producing “delayed rectifier” currents, expressed in mammalian cells.

I studied two Kvi .2 clones: NGK1 (rat Kvi .2) expressed in mouse fibroblasts, and MK2

(mouse brain Kvi .2) expressed in Chinese hamster ovary (OHO) cells. Kvi.2 showed relatively positive activation that shifted negatively on repeated activation, some inactivation, block by dendrotoxin and various cations, and activation by niflumic acid. It had persuasive differences from the M-like current. mErg is the mouse homologue of the cardiac ether-a-go-go-related gene channel HERG. Having discovered that the HERG blocker WAY-123,398 partially inhibited the M-like current, I studied mErgla in OHO cells. mErg la , as previously reported, undergoes intense inactivation, removed on hyperpolarisation to give a transient increase in current preceding deactivation. I report that it can also produce a sustained current with kinetic and pharmacological correspondences to the slow component of the M-like current.

Blocking the presumed mErg component in NG108-15 left a current very similar to the sympathetic ganglion M-current. The latter was recently suggested to comprise a heteromultimer of KCNQ2 and KCNQ3, two neuronal relatives of the cardiac KvLQT 1. I expressed KCNQ2 and KCNQ3 in CHO cells, separately and together. Their kinetics match those of the M-current. The M-current blocker linopirdine blocks these channels indiscriminately, while tetraethylammonium blocks KGNQ2 > KCNQ2/KCNQ3 > KCNQ3. Since the selective blockers linopirdine and WAY-123,398, applied together, eliminate the total M-like current, Kvi .2 and other channels are unlikely to contribute. I conclude that blockers, when backed up by kinetic and expression data, were a particularly valuable tool for identifying M-like current components. Table of contents

Title...... 1 Abstract...... 2 Table of contents...... 3

Additional material not bound in...... 6

List of figures...... 6 List of tables...... 7

List of abbreviations and constants...... 8 Nomenclature and homologues of potassium channels...... 9 Acknowledgements...... 10 Preface: the M-current...... 11 Chapter 1 : Introduction - Sympathetic M-currents and the M-like current in NG108-15 cells.

1 2 Background to the M-current...... 12 Discovery of the M-current...... 13 Properties of the M-current in sympathetic cells...... 13 The NG108-15 neuroblastoma-glioma hybrid line ...... 14 Properties of the NG108-15 M-like current...... 15 Pharmacology: potassium channei blockers suggest heterogeneity of “M-currents”...... 18 Receptor effects on the M-like current...... 18 Potassium channels in the cloning era...... 19 Molecular identification of M- and M-like channels...... 22 My work...... 22 Chapter 2: Methods...... 23 Cell culture...... 23 T ransfection...... 24 Electrophysiology ...... 25 Patch pipettes...... 25 Considerations relating to internal solutions...... 25 Compositions of internal solutions...... 27 Drug addition...... 28 Hardware and software...... 28 Electrodes...... 28 Instrumentation: (1) Axoclamp-2...... 29 Recording procedure with the Axoclamp-2...... 29 Instrumentation: (2) Axopatch-200 recordingprocedure ...... 30 Compensation of series resistance...... 31 Analysis ...... 33 Leak-subtraction of data records...... 33 Fitting deactivation tails...... 33 More detail on fitting equations...... 35 Chemicals...... 36 Stably transfected cell lines and clones...... 36 Chapter 3: Kvi .2 ...... 37 3.1 : Kvi .2 - a candidate for the M-iike current?...... 37 Sites of expression of Kvi .2...... 37 NG108-15 ceiis: NGK1 and NGK2 ...... 38 Kinetics of Kvi .2...... 38 Potassium channel blockers acting on Kvi .2...... 39 Neurotransmitter modulation of Kvi .2...... 40 Potential complications: heteromultimers...... 40 Beta subunits...... 40 Other molecules associated with Kvi .2 channels...... 41 Summary of the case for Kvi .2 as an M-current constituent...... 42 3.2: Results...... 43 rKvl .2: the CL1023 line...... 43 Conductance curves...... 43 Activation kinetics...... 46 Subthreshold inactivation phenomena...... 56 Deactivation...... 56 Pharmacology ...... 61 mKvl .2: the MK2 line...... 69 Conductance curves...... 69 Kinetics...... 74 Pharmacology ...... 79 3.3: Discussion...... 84 Activation and conductance data...... 84 Deactivation...... 85 Dendrotoxin block...... 85

Enhancement by repeated activations, temperature and niflumic acid...... 8 6 Kvi .2 as an M-iike current constituent?...... 87 Chapter 4: mErgla and the slow NG108-15 M-like current...... 89 4.1 : Introduction: ERG channels...... 89 HERG homologues and accessory subunits...... 90 ERG channels and the M-like current...... 91 4.2: Results...... 93 Heterologously-expressed mErgla and the NG108-15 slow M-like current...... 93 Kinetics...... 93 Inactivation and de-inactivation...... 93 Current/voltage relationships: deactivation and its measurement...... 107 Reversal potential...... 107 Deactivation time constants...... 108 Activation...... 115 NG108-15 current dissection...... 118 Pharmacology ...... 122 Accommodation...... 128 4.3: Discussion...... 131 Kinetics...... 132 Blockers...... 133 Firing...... 134 Chapter 5: KGNQ2, KGNQ3 and the fast NG108-15 M-like current...... 135 5.1: Introduction - KGNQ channels...... 135 KGNQ channels in SGG and NG108-15 cells...... 136 5.2: Results...... 138 Heterologously-expressed KGNQ channels and the NG108-15 fast current...... 138 Kinetics...... 138 Reversal potential...... 141 Conductance...... 141 NG108-15 current dissection...... 149 Pharmacology ...... 159 5.3: Discussion...... 168 Kinetics...... 169 Blockers...... 170 TEA and stoichiometry ...... 171

Chapter 6 : General discussion and conclusions...... 173 Selectivity of drugs...... 176 Functional relevance of KGNQ and ERG currents...... 177 ERG currents and neuronal differentiation...... 179 ERG currents and the cell cycle...... 179 Stoichiometry, assembly and membrane complexes of KGNQ channels...... 180 Future directions...... 181 References...... 183

Additional material not bound In

Hadley, JK, Noda, M, Seiyanko, AA, Wood, 1C, Abogadie, PC & Brown, DA (2000). Differential tetraethylammonium sensitivity of KCNQ1-4 potassium channels. Brit J. Pharmacol. 129, 413-415.

Seiyanko, AA, Hadley, JK, Wood, 10, Abogadie, PC, Delmas, P, Buckley, NJ, London, B & Brown, DA (1999). Two types of K"^ channel subunit, Ergi and KGNQ2/3, contribute to the M-iike current in a mammalian neuronal cell. J. Neurosci. 19, 7742-7756.

Seiyanko, AA, Hadley, JK, Wood, 10, Abogadie, PO, Jentsch, TJ & Brown, DA (2000). Inhibition of K0NQ1-4 potassium channeis expressed in mammalian ceiis via Mi muscarinic acetyicholine receptors. J. Physiol. (Lond.) 522, 349-355.

List of figures

Figure 1.1: The NG108-15 M-like current...... 17 Figure 1.2: Genetic relatedness of potassium channels (taken from Ooetzee et al. 1999).... 21 Figure 3.1 : Conductance of rKvl .2 in mouse fibroblast ceils...... 45

Figure 3.2: Activation is for rKvl .2 ...... 48 Figure 3.3: Acceleration of rKvl .2 by repeated activation...... 50 Figure 3.4: Effects of temperature on rKvl .2 activation...... 53 Figure 3.5: “Hysteresis” in rKvl .2 suggests complex gating behaviour...... 55 Figure 3.6: Steady-state inactivation of rKvl .2 ...... 58 Figure 3.7: Deactivation of rKvl .2 from +20 m V...... 60 Figure 3.8: Dendrotoxin consistently blocks Kvi .2...... 63 Figure 3.9: Effect of nifiumic acid on rKvl .2 activation ...... 65

Figure 3.10: Niflumic acid evokes negligible dendrotoxin-sensitive currents...... 6 8 in NG108-15 ...... 6 8 Figure 3.11 : Exampie of an MK2 current...... 71 Figure 3.12: Conductance of Kvi .2 in MK2 CHO cells...... 73 Figure 3.13: Activation time constants for mKvl .2...... 76 Figure 3.14: Deactivation of mKvl .2 currents from +12 m V...... 78 Figure 3.15: Concentration/response curves on mKvl .2...... 81 Figure 3.16: Effect of niflumic acid on mKvl .2...... 83 Figure 4.1 : Two examples of mErgI a in CHO cells...... 96 Figure 4.2: Separation of NG108-15 M-like current constituents...... 101 Figure 4.3: a direct comparison of mErgla and the NG108-15 slow current...... 104 Figure 4.4: Selectivity and concentration-dependence of WAY-123,398 block...... 106 Figure 4.5: Fitting methods for mErgla and the NG108-15 slow current...... 110 Figure 4.6: Voltage-sensitive de-inactivation of mErgla and the NG108-15 slow current... 112 Figure 4.7: mErgla and NG108-15 slow current kinetics compared...... 114 Figure 4.8: mErgla activation and deactivation compared...... 117 Figure 4.9: Dissection and reconstitution of the M-like current...... 121 Figure 4.10: Concentration/response curves for four blockers on the mErgla current 124 Figure 4.11: Concentration/response curves for four blockers on the NG108-15 slow current 126 Figure 4.12: Effects of KGNQ and ERG block on NG108-15 firing...... 130 Figure 5.1 : Examples of KGNQ2+KCNQ3 currents...... 140 Figure 5.2: Two conductance/voltage curves for KGNQ2+KGNQ3...... 144 Figure 5.3: Two conductance/voltage curves for KGNQ2...... 146 Figure 5.4: Two conductance/voltage curves for KGNQ3...... 148 Figure 5.5: KGNQ2+KGNQ3 and the NG 108-15 fast current compared...... 151 Figure 5.6: Partial current/voltage relationships for KGNQ2+KGNQ3 and the NG108-15 fast current...... 153 Figure 5.7a: Deactivation traces and fits for KGNQ2+KGNQ3 and NG108-15 fast currents 155 Figure 5.7b: Deactivation kinetics of KGNQ2+KGNQ3 and NG108-15 fast currents...... 156 Figure 5.8: Deactivation as 1/xfor KGNQ2+KGNQ3 and NG108-15 fast currents ...... 158 Figure 5.9: Selectivity of linopirdine ...... 162 Figure 5.10: Goncentration/response curves on KGNQ channels...... 164 Figure 5.11 : Goncentration/response curves on the NG 108-15 fast current...... 166

List of tables

Table 1 : Deactivation time constants for the fitted curves shown in Figure 4.2...... 102

Table 2 : IG50 values for inhibition of mErgla and NG108-15 slow current by various drugs 127 Table 3: Block of the NG 108-15 fast current and of KGNQ currents by linopirdine, azimilide and TEA...... 167 List of abbreviations and constants

ATP: adenosine 5’-triphosphate cAMP: cyclic adenosine monophosphate CHO: Chinese hamster ovary DMSG: dimethylsulphoxide D IX : dendrotoxin ERG: ether-a-go-go-related gene e: exponential constant, 2.71828... F: Faraday constant, 96480 C/mol G: conductance G A/: conductance/voltage curve G protein: GTP-binding protein GTP: guanosine 5’-triphosphate HEK: human embryonic kidney I: current

IC50: half-maximal inhibitory concentration

IP3: inositol trisphosphate I A/: current/voltage curve MW: molecular weight PKA: protein kinase A PKC: protein kinase C PLC: phospholipase C R: gas constant, 8.314 JK^mol'^ SCG: superior cervical ganglion s.e.m.: standard error of the mean I : absolute temperature, taken as 293K in this work TEA: tetraethylammonium THA: 9-amino-1,2,3,4-tetrahydroacridine UTP: uridine triphosphate V: voltage (either a command voltage imposed by pCIamp, or a recorded voltage between the recording electrode and the bath solution) Vrev: reversal potential (at which ion flow through channels changes direction) V%: half-maximal activation voltage (at which half of the available channels are active) t : time constant for exponential growth or decay Z: charge on an ion, +1 for potassium. Nomenclature and homologues of potassium channels

The KCN... terminology follows the system supported by the HUGO Gene Nomenclature Committee for human genes, and is being adopted for homologues in other species. For K channels, the gene names begin with KCN. Aii Kvi channel gene names have the stem KCNA; all members of the Eag family have KCNH, and so on. Clones I have used are shown in bold text.

Channei name Species Synonym(s), Reference

Kvi .2 KCNA2 OMIM database* mouse MK2 Pongs (1992) rat NGK1 Pongs (1992) rat RCK5 Pongs (1992) rat RK2 Pongs (1992) rat BK2 Pongs (1992) rat RAK Paulmichi et al. (1991) Eag (ether-à-go-go) KCNH1 OMIM database* HERG (human ether-à-go-go-related gene) human KCNH2 OMIM database* mErgla mouse London et al. (1997) minK human IsK, KCNE1 OMIM database* KCNQ famiiy KCNQ1 human KvLQTI OMIM database* KCNQ2 human Wang et al. (1998) KCNQ3 rat Wang et al. (1998)

*OMiM database: Oniine Mendelian Inheritance in Man, Internet database from the National Center for Biotechnology Information: http://www.ncbi.nim.nih.gov/omim/. 10

Acknowledgements

This thesis is dedicated to the memory of my mother, Kathleen Maud Hadley, 1919-1999.

I would like to express my appreciation to all members (past, present and visiting) of David Brown’s lab and the Wellcome Laboratory for Molecular Pharmacology who have helped in the work described here. Particular thanks are due to David Brown for his efforts in obtaining financial support from diverse sources. As a result, the work described here has been supported by Merck Ltd., Sigma-Aldrich Company Ltd., UGL Graduate School, and the Medical Research Council. I also thank Alex Seiyanko for collaboration and discussion, and Sacha Filippov for sympathy and light relief. I gratefully remember the contributions of Brenda Browning, Misbah Malik-Hall and Yvonne Vallis to tissue culture work and advice. Joanne Leaney gave helpful guidance on techniques, and her request to test niflumic acid on Kvi .2 encouraged me to be adventurous with drug experiments. Among other members of the UCL Pharmacology department, Don Jenkinson and Tony Dickenson have provided occasional but valued reassurance, Tina Bashford has mediated splendidly in various administrative problems, and Mike Bovingdon has provided moral encouragement and gas- cylinder expertise. 11

Preface: the M-current

Neurones owe their electrical behaviour to the currents passed by a wide repertoire of ion channels, including many types of potassium channel. Some of these help to set a resting potential across the neuronal membrane, and some shorten action potentials or prevent their occurrence. The M-current is a voltage-dependent potassium current that contributes to regulating excitability by halting firing after a short burst of action potentials. It was described first in sympathetic neurones, and subsequently in some neuronal cell lines such as the NG108-15 mouse neuroblastoma x rat glioma hybrid line. When I started this work, some suggestions had been made about the molecular identity of the channels underlying this current, but there was no conclusive evidence. Thus, my aim was to identify a molecular substrate for the M-like current in NG108-15 cells.

During the course of my study, important evidence was published relating to the molecular basis of the rat sympathetic M-channel. I used this information for some of my experiments. Of necessity, much of this work was done in collaboration with others, and some of the main conclusions have been published (Seiyanko et al. 1999). The illustrative and descriptive material in this thesis is drawn almost entirely from my own results. However, clarity demands the incorporation of some results obtained in collaboration with others, which I have specifically indicated as collaborative results. 12

Chapter 1: Introduction - Sympathetic M-currents and the M-iike current in NG108-15 ceiis.

Background to the M-current

Potassium currents were first observed by Hodgkin, Huxiey and Katz in the 1950s as a component of the conductance evoked in the squid axon membrane by a voltage step (Hodgkin et ai. 1952, Hodgkin & Huxley 1952a, b, & c). Progress in establishing the physical basis of such conductances was at first slow, as the concept of a “channel” became established, and methods were found for isolating membrane proteins (Hille 1984, p.354). Recording techniques gradually advanced from the impalement of robust amphibian and invertebrate cells to patch-damping of delicate mammalian neurones.

Still, for many years, currents could only be characterised in terms of their properties: kinetics, modulation and pharmacological blockade. As Rudy (1988) wrote: “...a proper classification is not possible today. However, with the molecular characterisation of K channels and the advent of new specific toxins and drugs, a classification may become possible.” The most advanced molecular biology cited in Rudy’s 1988 review is from his own group: Xenopus oocytes injected with total rat brain messenger RNA showed mixed potassium current types. So it was just becoming clear that potassium currents were produced by geneticaiiy-coded proteins, rather than being some undefined property of cell membrane lipids. The rampant proliferation of potassium channei genes had yet to become apparent.

Functional terminology from this era is still widely used to describe potassium currents in native cells. Thus, when a voltage step is imposed, a curve in the evoked current trace represents the development or fall-off of a current. A potassium current which rises (activates) almost instantly and then falls off rapidly (inactivates) on a depolarising step is termed an A-current. A current that activates more gradually, with or without subsequent inactivation, is recognisable as a delaved rectifier. A current that is proportionally larger at strongly negative membrane potentials, leading to potassium entry rather than exit from the cell, is an inward or anomalous rectifier. A current that appears after an action potential, and can be eliminated by omitting extracellular calcium (or introducing a calcium chelator into the cell), is clearly a calcium-activated current. A current which remains steady at a moderately depolarised potential of -2 0 or -3 0 mV, declines with an exponential time-course during hyperpolarising steps, is relatively insensitive to most blockers, and can be suppressed through muscarinic receptors, acquired the name M-current. 13

Discovery of the M-current

The first M-current was reported in 1980 in the frog (Brown & Adams 1980). Frog sympathetic neurones, voltage-clamped at -3 0 mV, evinced a steady outward current which relaxed in a time-dependent fashion during hyperpolarising voltage steps. This type of recording protocol, covering a limited voltage range and using negative steps to deactivate the current, has become standard for M-currents. Steps at potentials positive to about -2 0 mV are not feasible, as these would evoke contaminating delayed rectifier currents. Depolarising steps from more negative holding potentials would activate A-currents and sodium- and calcium currents, whereas these are inactivated during prolonged depolarisation. Because of these constraints, kinetic information on native M-currents covers only part of their activation range, and it has never been clear whether they are capable of inactivation at positive voltages. Brown & Adams (1980) found their M-current relatively insensitive to the delayed-rectifier blockers tetraethylammonium and 4-aminopyridine, but susceptible to inhibition by muscarinic agonists. This and subsequent work suggested that the M-current’s physiological function was as a brake on action potential generation by the cell. This brake could be released by cholinergic inputs (Adams et al. 1982). The M-current was subsequently found in mammalian sympathetic ganglia (Constanti & Brown 1981). Superior cervical ganglion (SCG) neurones, being easily accessible, became a standard preparation for M-current study.

Properties of the M-current in sympathetic ceils

Constanti & Brown (1981) recorded from cells in isolated rat superior cervical ganglia at 29°C, using sharp microelectrodes. The rat M-current appeared as a standing current at holding potentials of -3 0 to -5 0 mV. The current was non-inactivating, with a threshold around -7 0 mV and half-maximal activation at about -4 5 mV. Single deactivation time constants were measured from 1 -second hyperpolarising steps; the curve of 1 /x against voltage was U-shaped (centred on V%), relaxations becoming markedly faster with stronger hyperpolarisations.

The M-current is susceptible to few blockers. The anticholinesterase 9-amino-1,2,3,4- tetrahydroacridine (THA; tacrine) at high concentrations of 0.5-1 mM appeared to inhibit the M-current in SCG neurones (Marsh et al. 1990). Barium blocked M-currents in rat SCG with an IC50 of 300 pM (Stansfeld et al. 1993). 14

Linopirdine (DuP 996) is an experimental cognition enhancer, capable of increasing acetylcholine release from brain tissue, presumably through potassium channel block. In rat brain preparations, it also enhanced dopamine and serotonin release (Aiken et al. 1996). Two more potent linopirdine analogues, XE 991 and DMP 543, have also been identified (Zaczek et al. 1998). In SCG neurones, linopirdine blocks the M-current directly from the extracellular side without involvement of receptors (Costa & Brown 1997), the IC50 being reported as 3.4 pM (Lamas et al. 1997) or 7.0 pM (Costa & Brown 1997).

Neurotransmitter inhibition of the M-current is iargely responsible for the increased excitability of ganglionic cells in response to the stimulation of endogenous muscarinic receptors (Brown & Seiyanko 1985). However, receptor-mediated moduiation of the classic M-current is still only partially understood. Calcium has long been recognised as an effective inhibitor of sympathetic M-channels (Seiyanko & Brown 1996). Bradykinin inhibition of the sympathetic M-current appears to be mediated by caicium release resulting from phospholipase C activation (PLC) and inositol trisphosphate (IP3) generation (Cruzblanca et al. 1998). Endogenous uridine nucieotide-sensitive receptors share this transduction pathway (Bofiil- Cardona et ai. 2000).

However, muscarinic inhibition cannot be attributed to calcium (Cruzblanca et al. 1998; del Rio et al. 1999; Bofill-Cardona et al. 2000). Seiyanko et al. (1992) provided evidence from cell-attached patch recordings of SCG M-currents that, when muscarine was applied to the membrane outside the patch electrode, channels within the patch electrode area were inhibited. This indicates that a messenger generated in the vicinity of the extra-patch receptors is capable of diffusing (or propagating) to the site of the channels being recorded. Because M-channel activity persisted when muscarine was included in the pipette soiution, a local modulation (for example, direct G-protein action on the M-channel) is aiso excluded.

The NG108-15 neuroblastoma-glioma hybrid line

Currents resembling M-currents aiso occur in tumour ceiis derived from the sympathetic nervous system. Cultured C-1300 neuroblastoma cells, derived from a mouse abdominai tumour, had been previously noted to retain neuronal characteristics such as neurotransmitter synthesis (Amano et al. 1972) and second messenger production (Hamprecht & Schultz 1973). Marshall Nirenberg’s laboratory created the NG108-15 neuroblastoma x glioma hybrid by fusing C-1300 neuroblastoma cells (subclone N18TG-2) with rat giioma cells (C6, subclone C6Bu-1) (Klee & Nirenberg 1974, Amano et al. 1974). Daniels & Hamprecht (1974) described the differentiation of the cells by the application of a 15 cyclic adenosine monophosphate (cAMP) analogue, and gave a detailed account of the morphology of differentiated cells. Hamprecht (1977) enthused that “...the homogeneity of cultured cell material and its almost unlimited capacity for proliferation allow experiments with large numbers of replica cultures. No wonder one is tempted to call...cells like the neuroblastoma-glioma hybrids, the Escherichia coiioi neurobiology.”

In 1988, NG108-15 cells were reported to have potassium currents similar to those in sympathetic neurones, including an “M-current” (Brown & Higashida 1988a). The idea of NG108-15 cells as a model for real neurones was further strengthened when Robbins & Sim (1990) sought and found a functional analogue of the action potential-shortening A-current (for comparison, see Marsh & Brown 1991).

Properties of the NG108-15 M-like current

While the NG108-15 current resembled the sympathetic M-current functionally, it had certain kinetic and pharmacological differences which led to its being termed “M-like”. the term I shall use from now on.

Robbins et al. (1992) found the M-like current to be confined to chemically-differentiated NG108-15 cells, and recorded it 4-9 days after differentiation using the patch-clamp technique. Like the sympathetic M-current, the NG108-15 M-like current manifested as a standing current at -3 0 mV, with increasingly rapid relaxations on stepping to progressively more negative potentials. Working at 35°C, Robbins et al. (1992) found this relaxation to be bi-exponential, but with the faster component of relaxation being more consistent. Like the sympathetic M-current (Constanti & Brown 1981), it had a threshold around -7 0 mV and half- maximal activation at -4 4 mV. Its relaxations could be fitted approximately with a single­ exponential fit, giving a U-shaped relationship of 1/x to voltage with a minimum around V%. Nevertheless, the relaxation kinetics were somewhat slower than the sympathetic M-current, despite a higher recording temperature: a (mono-exponential) time constant for the relaxation at V%, was 230 ms, compared with Constanti & Brown’s (1981) value of 100 ms.

Figure 1.1 shows a typical NG108-15 M-like current from my own records, and a mouse sympathetic M-current recorded by A.A. Seiyanko. The NG108-15 current shows characteristically slow deactivation kinetics, but the voltage-dependence of the tail-currents is clearly similar to that of the sympathetic M-current. 16

Figure 1.1: The NG108-15 M-like current

The voltage protocol (top) comprises holding at -2 0 mV and applying hyperpolarising steps, incrementing by -1 0 mV each time. A 20-40s interval was allowed between steps for the current to recover.

The NG108-15 M-llke current (centre) shows slow deactivations, requiring a step duration of 6s to include most of the tail. By contrast, deactivations of the classic M-current (bottom), recorded with a similar but shorter step protocol and shown to the same scale, are complete within a 1-second step. (Mouse SCG M-current record kindly provided by A.A. Seiyanko). -20 mV

110 mV

NG108-15 M-like current

200 pA

Mouse SCG M-current (courtesy of A.A. Seiyanko)

Figure 1-1 18

Pharmacology: potassium channel blockers suggest heterogeneity of “M-currents”

Like the sympathetic M-current, the NG108-15 M-like current was readily blocked by barium (Brown & Higashida 1988a). Robbins et al. (1992) found it to be insensitive to 100 nM dendrotoxin, and comparatively insensitive to tetraethylammonium. It was also blocked by a range of divalent and trivalent cations with, for example, IC50S of 2.13 mM for lanthanum and 0.011 mM for zinc (Robbins et al. 1992).

Other channel blockers revealed differences between the NG108-15 M-like current and the classic M-current. The M-like current was inhibited by 9-amino-1,2,3,4-tetrahydroacridine

(THA) with an IC50 of 8 |liM (Robbins et al. 1992). By contrast, the M-current in SCGs was only inhibited at above 100 ^M THA (Marsh et al. 1990). Another blocker that discriminates between NG108-15 and SCG currents is a toxin from the scorpion Buthus eupeus. 50 |Lig/ml of the whole venom inhibits the NG108-15 M-like current by 44%, and the SCG M-current by 14% (Filippov et al. 1996). The NG108-15 M-like current is less sensitive to linopirdine than the sympathetic M-current: Noda et al. (1998) measured an IC50 of 24.7 pM.

Receptor effects on the M-like current

Early work on the line disclosed that NG108-15 cells were endowed with receptors responding to adrenergic, cholinergic and opioidergic agonists. These gave scope for continuing research using cAMP levels to follow the actions of neurotransmitters (e.g. Traber et al. 1975a, b). NG108-15 cells also synthesise acetylcholine. After several days of exposure to dibutyryl-cAMP, they release acetylcholine in response to various stimuli (McGee et al. 1978). An electrical response of NG108-15 cells to bradykinin application was first noted in 1978 (abstract cited in Brown & Higashida 1988b). Bradykinin application to NG108- 15 somata stimulated acetylcholine release from a synapse formed with a muscle cell (Higashida 1988).

The M-like current was inhibited modestly by the action of muscarinic agonists (Brown & Higashida 1988a) and more effectively by bradykinin (Brown & Higashida 1988b). The bradykinin effect could be mimicked by applying phorbol esters or 1 -oleoyl- 2 -acetylglycerol, implicating PKC in the transduction mechanism (Brown & Higashida 1988c).

Robbins et al. (1992) found no significant muscarinic inhibition of the M-like current in normal

NG108-15 cells, attributing this to the absence of Mi or M3 receptors. They therefore studied muscarinic inhibition in a line of NG108-15 stably transfected with the Mi receptor. Although 19

only PLC-stimulating receptors inhibited M-like currents, no participants in the known signalling pathway (guanosine 5’-triphosphate (GTP), kinases, phosphatases,IP3, calcium) could be conclusively implicated in muscarinic inhibition (Robbins et al. 1993).

NG108-15 M-like currents showed the same “diffusible messenger” phenomenon as SGG M- like currents. That is, channels in a cell-attached patch were inhibited by bath-applied muscarine (Brown et al. 1993). This was confirmed for both muscarine and bradykinin by Seiyanko et al. (1995), where thorough analysis of M-like channel data also reveals single­ channel conductances centred on 7.5 and 13.5 pS, comparable with sympathetic M-current values (e.g. Seiyanko et al. 1992). Another PLC-linked receptor, thePzu or “UTP receptor”, is also present in NG108-15 cells and inhibits the M-like current (Filippov et al. 1994).

Potassium channels in the cloning era

It is now common knowledge that potassium channels consist of tetrameric assemblies of protein subunits. Enough potassium channel genes have been sequenced, from bacteria, yeasts, plants and animals, to allow detailed classification and speculation about their evolution (Derst & Karschin 1998). The first potassium channel to be studied was from the fruit-fly Drosophila] it was named Shaker aHer the effect of a mutation which caused the fly’s legs to shake when it was anaesthetised with ether (Pongs 1992). All the channel subunits I shall deal with in this work are structurally analogous toShaker, having six hydrophobic transmembrane segments termed S1-S6. S4 contains positively-charged amino acid residues and moves in response to a build-up of positive charge under the plasma membrane, leading to channel opening. S5 and S6 are linked by a peptide chain including a short sequence acting as a selectivity filter for potassium ions. The N- and C-termini of the protein are on the cytoplasmic side of the membrane. These typically contain several consensus sites where phosphate groups can be added by various protein kinases, as well as regions mediating channel assembly as tetramers, the attachment of auxiliary subunits, and interaction with the cytoskeleton.

The first mammalian potassium channels were cloned using the gene sequence of the Sha/rerchannel. As molecular biology identified separate genes for potassium channels, clones were initially named according to their provenance. For instance, Kvi .2 was termed RCK5 (Rat Cortical K channel 5), and NGK1 (Neuroblastoma-GIioma K channel 1) (Pongs 1992). A systematic classification (Kvi .1, Kvi .2, etc.) was gradually adopted, but is now being superseded by a nomenclature based on human gene names (see “Nomenclature and homologues of potassium channels”, p.10). Figure 1.2 is a phylogenetic tree showing the 20 relatedness of mammalian voltage-gated (and calcium-activated) potassium channels, based on homology over the 6-transmembrane region. 21

ECOK elkl elk2 ergi erg2 erg3 eagi eag2 SK2 SK3 SKI SK4 slol slo3 Kv9.2 Kv9.1 Kv9.3 KvS.I Kv6.1 KV2.1 Kv2.2 KvS.I Kv3.1 Kv3.3 Kv3.4 Kv3.2 Kvl.1 Kvi .2 Kvi .3 Kvi .6 Kvi .5 Kvi .4 Kvi.7 KV4.2 KV4.3 Kv4.1 KQT2 KQT3 KQT1

Figure 1.2: Genetic relatedness of potassium channels (taken from Coetzee et al. 19991 22

Molecular Identification of M- and M-llke channels

As previously indicated, the identity of the proteins forming M- and M-like channels was not established when I started work on candidate channels for the NG 108-15 M-like current. However, I had several leads resulting from suggestions made by others, which are outlined below and described in more detail in the Introduction sections of Chapters 3, 4 and 5.

(1 ) Kvi .2. Yokoyama et al. (1989) tried to identify the molecular basis of the NG108-15 M- like current. They isolated two cDNAs: one, which they termed NGK1, coded for the Sha/cer-related Kvi .2 subunit, while the other, named NGK2, corresponded to the Shaw- like subunit KvS.I. The latter probably produces the delayed rectifier current found in NG108-15 cells by Robbins & Sim (1990). Kvi .2 shows several differences from the M- like current (elaborated in Chapter 3), but was suggested to form part of the underlying channel, based on its presence in NG108-15 cells and its inhibition by the stimulation of coexpressed Mi muscarinic receptors inXenopus oocytes (Huang et al. 1993; Morielli & Peralta 1995). (2) Ether-à-go-go (Eag) and related genes emerged as contenders during the course of my work. Stansfeld et al. (1997) proposed that a member of the ether-à-go-go family might generate the neuronal M-current. While there are strong arguments against a role in M- currents for Eag itself (Mathie & Watkins 1997; Marrion 1997), sympathetic neurones express cDNAs for several members of the Eag, ERG and Elk channel families (Shi et al. 1997,1998). Moreover, Para veil i et al. (1996) pointed out the similarity to HERG of an “inward rectifying” current in neurobiastoma-derived cell lines, and Hu & Shi (1997) reported a similar (hence also HERG-like) current in NG108-15 cells. (3) KCNQ2 and KCNQ3. While I was working on the ERG channel hypothesis, Wang et al. (1998a) advanced strong evidence that the M-channel in sympathetic neurones is formed by a heteromuitimer of two recently-discovered members of the KCNQ famiiy, KCNQ2 and KCNQ3.

My work

In the work that follows, I have tried to assess the contributions made by each of the above potassium channel proteins, Kvi .2, ERG and KCNQ2+3, to the M-like current in NG108-15 cells. I have done this mainly by comparing the kinetic and pharmacological properties of the channels, expressed in mammalian cell lines, with those of the M-like current recorded from differentiated NG108-15 cells. 23

Chapter 2: Methods

Cell culture

CL1023 cells are B82 mouse fibroblasts, stably transfected with rat Kvi .2 and a neomycin- resistance gene (Werkman et al. 1992). They were maintained in Dulbecco’s Modified Eagle Medium (DMEM) containing 1% foetal calf serum (PCS) and optionally 20 mg/l geneticin

(G418), in 50 ml plastic flasks in an incubator at 37°C and 10% CO2 . Cells were subcultured at a dilution of 1:5, 1:10 or 1:20 every 3-6 days as convenient for plating and recording. Subculturing involved a wash with phosphate-buffered saline (PBS) or Hank’s balanced salt solution (HBSS), followed by treatment for 2-3 minutes with trypsin-EDTA (containing 0.5g/l trypsin). Once the cells had rounded up, they were dislodged by smacking the flask, fresh medium was added to halt trypsin action, the cells were centrifuged at 1000 r.p.m. for 5 minutes, the supernatant was tipped off, and the cells were resuspended in fresh medium. This suspension was diluted into new flasks, and into a stock solution for plating. Cells were plated in 35 mm plastic dishes, and recordings were made when they approached confluence. The selection medium geneticin (when used) was present throughout the culture cycle, without appearing to have any influence on the properties of the Kvi .2 current. It did not seem to be necessary to maintain expression of the current.

MK2 cells are CHO cells stably expressing the Kvi .2 clone derived from mouse brain, and a neomycin resistance gene. Their culture was similar to that of CL1023, but the incubator was maintained at 5% CO2 . Other differences were that the growth medium comprised RPMI with 10% PCS, 1% L-glutamine (Sigma; giving 2 mM) and 1% penicillin/streptomycin solution

(Gibco-BRL; giving 50 units penicillin and 50 pg streptomycin per ml). As a rule, geneticin was added at 100 mg/l. Again, geneticin did not affect measurable properties of the current. Subculture was at 3-4 day intervals.

CHO hml cells are a human muscarinic M r receptor-expressing derivative of the CH0-K1 line, described in Mullaney et al. (1993). They were cultured as for the MK2 line, but in MEM alpha medium containing 10% PCS, 1% L-glutamine and 1% penicillin/streptomycin. No selection medium was used. Cells were split at 3-4 day intervals, and channels were introduced by transfection 1 -2 days after plating. Recordings were made one or more days after transfection, at an interval determined by experience of the particular current expressed.

Two lines of NG108-15 neuroblastoma-glioma hybrid cells were used: one was an unmodified line, while the other stably expressed pig brain Mi receptor. These were grown at 24

10% CO2 in a growth medium comprising DMEM plus 5% PCS and 6 ml/l HAT supplement

(Sigma) giving 3 pM hypoxanthine, 12 nM aminopterin and 480 nM thymidine. Cells were split at a ratio of 1:3 by mechanical trituration, centrifugation and resuspension in fresh medium. Ceils were plated in this medium at a 1:6 dilution, on dishes pretreated with a solution of 0.01% poly-L-ornithine (in 0.15 M boric acid adjusted to pH 8.3 with NaOH). Cells were allowed to grow for 24 hours. Chemical differentiation took place in a piating medium containing DMEM with only 1% PCS (to slow the growth of undifferentiated cells) and 6 ml/l

HT supplement (Sigma) giving 3 pM hypoxanthine and 480 nM thymidine. This was supplemented initially with 10 pM prostaglandin El (PGE1) in DMSO and 50 pM 3-isobutyl-1- methylxanthine (IBMX) in ethanol, to raise cyclic AMP. The cells were left for three days, and then refed with plating medium. Recordings were made 7-21 days after plating, on cells showing neuronal morphology: large somata, processes, and the conspicuous vacuoles characteristic of differentiated NG108-15 cells. A majority of cells in these dishes continued to proliferate and showed a less differentiated morphology, being small and having short or no processes.

As a rule, the kinetic and blocker data shown here were obtained from the unmodified type of NG108-15 cells; the muscarinic receptor-transfected line was used on occasions according to avaiiabiiity. There was no apparent difference in the data from the two lines.

Transfection

CHO cells were transfected with channel DMAs using LipofectAmine Plus (Gibco-BRL) according to the manufacturer’s recommendations, 1-2 days after plating. Thus, 1 mg DMA was used per dish of cells. This 1 mg was divided into 10 parts DMA for channel(s) and 1 part marker, which was a plasmid coding for the T-cell receptor molecule GD8. All plasmids (supplied by Dr. Ian Wood and Dr. Fe Abogadie) were driven by cytomegalovirus (CMV) promoter. At least one day after transfection, cells for patch-clamping were identified by adding Dynabeads (Dynal (U.K.) Ltd.): microscopic polystyrene beads with a magnetic coating and an anti-CD8 antibody, which binds the bead to ceils expressing the CD8 molecule. Dynabeads were first washed by adding 50 ^il of bead suspension to 5 ml of pure water, vortexing, placing on a magnetic stand to pull down the beads, and removing the wash water. After resuspending the beads in 1 ml pure water, 0.2 ml of beads was distributed dropwise over each dish of transfected cells. Thus treated, the cells bearing beads continued to survive and proliferate for several days. 25

Electrophysiology

Dishes of cells were transferred to the stage of a Nikon Diaphot-TMD inverted microscope supported on a pneumatic air-table (Kinetic Systems), and superfused by a gravity-driven system at 10-13 ml/minute. The perfusing solution contained (mM): 144 NaCI, 2.5 KCI, 2

CaCl2 , 0.5 MgCb, 5 N-2-hydroxyethylpiperazine-N -2-ethanesulphonic acid (HEPES), and 10 glucose, plus Tris base to pH 7.4.

Patch pipettes

Patch pipettes were pulled on a vertical puller (List or Narishige). Filamented borosilicate glass capillaries (Clark Electromedical) were used. For basic whole-cell recordings, standard-wall (1.5 mm outside diameter; 0.86 mm inside diameter) glass was used to pull electrodes of around 8 MQ resistance when filled with internal solution. For perforated- patch recording, thin-wall glass (1.5 mm outside diameter, 1.17 mm inside diameter) was used to give controllable final electrode resistances of 2-3 MQ. The region near the pipette tip was coated with Sylgard (Dow Corning) to reduce capacitance between the electrode and the bath, and the tip was fire-polished on a microforge (Narishige) to facilitate sealing to the cell membrane. Pipettes for whole-cell recording were filled with internal solution using a syringe and a fine cannula. For perforated-patch recording, pipettes were dipped in normal internal solution for about 15s to fill the tip, and then filled up by cannula with internal solution containing the pore-forming compound amphotericin B (Rae et al. 1991). This solution was made daily by dissolving 1 mg amphotericin B in

20 pi dimethylsulphoxide (DMSG), sonicating for 10-20s to dissolve, adding 4 pi per 1 ml of internal solution, shaking and sonicating again. The final amphotericin B concentration was 200 pg/ml, a little weaker than the 240 pg/ml used by Rae et al. (1991).

Considerations reiating to internai soiutions

Direct electrical recordings from cells involve the use of a pipette solution differing in ionic composition from both the bath solution and the cytosol. This gives rise to two sources of deviation from the transmembrane voltage nominally imposed on the cell. These are the Donnan equilibrium junction potential and the liquid junction potential.

The Donnan equilibrium iunction potential arises because, while charge-carrying ions such as potassium and chloride equilibrate between the electrode pipette and the cell interior, there remains a trapped population of cell proteins carrying negative charges. This generates an extra potential difference between the pipette solution and the cell 26 interior. If all the intracellular negative charge is fixed, this potential difference can make the cell interior -1 2 mV negative relative to the pipette, where the electrode wire which measures and maintains the command potential is located. This can be balanced to some extent by having non-diffusing anions replacing chloride in the pipette solution (Horn & Marty 1988; Marty & Neher 1995, p.49). In the case of whole-cell recording, the less mobile anions gradually diffuse between the cell interior and the pipette solution, discharging the potential difference. With perforated-patch recordings, however, a stable potential difference is generated, which will affect the apparent activation range of a current, but not its reversal potential. The Donnan potential can be measured approximately during whole-cell recording by monitoring the shift in a current’s activation curve (Marty & Neher 1995, pp. 49-50). In the case of M-currents, which suffer from run­ down in the whole-cell configuration, such measurements are not feasible, and it can only be assumed that measured kinetic parameters may be displaced by up to 12 mV positive to their true positions.

While the Donnan potential arises from immobile ions, the second source of error is determined by the rates of movement of mobile ions, and can be measured and corrected without reference to current properties. A liquid iunction potential is a potential difference across the boundary between two solutions. Because the electrode potential is typically standardised by setting to zero when immersing the pipette in the bath, the junction potential between the pipette- and bath solutions is made invisible to the recording apparatus. When the pipette solution abuts the cvtoolasm of the cell being patched, this potential difference is no longer present, but the quantity is still being subtracted instrumentally.

If the pipette/bath junction potential is measured, a retrospective correction can be made to the command voltages imposed on the cell. The larger (less mobile) the ions in the pipette solution, the more slowly they diffuse into the adjacent bath solution, and the bigger the potential difference that builds up between the two solutions. Thus, pipette solutions rich in mobile chloride ions create only a small junction potential, but those containing high proportions of acetate generate a significant junction potential.

For Kvi .2-expressing cell lines, I made liquid junction potential measurements in current- damp mode on the Axoclamp-2, and the same bath (ground) electrode used for recording in each case (see Electrodes section, p.29). I immersed a low-resistance (1-3 M^) electrode in a bath perfused with the pipette solution, zeroed the pipette potential, changed to the standard external solution, and waited until the voltage readout settled at 27

a stable value. When I obtained two or three measurements in good agreement, I took an average as the junction potential correction to be applied to the command voltage.

For the internal solution used with mErgla, KCNQ channels and NG108-15 cells, I measured the liquid junction potential using the Axopatch-200 and the agar-bridge ground electrode arrangement. I immersed an electrode (2-3 MQ) in a bath perfused with the pipette solution, set the amplifier to voltage-clamp mode, and used the seal test function in Clampex to zero the electrode potential and check the electrode resistance. With the potential set at 0 mV, I then closed Clampex, switched the Axopatch-200 to current-clamp mode, changed the perfusing solution from the pipette solution to the external, and waited for the voltage reading to stabilise. As before, I calculated an average value.

Compositions of internai solutions

Early CL1023 recordings were made with an internal solution based on that used by Beech et al. (1991) and intended to approximate to background cell calcium-buffering activity. Its composition was (in mM): 175 KCI, 5 HEPES, 0.1 tetrapotassium salt of 1,2- bis(2-aminophenoxy)ethane-N,N,N’,N’-tetraacetic acid (BARTA), 3.3 MgCb, 3 NagATP, 0.3 GTP, and NaOH to pH 7.2. The pipette/bath junction potential is expected to be low for a chloride-based internal; I measured a value of less than - 2 mV, so I have not made any adjustment to voltages in my results for CL1023 cells.

Experiments to improve the recording lifetime of CL1023 cells used an acetate-based internal solution, based on Robbins et al. (1992,1993). Dendrotoxin experiments on CL1023 were done with a solution of (mM) 130 K acetate, 20 KCI, 40 HEPES, 3 MgCb, 3 ethylene glycol-bis(p-aminoethyl ether)N,N,N ,N -tetraacetic acid (EGTA), and NaOH to pH 7.2.

The formulation chosen for the MK2 line contained (mM): 90 K acetate, 20 KCI, 40

HEPES, 3 MgCl2 , 3 ethylene glycol-bis(p-aminoethyl ether)N,N,N,N-tetraacetic acid

(EGTA), 3 Na2ATP, 0.3 Na-GTP, and NaOH to pH 7.4. An alternative version with the addition of 1 mM CaCb caused such instability of Kvi .2 recordings that caicium was not routineiy added. Selyanko et al. (1995) obtained a junction potential of -4 mV for a similar solution against the standard external solution. I established a junction potential of - 8 mV (n=2 identical values) for this solution against the standard external, and I have used this to adjust all my results on the MK2 line. 28

For recording NG108-15, mErgla and KCNQ channel currents, the pipette solution contained (mM): 80 K acetate, 30 KCI, 40 HEPES, 3 MgCIa, 3 EGTA, 1 CaCb and NaOH to pH 7.4. The amphotericin B perforated-patch technique (Rae et al. 1991) was used for all these experiments. The junction potential should be slightly lower than that for the acetate-based solution used on mKvl .2, because of increased pipette [chloride] relative to [acetate]. The measured value was -7 .3 mV (n=4), but I have left my NG108-15, mErgla and KCNQ data uncorrected to maintain consistency with our published reports.

Drug addition

All drugs were applied cumulatively, by addition to the reservoir of perfusing fluid. Niflumic acid was applied as a stock solution in dimethylsulphoxide (DMSO), prepared on the day of the experiment. Forskolin and dideoxyforskolin were dissolved in DMSO and subdiluted in water. Linopirdine stock solutions were made up in 0.1 N hydrochloric acid, taking care not to disturb the pH of the perfusing solution by adding more than 0.1% of the acidic solution. Concentration/response curves for linopirdine were completed by washing off the drug and starting again with stronger stock. All other drugs were dissolved in deionised water, sonicating to dissolve if necessary.

Hardware and software

Electrodes

The bath (ground) electrode for CL1023 experiments consisted of a chloride-coated silver wire, connected to the back of the amplifier headstage and positioned in the dish of cells. Experiments on MK2, OHO hml and NG108-15 cells used an agar bridge filled with 3M KCI to connect the dish of cells to a pool of KCI contacting a silver/silver-chloride pellet connected to the headstage.

The recording electrode was a chloride-coated silver wire mounted in a Perspex holder with an outlet for suction tubing, and connected to the amplifier headstage. Electrode pipettes, once filled with sufficient internal solution to immerse the tip of the wire, were inserted over the wire, and the assembly was manoeuvred over the dish using a mechanical micromanlpulator (Goodfellow Technology). 29

Instrumentation: (1) Axoclamp-2

For discontinuous voltage clamp and current-clamp experiments, an Axociamp-2 amplifier (Axon Instruments) was used. This received an input signal from a computer (Dell) running Clampex 5 (for early CL1023 recordings) or Clampex 6.0 versions (software from Axon Instruments), via a TL-1 analogue-to-digital converter (Axon Instruments). The voltage output of the amplifier was connected through an oscilloscope to a pre-amplifier and thence to the interface. The current output was connected through another pre­ amplifier to the oscilloscope, and in turn to the interface. The Monitor and Sample clock outputs of the amplifier were connected to a second oscilloscope (used for making instrumental adjustments and monitoring recording). A pen recorder (Gouid 24008) received voltage and current signals out of the first oscilloscope. Additionally, a pulse generator for generating test pulses for sealing was connected to the Axoclamp-2, the interface and the first oscilloscope.

Recording procedure with the Axociamp-2

Once the pipette tip was below the surface of the bath solution, the pipette potential was zeroed using the Input Offset knob. Next, -11.1 nA pulses (15 ms duration) were applied from the pulse generator, giving an approximately square downward-going voltage wave on the first oscilloscope (Tektronix). The Bridge control on the Axoclamp was used to level the oscilloscope trace, giving a readout of the electrode resistance. The electrode was positioned just above the chosen cell, and lowered carefully until the pipette made contact with the cell membrane and the trace dipped, indicating increased resistance. Suction was applied through a fine tube connected to the electrode assembly. When the “Vm” readout on the amplifier reached approximately -2 0 mV, the pulse size was reduced to -1.1 nA. A combination of extra negative current (applied via the DC current command knob) and further suction achieved a seal. Seal resistance could be estimated from the

“Vm” and “I” readings on the amplifier, using Ohm’s law. V=IR .*. R = V/l, thus a seal in the GQ range (10® Q) was indicated when (say) an extra applied current of 0.1 nA (10’^° A) caused a change of 100 mV (10'^ V). These applied currents and voltages sound drastic, but worked reliably.

At this point, either additional suction was used to rupture the patch of membrane enclosed by the pipette tip, so going whoie-cell, or amphotericin B was aliowed to permeabilise the membrane by waiting for several minutes. The trace on the oscilloscope then returned to a shallow trough shape, indicating a lowered resistance. 30

Switching to direct current clamp (DCC) mode, the capacitance neutralisation knob was used to square off the trace on the second oscilloscope (Telequipment). The Holding Position knob could be used at this point to balance the resting membrane potential of the cell and obtain a readout. A suitable holding potential was then set.

Switching to single-electrode voltage clamp mode, the sampling rate was increased from around 3 to 6.2 kHz to get three cycles on the second oscilloscope. The gain was turned up progressively as high as possible (routinely 8-25 nA/mV) by alternate adjustment of the gain and capacitance neutralisation knobs, to obtain the maximum amplitude of traces without oscillation. This constituted the optimal amount of feedback to achieve the voltage specified by the command from the computer. Clampex version 6 enabled simultaneous recording of both current and voltage traces with the Axoclamp. This showed that good clamp was regularly obtained: the voltage traces were square, and (as an example) a command to a nominal +60 mV achieved nearly +50 mV even when the evoked current was around 10 nA. Thus, current activation kinetics were probably well represented, and current amplitudes could be plotted against actual voltages achieved.

Some recordings on NG108-15 cells (see mErgla section) were made in current-clamp mode on the Axoclamp. There were only minor differences from voltage-clamp experiments. The signal from the computer was routed through the interface to the EXT. ME1 COMMAND (rather than EXT. VC COMMAND) connection on the Axoclamp. The Axoclamp DESTINATION knob was set to ME1 (rather than VC), and the instrumental adjustments went only as far as those carried out in DCC mode. A small current command was applied as necessary to maintain a membrane potential of -9 0 mV. Rather than break the circuit to switch to voltage-clamp mode, I have assumed that the properties of the M-like current in these cells were typical.

Instrumentation: (2) Axopatch-200 recording procedure

The apparatus used for continuous voltage clamp was an Axopatch-200 amplifier (Axon Instruments), for some recordings a EF5-01 lowpass/highpass filter (Barr & Stroud) set to 1 kHz, chart recorder (Gould), interface and computer as above.

Recordings on the Axopatch-200 were made as follows. Once the pipette tip was below the surface of the bath solution, the pipette potential was zeroed using the Pipette Offset knob. Pulses of -0 .7 mV were applied using the TestSeal command in Clampex. This gave an approximately square downward-going current wave on the computer screen. The electrode resistance could be read out on screen. The electrode was lowered 31 carefully towards the chosen cell, until the pipette made contact with the membrane and the square trace became shallower, indicating increased resistance. Suction was applied to the eiectrode assembly, and the seal resistance was monitored on screen. When it reached 30-40 MQ, additional negative voltage was applied using the keyboard controi.

Once a vaiue approaching 1 GQ was attained, test puises were switched off and the amphotericin B was allowed to permeabilise the membrane by waiting for several minutes. The progress of current development was observed on the amplifier current window and the Gould chart recorder.

Compensation of series resistance

With the Axoclamp-2, capacitance and gain adjustments were an intrinsic part of recording, and success in achieving the command voltages could be evaluated from recordings of voltage simultaneously with current. With continuous single-electrode voltage-clamp, the only voltage-clamp option with the Axopatch-200, adjustments are not automatic. I did not routinely use capacitance/resistance compensation. It was assumed during coilaborative work on the M-like current that, as most NG108-15, m Ergla and KCNQ currents were less than 1 nA at -2 0 mV, errors would not be large. The error in voltage achieved is given by I.R a where I is the totai current and R a is the access resistance. Access resistance in a well-perforated patch can be assumed to be >3 times the original resistance of the electrode when immersed in the bath solution. I used electrodes of 2 MQ or slightly more. A 2 MQ electrode recording a current of 1 nA “loses” around 1 nA x 6 MQ = 6 mV across the electrode-cell contact. So, for the KCNQ2+KCNQ3 currents I recorded with the M-current protocol, which averaged 1 nA when holding at -2 0 mV, the maximum error should be below 10 mV. Average maximal mErgla current amplitudes, even when boosted by de-inactivation, were beiow 1.7 nA, giving the same order of error. The NG108-15 M-like current amounted to 0.5-1.5 nA at - 20 mV, so the expected errors for this current are simiiar to those for the cloned currents in CHO celis.

At the more positive voltages used to evaluate KCNQ2+KCNQ3 activation properties, errors are larger. A 2.5 MQ electrode recording a current of 3.5 nA at a nominal +50 mV gives rise to a voltage error of 3.5 x 7.5 = 26 mV, so the actual current properties reflect those at about +20 mV. I have included some such large currents in my caicuiations of KCNQ2+KCNQ3 activation curves, but the average maximum current size is around 2.3 nA at a nominai +50 mV and 1.6 nA at the +20 mV point I have used for normalisation, i therefore estimate an average maximum error of about 20 mV for the most positive 32

potentials used In activations of KCNQ2+KCNQ3 currents. KCNQ2 expressed separately produced currents of similar size, so similar errors are expected. For KCNQ3 expressed alone, the small currents (average 262 pA at +20 mV) are unlikely to be affected by this problem.

This type of shortfall in applied potential will not affect activation thresholds and reversal potentials, where there is little current, but can lead to inaccurate values for parameters such as the half-activation voltage. However, these can also be affected by inactivation when the true command potential is approached. i subsequently made some recordings of large KCNQ currents incorporating capacitance and series resistance compensation as follows. After ascertaining with a series of voltage steps that the current properties were satisfactory, filtering was bypassed by setting the amplifier’s intrinsic iowpass filter to 50 kHz. Using a test pulse from Clampex, at -6 0 mV (outside the activation range of the current), the Whole Cell Capacitance and Series Resistance knobs were adjusted until the initial transient was eliminated. The Series Resistance Compensation “prediction” knob was turned to 95%, and the “correction” knob was carefully turned as high as possible, reducing the Series Resistance setting to remove persistent transients as they appeared on the test pulse on-screen. I seldom managed to compensate by more than 70-80% without substantial decline in the currents (with or without a resistance drop suggestive of going whole-ceii). The Axopatch filter was reset to 1 kHz for recording. Adequate capacitance compensation was verified by eye, using the capacity transients visible on voltage steps.

Series resistance compensation increased the amplitude of KCNQ currents at -2 0 mV, indicating that an improved (more positive) voltage was attained. It also revealed a small amount of inactivation in KCNQ2 and KCNQ2+3 currents at high positive potentials. The scale of this inactivation is shown in Selyanko et ai. (2000): inactivation of KCNQ2+KCNQ3 currents at +50 mV is removed immediately on hyperpolarising, leading to a small increase in current on the first few steps negative from +50 mV. KCNQ3 current inactivation was obvious even without compensation.

Overall, therefore, I consider series resistance errors to have little impact on the NG1 OS- 15 M-like current and the candidate mErgla and KCNQ2+KCNQ3 channels when recorded with the M-iike current deactivation protocol. Activation properties measured on KCNQ2 and KCNQ2+KCNQ3 currents over a wider range of potentials, however, are approximate rather than definitive. 33

Analysis

Initial data analysis was done in Clampfit versions 5, 6 or 8 (Axon Instruments). Results of fitting and amplitude measurements were exported to Quattro Pro (Borland, version 5.0) or Microsoft Excel 97. Spreadsheet calculations and data records for illustration were exported to Microcal Origin (version 4.1 and 5.0), which was also used to fit conductance/voltage and concentration-response curves.

Leak-subtraction of data records

Before evaluating current amplitudes for current-voltage and conductance-voltage curves, records were usually leak-subtracted. In Clampfit 6, cursors were positioned near the end of the current step, and a set of Basic Statistics obtained. The mean amplitude between the cursors was plotted as a quasi-steady-state current/voltage (l/V) curve. A straight line was fitted to the linear portion of this, negative to the activation threshold of the channel, and the slope in pA/mV was noted from the Results table. The data file was then re­ opened, and the leak adjusted using the reciprocal of the slope. An alternative method (used with Clampfit 8) was to export the I A/ measurements to Excel and then use Excel’s AutoFill feature to create a leak line from sub-threshold data points. This allowed greater selectivity, as any three (or more) well-aligned points from -110 to -6 0 mV could be chosen, omitting less trustworthy points. The resulting set of extrapolated leak values was subtracted from the raw measurements.

Fitting deactivation taiis

The time constant x is the time for the current to fall by 63% of its initial magnitude (1-1/e).

Fits are only trustworthy if the fitted region occupies a time 86% (1 -1/e^) of the current relaxation occurs within two time constants, 95% (1-1/e^) within three, and 98% (l-l/e"*) within four.

For NG108-15, mErgla and KCNQ currents, accurate measurement of the deactivation tails in Clampfit required back-fitting of exponential curves. This served to exclude capacity transients, and the brief rising phase of mErgla currents as they de-inactivated at the beginning of the voltage step. Where multiple current constituents were present in NG108-15 cells, three exponential components were fitted, with the first fitting cursor at the earliest feasible position and the second cursor as late as possible in the step. The fastest component, x -50-100 ms, was taken as representing the KCNQ-like current, and 34

the two slower components with xs of -100-800 ms and -1000-2000 ms were attributed to the mErgla-iike current. Expressed mErgla currents in CHO cells were fitted with two exponentiai components. The residual fast current present after WAY-123,398 block of the NG108-15 slow fraction was fitted with one exponentiai. KONG currents expressed in CHO cells, however, gave visually better fits when two exponentials were fitted.

To standardise the amplitudes corresponding to these time constants, two methods were availabie. One invoived isoiating the fit curves calcuiated by Ciampfit, and refitting the fit curves with the fitting cursor repositioned at the first sample of the voltage step. Alternatively, the fitted amplitude was manipuiated using a calculation based on the exponential equation, it = lo.exp(-t/x), where It is the fitted current amplitude (reiative to the fitting offset) at time t, while lo is the current amplitude at time zero (i.e. the start of the voitage step). By rearrangement, lo = lt/exp(-t/x). This equation was used on the exported Ciampfit Results Table fit data in the spreadsheet program, using Amplitude #1 as It, (Start Fit - episode start time) as t, and Tau #1 as x (and similariy for Amplitude #2, Tau #2, etc.). Start Fit is the position of the first fitting cursor. The time of the first sampie in the voltage step (episode start time) could be verified as foiiows. In Clampex. the number of samples in the first hoiding intervai, plus any steps preceding deactivation, pius 1, gives the sample number corresponding to the start of the deactivation step. In Clampfit version 6, the cursor position readout gives both the sample number and the time in miiliseconds, so locating the active cursor on this first sample reveals the episode start time.

Adjusted ampiitudes obtained by re-fitting fitted curves, and by calcuiation, matched very closely (commonly to at least 5 significant figures), validating both methods.

Where only one current was present, as in transfected CHO hml celis, Ciampfit version 6 allowed a quicker method of obtaining the totai back-fitted deactivation amplitude for kinetic and concentration-response anaiysis. Clampfit’s Cursor #3 was positioned on the first sample of the voltage step (determined as above). Fitting cursors were positioned, avoiding transients, and the requisite number of exponential components was fitted. The exported Results Table includes the back-fitted current amplitude as Cursor #3 intercept, and the level of the asymptote to which the curve wili eventuaiiy fall as Offset. The total deactivation ampiitude is obtained simpiy by subtracting one from the other. 35

More detail on fitting equations

In Clampfit, I fitted standard exponential growth and decay functions to raw current data, using the Chebyshev method for speed. I did preliminary analysis of time constants, current amplitudes, conductances and inhibition by blockers In Quattro Pro or Excel. I used Origin to graph the results and fit Boltzmann and Hill functions, using the least- squares method.

Normalised current/voltage (l/V) and conductance/voltage (G/V) plots were fitted with the Boltzmann equation: l/lmax = 1/(1 + exp(V% - V)/k) or G/Gmax = 1/(1 + exp(V^ - V)/k Here, Lax is the maximal value of the current (often, but not always, the value to which currents were normalised), and G max is the maximum value of the conductance. V% is the membrane potential, V, at which the current I (or conductance G) is half of Lax (or G max), k is the slope, measured in mV, and quantifies the voltage range over which there is an e- fold change in current or conductance.

For concentration-response data, amplitudes of drug-inhibited currents were normalised to the control current amplitude, and converted to proportional inhibition of the control current. Inhibition was plotted against concentration on a logarithmic scale, and fitted with Origin’s “logistic” equation, which corresponded to the Hill equation:

Y = Y„»/x" V + IC5o*). Here Y is the inhibition, and Ymax is the maximal inhibition. Ymax was set at 100% unless this was obviously inappropriate to the data, x is the concentration of blocker, and the

IC50 is the concentration at which Y is half of Ymax- Hh is the slope (“Hill slope” or “Hill coefficient”), which characterises the channel-blocker interaction. For a simple interaction of a blocker molecule with a single binding site on the channel protein, is expected to be 1, but I did not constrain it to 1 when fitting, nn values of <1 may indicate that two or more types of channel are present, with different affinities for the blocker. Oh values of >1 suggest co-operativity of binding. This is a recognised phenomenon in receptors for neurotransmitters, where the binding of one agonist molecule facilitates binding of a further agonist molecule. It can be hard to explain in the case of potassium channel block, where it is assumed that a single blocker molecule obstructs the pore of each channel.

The results of data analysis are quoted as mean + standard error of the mean (s.e.m.). I have made limited use of statistics for comparison of native currents with their presumed 36 cloned counterparts; statistical analysis would ideally require comparison of equivalent data for a wider range of channels, and with larger sample numbers.

Chemicals

WAY-123,398-A-5 was a generous gift from Wyeth-Ayerst Research (Princeton, New Jersey, USA), and azimilide from Dr. A.E. Busch (Hoechst Marion Roussel, Frankfurt am Main, Germany). Linopirdine (DuP 996) came originally as a gift from Du Pont (Wilmington, Deiaware, USA); later it was purchased from Research Biochemicals International (Natick, Massachusetts, USA). Alpha-dendrotoxin was provided by Prof. J.O. Doliy (Imperial College, London, UK). TEA was from Lancaster Synthesis (Morecambe, UK), and all other chemicals from Life Technologies (Paisley, UK), Sigma- Aldrich (Gillingham, UK) or BDH Chemicals (Poole, UK).

Stably transfected cell lines and clones

The NG108-15 cells derived originally from Nirenberg's laboratory, and reached us from Professor H. Higashida (Kanazawa University, Japan). The Kvi .2-expressing cell line CL1023 was also kindly supplied by Prof. Higashida. The MK2 line was made available by Dr. David Owen (now at CeNeS Pharmaceutical PLC, Cambridge, UK), as a welcome gift from the laboratory of Dr. Bruce Tempel (University of Washington, Seattle, USA). DNA for mErgla was generously provided by Dr. B. London (University of Pittsburgh, USA), and DNAs for human KCNQ2 and rat KCNQ3 by Dr. D. McKinnon (State University of New York, USA). 37

Chapter 3: Kv1.2

3.1 : Kvi .2 - a candidate for the M-iike current?

Sites of expression of Kvi .2

Kvi .2 was initially cloned from rat brain by using Drosophila Shaker as a probe. As the second brain potassium (K^) channel, it was described as BK2 (McKinnon 1989). Alpha- dendrotoxin was used to extract the Kvi .2 protein from mammalian brain (Rehm et al. 1989). in-situ hybridisation (detecting mRNA) and immunocytochemistry (detecting proteins) show Kvi .2 to be localised in various regions of the mouse brain (Wang et ai. 1994). Using similar techniques, rat hippocampus, thalamus, cerebral cortex and cerebellum show high levels of Kvi .2 (Tsaur et al. 1992, Sheng et al. 1994). Collectively, proteins of the Kvi subfamily have a complex distribution in rat brain, with clear colocalisations in some types of neurone (Veh et ai. 1995). Southan & Robertson (1998) recently revealed the presence of functional a- dendrotoxin-sensitive channels by patch-clamping terminals in rat cerebellar pinceaux (brushiike structures formed by cerebellar basket cells synapsing on Purkinje cells). Dendrotoxin-sensitive potassium currents (Kvi .1, Kvi .2 or Kvi .6) are also present in ouinea- pio hippocampus (Simmons & Chavkin 1996). Kvi .2 is therefore a major feature in the electrical repertoire of the mammalian brain.

Dixon & McKinnon (1996) surveyed mRNA abundance in rat sympathetic ganglia. The coeliac and superior mesenteric ganglia lack M-currents, but express Kvi .2 mRNA at higher levels than the M-current containing superior cervical ganglion. Thus, Kvi .2 message in autonomic ganglia correlates inversely with the M-current, although this does not guarantee that Kvi .2 protein levels also do so.

Outside the nervous system, Kvi .2 is present as mRNA and protein in the rat heart (Nakamura & lijima 1994; Xu et ai. 1996; Barry et al. 1995). Kvi .2 mRNA is also abundant in ferret heart (Brahmajothi et al. 1996). Additionally, Kvi .2 occurs in airway smooth muscle (Adda et al. 1996), cultured oligodendrocytes and their progenitor cells (Attali et al. 1997), and in the PCI 2 cell line derived from a tumour of the rat adrenal medulla, a region developmentaily related to sympathetic ganglia (Conforti & Millhorn 1997; Confort! et al. 2000). Canine colonic smooth muscle expresses another variant of Kvi .2 , with a protein sequence 90-91% homologous to the known rodent Kvi .2 channels (Hart et al. 1993). 38

NG108-15 cells: NGK1 and NGK2

Kv1.2 was cloned from NG108-15 neuroblastoma-glioma hybrid cells, and named NGK1 ; when expressed in Xenopus oocytes, it showed delayed rectifier behaviour with rapid activation and very slow inactivation at 0 mV (Yokoyama et al. 1989). An accompanying gene, KvS.I, was later suggested as the basis of the inactivating current characterised in NG108-15 cells by Robbins & Sim (1990). The discrepancy between the low tetraethylammonium sensitivity of the NG108-15 inactivating current, and the high TEA sensitivity of KvS.I, might be resolved by invoking heteromultimerisation (Ito et al. 1992; Yokoyama et al. 1993). In support of this, a new single-channel conductance emerged in oocytes co-injected with Kvi .2 and KvS.I mRNAs (Shahidullah et al. 1995a, b). Selyanko et al. (1995) measured single-channel conductances in NG108-15 cells. Their slope conductance of about 18 pS for the delayed rectifier agrees closely with Shahidullah et al.’s (1995a) figure of 18.2 pS for the presumptive Kvi .2/Kv3.1 heteromultimer.

Kinetics of Kvi .2

The activation properties of Kvi .2 are covered at greater length in the Discussion (section 3.3). A threshold in the range of -3 0 to -4 0 mV is widely reported for Kvi .2 expressed in Xenopus oocytes (Hart et al. 1993; Ito et al. 1992; PaulmichI et al. 1991 ; Sprunger et al. 1996; Stühmer et al. 1989). A rat cardiac Kvl .2 expressed in Xenopus oocytes showed an activation threshold of -5 2 or -5 7 mV according to the amount of cRNA injected (Guillemare et al. 1992). In the mammalian cell line CL1023. Werkman et al. (1992) observed Kvi .2 currents at potentials positive to -2 0 mV.

Grissmer et al. (1994) reported a kinetic curiosity: Kvi .2 currents evoked by steps to +40 mV increased in size with repeated activations. They suggested that this was caused by the existence of successive closed states of the channel. Thus, after deactivation, the channels are initially in a closed state capable of rapid opening, but convert over seconds into a “resting” state from which they can open only slowly.

Russell et al. (1994) described the behaviour of Kvi .2 at near-physiological temperatures. At 32-35°C, canine colonic smooth muscle Kvi .2 currents show a more negative activation range and greater inactivation than at room temperature. 39

Potassium channel blockers acting on Kv1.2

Many blockers of Kvi .2 have been reported, the best-characterised being peptide toxins. Charybdotoxin (from a scorpion), dendrotoxin (from the green mamba Dendroaspis angusticeps) and mast cell degranulating peptide (from bee venom) are all potent blockers of Kvi .2 expressed in B82 mouse fibroblast cells (Werkman et al. 1992). Alpha-dendrotoxin generally blocks Kvi .2 (in both oocytes and mammalian cells) with half-maximal current inhibition at 2-4 nM dendrotoxin (Werkman et al. 1992; Stühmer et al. 1989; Guillemare et al. 1992). Dendrotoxin block of Kvi-subfamily channels depends on three amino acid residues in the 85-86 linker of the channel, which also contains the pore loop (Hurst et al. 1991); dendrotoxin effects are summarised in Harvey (1997). Besides charybdotoxin, several other scorpion toxins also block Kvi .2 (Werkman et al. 1993; Grissmer et al. 1994; Kharrat et al. 1996,1997; Rogowski et al. 1996). Further, several peptides from sea anemone species inhibit Kvi .2 currents (Schweitz et al. 1995; Pennington et al. 1996a, b; Gendeh et al. 1997; Cotton et al. 1997).

The existence of so many Kvi .2-blocking peptides in animal toxins may reflect the prevalence of Kvi .2 in non-CN8 regions of mammals. As Veh et al. (1995) describe, Kvi .2 is found in motor and sensory regions of the spinal cord, while Kvi .2 channels accessible to blood-borne toxins are present in the heart and in some types of smooth muscle. Blocking Kvi .2 channels can be expected to contribute to cardiac failure and general paralysis, adding to the poisonous effect of animal venoms.

While several cardiac drugs affect Kvi .2 currents, the Class III antiarrhythmics d-sotalol, E- 4031 and M8-551, do not block Kvi .2 (Yamagishi et al. 1993). The experimental cognition enhancer linopirdine (DuP 996) blocks Kvi . 2 with an IC50 of about 50 |iM (D.G. Owen, personal communication), which is much higher than its I C 5 0 of 3.4 pM on the sympathetic M- current (Lamas et al. 1997).

Kvi .2 can be blocked by the polyunsaturated fatty acids docosahexaenoic and arachidonic acid, their ethanolamides, and A®-tetrahydrocannabinol (Poling et al. 1996a, b). This appears to be a direct block of the channel. In contrast, the reported effects of arachidonic acid and cannabinoids on M- and M-like currents are receptor- or enzyme-mediated. Thus, arachidonic acid increased and then decreased the NG108-15 M-like current amplitude, with a decrease in relaxation time constant (Béhé & Meves 1992; Béhé et al. 1992), apparently through protein kinase 0 8( chmitt & Meves 1993). Mammalian hiopocamoal neurones show a native M-current resembling that of sympathetic neurones. This M-current was increased by arachidonic acid ( 8 chweitzer et al. 1990), whose lipoxygenase metabolite leukotriene C4 40 was suggested as the final mediator in the somatostatin receptor transduction cascade (Schweitzer et al. 1993; Lammers et al. 1996). Schweitzer (2000) inhibited this hippocampal M-current by activating native CB1 cannabinoid receptors with the agonist WIN-55,212. However, WIN 55,212-2 did not inhibit the M-current in rat SCGs transiently expressing CB1 cannabinoid receptors, though calcium channels were affected (Pan et al. 1996).

Neurotransmitter modulation of Kvi .2

The RAK (Rat Atrial K channel) differs by one amino acid from BK2 (PaulmichI et al. 1991). Huang et al. (1993) expressed RAK in Xenopus oocytes, and identified a tyrosine kinase- dependent pathway for inhibition of the current following activation of coexpressed Mi muscarinic receptors. On this basis, they identified as Kvi .2 a tyrosine kinase-inhibitable outward current which they obtained (during steps from -7 0 to 0 mV) in NG108-15 cells. The same group subsequently raised the possibility that “the RAK protein comprises at least part of the M-current channel” (Morielli & Peralta 1995). Huang et al. (1994) also reported a protein kinase A-mediated enhancement of the RAK clone, reproducing the effect of p2 modulation on the cardiac myocyte delayed rectifier.

Potential complications: heteromuitimers

The ability of channels to form heteromuitimers within and between subfamilies of voltage- gated potassium channels might explain differences between cloned and native currents. There is evidence for the association of different types of Kvi subunits inXenopus oocytes (Po et al. 1993), rat brain (Sheng et al. 1993; Wang et al. 1993; Deal et al. 1994), and bovine cerebral cortex (Scott et al. 1994a; Shamotienko et al. 1997). It is now widely accepted that Kv channels can only heteromultimerise within subfamilies. (Xu et al. 1995; Shen & Pfaffinger 1995; Tang et al. 1998). However, Shahidullah et al. (1995a, b) claimed Kv1.2/Kv3.1 heteromultimerisation, and Chen et al. (1996) reported distinctive current properties when Drosophila Shaker ar\6 eag channel RNAs were injected simultaneously into Xenopus oocytes. Thus, Kvi-subfamily channel subunits have an undisputed affinity for their close relatives, but at best coassemble less reproducibly with subunits from other groups. It is conceivable that, in certain cellular environments, a factor such as a chaperone or beta- subunit might allow “incompatible” channel subunits to cohabit.

Beta subunits

The “dendrotoxin receptors” cloned from rat and bovine brain contain not only conventional 6- pass transmembrane potassium channel subunits, but also smaller intracellular protein 41

molecules. The pore-forming proteins are now considered as a-subunits, and the accessory molecules as p-subunits. The Kvp subunits are intracellular proteins with consensus phosphorylation sites (Scott et al. 1994b), which associate in rat brain with Kvi .2 and Kvi .4 channels (Rhodes et al. 1995). This association is dependent on a defined molecular region of Kvi family subunits (Yu et al. 1996; Sewing et al. 1996). Kvpi .2 causes substantial inactivation of oocyte-expressed human Kvi .2 currents, depending on the redox state (Wang et al. 1996c). The Kvpi .2 subunit may also have a chaperone function (Accili et al. 1997).

Working on the related channel Kvi .5. Uebele et al. (1996) found that its activation range was shifted negatively in mouse L-cells and CHO cells, compared with HEK 293 cells or Xenopus oocytes. Northern blotting correlated this with the presence of an endogenous p- subunit, which was cloned from L-cells as mouse Kvp2.1 and appeared also to be present In

CHO cells. In COS cells, Kvp2 subunits appear to act as chaperones and stabilise Kvi .2 (Shi et al. 1996). Li et al. (2000) provide an explanation of why Kvi .2 channels may not be well expressed in some cell types. They identify a VXXSL motif in the C-terminal region of Kvi .4 channels, which promotes membrane targeting of Kvi-family channels in HEK 293 cells. Kvi .2 has VXXSN in the equivalent position; its sub-optimal expression in these cells can be improved by an N ^ L substitution in the C-terminal sequence, or by coexpressing

Kvp2 subunits with wild-type Kvi .2.

Other molecules associated with Kvi .2 channels

The targeting of Kvi .2 channels to specific cellular regions implies the existence of control mechanisms, both for directing the protein to the appropriate part of the cell (as in the above paragraph), and for keeping it in place when it gets there. Kim et al. (1995), using a yeast two-hybrid screen of a human brain cDNA library, found that the C-terminal region of Kvi .4 associated with three intracellular proteins having characteristic 90-amino-acid sequences known as PDZ domains (Gomperts 1996). Numerous partners for PDZ domain proteins have been suggested, including neuronal nitric oxide synthase (Brenman et al. 1996), NMDA receptors (Kim et al. 1996), sodium channels and G-protein-coupled receptors, as well as potassium channels. These can potentially create synaptic assemblies of receptors, ion channels and enzymes, probably in specific combinations according to the cell type. Kvi .2 was found colocalised with Kvi .1 and PSD-95 at septate-like junctions in cerebellar pinceaux (Laube et al. 1996). In COS cells, however, Kvi .2 did not cluster efficiently with PSD-95 (Kim

& Sheng 1996) unless Kv(32 was present (Tiffany et al. 2000), suggesting that Kvi .2/PSD-95 binding is secondary to surface availability of the Kvi .2 channels. By contrast, the 42

intracellular location of SAP97, and its ability to prevent channels reaching the membranes of COS celis, may suggest a role in trafficking (Tiffany et ai. 2000).

Summary of the case for Kvi .2 as an M-current constituent

The presence of Kvi .2, at least as mRNA, in cells with M-currents and M-like currents is a necessary precondition for a roie in generating M-like currents. Kvi .2 message has been found in SCO neurones (aibeit at a low level), and in NG108-15 cells. In the central nervous system, M-like currents have been studied only in selected regions, while Kvi .2 is so widespread that a fuil correlation might find appreciable overlap.

Regarding kinetic properties of Kvi .2, the possibie presence of p2-subunits in the CHO and fibroblast expression systems I have used may affect Kvi .2 current properties. The absence of such chaperone/modulatory subunits in other ceils, such as NG108-15, might lead to lower surface expression and an altered voltage threshold for activation of Kvi .2. Alternatively, the presence of some other accessory subunit in neuronai ceils might change current characteristics.

The Dharmacoloov of Kvi .2 as previously reported does not match that of M- and M-like currents, especially since the latter are insensitive to dendrotoxin. Conceivably, though, post- translational modifications such as glycosylation might operate more extensively in native cells, and interfere with the access of a toxin to a binding site.

Muscarinic modulation of Kvi .2 has been shown, as have other receptor-mediated effects, with some discrepancies between Kvi .2 and M-like currents. A proviso is that cell-type- specific cytoskeletal associations with receptors and kinases, or the presence of a p-subunit obstructing channel phosphorylation sites, might lead to differences in the capacity for modulation by particular transduction pathways.

Thus, while Kvi .2 showed little resemblance to the sympathetic M-current, it could not be completely ruled out as a constituent of the NG108-15 M-like current when I started this work. There was no way to define how broad a range of effects beta-subunits, heteromultimerisation and other cell-specific post-translational modifications might have. Rodent celis expressing rat and mouse ciones of Kvi .2 seemed an appropriate model: as mammalian cell lines, they might be expected to approximate more closeiy than amphibian oocytes to the rodent neuronal cells which express M- and M-like currents. 43

3.2: Results rKv1.2: the CL1023 line

The rat homologue of Kv1.2 (rKv1.2). cloned from neuroblastoma-glioma cells and otherwise known as NGK1, was expressed reliably by CL1023 fibroblasts. Figure 3.1 illustrates its properties during a series of 1 -second steps from -7 0 mV to progressively more positive potentials. rKvl .2 shows slow, slightly sigmoidal activation starting around -3 0 mV, and visible time-dependent inactivation during 1 -second steps to above +30 mV.

Conductance curves

I constructed the rKvl .2 conductance curves in Figure 3.1 by recording currents with the protocol above, leak-subtracting, and measuring the “peak” and end-of-step amplitudes in Clampfit. These are the same measurements made by Werkman et al. (1992). “Peak” values were the maximum amplitudes between cursors enclosing the whole episode; because activation was not complete on the smaller depolarisations, the “peak” value should not be taken to represent a steady-state current amplitude. End-of-step values were the mean amplitudes between two cursors placed close together near the end of the voltage step, at about 1 second after the start of activation.

To convert current into conductance, it is necessary to calculate the driving force on potassium ions at any given potential, and this requires a reversal potential. Theoretically, the 175 mM potassium internal and 2.5 mM potassium external solutions should give a reversal potential Vrev of -107.3 mV, according to the equation Vrev = (ln([K'^]o/[K^j)).RT/ZF. In practice, I obtained a less negative value. I recorded tail-currents using a deactivation protocol (see later) starting at 0, +10 or +20 mV and stepping down by 10 mV steps to -110 mV. When the initial and final amplitudes of the currents are plotted, the point where the “initial” and “final” curves cross is the actual reversal potential of the current in each cell. An average of 8 cells gave a reversal potential of -9 0 mV, which is the value used here.

The conductances plotted in the lower part of Figure 3.1 are given by the equation G = l/(V- Vrev), where I is in amperes and V is in volts. Conductances are normalised to the highest value in each case, namely +60 mV for the “peak” conductance and +30 mV for the end-of- step conductance. “Peak” and end-of-step conductances are fitted (up to the point used for normalisation) with Boltzmann curves constrained to start at 0 and finish at 1. The slope of the curve is similar for “peak” and end-of-step rKvl .2 currents, 9 and 7 mV, respectively, per e-fold change in conductance. The half-activation voltages measured from these data are +4.5 and -0.2 mV for “peak” and end-of-step currents, respectively. 44

Figure 3.1 : Conductance of rKvl .2 In mouse fibroblast cells

(Top) Activation protocol;

(Upper part) rKvl .2 current family evoked by voltage steps from -7 0 mV, starting at -9 0 mV and incrementing by +10 mV intervals

(Lower part) Calculated conductances + s.e.m., for “peak” and end-of-step currents in rKv1.2. Fitted parameters for Boltzmann equation: end-of-step (black): V% = -0.22 ± 0.25 mV k = 7.34 + 0.22 mV

“peak" (grey): V% = 4.55 ± 0.29 mV k = 9.06 + 0.27 mV +60 mV 45

-70 mV -90 mV

1 0 -

< c c

3

0.0 0.5 1.0 1.5 time (s) 2.0

1.0 -

Q) Ü 0 . 8 - ^ "peak"(n=16) Boltzmann fit I3 ■D ■ end(n=17) C Boltzmann fit 0 Ü 0.6 - T 3 (1) W 10 1 0.4-1 o c c CO (D E 0 .2 -

0.0 -

“T“ -f— —|— -80 -60 -40 -20 0 20 40 60 membrane potential (mV)

Figure 3.1 46

Activation kinetics

I attempted a fit of activation time constants to compare with the literature data of PaulmichI et al. (1991) for the RAK channel expressed in Xenopus oocytes. One problem is the apparently sigmoidal nature of rKvl .2 activation, and another is that the Kvi .2 current starts to inactivate before it has finished activating. I tried two approaches. One involved selecting the aooarentlv exponential portion of the current evoked by depolarisation (avoiding both the very beginning and the region where the current starts to decline on large steps), and fitting a single exponential growth equation. The second method was to fit three exponential components to long depolarising current steps. Typically, this gave two time constants with a positive amplitude (representing current decline by inactivation), and one with a negative ampiitude (representing activation) which I plotted. Figure 3.2 shows the results of both fitting procedures. Both methods gave similar activation time constants which decreased with increasing levels of depolarisation, values being 1-3 s at 0 mV, decreasing to 30-50 ms at +50 or +60 mV.

When recorded at room temperature, rKvl .2 activated more rapidly on repeated depolarisations. Figure 3.3 shows this effect on two different cells using different voltage steps. The records in the upper portion of the figure inciude a deactivation from +40 mV to -1 0 mV. The superimposition of the tail-currents at -1 0 mV demonstrates that deactivation behaviour is independent of the preceding activation rate. The lower portion of Figure 3.3 shows a more abrupt activation of the current on the second trace. In all cells tested, the effect had substantially disappeared 20 seconds later.

Inactivation accumulates with prolonged or repeated activation of Kvi .2, and recovery from the inactivated state also proceeds on a time course of many seconds. An interval of at least 20 seconds between activations from rest is necessary both to maintain the full current amplitude and to avoid changes in activation kinetics. 47

Figure 3.2: Activation xs for rKvl .2

(Top) X + s.e.m. for a single-exponential fit to rKv1.2 currents activated from -7 0 mV.

(Bottom) X + s.e.m., using x values extracted from a three-component exponential fit to rKvl .2 currents activated for 6 seconds from -7 0 mV and showing inactivation during the step. 3.5-1 48

3 .0 -

activation x from 1-exponential fit 2 .5 -

2 .0 -

1 .5 -

1.0 -

0.5 - m m

0.0 -

-0.5 T" T“ 10 20 30 40 50 60 c o 3.5-1 g •g OJ 3 .0 - ♦ activation x from 3-exponential fit of long current steps 2 .5 -

2 .0 -

1 .5 -

1.0 - I

0 .5 -

0.0 -

-0.5 T 10 20 30 40 50 60 membrane potential (mV)

Figure 3-2 49

Figure 3.3: Acceleration of rKv1.2 by repeated activation

(Top) Activation/deactivation protocol and four pairs of rKv1.2 currents at progressively longer Intervals. The black trace Is the current activated from rest; the grey trace Is the current re­ activated after the specified delay. After a delay of 15 seconds, the current traces coincide.

(Bottom) Activation protocol and four pairs of rKvl .2 currents In a different CL1023 cell, at progressively longer Intervals. Here, currents are nearlv superimposed after a 15-second delay. 50

+40 m V Trace 1 1 second ■10 mV T race 2 ■70 m V Zero delay 5 seconds delay between episodes between episodes

10 seconds delay 15 seconds delay

1 nA

2 5 0 m s

0 m V

1.8 seconds

-70 m V 1 nA

Zero delay 5 seconds delay 2 5 0 ms

10 seconds delay 15 seconds delay

Figure 3-3 51

To check current behaviour when a full step protocol would take too long, such as during drug applications or temperature alterations, I devised a ramp protocol. This Is Illustrated In the top record of Figure 3.4. Four posltlve-going voltage ramps were applied at the minimum Interval allowed by Clampex. The effects are striking, though not quantitative because of an unknown time-lag In the manifestation of Inactivation phenomena. Still, It Is clear that the first activation of the current from rest displays a substantially sigmoidal current/voltage relation, whereas subsequent steps show extra Inactivation starting below 0 mV. For all such ramp records, I have shown the voltage traces recorded with the Axoclamp-2 amplifier, as evidence that poor clamp Is not responsible for the sudden dips In current. Where appropriate, I have annotated the time-scale for the current records with some voltage landmarks, based on the achieved (rather than command) potential.

The temperature series In Figure 3.4 shows that the discrepancy between Initial and accelerated activations of current Is greater at low temperatures. The first activation from rest has a higher voltage threshold at 20°C than at room temperature, while the accelerated currents have a similar starting-point. The difference between first and subsequent activations appears to be negligible In the physiological temperature range above about 30°C, where all activations occur at similarly negative-shifted potentials and Inactivation Is pronounced. Thus, accelerated activation on repeated steps Is an Interesting “fingerprint” for the expressed Kvi .2 channel In a cell line, but Is unlikely to be physiologically relevant.

The negative shift In activation on the second ramp Is not visible unless the first ramp has reached a voltage more positive than 0 mV, and Is maximal after activation at a nominal +50 mV (not shown). This effect, therefore, seems to follow activation at voltages producing visible rapid Inactivation. Figure 3.5 shows an Intriguing phenomenon elicited by a series of mirror-lmage ramps: with a small time lag, the rapid Inactivation seen at Increasing positive voltage Is reversed as voltage falls again. Thus, depolarisation Induces both activation and Inactivation, and hyperpolarisation leads to both deactivation and removal of Inactivation.

On the ramp currents of Figure 3.5 (as with the tall-currents In Figure 3.3), current decline on hyperpolarlsing follows a similar time-course Irrespective of the activation history of the channel. This Is apparent In the bottom section of Figure 3.5, where I have plotted the first and second ramp currents against the actual membrane potential recorded by Clampex. A hysteresis loop emerges. In which the current/voltage slope during the latter part of the hyperpolarisation coincides for both currents, and parallels that on the start of the second depolarisation, but Is displaced to more negative potentials. This shift probably reflects a time-lag In actlvatlon/deactlvatlon related to the ramp speed. 52

Figure 3.4; Effects of temperature on rKv1.2 activation

(Top) Recorded voltage (left) and current traces (right) using four ramp commands at the minimum interval permitted by Clampex. In each record of this figure, traces are numbered on the far right, and the first activation from rest is indicated in grey. Approximate voltage landmarks (dotted grey vertical lines) are derived from the voltage plots on the left. This (top record) is a control record at room temperature (24®G).

(Centre) Voltage (left) and current (right) traces in the same ceil, but during perfusion with chilled Krebs solution, reservoir temperature 20°C. The first (grey) current trace activates more positively than in the control record. The subsequent currents have a similar negative shift in threshold to that in the control record, but show less inactivation.

(Bottom) Voitage (left) and current (right) traces in the same ceii during perfusion with heated Krebs, reservoir temperature 33°C. All current activations occur more negativeiy than in the control condition, the difference in threshoid is negligible between the first and subsequent current activations, and inactivation is early and pronounced. 53

-50 mV -25 mV +2 0 m V 75-1 Control

5 0 - 1 0 -

2 5 - 3,4 D) < C 0 - c

Q) -2 5 - I = 3 O

-5 0 -

■75 0 2 3 0 1 2 3

7 5 -, Cold, 20°C

5 0 - 1 0 -

2,3,4 2 5 -

0 -

-2 5 -

-5 0 -

-75 0 1 2 3 0 1 2 3

7 5 -,

5 0 - 1 0 - Hot, 33°C 2 5 -

0 -

-2 5 - 3,4

-5 0 -

-75 0 1 2 3 0 1 2 3 Time (seconds) Time (seconds)

F ig u re 3-4 54

Figure 3.5: “Hysteresis” In rKvl .2 suggests complex gating behaviour

(Top) Recorded voltage traces for four mirror-image ramps applied with the minimum delay permitted by Clampex. The first trace is picked out in grey. Ail voitage traces are substantially superimposed, and (despite the large currents evoked) a command of +60 mV achieves an actual recorded voitage of about +50 mV. Notably, there is no evidence for a serious failure of membrane clamp at positive potentials.

(Centre) rKv1.2 currents evoked by the voitage ramps illustrated above. The left-hand side (ascending ramp command) shows typical rKv1.2 behaviour. On the right-hand side, as voltage falls from +50 mV, the dip in current amplitude is temporarily reversed. Finally, all current traces converge as they finish deactivating.

(Bottom) The first (grey) and second (black) current traces from the current recording (above), plotted against the corresponding actual voltage (top). Positive to -1 0 mV, currents show different relationships to voltage on depolarising and hyperpolarising, suggesting complex gating behaviour. However, at negative potentials, the current response to hyperpolarisation (both traces) parallels the onset of the accelerated (second) current during depolarisation (black). Allowing for a time-lag In channel opening and closure, these portions of the record show similar voltage-dependency, and therefore could represent a single gating transition going in opposite directions. 55

Membrane potential (mV) 80 -n 40 -

-40 — -80 0 1 2 3 10-1 < c

6 - 3 ü ■a

0 1 2 3 time (seconds)

< Hyperpolarising c 10 n (1st trace) c Hyperpolarising 2 (2nd trace) \ 3 O ■o î £ 4 - S 3 Depolarising (1st trace) CO 05 Depolarising (2nd trace) 0

0 - TT -80 “60 -40 -20 0 20 40 60 80 Actual membrane potential (mV)

Figure 3-5 56

Subthreshold inactivation phenomena

Figure 3.6 demonstrates that prepulses at -5 0 to -1 0 mV, which themselves evoke little current, caused a reduction of the current amplitude on a subsequent positive voltage step. Thus, there is an inactivation process occurring even at -2 0 mV that can reduce the current amplitude reached during an ensuing depolarisation, without accelerated activation being apparent: activation xs change by a factor of 2 at most.

Deactivation

One feature of Kvi .2 behaviour is quite consistent whatever the history of current activation. Kvi .2, identified as mRNA in NG108-15 cells during a search for the M-current channel, shows deactivations which visually resemble those of IVI-currents. Figure 3.7 shows an example of deactivations following a step to +20 mV, and plots of the time constants obtained from eight such records, using a two-exponential fit to relaxations in the M-current voltage range. As indicated in the Methods chapter, fitted xs exceeding the step duration, such as those of two seconds or more plotted in Figure 3.7 (bottom left), are not quantitatively reliable. However, a slow component of current decline is evident, and necessitates a two- exponential fit. A clearly visible, relatively fast deactivation process is responsible for the xs of 60-300 ms (Figure 3.7, bottom right), which are similar to M-current values (see Chapter

5). The U-shape of both %N curves is unexplained. I suspect the upturn in x at negative potentials is an artefact, possibly arising from fitting errors. A very slow drift in current would produce a large x, and probably one rather than two components would suffice at highly negative potentials. Another possible explanation is the presence of endogenous channels active at -7 0 mV and more negative, though fibroblast cell types are generally agreed not to have prominent ionic currents of their own. 57

Figure 3.6: Steady-state inactivation of rKvl .2

(Top) Voltage protocol, with a 30-second moderate depolarising step followed by a 1-second step to 4-30 mV.

(Centre) rKv1.2 currents evoked by the protocol above. A small current is evoked during 30- second steps (prepulses) to -2 0 and -1 0 mV. The subsequent current activation at 4-30 mV is annotated (right-hand side) with the voltage command for the preceding 30-second prepulse. It is clear that the current at 4-30 mV is reduced by inactivation taking place at -2 0 or -1 0 mV, although inactivation is not visible during this step.

(Bottom left) A plot of rKvl .2 current amplitude + s.e.m., at 4-30 mV, against prepulse potential indicates that some inactivation also takes place during steps to -4 0 and -3 0 mV, where negligible currents are evoked. (Bottom right) A plot of the time constant, ± s.e.m., for activation of the rKvl .2 current at 4-30 mV shows little dependence on prepulse potential up to -1 0 mV. ■vj o average normalised amplitude at +30 mV 3 < o 3 <

CD "g

CD S -

1 ^ ' 3 6.

3 <

average t (ms) at +30 mV

T3 ■OCD C w CD "D O CD3

3 < 4- ro CO o o O ■o c 3 3 CaD 3 c/) < < O < CD en 3 00 < 59

Figure 3.7: Deactivation of rKvl .2 from +20 mV

(Top) Voltage protocol used to study deactivation of rKvl .2.

(Centre) An rKvl .2 current family resulting from this protocol. The initial step to +20 mV, evoking a substantial current without pronounced inactivation, is followed by hyperpolarising steps at -1 0 mV increments.

(Bottom) A plot of T + s.e.m., using x values obtained from a 2-exponential fit to deactivation tails obtained using the above protocol. The large (slow) x values indicate a trend, but are not quantitatively reliable (see text). The faster xs obtained are -60-400 ms. +20 mV 60 -20 mV

-70 mV -90 mV

1 nA

2 5 0 ms

400-1

— ♦ — fast T (n=8)—■— slow T (n=8) 3 5 0 - —♦ — fast T (n=8)—■—

3 0 0 -

2 5 0 -

.2 0 0 -

1 5 0 -

1 0 0 -

5 0 -

1 0 TTTTTT T ■50 -40 -30 membrane potential (mV) 61

Pharmacology

One of the most consistent characteristics of Kvi .2 channels is their binding and blockade by the snake toxin a-dendrotoxin. The only doubt ever raised was by Guillemare et al. (1992), who found reduced dendrotoxin block inXenopus oocytes over-expressing Kvi .2. At the same time, the insensitivity of M-currents to dendrotoxin was a major bar to their identification with Kvi .2. It was thus relevant to check the sensitivity of Kvi .2 to dendrotoxin in the environment of the mammalian fibroblast ceil. Figure 3.8 provides examples of raw data obtained with 150 mM K^ internal solution (see Methods) and the perforated-patch recording configuration. Ail inhibitions are calculated on the maximum current amplitudes attained during the step. Currents evoked at a nominal 0 mV are typically about half-inhibited by 3 nM dendrotoxin, as in the top record in Figure 3.8. Mean data for four GL1023 cells (not shown) gave an IC50 of 3.6 nM. I also applied dendrotoxin under conditions enhancing Kvi .2 activation. In the centre section of Figure 3.8, an rKvl .2 current underwent similar block by 3 nM dendrotoxin, whether currents were allowed to return to their resting state (depolarisations at 40-second intervals) or accelerated by frequent activation (5-second intervals). The bottom part of Figure 3.8 shows that, in the presence of niflumic acid (whose effect I describe below), the greatly enhanced current was even more strongly blocked by dendrotoxin than the control current in the same cell. Because of the different extents of inactivation under control and niflumic acid-treated conditions, an exact assessment of differences was not pursued. Thus, Kvi .2 was consistently sensitive to dendrotoxin.

Niflumic acid is a known blocker of chloride channels (reviewed by Greger 1990). Along with related chloride channel blockers, it has also been reported to affect potassium channels, both calcium-activated (Gribkoff et al. 1996) and KCNQI/minK (Busch et al. 1994a; Busch et al. 1997a; Abitbol et al. 1999). More recently, a niflumic acid action on KCNQ5 was documented by Schroeder et al. (2000). The effect of niflumic acid on Kvi .2 is therefore interesting both for comparison with the effect on other potassium channels, and as a potential source of interference when it is used to eliminate chloride currents from native cells with a mixed complement of ion channels.

I applied a standard chioride-channel-biocking dose of 200 pM niflumic acid while running my usual Kvi .2 voitage protocols. Figure 3.9 shows the effect of niflumic acid on ramp currents, which resembles the effect of raised temperatures. The differential rate of activation between de novo and repeated current activations disappears, and the whole activation curve shifts to more negative voltages. This effect is substantially reversed on 3 minutes washout. 62

Figure 3.8: Dendrotoxin consistentiy blocks Kv1.2

(Top) rKvl .2 current activated at a nominal 0 mV (from -7 0 mV), showing that 50% inhibition in this cell occurs at a concentration between 3 and 10 nM dendrotoxin (D IX ).

(Centre) rKvl .2 currents activated once every 40 seconds (left) and accelerated by activation every 5 seconds (right), at 0 mV, with and without 3 nM dendrotoxin; both show similar inhibition.

(Bottom) rKvl .2 currents activated at 0 mV in the absence (left) and presence (right) of 200 pM niflumic acid. The activated current (right) is even more intensely blocked than the control by 10 nM dendrotoxin. 63

control at 0 mV

3 nM DTX: 35.7% inhibition

10 nM DTX: 70.8% inhibition

30 nM DTX: 86.1% inhibition

0.025 Hz control 0.2 Hz control

0.025 Hz: 3nM DTX, 52% inhibition 0.2 Hz: 3nM DTX 61% inhibition

0.5 nA

250 ms

200 nM niflumic acid at 0 mV in the same cell 1 nA

250 ms

10 nM DTX in 200 nM control at 0 mV niflumic acid: 82% 10 nM DTX at 0 mV: inhibition 54% inhibition

Figure 3-8 64

Figure 3.9: Effect of niflumic acid on rKvl .2 activation

Recorded voltage (left) and current traces (right) using four ramp commands at the minimum interval permitted by Clampex. Traces are numbered on the far right, and the first activation from rest is indicated in grey. Approximate voltage landmarks (dotted grey vertical lines) are derived from the voltage plots on the left.

(Top) a control recording before niflumic acid application.

(Centre) The same cell, 90 seconds after starting perfusion with 200 pM niflumic acid. All current activations begin at the same negative voltage, and inactivation is increased relative to control.

(Bottom) The effect is reversed after 3 minutes washing out. 65

-50 mV -25 mV +20 mV 12 Control 10

8 O) 6

T3 4 0 -2 5 - ■o 2

0

0 1 2 3

75 12 niflumic acid 10 50 200 ^M, 90s 8 25

6 0 3,4 4 ■25 2 -50 0

-75 01 2 3 0 1 2 3

75 12 3 mins wash

50 10

8 25

6 0

4 -25 2 -50 0

-75 0 1 2 3 0 21 3 Time (seconds) Time (seconds)

Figure 3-9 66

I tested the availability of Kv1.2 current in NG108-15 ceils by applying niflumic acid while running a protocol based on Filippov et al. (1994). From a holding potential of -2 0 or -3 0 mV, a hyperpolarisation to -6 0 mV was given in order to monitor the M-iike current relaxation, followed by a further step to 0 or +20 mV to evoke the delayed rectifier.

Figure 3.10 shows a cell in which the standing current increase in 200 pM niflumic acid was comparatively pronounced. Dendrotoxin, added in the continuing presence of niflumic acid, had little effect on the enhanced current at -3 0 mV and the delayed rectifier at 0 mV, and none on the enhanced relaxation at -6 0 mV.

In total, nine of eleven NG108-15 cells had an increased holding current at -2 0 or -3 0 mV in

200 pM niflumic acid, accompanied by an increased relaxation amplitude at -6 0 mV. The delayed rectifier was not consistently affected by niflumic acid, showing a slight increase or decrease or no change. Four ceils (still in niflumic acid) treated with 100 nM dendrotoxin had a slight decrease in the delaved rectifier, but no change in the enhanced current relaxation at -6 0 mV. Thus, the NG108-15 M-iike current included no dendrotoxin-sensitive Kvi .2 current even after niflumic acid treatment.

In two cells, I tested for a contribution of a classic M-current to the niflumic acid effect, by applying the M-current blocker linopirdine on top of niflumic acid. In these cells, 30 pM linopirdine, which would be expected to produce full block of a sympathetic M-current, produced only a fractional change in the niflumic acid-enhanced NG108-15 current at -3 0 and -6 0 mV, the size of the reduction being less than the original standing current and relaxation amplitudes. Thus, it appeared that neither the original standing current nor the additional niflumic acid-evoked current included much of the channel type forming the sympathetic M-current (unless niflumic acid blocked access to the linopirdine binding site).

I also considered a calcium-activated potassium current as a possible cause of the increase in standing current, since niflumic acid was stated to enhance currents through cloned Sio channels (Gribkoff et ai. 1996). However, five of the eleven cells were tested in caicium-free Krebs solution, and four showed a standing current increase in these iow-calcium conditions. The niflumic acid-evoked current in NG108-15 cells therefore remains unidentified. 67

Figure 3.10: Niflumic acid evokes negligible dendrotoxin-sensitive current in NG108-15

(Top) Voltage protocol used on NG108-15 cells, designed to demonstrate standing current at -3 0 mV, M-like current deactivation at -6 0 mV, and delayed rectifier activation at 0 mV.

(Main picture) Currents elicited by this protocol on an NG 108-15 cell showing a clear response to niflumic acid. The control trace (black) shows a small M-like current relaxation at

-6 0 mV. 200 pM niflumic acid (medium grey) causes an increase in standing current at -3 0 mV, with an increased deactivation amplitude at -6 0 mV. This cell also shows an increase in the delayed rectifier at 0 mV. Only the delayed rectifier increase was reversed by the Kvi .2 blocker dendrotoxin (pale grey). 68

0 mV -30 mV -60 mV

control

in 2 0 0 )a M niflumic acid

in niflum ic acid + 100 nM dendrotoxin

0.25 nA

0.5 s

0 nA

Figure 3-10 69

mKv1.2: the MK2 line

The CL1023 line developed problems with recording stability, but fortunately Dr. Bruce Tempers MK2 line of Chinese Hamster Ovary cells was available. This line was stably transfected with the MK2 (mouse brain) clone of Kv1.2, which I refer to here as mKv1.2. An example of a current family is shown in Figure 3.11. The current was essentially similar to rKvl .2 in its activation, deactivation, inactivation and acceleration phenomena. The minor differences may be attributable to the largely acetate-based internal solution, which I chose because it had partially alleviated the instability of CL1023 recordings. Allowing for a measured junction potential correction of -8 mV with this internal solution, the current (when activated from -7 8 mV) usually began to activate around -2 0 mV, and had visible time- dependent inactivation above +30 mV.

Conductance curves

I constructed conductance curves as for rKvl .2, after making a junction potential correction of -8 mV to the recording voltages and the reversal potential. The 110 mM potassium internal and 2.5 mM potassium external solutions theoretically give a reversal potential of -95.5 mV, according to the equation given previously. The value used here, obtained empirically from a plot of the Initial and final amplitudes of 12 tail-currents, is -8 0 mV (which incorporates the junction potential correction).

I normalised conductances to the maximum in each data set, plotted normalised values against voltage, and fitted them (up to the point used for normalisation) with Boltzmann curves constrained to start at 0 and finish at 1. Figure 3.12 shows conductances for (top) currents activated after a resting period of at least 20 seconds, and (bottom) currents activated 200 ms after the end of a 1-second prepulse to +52 mV, which produced maximal acceleration. The maximum conductance, to which all values were normalised, was at +32 mV for activations from rest, +12 mV for end-of-step accelerated currents, and +22 mV for “peak” accelerated currents. For mKvl .2 activations from rest, the slope of the resulting curve was 7 mV per e-fold change in conductance, and the half-activation voltage was +5 mV, for both “peak” and end-of-step currents. These values are close to thoseI obtained for rKvl .2 in the CL1023 line: the “peak” rKvl .2 current had a slope of 9 mV and a of +4.5 mV, and the end-of-step rKvl .2 current had a slope of 7 mV and a Wa of -0.2 mV. mKvl .2 currents accelerated by a prepulse had a slope of 6-7, and half-activation voltage - 18 mV; I did not quantify the properties of accelerated currents for rKvl .2. 70

Figure 3.11 : Example of an MK2 current (Above) Activation protocol;

(Below) mKvl .2 current family evoked by voltage steps from -7 8 mV, starting at -9 8 mV and incrementing by +10 mV intervals. +52 mV 71

-78 mV -98 mV

7

6

5

4 < c c 0) 3

Ü

2

1

0

0.0 0.2 0.4 0.6 0.8 1.0 1.2

Figure 3-11 time (s) 72

Figure 3.12: Conductance of Kvi .2 In MK2 OHO cells

(Top) “Peak” (grey) and end-of-step (black) conductance, + s.e.m., plotted against voltage for mKvl .2 currents activated for Is from -7 8 mV. Boltzmann fit parameters (“peak”): V% = 5.21 ± 0.52 mV k = 7.16 ±0.26 mV

Boltzmann fit parameters (end-of-step): V% = 4.49 ± 0.53 mV k = 7.04 + 0.24 mV

(Bottom) “Peak” (grey) and end-of-step (black) conductance, ± s.e.m., plotted against voltage for mKvl .2 currents following a step to +52 mV. Boltzmann fit parameters (“peak”): V% = -17.64 ±1.80 mV k = 6.89 + 0.96 mV

Boltzmann fit parameters (end-of-step): V% = -17.87 ± 1.55 mV k = 6.12 + 0.82 mV 73

0 . 8 -

Activating from rest ■ end (n=25) 0.6 - Boltzmann fit ♦ peak (n=25) 0.4 - — Boltzmann fit

0.2 -

■o 20 40 60 80 O

2 0.8 - After a +52 mV prepulse ■ end (n=6) Boltzmann fit c 0.6 - ♦ peak (n=6) Boltzmann fit

0.4-

0 . 2 -

0 . 0 -

120 -100 -80 -60 -40 -20 0 20 40 60 80 membrane potential (mV) Figure 3.12 74

Kinetics

Activation of the mKv1.2 current was less sigmoidal than that of rKv1.2, and was readily fitted by two exponential components. Figure 3.13 plots the resulting time constants against voltage. Activation xs were about 120 and 980 ms for a step to +2 mV, and 8 and 78 ms at +52 mV, the maximum command potential used when the junction potential was allowed for.

Acceleration of mKvl .2 activation occurred with repeated depolarisations and at raised temperature, as for rKvl .2. An example of this behaviour by mKvl .2 is shown in the upper left corner of Figure 3.16, where it serves as a control for the effect of 200 pM niflumic acid. Acceleration of the current is only partial because this protocol has two successive steps to the same potential rather than a consistent positive prepulse. However, as described for rKvl .2, prompt re-activation of the current causes a faster rise in amplitude and the appearance of pronounced inactivation, which was previously masked by slow activation of the current.

Inactivation of mKvl .2 during strong depolarising steps resembles that of rKvl .2 (compare Figure 3.11 with Figure 3.1).

Deactivation of mKvl .2 is also superficially similar to that of rKvl .2. Figure 3.14 (upper part) shows an example current family, resulting from an initial step to +12 mV and deactivations at -1 0 mV increments. Deactivation time constants from records of this type (n=13) gave the plots in the lower section of Figure 3.14, of x against membrane potential. The slower deactivation xs, around 100-300 ms, are mathematically sounder than those obtained on rKvl .2 because they fall well within the 1-second duration of the step (see Methods chapter and p.56). Fast xs are 10-40 ms over an M-current-appropriate range of voltages. All xs, slow and fast, for deactivation potentials negative to -3 0 mV are below about 200 ms, which would not rule out Kvi .2 as an M-like current constituent. 75

Figure 3.13: Activation time constants for mKv1.2 xs from a two-exponential fit of Kvl .2 activations from -7 8 mV, plotted separately:

(Above) slow x ± s.e.m.;

(Below) fast x ± s.e.m. 76 1.2 -1

1.1 - m ean slow t

1 . 0 -

0.9 -

0 .8 — (/) 0 . 7 - c 0.6 H g CO 0.5 - > o 0 . 4 - co 0 . 3 -

0 .2 -

0.1 -

0 .0 I n r "T" “T" 0 10 2 0 30 40 50

160 -

1 4 0 - ■ mean fast x

1 2 0 -

^ 100-

c5 60 — > ü CO 40 -

2 0 -

0 10 20 30 40 50 membrane potential (mV) Figure 3-13 77

Figure 3.14: Deactivation of mKv1.2 currents from +12 mV

(Top, centre) Voltage protocol used to study mKv1.2 deactivation, and corresponding current family.

(Bottom left) slow x + s.e.m., plotted against voitage;

(Bottom right) fast x + s.e.m., plotted against voltage. Both data sets from 13 cells. +12 mV -18 mV 78

-78 m V -1 0 8 m V 7000 -,

6000 -

5 0 0 0 -

4000 -

3 0 0 0 -

1 nA 2000 - 0.5 s 1000 -

-1000 0 1000 2000 3000 4000 5000

1.0-1 50 -

♦ slow T 45 - fast T 0.8 40 -

35 - 0.6

30 - $

(/) ^ 0.4 E 25 - H 20 - 0.2 $ ♦ 15 -

10 - 0.0

5 -

- 0.2 1 u -i— '— I— '— I— '— I— '— I— '— I -120 -100 -80 -60 -40 -20 0 -120 -100 -80 -60 -40 -20 0 membrane potential (mV) 79

Pharmacology

Figure 3.15 summarises some concentration/response data on mKvl .2 for dendrotoxin, as well as three substances not previously characterised as Kvl .2 blockers. Dendrotoxin blocked the mKvl .2 current with an IC50 of 4.5 nM (close to the value of 3.6 nM that I found for rKvl .2, see p.62). Forskolin, tested with the aim of reproducing the reported enhancement of Kvl .2 currents by protein kinase A activation, proved instead to block the mKvl . 2 current in a concentration-dependent fashion, with an I C 5 0 of 25 pM. Dideoxyforskolin, an analogue which does not activate adenylate cyclase, also had a blocking effect at 10 pM (n=1 ; not shown). 2 mM dibutyryl cyclic AMP, a membrane permeable cAMP analogue, had no effect (n=3; not shown). Taken together, this supports the non-involvement of a protein kinase A mechanism in the inhibitory action of forskolin.

Lanthanum had an I C 5 0 of 25 pM on mKvl .2, for a maximum inhibition of 74% at + 1 2 mV.

Zinc, potentially a physiological blocker of neuronal ion channels, had an I C 5 0 of 0.43 mM on mKvl . 2 at + 1 2 mV. Barium (not shown) had an I C 5 0 of around 0.8 mM averaged over a range of voltages from - 8 to +52 mV; there was no great difference in sensitivity between- 8 and +52 mV.

Niflumic acid at 200 pM affected mKvl .2 in the same way as rKvl .2. This is illustrated in Figure 3.16, using an activation protocol with two successive steps to the same potential.

The current elicited at - 8 mV is labelled, to demonstrate the increased amplitude and activation rate in niflumic acid compared with the control current. Using this step protocol on 7 cells, I constructed conductance curves for the niflumic acid-enhanced current. These are shown in the lower part of Figure 3.16. The half-activation voltages of -1 7 (“peak”) and -2 0 mV (end-of-step) in niflumic acid closely resemble the value of -1 8 mV for both “peak” and end-of-step currents after a +52 mV prepulse (Figure 3.12). Currents elicited by depolarising ramps in 2 0 0 pM niflumic acid (not shown) showed a negative shift in activation threshold, like that for rKvl .2 but less pronounced; the first mKvl .2 ramp current from rest was not fully left-shifted. Deactivation showed no major changes in the presence of niflumic acid (not shown).

Two other chloride channel blockers, DIDS and SITS at 100-1000 pM, did not accelerate or activate the mKvl .2 current (not shown). This supports a mechanism involving a direct interaction of niflumic acid with the Kvl .2 channel, rather than via endogenous chloride channels. 80

Figure 3.15: Concentration/response curves on mKvl .2 All data points are plotted as mean inhibition ± s.e.m.

(Top left) Alpha-dendrotoxin (n=4) Hill equation fit parameters:

Y m a x = 100% ICso = 4.47 ± 0.82 nM nH = 0.72 ±0.08

(Top right) Forskolin (n=6) Hill equation fit parameters:

Ymax = 100%

ICso = 24.58 ± 0.82 pM nH = 1.10 + 0.03

(Bottom left) Lanthanum (n=7) Hill equation fit parameters:

Y m a x = 74.38 ±0.13%

ICso = 25.38 ±11.83 pM nn = 1.04 ± 0.20

(Bottom right) Zinc (n=6) Hill equation fit parameters:

Y m a x = 100% ICso = 0.43 ± 0.04 mM nn = 1.23 ±0.14 81

100-1 100-1 Alpha-dendrotoxin: Forskolin: 2 steps, 2 ramps 90- n=6, ramps to +22 mV ? to +22 mV C 80- 80- g 70-

2 60- 60- c o o 50- o 40- 40- c o 30-

2 0 -

1 0 -

0.01 0.1 1 10 100 1000 1 10 100 1000 [DTX] (nM) [Forskolin] (pM)

100-1 100-1 Lanthanum: Zinc: 90- 90- n=7, steps n=6, steps 80- to +12 mV 80- to +12 mV

70- 70-

60- 60-

50- 50-

40- 40-

3 0 - 30-

2 0 - 2 0 -

1 0 - 1 0 -

1 10 100 1000 0.01 0.1 1 10 [Lanthanum] (pM) [Zinc] (mM)

Figure 3-15 82

Figure 3.16: Effect of niflumic acid on mKvl .2

(Top) Voltage protocol with two successive steps to the same potential.

(Above) mKvl .2 current families evoked by this protocol, in the absence (left) and presence (right) of 200 pM niflumic acid. The current at -8 mV is indicated in grey, to demonstrate the kinetic changes caused by repeated activation and niflumic acid. Note that the effect of niflumic acid on the current profile resembles that of prior activation.

(Below) “Peak” (grey) and end-of-step (black) conductance + s.e.m., against voltage for activations from rest in 7 MK2 cells in 200 niflumic acid. Boltzmann fit parameters (“peak”, grey): = H—16.77 + 1.13 mV k = 5.50 4^0.48 mV

Boltzmann fit parameters (end-of-step, black): V% = + -20.43 ± 1.41 mV k = 6.12 + 0.80 mV 83

+52 mV

-8 mV

-98 mV

0.5 nA

0.5 s

8 mV

-8 mV

control in 200 |iM niflumic acid

8 1 0 - c 2 Conductance for Kv1.2 u 0.8 - in MK2 OHO cells in ■o c 200 luM niflumic acid o o 0.6 - ■O 0 0 0.4 - ■ end (n=7) in niflumic acid 0 E Boltzmann fit . - peak (n=7) in o 0 2 ♦ c niflumic acid c Boltzmann fit 03 0 .0 ^ 0 E 100 -80 -60 -40 -20 0 20 40 60 membrane potential (mV) Figure 3-16 84

3.3: Discussion

Activation and conductance data

I find an activation threshold of approximately -3 0 mV for rKv1.2 in CL1023 cells, and -2 0 mV for mKv1.2 in the MK2 line. The half-activation voltage is 0 mV for rKv1.2 measured at the end of a 1-second step, and +5 mV for the “peak” rKvl .2 current (in this case, the highest current measured during a 1-second step). The half-activation voltage is +5 mV for mKvl .2 for both “peak” and end-of-step measurements. Pre-activation of the mKv1.2 current by a large positive pulse caused a shift of the half-activation voltage to -1 8 mV.

Workman et al. (1992) observed rKvl .2 currents at potentials positive to -2 0 mV in CL1023 cells. Their half-activation voltages, 0 mV for the “steady-state” current and +5 mV for “peak” current, were identical to my figures for this cell line. Using the same cells, Grissmer et al. (1994) reported a half-activation voltage of +27 mV for the “peak” conductance.

Kvl .2 has often been expressed inXenopus oocytes. The activation threshold in this cell environment is generally reported to be somewhere negative to -3 0 mV. Stühmer et al. (1989) report a threshold negative to -4 0 mV for rKvl .2 (RGK5 clone) in oocytes, with a half­ activation voltage of -3 4 mV. Ito et al. (1992) give the threshold for rKvl .2 (NGK1) activation in oocytes as -3 6 mV. Hart et al.’s (1993) canine colonic Kvl .2 produced currents at -3 0 mV in oocytes. rKvl .2 (RAK clone) in oocytes began to activate positive to -4 0 mV (PaulmichI et al. 1991). Sprunger et al. (1996) also worked on RAK. observing that substantial activation occurred around -3 5 mV; their activation curve (from tail currents in symmetrical K^) yielded a half-activation voltage of -21 mV. Guillemare et al. (1992) cloned a rat cardiac Kvl .2 and expressed it in Xenopus oocytes, finding an activation threshold of -5 2 or -5 7 mV according to the amount of cRNA injected (low and high levels, respectively). Conductance curves from tail currents showed a half-activation voltage around -2 0 mV in this case.

In order to have some data for comparison with the literature, I attempted to measure activation time constants for Kvl .2 clones. Despite signs of sigmoidal behaviour on the smaller depolarisations, some fairly consistent values emerged. Single activation xs fitted to rKvl .2 currents in GL1023 cells were 1 -3 s at 0 mV, and 30-50 ms at +50 or +60 mV. Activation xs fitted to mKvl .2 showed two components of about 120 and 980 ms at +2 mV, and 8 and 78 ms at +52 mV. Thus, the slow x values I obtained for mKvl .2 were similar in size to the single xs for rKvl .2. The fast xs in mKvl .2 resemble those reported by PaulmichI et al. (1991) for rKvl .2 (RAK clone) in oocytes: this (when stepped from -7 0 mV) had 85

activation time constants ranging from 58 ms at -2 0 mV to 6 ms at +60 mV. For all the limitations of fitting exponential growth functions to a partially sigmoidal process, there is some consistency between Kv1.2 properties in oocytes and the MK2 line of CHO cells.

Deactivation

Deactivation kinetics were the point of maximum resemblance between Kvl .2 and the M-like current, so It was necessary to study Kvl .2 deactivation over the range of voltages conventionally used for M- and M-like currents. For rKvl .2, fitted fast xs were 60-300 ms, with a slow component in the range of seconds. For mKvl .2, fast xs were 10-40 ms, and slower xs around 100-300 ms. Disregarding the very slow component seen with rKvl .2, the

10-300 ms spread of Kvl .2 deactivation xs in the mammalian cell environment is similar in magnitude to M-current and M-like current values. Constant! & Brown (1981) reported a maximum deactivation x of -200 ms for the SOG M-current at 22°G. Robbins et al.’s (1992) value on the NG108-15 M-like current was 230 ms at 35°C.

Dendrotoxin block

My ICso of 3.6 nM for dendrotoxin on rKvl .2 is close to literature values. Werkman et al. (1992) reported a dissociation constant of 2.8 nM for dendrotoxin on rKvl .2 in the same

CL1023 cell line. Stühmer et al. (1989) obtained an ICso of 4 nM on rKvl .2 (RCK5) in oocytes. Guillemare et al. (1992) found an ICso of 2 nM in oocytes injected with low cRNA concentrations of a rat cardiac Kvl .2 clone. Apart from Guillemare et al.’s (1992) anomalous finding of reduced dendrotoxin block at high Kvl .2 expression levels, these values are very consistent. If the NG108-15 M-like current were based on Kvl .2, it ought to be sensitive to dendrotoxin.

I tested forskolin and dibutyryl cyclic AMP hoping to replicate the protein kinase A-mediated enhancement of the RAK clone reported by Huang et al. (1994). This PKA effect depends on an amino-terminal threonine residue, which appears to be conserved at position 46 in all mammalian Kvl .2 isoforms for which sequences are available (data from the Swiss Institute of Bioinformatics SWISS-PROT protein sequence database). In my experiments, these membrane-permeant activators of adenylate cyclase and protein kinase A failed to induce an increase in current, and forskolin consistently blocked the current in a dose-dependent fashion with an ICso of 25 pM. 86

I tested lanthanum and zinc to compare Kvl .2 with the NG108-15 M-like current. For lanthanum on mKvl .2, my IC 50 was 25 pM, compared with 2.13 mM on the M-like current

(Robbins et al. 1992). My I C 5 0 for zinc on mKvl .2 was 0.43 mM, compared with 0.011 mM on the M-like current (Robbins et al. 1992). These figures are sufficiently different to argue against Kvl .2 being a major constituent of the M-like current.

Enhancement by repeated activations, temperature and nifiumic acid

The enhancement of Kvl .2 currents by repeated activations was plausibly attributed by Grissmer et al. (1994) to the existence of successive closed states of the channel. Thus, the channels open slowly on the first depolarisation following a resting period. After deactivation, the channels are initially in a closed state capable of rapid opening, but convert over seconds into a “resting” state. I confirmed this activation behaviour, and found additionally that deactivation tails were consistent whether activation during the preceding step was fast or slow. This suggests that the channels arrived at a uniform open state during the preceding step. The phenomenon illustrated in Figure 3.5, where an apparent inactivation is reversed during an ensuing gradual hyperpolarisation, might indicate either closure of some channels or a voltage-dependent substate of the open channels. Thus, these simple protocols provide evidence for at least two closed states of the channel, and probably one major open state (except perhaps at extreme positive potentials). The presence of inactivation complicates any attempt to devise a gating scheme for Kvl .2, and single-channel recordings would be needed to establish any further refinements.

My finding of a leftward shift in the Kvl .2 conductance/voltage curve and increased inactivation at raised temperatures accord with results for canine colonic smooth muscle Kvl .2. These latter currents, recorded at 32-35°C (an unusual condition inXenopus oocytes), show a leftward shift in the conductance/voltage curve and a big increase in inactivation, compared with data obtained at room temperature (Russell et al. 1994).

A minimal explanation for the effects of repeated activation, raised temperature and niflumic acid is that all these conditions cause molecular adjustments pushing the channel towards the fast-opening closed state hypothesised above. Notably, the niflumic acid effect reversed rapidly on washout. As this only usually happened with drugs incapable of crossing the plasma membrane, I suggest that niflumic acid exerts its effect from an external binding site on the channel protein. 87

Niflumic acid is reported to increase some other potassium channel currents. 10 pM niflumic acid increased the xKCNQI/minK current produced by injecting minK RNA into Xenopus oocytes, while 100 pM had a blocking effect (Busch et al. 1994a). The enhancement of KCNQI/minK currents in oocytes by two related chloride channel blockers, mefenamic acid and DIDS, was mediated by the minK subunit rather than KCNQ1 (Busch et al. 1997a). 1-

500 pM niflumic acid increased BK-type calcium-activated potassium currents through cloned

Slo channels (Gribkoff et al. 1996). 500 pM niflumic acid activated KCNQ5 by inducing a -2 0 mV shift in voltage-dependence (Schroeder et al. 2000). Therefore, “activation” by niflumic acid is also found among other potassium channels of the voltage-gated superfamily.

Kvl .2 as an M-like current constituent?

Like Selyanko et al. (1995), but in contrast to Huang et al. (1993), I did not see any appreciable Kvl .2 current in NG108-15 cells. It is possible that NG108-15 cells used by different laboratories may develop different properties in culture. In fact, the NG108-15 traces shown by Huang et al. (1993) resemble the very square currents in Shahidullah et al. (1995a, b), apparently produced by a Kvl .2/Kv3.1 heteromultimer. Zhou et al. (1998) generated a Kvl .2 blocking antibody which inhibited this NG108-15 current, supporting the presence of Kvl .2 in the NG108-15 line studied by Peralta’s group. They did not report whether the equivalent anti-Kv3.1 antibody had any effect on the NG108-15 current.

Kvl .2 deactivations occur on a time-scale similar to those of M- and M-like currents. However, M-currents and M-like currents have been described as non-inactivating over a voltage range from -2 0 mV to -6 0 mV. It is true that Kvl .2 currents (activated from rest) do not show strono inactivation over a similar voltage range of about 40 mV above the activation threshold. It is possible to lower the Kvl .2 current activation threshold in fibroblast and CHO cells to about -5 0 mV, using raised temperatures, niflumic acid treatment or pre-activation. However, this causes extra inactivation to set in below 0 mV. Kvl .2 is an unlikely candidate for an M-like current constituent unless its eccentric acceleration and inactivation behaviours can be shown to be an artefact of the expression systems used.

There is virtually no correspondence in the effects of blockers on Kvl .2 and on the M-like current. The insensitivity to dendrotoxin of the NG108-15 M-like current argues strongly against a contribution from Kvl .2 channels. This has also been commented on by Selyanko et al. (1995). There are two minor cautions regarding the use of dendrotoxin for identifying native Kvl .2 currents. One is the anomalous finding of Guillemare et al. (1992) that Kvl .2, when expressed at a high cRNA level inXenopus oocytes, lost its tendency to inactivate and 88 its sensitivity to dendrotoxin. However, this has not been duplicated in other cell environments, and my mammalian cells expressed substantial Kv1.2 currents with no loss of dendrotoxin sensitivity. The other proviso relates to heteromultimers. For Kv1.1 at least, four channel subunits must contain the appropriate dendrotoxin-binding residues to allow high- affinity binding (Tytgat et al. 1995). However, even if a Kvl .2 heteromultimer with low dendrotoxin sensitivity were implicated in the NG108-15 M-like current, the most likely partner in the heteromultimer would be another Kvl subunit, and one would expect this to have shown up during the original cloning of NGK1.

In summary, there are several lines of evidence that Kvl .2 is unlikely to contribute to the NG108-15 M-like current. The only kinetic data which are comparable between Kvl .2 in non­ neuronal mammalian cells and the M-like current in NG108-15 cells are the rates of deactivation. The oharmacolocv of Kvl .2 and the M-like current is markedly different in respect of the blockers dendrotoxin, zinc, and lanthanum. In the case of linopirdine, there is a spread of sensitivities with the IC50 for the M-like current (24.7 pM: Noda et al. 1998) falling between that for the sympathetic M-current (3.4 pM: Lamas et al. 1997) and that for Kvl .2

(about 50 pM: D.G. Owen, personal communication). Bearing in mind the low selectivity of linopirdine (Lamas et al. 1997), this is insufficient basis for postulating a role of Kvl .2 in the M-like current. My provisional conclusion at the end of this work was that Kvl .2 was unlikely to form even part of the M-like current in the NG108-15 cell line I was using. 89

Chapter 4: mErgla and the slow NG108-15 M-like current

4.1: Introduction: ERG channels

Ether-à-go-go {eag) potassium channels, likeShaker, were first detected from a mutation in the fruit fly Drosophila. Eag mutations cause leg shaking when the fly is anaesthetised with ether, resulting from hyperactivity of motor neurones (Schwartz & Bauer 1999). Warmke & Ganetzky (1994) identified a series of homologous genes, including h-erg (human eag- related gene) from human hippocampus. ERG currents, however, were first identified in the heart.

The human electrocardiogram features a series of waves designated P to I , reflecting the passage of depolarising and repolarising currents across the heart muscle. The landmarks of the cardiac action potential are designated P,Q,R,S and T waves. Particular abnormalities of the EGG are often diagnostic of clinical disorders (Schauf et al. 1990, p.333). The QRS complex, caused by depolarisation spreading rapidly across the ventricles, and the longer- lasting T wave reflecting ventricular repolarisation, give the name of the cardiac disorder long QT syndrome (LOTS). This may be inherited, but can also arise from treatment with certain antiarrhythmic and other drugs. It is identified by a lengthening of the Q-T interval, reflecting abnormal repolarisation of the ventricles. It is frequently accompanied by "torsade de pointes”, a distortion of the QRS axis, which may be a forerunner of ventricular fibrillation, causing sudden death in apparently healthy individuals. Curran et al. (1995) identified one of the genetic loci linked to LOTS as the newiy-discovered potassium channel gene HERG, whose mRNA is strongly expressed in the heart.

Many blockers of HERG and its analogues have been discovered in the course of clinical drug development and monitoring. Class III antiarrhythmic drugs, which act by prolonging cardiac action potential duration and effective refractory period (Wang et al. 1999), include several known to induce LOTS. The benzenesulphonamide compound E-4031 selectively blocks the fast (iKr) portion of the “delayed rectifier” currents which are found in cardiac myocytes and were later attributed to analogues of HERG. iKr has the same bell-shaped current/voltage relationship (Sanguinetti & Jurkiewicz 1990) as reported for HERG.

WAY-123,398 is a 4-[(methylsulphonyl)amino]benzenesulphonamide having a 2 - aminobenzimidazole substituent, synthesised by Wyeth-Ayerst Research as part of a programme to create improved Class III antiarrhythmics (Ellingboe et al. 1992). WAY- 123,398 also acts on the “delayed rectifier” (Spinelli et ai. 1993). Azimilide (NE-10064), 90 initially reported as a blocker of the slow portion of the “delayed rectifier”, Iks (Busch et al.

1994b), also inhibits iKr and cloned HERG channels with an IC 50 in the micromolar range (Busch et al. 1998). The susceptibility of HERG to a wide range of drugs is reviewed in Taglialatela et al. (1998). From the natural world, Gurrola et al. (1999) identified a scorpion toxin seiective for ERG currents (although the unpredictability of human LOTS suggests that cardiac ERG block would not contribute much to the incapacitating effect of scorpion venoms).

HERG (expressed in Xenopus oocytes) has a bell-shaped current/voltage relationship showing a peak at 0 mV, and undergoes block by lanthanum (Sanguinetti et al. 1995), caesium, barium and E-4031 (Trudeau et al. 1995). A short period of hyperpolarisation relieves inactivation of the channels, so any sudden return to depolarised potentials causes a large transient HERG current to appear. This mimics the situation in heart muscle undergoing premature depolarisation, where the HERG current serves to suppress the incipient arrhythmia (Miller 1996). HERG currents expressed in oocytes are similar to currents in cardiac myocytes. Single-channel properties closely resemble the native ones (Zou et al. 1997). HERG complexes in vitro with a new accessory subunit, MiRPI (MinK- related peptide 1), in preference to minK, to give native-type currents (Abbott et al. 1999).

HERG homologues and accessory subunits

In the fruit-fly Drosophiia, the channel homologous to HERG is the product of a gene at a locus previously named for the seizure mutant. The seizure mutant flies show spontaneous flight activity at high temperatures, parallelied by increased spontaneous firing in motor neurones when studied electrophysiologically. The Drosophiia channel is therefore likely to participate in action potential repolarisation, and/or in setting a resting potential (Titus et al. 1997; Wang et al. 1997b).

HERG homologues (ergi) are prominently expressed in the hearts of several mammalian species besides man, in accord with their role as the iKr constituent of ventricular potassium currents (Wymore et al. 1997). The mouse heart contains mRNA coding for two principal isoforms of the protein (Lees-Miller et al. 1997). The protein isoforms, mErgla and mErgIb, coassemble in Xenopus oocytes to duplicate the kinetics of the native iKr (London et al. 1997), though other researchers find no evidence for ERGI b protein (as opposed to mRNA) in aduit rat, mouse or human heart (Pond et al. 2000). The mErgla cDNA cloned by London et al. (1997) is derived from 15 exons of the mErg gene, and has an open reading frame of 1162 amino acids. The amino acid sequence is 96% similar to that of HERG. At an average 119.4 daltons per amino acid (Creighton 1993, p.4), the mErgla channel subunit can be 91

expected to have a molecular weight of some 139 kD, and a tetramerlc channel 555 kD, plus extra mass accounted for by glycosylation.

All three known members of theerg family are expressed in the rat nervous system. Rat ergS, present in the SCG as well as in the coeliac and superior mesenteric ganglia, displays a strongly-inactivating current with a negative threshold when expressed in oocytes (Shi et al. 1997).

ERG channels and the M-like current

The rationale for testing a Class III antiarrhythmic drug on the NG108-15 M-like current in September 1997 requires a little explanation. Similarities had been noted between the current complement in NG108-15 cells and that in certain pituitary tumour-derived cells. Hu & Shi (1997), studying NG108-15 cells, found no M-current, but did detect an inwardly- rectifying current with similar partial block by TEA, cobalt and cadmium. The GH3 line of rat pituitary adenoma cells expresses a potassium current, modulated by thyrotropin-releasing hormone, which is blocked partially by 5 mM caesium and totally by 10 mM TEA. Bauer et al. (1990) described it as inwardly-rectifying; Sankaranarayanan & Simasko (1996) as M-like.

Thus, it appeared possible that a current with particular pharmacological characteristics could manifest as either an M-like current or an inward rectifier, according to recording conditions and in the hands of different researchers. A possible channel identity for one or both of these currents emerged when Weinsberg et al. (1997) reported that the “inward rectifier” in the GH3/B6 subclone, recorded in high external potassium, was inhibited by the HERG-blocking drugs E-4031 and WAY-123,398. The GH 3 “inward rectifier” current has recently been confirmed as rat ERG (r-ergi), which has 96% amino acid identity to HERG (Bauer et al. 1998).

Since it appeared that the M-like current in GH3 and NG108-15 cells might be an alternative form of the “inward rectifier”, it seemed appropriate to test the M-like current found in our laboratory's NG108-15 cells with the “inward rectifier” blocker WAY-123,398. On the rat SCG M-current, WAY-123,398 had no effect. However, I saw a substantial inhibition of the NG108-15 M-like current at -2 0 mV. The inhibition was not total: closer observation revealed that the residual fraction of the M-like current had fast relaxations like the sympathetic M- current, while the inhibited fraction had slow relaxation kinetics. I therefore refer to these pharmacologically and kinetically distinct currents as the NG108-15 fast current and the NG108-15 slow current, respectively, or as the fast and slow portions, fractions, or 92

constituents. I have tried to avoid referring to them as “components” to avoid confusion with the two kinetic components of deactivation measured on the slow current.

Therefore, I and other members of the laboratory sought further evidence for an ERG current in NG108-15 cells. Polymerase chain reaction results obtained by collaborators in the Wellcome Laboratory for Molecular Pharmacology Indicated that a message was present in NG 108-15 cells for the mouse homologue of HERG. Experiments with the two known isoforms of mErgI later narrowed down the likely identity of the NG108-15 slow current to the longer isoform, mErgla (Selyanko et al. 1999). In collaboration with A.A. Selyanko, I compared the mErgla current kinetically and pharmacologically with the slow portion of the NG108-15 M-like current, as described below. In addition to my own data, I cite some joint findings published in Selyanko et al. (1999). 93

4.2: Results

Heterologously-expressed mErgla and the NG108-15 slow M-like current

I expressed the mErgla channel in CHO hm1 cells (see Methods) by transfection with LIpofectAmine Plus. The choice of this cell type was a result of (I) success with the MK2 line (also a CHO cell type) and (II) the availability of cells already stably expressing muscarinic receptors.

The mErgla currents were recorded 1-3 days after transfection, using the perforated-patch recording configuration. Functional channels were sparsely expressed during the whole of this period, with only about one cell In every nine having a current big enough to give reliable data. NG108-15 currents were normally recorded 7-14 days after plating, which regularly gave a standing M-llke current of about 0.5-1.5 nA at the nominal holding potential of -2 0 mV. I have left the junction potential (-7.3 mV) uncorrected for consistency with the published data.

Kinetics

Activating mErgla currents from rest Is not an optimal way to study this current. Depolarisations from below the activation threshold gives current families (see Figure 5 In Selyanko et al. 1999) with untidy crossovers, as Inactivation becomes Increasingly dominant at more positive potentials. A version of the traditional M-current deactivation protocol was more Informative, and was used for both NG108-15 slow currents and transfected mErgla currents. For NG108-15, the holding potential was nominally -2 0 mV to give a maximal M- llke current amplitude without activating other currents. With mErgla, a holding potential of 0 mV was necessary to give a sufficiently large and stable standing current. A series of hyperpolarlsing voltage steps, at -1 0 mV Increments, was given. The steps were6 seconds In duration, to Include most of the slow deactivation, and an Interval of at least 40 seconds was normally allowed between pulses to allow for current recovery.

Inactivation and de-inactivation

With mErgla, this long-step variant of the M-current protocol demonstrates the de- Inactlvatlon phenomenon seen also In HERG. Figure 4.1 shows m Ergla current families from two different cells. It should be remembered that these are responses to hvoerpolarlslna voltage steps, and that mErgla owes Its paradoxical kinetic behaviour to very fast voltage-dependent Inactivation and de-lnactlvatlon. 94

Thus, moderate hyperpolarlsing steps from 0 mV cause a transient increase in the current. This is because a large number of available channels have been inactivated at 0 mV. On stepping to a more negative potential where the voltage-dependent inactivation is lower, many of these channels lose inactivation almost instantaneously and the current rises to a peak. The ensuing decline in current (while resembling the inactivation of an A-current) is caused by voltage-dependent deactivation.

The extent of steady-state inactivation at 0 mV varies in CHO cells, illustrated by the difference between the two cells in Figure 4.1. In the upper record, steady-state inactivation is high, and a hyperpolarlsing step to -5 0 mV initially releases considerably more current than was visible at 0 mV. The slow deactivation means that the current amplitude at the end of a 6 -second hyperpolarisation to -3 0 mV is still greater than the holding current at 0 mV. With larger hyperpolarisations, to -4 0 or -5 0 mV, the transient peak level of the current is still a long way above the holding current, but the deactivation process is becoming rapid enough to reduce the current at the end of the step to below the holding current.

The lower record in Figure 4.1 is an example of a mErgla current with comparatively low steady-state inactivation, similar to that seen on the NG108-15 slow current. In this case, only the -1 0 mV step gives a current exceeding the holding current after6 seconds, and the amplitude of the deactivation tail elicited by the -5 0 mV step never exceeds the holding current. This current family also shows, at the beginning of the step back to 0 mV following the hyperpolarising step, a sudden drop in current that probably represents the inactivation process induced by depolarisation, after which the current slowly increases back towards the holding level.

Thus, this protocol largely separates Inactivation, deactivation and activation. Inactivation can be observed retrospectively from the jump in the current on hyperpolarising, while deactivation accounts for the time-dependent decline during the voltage step. The onset of inactivation on returning to a depolarised potential is so rapid that it barely contaminates the slow activation process. 95

Figure 4.1: Two examples of mErgla in CHO ceils

Current families in two different CHO cells evoked by a protocol (top) of hyperpolarising steps, 6 seconds in duration, from a 0 mV holding potential. On the upper record, currents are increased even at the ends of the first two steps, to -1 0 and -2 0 mV. At -3 0 mV and more negative potentials, the current deactivates from an initial peak which is much higher than the holding current at 0 mV.

Inactivation was variable; the lower record shows a m Ergla current family with comparatively low steady-state inactivation at 0 mV.

Whatever the extent of inactivation, the deactivation at -5 0 mV (picked out in grey) reliably had a large amplitude: this was the potential chosen for drug tests. 96

0 mV

-90 mV

-30 mV

-10 mV

-20 mV

-40 mV

-50 mV

0.2 nA

2.5 s mV

50 mV

Figure 4-1 97

The presence of a mErg1a-like current in NG108-15 cells Is demonstrated In the following Figures: Figure 4.2, 4.3 and 4.4. Figure 4.2 (upper record) shows a representative NG108-15

M-llke current, recorded In response to6 -second steps from -2 0 mV to progressively more negative potentials, by -1 0 mV Increments. I have fitted the current traces (grey) with a three-exponentlal fit (dotted black lines), which are extrapolated beyond the beginning and end of the deactivation steps. Table 1 lists the time constants. These fit lines demonstrate that most but not all of the deactivation occurs within 6 seconds on steps to -3 0 to -6 0 mV; the slowest xs fitted being of the order of seconds. Thus, the typical NG 108-15 M-llke current contains an element which deactivates very slowly Indeed. A medlum-length component has a X of several hundred milliseconds, and the fastest Is around 100 ms for all but the most negative voltages. The centre record of Figure 4.2 shows the currents In the same cell after applying 10 i^M WAY-123,398. The current deactlvatlons here are fitted with a single exponential equation, which describes a current which settles rapidly on hyperpolarlsing, with time constants of around 1 0 0 ms (Table 1 ), as found for the fastest of three components fitted to the whole current. The traces In the bottom panel of Figure 4.2 are obtained by digital subtraction of the current In WAY-123,398 from the whole current. This current constituent has time constants (Table 1) of several hundred milliseconds and several seconds, corresponding to the two slower components present In the whole current.

Figure 4.3 compares the mErgla current from the bottom of Figure 4.1 and the WAY- 123,398-sensltlve subtraction current from Figure 4.2. While the mErgla current deactlvatlons are from a holding potential of 0 mV and the NG108-15 WAY-123,398-sensltlve current Is from -2 0 mV, there Is a close resemblance. Both currents relax on a similar tlme- scale, and the smaller hyperpolarisations cause a transient Increase In current before the deactivation process Is well under way.

Figure 4.4 demonstrates the selectivity of WAY-123,398 for mErgla and the NG108-15 slow current, compared with the other currents I have studied In this work. Thus, 10 ^iM WAY- 123,398 blocks much of the mErgla standing current and only a minimal deactivation tall remains. On the whole NG108-15 current (recorded In this case from a holding potential of 0 mV), the standing current Is appreciably decreased and only a small, fast deactivation Is seen. In both cases an Intermediate concentration of W AY-123,398,1 |liM, gave a partial Inhibition of the slow tall-current. By contrast, the KCNQ24-KCNQ3 current In CHO hml cells and the mKvl .2 current In MK2 CHO cells are not Inhibited at all by 10pM WAY-123,398.

The same result was obtained on a total of three cells for both KCNQ2+KCNQ3 and mKvl2 .. 98

Figure 4.2 therefore indicates one way in which the NG108-15 slow current, otherwise masked both by the fast current and by the capacity transients of these large and non- spherical cells, can be singled out. 10 pM of WAY-123,398 gave about 95% block of both currents (see later Figures 4.10 and 4.11) and 20 pM gave no additional block. It was therefore considered that the current blocked by 10 pM WAY-123,398 constituted the mErgla-like component of the NG108-15 current, while the residual fast current in 10 pM WAY-123,398 had negligible contamination by the slow current. Digital subtraction of the current in 10 pM WAY-123,398 from the pre-drug control makes it possible to visualise the slow current in isolation, freed even from the usual transients. (Another method is by selection of exponential fit components: see later).

Such a direct comparison, of a difference current with a cloned channel recorded in the absence of a blocker, is subject to some limitations. WAY-123,398 block of the mErgI a-like current does not appear to be strongly voltage-, state- or time-dependent. However, with a deactivation protocol as used here, and a mErgla-like current component, several changes might appear in the profile of the difference current according to the behaviour of the blocker. A voltace-deoendent blocker with a stronger effect at more positive voltages could raise the apparent voltage threshold: block might be strong at -2 0 mV, moderate at -3 0 mV, and weak at -5 0 mV, giving a negligible difference current at -5 0 mV. A state-dependent blocker could favour open, closed or inactivated channels. Selective block of open channels would suppress the standing current, while still allowing inactivated channels to be released by de­ inactivation. The difference current after subtraction would show a dominant standing current and less evidence of the large Erg de-inactivation tails, and look more like a delayed rectifier. A preference of the blocker for inactivated channels would cause little or no decline in standing current (depending on the equilibrium between active and inactivated channels at the holding potential), but tails would be smaller in the drug-treated current, therefore larger in the difference current, exaggerating the normal Erg pattern. If the onset or removal of block is time-dependent over the same time-scale as the deactivation process, this could add a further distortion of the current behaviour by accelerating or slowing the drug-treated tail currents. The corresponding difference currents would tend to show, respectively, a rise followed by a downward creep, or a fall followed by an upward creep - quite unlike the normal profile of the current.

In such a case, where block by a channel-specific drug gave rise to a difference current whose kinetics did not match those of the cloned channel, it would be necessary to compare the cloned channel current and the native current component under identical conditions. Recording the cloned channel current with and without the blocker, and using the same 99 subtraction method as for the native current, should duplicate the difference current if the underlying channels are the same.

In the NG108-15 slow current, the initial increase in amplitude on hyperpolarising is smaller than for mErgla in CHO cells. This can probably be explained by the less depolarised holding potential used on NG108-15, which produces lower inactivation (although a small state-dependence of WAY-123,398 block might contribute). Also as in mErg la-transfected CHO cells, I found some variation in the amount of inactivation removed by hyperpolarisation in NG108-15 slow currents: among 19 cells, the transient increase in current on modest hyperpolarisations was usually small but visible. In one case it was non-existent, and one cell had peak currents several times the standing current amplitude. The current I have used as an example in Figure 4.2 and Figure 4.3 has slightly more steady-state inactivation than average for these 19 cells. (Note also the error-bars on the mean normalised peak amplitudes in Figure 4.6, which indicate the variation in the proportion of current released by de-inactivation). 100

Figure 4.2: Separation of NG108-15 M-iike current constituents

(Top) The standard protocol used on NG108-15 cells was 6s long, as for mErgla, but the holding potential was -2 0 mV.

(Upper record) A specimen of a mixed M-like current, with a three-exponential fit superimposed.

(Centre record) The same cell, during perfusion with 10 pM of the HERG blocker WAY- 123,398, has lost the slow components of the deactivations; the superimposed single­ exponential fit is adequate.

(Bottom record) Difference current, obtained by digital subtraction of the centre record from the upper record, with superimposed two-exponential fit. This current shows typical ERG- type de-inactivation during the hyperpolarising steps. -20 mV 101

-100 mV

2 ,0 -1

Whole NG108-15 M-like current, fitted with three exponential components

0.5 -

0.0-

-0.5 0 5 10 15

2.0-1

M-like current in 10 |nM WAY-123,398, fitted with one exponential component

<[10-

0.5 -

0.0 -

-0.5 0 5 10 15

WAY-123,398-inhiblted current by subtraction,

0 .5 0 - fitted with two exponential components

0.25 -

0.00-

0 510 time (s) 15

Fig u re 4 -2 102

Table 1 : Deactivation time constants for the fitted curves shown in Figure 4.2

1. Whole current.

Membrane potential (mV) Fast X (ms) Medium x Slow X (ms) (ms)

-30 143.9 707.9 4817.9 -40 115.1 796.2 3527.0 -50 111.1 673.6 3061.0 -60 121.8 610.7 2810.2 -70 98.5 451.4 1502.2 -80 83.5 211.2 845.5 -90 18.4 55.4 697.1 -100 30.8 64.8 129.4

2. Fast current (in the presence of 10 pM WAY-123,398).

Membrane potential (mV) X (ms)

-30 172.0 -40 152.5 -50 117.7 -60 106.5 -70 105.4 -80 39.8 -90 46.4 -100 41.0

3. Slow current (by subtraction of WAY-123,398-inhibited current from whole cur

Membrane potential (mV) faster x (ms) slower X (ms)

-30 869.5 4827.5 -40 974.0 4005.7 -50 849.9 4204.7 -60 359.0 1617.7 -70 191.1 762.1 -80 161.2 728.0 -90 110.3 791.8 -100 59.0 147.9 103

Figure 4.3: a direct comparison of mErgla and the NG108-15 slow current (Above) Voltage protocol (holding at G mV) and mErgla current deactivating from G mV.

(Below) Voltage protocol (holding at -2 0 mV) and WAY-123,398-sensitive current in NG1G8- 15. 0 m V 104

"90 mV

300 pA

mErgla in CHO hm1

-20 mV

-100 mV

100 pA

2s

WAY-123,398-sensitive current in NG108-15 Figure 4.3 105

Figure 4.4: Selectivity and concentration-dependence of WAY-123,398 block The effect of WAY-123,398 on four currents, with the relevant voltage protocol shown above each one.

(Above left) mErgla. Reduction of relaxation a t-5 0 mV by 1 pM WAY-123,398. Substantial decrease in the holding current at 0 mV, and virtual elimination of the relaxation at -5 0 mV, by 10|iM WAY-123,398.

(Above right) M-like current in NG108-15. Reduction in holding current at 0 mV, and reduction in the -5 0 mV relaxation, by 1 ^iM WAY-123,398. Further reductions in both by 10

^iM WAY-123,398, with the residual relaxation being faster.

(Below left) KCNQ2+KCNQ3 current expressed in a CHO hml cell. No effect of 10 pM WAY- 123,398. Same result in a further 2 cells.

(Below right) mKvl .2 in an MK2 cell. No effect of 10 pM WAY-123,398. Same result in a further 2 cells. 0 mV 0 m V 106 1

-50 mV -50 mV

m E rg la NG108-15 M-like current

200 pA

2.5 s

200 pA

2.5 s

------control ------control — ....1 |iM WAY-123,398 1 nM WAY-123,398 10 nM WAY-123,398 10 nM WAY-123,398

-20 mV -t-20 mV -20 mV

-50 mV -60 mV

KCNQ2+KCNQ3 m K vl .2

20 pA

0.5 s 200 pA

0.5 s

• control • control 10 nM WAY-123,398 10 hM w a y -123,398

Figure 4-4 107

Current/voltage relationships: deactivation and its measurement

Figure 4.5 clarifies the fitting procedure used for characterising mErgla and NG108-15 slow currents, using a mErgla current as an illustration. “End-of-step” values represent the mean current amplitude over a short time interval near the end of the voltage step, leak-subtracted either before analysis in Clampfit or retrospectively in Excel (see Chapter 2, Methods: Leak- subtraction of data records, p.34). “Instantaneous” or “peak” current values for both mErgla and the NG108-15 slow current, at the start of hyperpolarising voltage steps, are standardised by back-fitting or calculation (see Chapter 2, Methods: Fitting deactivation tails, p.34). For mErgla expressed in CHO cells, and WAY-123,398-sensitive subtraction currents in NG108-15, deactivation manifests as a bi-exponential decline of the current during the 6- second hyperpolarising step. Thus, the peak amplitude of the deactivation tail is obtained by back-extrapolating a two-exponential fit to the first sample of the hyperpolarising voltage step, and leak-subtracting as appropriate.

Figure 4.6 plots the combined “instantaneous” and end-of-step\N relationships for mErgla in CHO hml cells, and for the WAY-123,398-sensitive NG108-15 slow current as obtained by subtraction. When leak-subtracted mErgla current amplitudes are normalised, with the standing current at 0 mV defined as 1, the maximum averaged “instantaneous” current (at - 40 mV) is over 3. Thus, considering that there is a drop in driving force on hyperpolarising, more than 70% of the available channels are inactivated when holding at 0 mV. In the case of the NG108-15 slow current, when the standing current at -2 0 mV is set at 1, the maximum normalised “instantaneous” current (at -5 0 mV) is 1.7. Close inspection of Figure 4.6 also reveals that there is a small residual NG108-15 current even at -6 0 mV, whereas mErgla is barely active at -5 0 mV.

Reversal potential

The solutions used for NG108-15 and mErgla (and for KCNQ - see Chapter 5) channel recordings included 110 mM potassium internally and 2.5 mM potassium externally. This gives a theoretical reversal potential of -95.54 mV, according to the equation Vrev = (ln([K'"]o/[K'"]i)).RT/ZF. My actual values, obtained from the point where “instantaneous” and end-of-step l/V curves cross, were nearer -8 0 mV: -7 8 mV for m Erg la-transfected CHO hml cells and -8 4 mV for the WAY-123,398 subtraction current in NG108-15 cells. The discrepancy can be partially accounted for by the uncorrected liquid junction potential of -7.3 mV (see Chapter 2, Methods). It is possible that slow clearance of extracellular potassium build-up around the cell membrane may also pull the measured reversal potential in a positive direction. However, the holding potentials of 0 and -2 0 mV used for these 108 experiments produced stable standing currents; if the holding current drifted, I perfused with fresh bath solution until it was stable before continuing recording. Any effect of raised extracellular potassium was therefore rapid, and applied consistently to all current measurements.

I have used unmanipulated current/voltage data to characterise mErgla; conductance calculation would require a specialised protocol to avoid inactivation effects. Therefore, these reversal potentials serve mainly as evidence for the potassium selectivity of the channels.

Deactivation time constants

In addition to the current amplitude measurements used to construct the l/V curves, the above fitting procedure also provides another basis for comparing the two currents: deactivation time constants. These are shown in Figure 4.7. The upper graphs demonstrate that the two fitted xs are similar in magnitude for mErgla (deactivating from 0 mV) and the

NG108-15 slow current (deactivating from -2 0 mV). Both fast and slow x values decrease with increasing hyperpolarisation, as relaxations become faster. Further, the amplitudes corresponding to the two xs show a similar relationship to voltage in both currents (centre graphs). The slower component contributes more of the relaxation on the smaller hyperpolarisations, and the faster component becomes predominant negative to -4 0 or -5 0 mV. When each amplitude is plotted as a proportion of the total relaxation amplitude (lower graphs), the contribution of the two components is seen to be strongly voltage-dependent.

Most of the deactivation xs obtained for mErgla and NG108-15 currents occupy less than half of the 6-second step duration, allowing some confidence in the values obtained (see Methods). A two-exponential fit of WAY-123,398-sensitive slow M-like currents (19 cells) gave time constants with ranges 50-900 ms and 1800-4000 ms. Equivalent values for CHO hml cells expressing mErgla (23 cells) were 50-400 ms and about 500-4000 ms. 109

Figure 4.5: Fitting methods for mErgla and the NG108-15 slow current How I obtained end-of-step and “instantaneous” current/voltage relationships, using a mErgla current as an example. 350- 110 End-of-step current: 3 0 0 - measure average value between two cursors 2 5 0 - placed around here... 200-

1 5 0 - ...choose a linear sub- < threshold region of the l/V CL 100- A \ and extrapolate a leak line...

3t 5 0 - O 0- ...either correct for the leak in -5 0 - Clampfit and re-measure the l/V, or construct a leak line in the -100- analysis spreadsheet and subtract leak values from the current amplitudes. -1 5 0 -

Normalise end-of-step amplitudes against the holding current.

END-OF-STEP l/V Take the holding current amplitude and the leak line from the end-of-step l/V to use in the peak l/V.

place fitting cursors about here, and fit bi-exponential curves to tail-currents... —

...extrapolate back to an extra cursor here, at the start of deactivation, or calculate amplitudes in spreadsheet program.

The back-fitted current amplitude corresponds to the theoretical current level at the first sample of the deactivation step. Leak-subtract if necessary, and normalise to the holding current (from end-of-step l/V).

T — '— r -|-----1— I— 1— I---- 1— I— 1— I— I

tim e (s) INSTANTANEOUS l/V

F ig u re 4-5 111

Figure 4.6: Voltage-sensitive de-inactivation of mErgla and the NG108-15 slow current

(Top) Back-fitted peak (grey) and end-of-step (black) mErola amplitudes, + s.e.m., for currents evoked by the hyperpolarising step protocol (Figure 4.1), normalised to the amplitude at 0 mV, averaged, and plotted against voltage. Peak currents are obtained by back-fitting; also note that “end-of-step” currents are not steady-state currents, since the slow relaxations are not all completed within the 6-second step. A true steady-state current/voltage curve would probably show a less pronounced bell-shape.

(Bottom) Back-fitted peak (grey) and end-of-step (black) WAY-123.398-sensitive NG108-15 current amplitudes ± s.e.m., for deactivations in response to hyperpolarisations from -2 0 mV. Currents were obtained by digital subtraction of records in the absence and presence of the drug (as in Figure 4.2), normalised to the current amplitude at -2 0 mV, averaged, and plotted against voltage. As for mErgI a in CHO cells, a true steady-state current would show less of a bell-shape (the extrapolated fit lines in Figure 4.2 give an idea of the current amplitudes at

>15 seconds). 112

— end-of-step - ■ — peak

0 -

■D mErgla in 22 CHO cells

-4 -

03 -120 -100 -80 -60 -40 -2 0 0 E o c

A — end-of-step Q. A — peak E 03 4—» c CD Ï—

=3 O

WAY-123,398-sensitive com­ ponent of the M-like current in 19 NG108-15 cells

-2 -

-1 2 0 -100 -80 -60 -40 -2 0 0 membrane potential during step (mV)

Figure 4-6 113

Figure 4.7: mErgla and NG108-15 slow current kinetics compared

(Top) Average time constants, + s.e.m., from two-exponential fits to 23 mErgla currents in CHO cells (left) and 19 WAY-123,398-sensitive currents in NG108-15 cells (right).

(Centre) Average amplitudes, mathematically corrected, + s.e.m., corresponding to the two xs plotted above, for mErgla (left) and WAY-123,398-sensitive NG108-15 currents (right).

(Bottom) Amplitudes for the two deactivation components in each cell were summed, and each component was expressed as a proportion of the total amplitude. Averaged values + s.e.m., are shown for mErgla (left) and the WAY-123,398-sensitive NG108-15 current (right). 114

4000- 4000-

mErgla NG108-15 slow —■— mean fast x —A— mean fast x 3000- —■— mean slow x 3000- —A— mean slow x

2 2000- 2000 -

1000 - 1000-

0 - 0 -

-120 -100 -80 -60 -40 -20 0 -120 -100 -80 -60 -40 -20 0

6 0 0 - 6 00 -

4 0 0 - 4 00 -

2 0 0 - 2 0 0 - Q.

0 -

Q . -2 0 0- mErgla - 2 0 0 - NG108-15 slow - mean fast amplitude * ■ - - mean fast amplitude —■— mean slow amplitude —A— mean slow amplitude -4 0 0- -4 0 0-

-6 0 0- -6 0 0-

-120 -100 -80 -60 -20 0 -120-40 -100 -80 -60 -40 -20 0

mErgla amplitudes NG108-15 slow amplitudes —■— mean fast fraction —* — mean fast fraction —■— mean slow fraction —A— mean slow fraction

1 .0 -

0 .8 - 0 .8 -

C o 0.6 - ■c o Q. 0 .4 - 2 D.

0 .2 - 0 .2 -

0.0 - 0 .0 -

-120 -100 -80 -60 -40 -20 0 -120 -100 -80 -60 -40 -20 0

Figure 4-' membrane potential during step (mV) 115

Activation

I obtained a limited amount of activation data for mErgla as a by-product of the deactivation protocol. The reactivation traces from the various potentials to 0 mV were fitted by two exponential components. Although the smaller hyperpolarising steps commonly cause an overall increase in current (since inactivation has been removed), returning to 0 mV causes the current to drop below the original 0 mV current. This is followed by a slow rise back to the holding current amplitude. A very rapid drop in current at the beginning of the step back to 0 mV is visible on the lower record in Figure 4.1, and apparently represents the inactivation process induced by depolarisation. This initial region of the step was excluded from the fit, and the usual back-fitting methodology was used. The resulting xs for activation from voltages between -1 1 0 and -1 0 mV to 0 mV are about 700-1200 ms and 4400-7000 ms, measured over a duration of 15s. I have plotted these data on a logarithmic scale in Figure 4.8, along with a log-scale version of the mErgla deactivation xs. While the deactivation xs (lower graph) are voltage-dependent, this sample of activation data shows little effect of the initial voltage. This contrasts with the notable behaviour of the related channel r-eag, which activates more slowly from strongly negative potentials (Ludwig et al. 1994). Like deactivation, activation has two components. Unfortunately, the equivalent data for reactivations to -2 0 mV of the NG108-15 slow current, even when isolated on subtracted records, are too slow and noisy to fit with any confidence. 116

Figure 4.8: mErgla activation and deactivation compared

(Top) T ± s.e.m. from a 2-exponential fit to activations from hyperpoiarised potentiais back to the hoiding potential of 0 mV, plotted on a logarithmic y-axis against the potential for the previous step.

(Bottom) T ± s.e.m. for mErgla deactivation, as shown in Figure 4.7 (top left) but plotted on a logarithmic y-axis for comparison. 10000-] 117

$ : m M m m M m "

a - i 1000 - E H C O 03 > ü 03

100 -120 -100 -80 -60 -40 -20 0 previous membrane potential (mV)

10000-1

1 0 0 0 - i - î

100 -

"O -120 -100 -80 -60 -40 -20 0 Step potential (mV) Figure 4-8 118

NG108-15 current dissection

While a whole current constituent can be Isolated by subtracting currents recorded In the absence and presence of a drug such as WAY-123,398, one cannot apply additional drugs to the resulting mathematical abstraction. In order to assess the effects of other drugs on the slow fraction of the NG108-15 M-llke current, alongside heterologously-expressed mErgla, another method was necessary. This, more universally applicable, method Involved fitting three exponential components to the whole NG 108-15 M-llke current.

The top trace of Figure 4.2 shows an example of an entire M-llke current fitted with three exponential components; the fit lines are extrapolated beyond the ends of the deactivating voltage step. The lower traces show the current remaining in WAY-123,398, fitted with one exponential, and the WAY-123,398-sensltlve (presumed mErgla) difference current, fitted with two exponential components.

The WAY-123,398-reslstant current relaxed rapidly with time constants -50-100 ms. These are similar to the values found for the fastest component of the whole M-llke current, and for the classic M-current (see Chapter 5). The WAY-123.398-sensltlve current and the two slower components of the three-exponentlal fit had time constants -100-800 ms and -1000- 2000 ms. Summing the corrected amplitudes for these two slower components gave the amplitude of the proposed mErgla current. Thus, It was possible to assess the proportional contribution of the two channel types to the M-llke current.

Figure 4.9 compares the properties of the -2 0 to -5 0 mV relaxation for sixteen NG108-15 cells for which good separated fast and slow current records were available. These relaxations were fitted as described above: the original current with three exponential components, the WAY-123,398-reslstant current with one exponential, and the WAY- 123.398-sensltlve current with two. The results for the separated currents were reassembled and compared with those for the original current. The time constants (upper graph) and amplitude contributions (lower graph) show an Impressive resemblance, testifying that the fitting procedure was well-founded - and hence that WAY-123,398 Is a selective and complete blocker of the slow current.

This sample of 16 cells was chosen for WAY-123,398 application because of visibly slow relaxations; the contribution of the slow current was about 72%. Similar fits to my data from the (unselected) total of 65 cells established a slow current contribution of 70%. 119

The channels whose closure makes up the relaxation are those which were open at -2 0 mV, plus any released from inactivation by the hyperpolarising step. Figure 4.6 shows that this inactivation is less pronounced in NG108-15 cells than for mErgla in CHO cells under the conditions used. Thus, the slow current might be expected to comprise something under 70% of the standing current. I used current/voltage data to obtain a leak-corrected amplitude for the current at -2 0 mV, in the presence and absence of WAY-123,398. On 19 cells, the

-2 0 mV current amplitude was reduced 66 + 2% by 10 pM WAY-123,398. 120

Figure 4.9: Dissection and reconstitution of the M-iike current

(Top) Average x + s.e.m. from separate fits to WAY-123,398-reslstant and WAY-123,398- sensitive currents deactivating at -5 0 mV in sixteen NG108-15 cells (above). Compare xs from a three-exponential fit to whole M-like current deactivations (below).

(Bottom) The contribution of the amplitude for each deactivation component was calculated (± s.e.m.) as a proportion of the total, for M-like currents separated by WAY-123,398 (upper bar) and for whole M-like currents (lower bar). WAY-123,398-resistant WAY-123,398-resistant (1-exponential fit) + (1-exponential fit) 4- Whole M-like current WAY-123,398-sensitive Whole M-like current WAY-123,398-sensitive (3-exponential fit) (2-exponential fit) (3-exponential fit) (2-exponential fit)

o o - 1 b

O Ü1 o - O o 0 "D 00 O 1 3 . o o' o cn 1 3 Q. o o o c cn o o o CD r— 3 I— H ro (/> O o o o c b o

CD ro CD =3 Ol CL o o CD o H 00 C/) 3 H 00 "O O 3 8 c (/) Q_ CO CD o cn o o ro 122

Pharmacology

The multi-exponential fit method was used to study the effect of several potassium channel blockers on the slow portion of the M-like current. Drugs were applied cumulativeiy in doses following a 1,3,10,30... pattern, giving intervals of about 0.5 log unit. The reiaxation at -5 0 mV was assessed for a control trace, and after reaching a stable current amplitude at each dose of the blocker. For the NG108-15 current, three exponential components were back- fitted, and the mathematically-corrected amplitudes (see Methods) were summed for the two slower ones. For mErgla, the total amplitude of the relaxation from 0 to -5 0 mV was measured. The inhibition of the current was then calculated, plotted against the concentration of drug, and fitted with the Hill equation (using Microcal Origin’s “logistic” equation; see Methods) to obtain a Hill slope and a half-maximal inhibiting concentration.

Figure 4.10 shows concentration/response curves for WAY-123,398, the antiarrhythmic drug azimilide, THA and tetraethylammonium on mErgla currents expressed in CHO hml cells.

Figure 4.11 shows the corresponding data for the NG108-15 siow current. Table 2 summarises my results. Agreement between mErgla and the NG108-15 slow current is generally good. [There are some small discrepancies between the ICsoS I give here and the figures published in Selyanko et al. (1999) for the same sets of data, probably attributable to choice of fitting criteria. My fits, as given here, are all constrained to a Ymax of 100%].

Fitting curves to the drug inhibition data for each individual cell, and using a two-tailed Student’s t-test (in Excel) for comparison between the m Ergla and NG108-15 currents, gave the P values shown in the lower part of Table 2. In no case was a P value of less than 0.05 (which would suggest a significant difference between data for m Ergla and the NG108-15 slow current) obtained. The outcome was similar when a cross-comparison was made between data for azimilide and TEA inhibitions of mErgla and the NG108-15 fast current.

However, see also the point in Chapter6 (p.178), that the actions of several drugs considered as a group give a clearer indicator of channel identity than the individual potencies. Thus, the low linopirdine block of mErgla and the slow NG108-15 current, and the lack of WAY-123,398 block of KCNQ channels and the fast NG108-15 current, where

IC50S could not be obtained, form a persuasive part of the pattern. 123

Figure 4.10: Concentration/response curves for four blockers on the mErgla current All data are plotted as mean inhibition + s.e.m.

(Top left) WAY-123,398 (n=3) Hill equation fit parameters:

Y m a x = 1 0 0 %

ICso = 0.29 ± 0.01 pM nH = 0.88 ± 0.02

(Top right) Azimilide (n=4) Hill equation fit parameters:

Y m a x = 1 0 0 %

ICso = 5.84 ± 0.28 pM nH = 0.85 + 0.03

(Bottom left) THA (n=3) Hill equation fit parameters:

Y m a x = 1 0 0 %

ICso = 58.12 + 8.48 pM nn = 0.70 + 0.07

(Bottom right) TEA (n=3) Hill equation fit parameters:

Y m a x = 1 0 0 % ICso = 22.62 ± 2.97 mM nn = 0.55 + 0.04 % inhibition of control m Ergla current % inhibition of control m Ergla current

T1 è g § g ê g g § cq‘ c

CD 4:^

% ro > co CO 0 5 î i CD 0 5 00 co co 00 §

è g g g g 8 g è g g g

H > S- o

co

Q.

8 N) 125

Figure 4.11 : Concentration/response curves for four blockers on the NG108-15 slow current All data are plotted as mean inhibition + s.e.m.

(Top ieft) WAY-123,398 (n=3) Hill equation fit parameters:

Ymax = 100%

IC50 = 0.31 ± 0.03 pM nH = 1.05 + 0.12

(Top right) Azimilide (n=4) Hili equation fit parameters:

Y m a x = 1 0 0 %

iCso = 8.39 ± 0.47 pM nH = 0.96 + 0.05

(Bottom left) THA (n=3) Hill equation fit parameters:

Y m a x = 1 0 0 %

IC50 = 23.54 ± 2.72 pM nH = 0.91 + 0.09

(Bottom right) TEA (n=3) Hiil equation fit parameters:

Ymax = 100%

IC5o = 3 2 . 2 4 + 7 . 1 8 mM nH = 0.89±0.18 % inhibition of control NG108-15 slow current % inhibition of control NG108-15 slow current TI ê g § § § 8 o è 8 § 8 cq‘ § c o

CO CO H w % CO > 00

00

S

8 è 8 8 8 8 8

H Q. m > Q.

ro

Table 2 : IC50 values for Inhibition of mErgla and NG108-15 slow current by various drugs

All IC50 values are quoted + s.e.m., firstly, as calculated on averaged inhibition data: DRUG mErgla in NG108-15 CHO hml slow WAY-123,398 0.29+0.01 0.31+0.03 (pM) n=3 n=3

azimilide 5.84+ 0.28 8.39+0.47 (pM) n=4 n=4

THA 58.1 ±8.5 23.5±2.7 (pM) n=3 n=3

TEA 22.6+3.0 32.2+7.2 (mM) n=3 n=3

linopirdine >30 » 3 0 (pM) n=3 n=3 and secondly from fitting data individually for each cell, averaging and applying a t-test: DRUG mErgla in NG108-15 P value, P value, CHO hml slow mErgla vs. mErgla vs. NG108-15 NG108-15 fast slow WAY-123,398 0.40 + 0.14 0.71 ± 0.50 0.58 (pM) n=3 n=3

azimilide 6.88 ± 1.23 8.75 ± 3.10 0.59 0.13 (pM) n=4 n=4

THA 63.33 ± 22.05 11.83 ±1.88 0.08 (pM) n=3 n=3

TEA 41.67 ±29.3 14.25 ±4.64 0.33 0.24 (mM) n=3 n=4

Note: I obtained the results for WAY-123,398 early on in the programme of work, before a holding potential of -2 0 mV had been standardised for NG108-15 cells. Thus, for this blocker only, the NG108-15 currents were measured on relaxations at -5 0 mV from a holding potential of 0 mV rather than -2 0 mV. There is no reason to suppose that WAY-123,398 block is appreciably voltage-dependent, and the results are in this case recorded for identical voltage steps for CHO hml and NG108-15 cells. 128

Accommodation

I also studied the effect of blockers on NG108-15 firing. Figure 4.12 shows data obtained in current-clamp mode on two NG108-15 cells. I applied a series of incremental current steps, and selected cells showing consistent phasic firing for drug treatment. Several cells were discarded because of instability in the resting potential, or loss of firing over time. It Is unfortunate that the recording instrumentation could not be easily switched to voltage-clamp to check the properties of the M-like current. Nevertheless, I consider that the two cells illustrated in Figure 4.12 display the normal “braking” action of the total M-like current and its susceptibility to blocking drugs at room temperature. What emerges is that eliminating the mErgla constituent with 10 |iM WAY-123,398 only slightly increases the spiking rate. In a cell treated with WAY-123,398 alone, firing stops at the same time-point as in the control record, indicating that the duration of firing is controlled by some independent factor. This factor is revealed by applying the KCNQ channel blocker linopirdine at an M-current-blocking dose of 30|iM. Continuous spiking ensues, demonstrating that the major influence on NG108-15 spike accommodation resembles the sympathetic M-current in pharmacology. 129

Figure 4.12: Effects of KCNQ and ERG block on NG108-15 firing

(Left) Current-clamp protocol and voltage recordings from an NG108-15 cell to which the KCNQ channel-/M-current blocker linopirdine was applied, followed by WAY-123,398. Linopirdine induces continuous spiking; WAY-123,398 marginally increases the firing rate.

(Right) Protocol and voltage recordings made in current-clamp mode, on an NG108-15 cell in which the ERG blocker WAY-123,398 was applied, followed by linopirdine. ERG block has a minor effect on spiking rate, but not duration. 130

+0.23 nA # +0.22 nA - -0.07 nA I = -0.03 nA

Control Control

0 mV

5 spikes 5 spikes

-90 mV

+ 30 juilVI linopirdine + 10 |liM w a y 123,398

16 spikes 6 spikes

+ lOjuM WAY 123,398 + 30 |iM linopirdine

20 mV G

1 s 19 spikes 23 spikes

Figure 4-12 131

4.3: Discussion

The foregoing data make it ciear that the NG108-15 M-like current can be separated kineticaliy and pharmacologically into two parts, of which one corresponds closeiy to the current generated by mErgla channels expressed heterologously in CHO cells.

Four sets of criteria for identifying molecular counterparts of native currents are listed in Selyanko et ai. (1999). These characteristics are: (1) the presence of mRNA transcripts and protein; (2) biophysical properties, including various kinetic measurements; (3) sensitivity to blockers; and (4) modulation by agonist action. For m Ergla as a basis for the NG108-15 slow current, the expression of the channels (point (1)) is covered in Selyanko et al. (1999). Collaborators in the Wellcome Laboratory for Molecular Pharmacology confirmed by RT-PCR the presence of mErgI mRNA in both NG 108-15 ceils and mouse SCG neurones. Immunostaining for the mErgI protein was strong in morphologicaliy differentiated NG108-15 cells (which have large M-like currents), weak in undifferentiated cells (which have only very small M-like currents), and absent in mouse SCG neurones (which have oniy a standard fast M-current). Likewise, modulation by an agonist of Mi receptors (point (4)) is shown both for the NG108-15 slow current and for mErgI a in CHO ceiis (Seiyanko et ai. 1999). My own data, discussed here, address points (2) and (3): the kinetics and blocker sensitivities of mErgla and the NG 108-15 slow current.

Meves et al. (1999) have also separated the NG 108-15 M-like current into a conventional M- current and an ERG current. Their work duplicates the pharmacological dissection approach described here and in Seiyanko et ai. (1999), but at physioiogical temperature. Meves et al. (1999) show that the ERG blocker E-4031 also partially blocks the fast portion of the current, demonstrating that WAY-123,398 is so far the oniy truiy seiective blocker for use on mixed M- iike currents.

A novel aspect of this work (Meves et al. 1999; Selyanko et al. 1999) is that a HERG-reiated current, mErgla, appears to provide a kinetic and pharmacological counterpart to a standing current seen at somewhat depolarised potentiais in NG108-15 celis. It is also evident as a slow component of current relaxations. Channels in the ERG family have commonly been studied under conditions eliciting a transient current, and it is oniy recentiy that the effects of channel-blocking drugs have led researchers to look for other manifestations of ERG currents. The NG108-15 mErgla-like current shows the steady-state inactivation typical of the ERG family, but still has a substantial number of channels open at voltages negative to the action potential threshold. 132

Depending on how the measurement was made, some 70% of the current in NG108-15 cells over the voltage range -2 0 to -5 0 mV was produced by channels whose macroscopic properties closely resembled those of mErgla expressed In CHO cells. I found a 72% contribution of the slow current In a sample of 16 cells which had visibly slow relaxations of the whole current at -5 0 mV. In a total of 65 cells, I established an average slow current contribution of 70%. Pooled results for 86 cells. In Selyanko et al. (1999), yielded a 67% contribution of the slow current. A.A. Selyanko also found 15 differentiated NG108-15 cells, normal in other respects. In which onlv a slow current was present. For the standing current, a sample of 19 cells (Including the 16 referred to above) showed a 66% contribution of the slow component, as assessed by the decrease In standing current at -2 0 mV In the presence of lOpM WAY-123,398.

Because of gaps In the literature, much of the comparison data for mErgI relates to the better-studied HERG channel. mErgla Is 96% Identical to HERG In protein sequence, and >99% Identical over the region Including the transmembrane segments (London et al. 1997). Similarly, I have supplemented published data on NG 108-15 cells by Information on neuroblastoma lines analogous to the neuronal parent of NG108-15, and other neuroblastoma hybrids.

Kinetics

I studied deactivation of mErgla In CHO cells at -110 to -1 0 mV, from a holding potential of 0 mV. I found two time constants of 50-460 and 450-4000 ms (plotted In Figure 4.7). London et al. (1997) obtained closely similar results In Xenopus oocytes, for mErgla relaxations at -

100 to -3 0 mV from a holding potential of +20 mV. Their graph shows deactivation xs of about 100-600 and 300-4000 ms. London et al.’s (1997) deactivation protocol evokes tall currents with peak amplitudes several times larger than the current at +20 mV, just as my tall currents peak considerably higher than the holding current at 0 mV. It looks as If oocytes and CHO cells present a very similar mErgla channel complex In their membranes. Deactivation properties of mErgla In oocytes also closely resemble those of HERG (London et al. 1997).

Kineticaliy, a striking resemblance emerges between mErgla, expressed In CHO cells, and the slow fraction of the NG108-15 M-llke current. The match holds good for the pattern of Inactivation shown by the current/voltage curves, the magnitudes of the two xs fitted to the deactlvatlons, and the relative contributions of the deactivation processes corresponding to 133

these xs at different voltages (Figures 4.6 and 4.7). No channel outside the ERG family has displayed such pronounced inactivation/de-inactivation behaviour.

My recordings of mErgla were made using internal and external solutions with roughly physiological potassium concentrations. Numerous kinetic data have been reported for the human homologue HERG, but I shall not make comparisons with these because they were obtained at high external potassium concentrations (e.g. Trudeau et al. 1995). High external potassium levels shift the potassium reversal potential in a positive direction, enlarging and separating tail currents at strongly negative voltages, where properties can be studied with less hindrance from inactivation. High external potassium has been noted to affect inactivation (Wang et al. 1996b) and perhaps activation of HERG, leading to increased overall current (Wang et al. 1997a). Numaguchi et al. (2000) suggest that raised external potassium opposes a tonic channel block by external sodium, leading to increased HERG current amplitudes in high potassium. Physiological potassium concentrations highlight differences in current in the more positive voltage range relevant to action potential generation, and are found in papers discussing clinical (Sanguinetti et al. 1995) and pharmacological aspects of HERG (e.g. Busch et al. (1998) on azimilide).

Blockers

WAY-123,398 blocked the mErgla and the NG108-15 slow M-like current with an IC50 of 0.3 pM. There is no reported IC50 for cloned ERG channels from any species to compare with this. On native currents, researchers often cite effects of single doses. The only published concentration-response data relate to inhibition of the “delayed rectifier” in cat ventricular myocytes (presumably feline ERG) by WAY-123,398 with an IC 50 around 0.1 pM (Spinelli et al. 1993).

With azimilide, I obtained low-micromolar IC5 0 values, 6 pM on mErgla and 8 pM on the slow portion of the M-like current. For comparison, Busch et al. (1998) found that azimilide (NE-

10064) inhibited cloned HERG channels with an IC50 of 5.2 pM at a stimulation frequency of 1

Hz, and 1.4 pM at 0.1 Hz (Busch et al. 1998). This decrease of block with increasing channel activation amounts to reverse use-dependence; I did not test for such phenomena on mErgla or the NG 108-15 current.

Overall, the similarity between blocker sensitivities of mErgla and the NG108-15 slow current strongly supports the identity of the two currents. Four drugs, WAY-123,398, azimilide, THA and TEA, block the two currents at similar concentrations. Linopirdine at 30 pM produces a 134

small inhibition in both cases, but not enough to justify estimating an IC50. Additionally, I tested 10 |liM WAY-123,398 on two types of cloned channels in CHO cells: mKvl .2 and KCNQ2+KGNQ3 currents (n=3 for both) showed no inhibition, supporting the selectivity of this compound.

Firing

My results on NG108-15 cells suggest that only a small increase in spiking rate, with no increase in the duration of spiking activity, results from selective loss of the mErgla-like current. In these cells, a current apparently identical to the classic sympathetic M-current acts within a short time as the major influence on accommodation.

A potassium current comparable to the slow NG108-15 current was also reported from human and mouse neuroblastoma cells (Arcangeli et al. 1995). In a hybrid cell line derived from mouse neuroblastoma and rat dorsal root ganglion cells, this “inward rectifier” current shared the pharmacology of HERG, but had an even smaller outward current component (Faravelli et al. 1996). Chiesa et al. (1997) proposed that ERG channels play a role in spike frequency accommodation, since in this line, at room temperature, suppressing the ERG current with WAY-123,398 led to increased firing frequencies during a sustained depolarisation. Schonherr et al. (1999) extended this scenario with a human leukaemic cell line and HERG mutants expressed in oocytes. The time taken to achieve a steady-state ERG current during repetitive depolarisations depends on the voltage, and can be quite long. Conceivably, tens of spikes might occur before the sustained ERG current becomes large enough to inhibit further spiking. 135

Chapter 5: KCNQ2, KCNQ3 and the fast NG108-15 M-like current

5.1: Introduction - KCNQ channels

The KCNQ family of channels is closely related to the Shaker-V^pe Kv channels, with other potassium channels such as theeag group being more distant on the phylogenetic tree (Derst & Karschin 1998). I shall give a brief history of the discovery of this group of channels.

An untypical potassium channel subunit with one transmembrane region known as Uk or minK (minimal K channel protein), and later as KCNE1, was found to mediate the cardiac slow delayed rectifier (Folander et al. 1990). It emerged subsequently that minK does not form channels on its own. Genetic research determined that the commonest cause of long 0 7 syndrome (see Chapter 4) is mutation of a gene initially classified as a “Kv” channel because of a substantial amino acid similarity to Shaker (Wang et al. 1996a). This channel, KvLQT 1, interacts with minK in Xenopus and mammalian expression systems (Sanguinetti et al. 1996; Barhanin et al. 1996). The ability of Xenopus oocytes to give currents when the oocytes are injected with minK message is caused by the presence of aXenopus homologue of KvLQT 1 (Sanguinetti et al. 1996).

KvLQT 1, despite its typical 6-transmembrane-segment structure and close relationship to the Kv channels, shares little pharmacology with other channel types. The chromanol compound

2938 blocks the native Iks and cloned KvLQT 1 channels (Busch et al. 1996; Bleich et al. 1997), acting on the KvLQT 1 subunit (Loussouarn et al. 1997). 293B is inactive against HERG (Busch et al. 1997b). The HERG-blocking antiarrhythmics E-4031 and dofetilide are without effect on KvLQT 1 (Yang et al. 1997). The benzodiazepine compound L-364,373 activates the cardiac Ik$, and KvLQT1 (in the absence of minK) in oocytes (Salata et al. 1998).

The next family member was first identified in the nematode C. elegans and named nKQT2 (Wei et al. 1996). Geneticists studying epilepsy discovered a mammalian homologue, KGNQ2, and a further family member, KCNQ3 (Biervert et al. 1998; Charlier et al. 1998; Singh et al. 1998; Stoffel & Jan 1998). The effects of damage to the KCNQ2 or KCNQ3 genes are comparatively mild: a rare condition called benign familial neonatal convulsions (BFNC), with seizures during the first weeks of life and usually no continuing epileptic syndrome. KCNQ2 mutations do not appear to contribute to any other epileptic syndromes besides BFNC (Steinlein et al. 1999). 136

When expressed individually in Xenopus oocytes, KCNQ2 and KCNQ3 produce small potassium currents, but coexpressing KCNQ2 and KCNQ3 gives currents 15-foid larger than those from the individual subunits. Both genes are expressed principaiiy in the brain. In mouse brain, KCNQ3 expression lags behind KCNQ2 until postnatal day 30 (Tinel et al. 1998).

Pharmacologically, KCNQ2 is not affected by the KCNQ1 (KvLQT 1 ) blocker 2938 or the KCNQ1 activator L-364,373. However, the benzodiazepine L-735,821 blocks both KCNQ1 and KCNQ2 (Tinel et al. 1998). The anticonvulsant retigabine may owe its activity to opening of KCNQ2+KCNQ3 channels (Rundfeldt & Netzer 2000). Other agents Inhibiting KCNQ2 are linopirdine and tetraethylammonium (see next section).

A further member of the family, KCNQ4, is genetically linked to deafness. KCNQ4, like KCNQ2, produces larger currents when coexpressed with KCNQ3 inXenopus oocytes (Kubisch et al. 1999). A KCNQ5 has now been reported by two groups (Lerche et al. 2000; Schroeder et al. 2000). No disease has yet been firmly linked to KCNQ5 mutations, but a contribution to epileptic disorders is feasible.

KCNQ channels In SCG and NG108-15 cells

The fast portion of NG108-15 M-iike current closely resembles the mouse sympathetic M- current (Seiyanko et ai. 1999). Wang et ai.’s (1998a) identification of a KCNQ2+KCNQ3 heteromultimer as the molecular basis of the sympathetic M-current therefore also gave us a likely candidate for the NG108-15 fast current.

While McKinnon’s group (Dixon & McKinnon 1996) found no potassium channel mRNA expression levels that parallelled the pattern of M-current expression in rat sympathetic ganglia, they recently reported that the ganglionic expression of KCNQ2 ^ match the pattern of M-current expression (Wang et ai. 1998a). KCNQ3 mRNA was present in ail the ganglia tested, but higher levels of KCNQ2 mRNA distinguished the M-current-containing superior cervical ganglion from the coeiiac and superior mesenteric ganglia, which do not have M-currents (Dixon & McKinnon 1996). Further, the small currents produced by KCNQ3 expressed on its own in heterologous systems suggest that the presence of both channel proteins is necessary for the emergence of an M-current. Kinetic data supported Wang et al.’s (1998a) identification, since activation and deactivation kinetics of oocyte-expressed KCNQ2+KCNQ3 currents resembled those of the M-current. Pharmacological findings added persuasive evidence, since the KCNQ channels were sensitive to low concentrations of the M-current blocker linopirdine and its analogue XE991. Eag-family currents, and Kvi .2 and 137

Kv4.3 (exemplifying delayed rectifiers and A-currents, respectively) were much less sensitive to these drugs (Wang et al. 1998a). An initial proposal of KCNQ2 as the M-channel (Wang et al. 1998b) was revised in Wang et al. (1998a) to acknowledge that the TEA sensitivity of the native M-current could only be accounted for by invoking a KCNQ2+KCNQ3 heteromultimer. KCNQ2 is readily inhibited by 1 mM TEA; KCNQ3 is much iess sensitive, and the heteromultimer current is intermediate in sensitivity (Yang et al. 1998), like the native M- current.

Following Wang et al.’s (1998a) report, I examined the properties of the KCNQ2+KCNQ3 heteromultimer expressed in CHO cells, for comparison with the M-current (Seiyanko et al. 1999) and the NG108-15 fast current (present work). For this, I expressed the KCNQ channels used by Wang et al. (1998a), human KCNQ2 and rat KGNQ3, separately and together. The human KCN02 sequence is 841 amino acids in length, with molecular weight 92.5 kilodaltons. The rat KGNQ3 sequence is 873 amino acids long, with molecular weight 97 kilodaltons (data from the Swiss Institute of Bioinformatics TrEMBL database). Wang et al. (1998a) found 95% amino acid sequence identity between rat KGN03 and a partiai human KGNQ3 clone. NG108-15 cells probably express the mouse homologues, derived from the mouse neuroblastoma parent of this line. I do not have sequence information on these, but rat/mouse protein homology can be expected to be close. 138

5.2: Results

Heterologously-expressed KCNQ channels and the NG108-15 fast current

As described in Chapter 4, NG 108-15 currents recorded 7-14 days after piating had a standing M-like current of about 0.5-1.5 nA when holding at -2 0 mV. About 30% of this amplitude was composed of a fast-deactivating current, resistant to WAY-123,398, that resembled the classic sympathetic ganglion M-current. After WAY-123,398 block of the slow NG108-15 current, the fast current yields kinetic and pharmacoiogical data, which I compare here with a cloned counterpart. I expressed Dr. D. McKinnon’s clones of human KCNQ2 and rat KGNQ3 in CHO hml cells (see Methods) by transfection with LipofectAmine Plus. I expressed these subunits either separately or together in equal proportions by weight.

There was a striking difference in the sizes of currents produced by the KCNQ subunits. KCNQ2+KCNQ3 currents expressed readily: of the cells marked by CD8 expression (see

Methods), an estimated >90% gave currents of 0.5-1 nA at -2 0 mV, one day after transfection. Two and three days after transfection, currents were even bigger, impairing recording stability. KCNQ2 expressed alone gave currents in about 80% of marked cells, with amplitudes of 0.2-1 nA at -2 0 mV, when recorded t ^ days after transfection. Useful KCNQ3 current amplitudes were only achieved three days after transfection; under 50% of marked cells had even 50 pA of current, and currents much above 200 pA at -2 0 mV were extremely rare.

Kinetics

KCNQ2, KCNQ3 and KCNQ2+KCNQ3 currents show delayed rectifier behaviour. I have characterised this in several ways. Figure 5.1 shows examples of the three protocols used, from three different cells. The uppermost record in Figure 5.1 was obtained with the classic M-current protocol deactivating from -2 0 mV. Current-voltage and conductance-voltage curves derived from such records enable the KCNQ currents to be compared with native M- type currents (see later). The second record in Figure 5.1 illustrates a protocol activating the channels from -8 0 mV, giving a more complete current/voltage relationship. Most useful of all is a compound protocol following an initial activation to +50 mV with sequential deactivations to a range of more negative potentials, and finally to -7 0 mV. This forms the bottom record in Figure 5.1. The inset of Figure 5.1 shows a magnified view of the final relaxations in this protocoi. The amplitudes of these relaxations give a valuable “current/voltage”, or rather conductance/voltage, curve: all measurements are made at the same potential, so the driving force on the potassium charge-carrier is the same. 139

Figure 5.1 : Examples of KCNQ2+KCNQ3 currents

(Top) M-current voltage protocol and KCNQ2+KCNQ3 current family recorded in a CHO cell.

(Centre) Activation protocol and KCNQ2+KCNQ3 currents.

(Bottom) Double-deactivation protocol and KCNQ2+KCNQ3 currents.

(Inset) Magnified tail currents from the double-deactivation protocol, seen on stepping back to -7 0 mV. 140

-20 mV M-current protocol -110 mV

0 .5 nA

+50 mV Activation protocol -80 mV

25 pA

1 nA 100 ms

Double deacti- +50 mV vation protocol -50 mV -70 mV

0 .5 nA

r T T T T I 2 time (s) 3

Figure 5-1 141

Reversal potential

The solutions used for KCNQ channel recordings (110 mM potassium internally and 2.5 mM potassium externally) give a theoretical reversal potential of -95.54 mV, as described in the Results section of Chapter 4. The NG108-15 fast current reversed at an average of -74.5 mV (n=19). The reversal potentials I have used for conductance calculation came from the largest available data-set in each case. For KCNQ2+KCNQ3, Vrev was -78.2 mV for eight deactivations from -20m V (six deactivations from +50 mV gave a similar value of -76.8 mV).

For KCNQ2 and KCNQ3 expressed separately, Vrev values came from deactivations after a step to +50 mV. Vrev was -82.2 mV (n=6) for KCNQ2, and -73.8 mV (n=7) for KCNQ3. In the cases of KCNQ2 and KCNQ2+KCNQ3, the overall deviation from the calculated value is similar to that for mErgla, and presumably reflects the same sources of error, including some 7.3 mV of liquid junction potential. KCNQ3, recorded under the same conditions as the other channels, showed a more positive reversal potential, commonly evident during recording as a change in the direction of tail currents at a nominal step potential of -7 0 mV. While the influence of potassium build-up around the cell would be less with the small currents produced by this channel type, contamination by leak currents is proportionally greater, and may explain a reversal potential error. Tiny outward current relaxations close to the true reversal potential of KCNQ3 could be concealed by even a small transient inward leak, such as might occur through non-potassium-specific channels that reverse more positively.

Conductance

I have plotted two conductance/voltage curves for each type of KCNQ channel tested. One curve is derived from the back-fitted amplitude of relaxations at -7 0 mV. The other is calculated from mean end-of-step amplitudes of currents activated from -8 0 mV, corrected for driving force by the formula given in Chapter 3.2. All data-sets are normalised to the value at +20 mV, for convenience with some incomplete data sets and to standardise properties with respect to KCNQ3, which shows inactivation at more positive potentials.

Figure 5.2 shows the data for KCNQ2+KCNQ3 coexpressed currents. Maximum conductance in each case reaches 1.1, where the value at +20 mV is set to 1. The slope of the curve (mV/e-fold change in conductance) is 12.5 for the activation steps, and 12.7 for the deactivation steps. The largest discrepancy is the less negative half-activation potential (-8 mV) measured for the activation steps, compared with that (-20 mV) for the deactivations. The problem is that in both sets of data, the lack of series resistance compensation on the large KCNQ2+KCNQ3 currents will have prevented the more positive command potentials 142

being reached. Some later recordings with series resistance compensation (not shown) do however provide evidence for slight inactivation of the current at very positive potentials, which is however promptly removed on modest hyperpolarisations. The deactivation steps are measured at -7 0 mV and inactivation is expected to have been removed from all but the most positive. A V% of -2 0 mV, obtained from this data set, should be regarded as an interim approximation.

Figures 5.3 and 5.4 show the equivalent data for KCNQ2 and KCNQ3, respectively. Here, the two methods of calculating conductance give closely similar results. KCNQ2 shows half­ activation voltages -1 6 and -1 4 mV, and slope 13 and 12 mV, from deactivation and activation protocols, respectively. In Figure 5.4, because of the inactivation of KCNQ3 currents, I fitted Boltzmann curves only up to the normalisation point at +20 mV, and obtained a correspondingly negative half-activation voltage of -26 or -2 3 mV. The slopes were 11 and 8 mV for deactivation and activation data, respectively. Slope factors relate to the gating behaviour of the channels, in particular to the movement of charges in the voltage sensor. Given that both KCNQ2 and its heteromultimers with KCNQ3 show slopes around 12, this might be the expected slope for KGNQ3 as well. However, KCNQ3 currents were very small, and the relaxation amplitudes at -7 0 mV (source of the 11 mV slope factor) were often only a few picoamps. The larger currents measured during activation steps (producing the 8 mV slope) gave a much smoother, and probably more reliable, curve. 143

Figure 5.2: Two conductance/voltage curves for KCNQ2+KCNQ3

(Top) Averaged normalised amplitude, + s.e.m., of KCNQ2+KCNQ3 deactivation tails at -7 0 mV (see Figure 5.1 inset), against potential during the preceding step. Boltzmann fit parameters: normalised Gmax = 1.07 ± 0.01 V% = -19.79 ±0.52 mV k = 12.72 ±0.44 mV

(Bottom) Averaged normalised conductance, ± s.e.m., calculated from KCNQ2+KCNQ3 activations (see Figure 5.1 centre), against step potential. Boltzmann fit parameters: normalised Gmax = 1.12 ± 0.01 V% =-8.32 ± 0.65 mV k = 12.48+ 0.53 mV 144

KCNQ2+KCNQ3 (n=4), deactivated to -70 mV

(= 0 .8 - o OJ o 0.6 - ■D % 0 .4 -

0 .2 -

“ 0 . 0 -

-80 -60 -40 -200 20 40 60

KCNQ2+KCNQ3 (n=8), activated from -80 mV o 0.8 -

2 0.6 - "O

0.2 -

0.0 -

-80 -40 -20 0 20 40 60-60

membrane potential (mV) Figure 5-2 145

Figure 5.3: Two conductance/voltage curves for KCNQ2

(Top) Averaged normalised amplitude, + s.e.m., of KCNQ2 deactivation tails at -7 0 mV (as in Figure 5.1 inset), against potential during the preceding step. Boitzmann fit parameters: normalised Gmax = 1 • 10 ± 0.01 V% = -15.51 + 0.56 mV k= 13.07 + 0.47 mV

(Bottom) Averaged normalised conductance, ± s.e.m., calculated from KGNQ2 activations (as in Figure 5.1 centre), against step potentiai. Boitzmann fit parameters: normalised Gmax = 1.06 ± 0.01 Vy, = -14.01 ±0.67 mV k = 11.74 + 0.56 mV 146

1.0 - KCNQ2 (n=5) deactivated to -70 mV o 0.8-

0.6 -

0 .4 - o 0.2-

0.0 -

-80 -60 -40 -20 0 20 40 60

KCNQ2 (n=9) activated from -80 mV o 0 -8 -

2 0.6 - ■o 0) B 0 .4 -

0.2 -

0.0 -

-80 -60 -40 -20 0 20 40 60 membrane potential (mV) Figure 5-3 147

Figure 5.4: Two conductance/voltage curves for KCNQ3

(Top) Averaged normalised amplitude, ± s.e.m., of KCNQ3 deactivation tails at -7 0 mV (as in Figure 5.1 inset), against potential during the preceding step, fitted up to the normalisation point at +20 mV. Boltzmann fit parameters: V% = -26.43 ± 1.46 mV k = 11.43 ± 1.31 mV

(Bottom) Averaged normalised conductance, + s.e.m., calculated from KCNQ3 activations (as in Figure 5.1 centre), against step potential. Boltzmann fit parameters: V% = -22.65 ± 1.03 mV k = 7.59 + 0.91 mV 148

KCNQ3 (n=5), deactivated to -70 mV

§ 0.8 -

■Q 0 . 6 -

CO 0.4 - c 0.2-

0 .0 - T -80 -60 -40 ■20 0 20 40 60

KCNQ3 (n=6) activated from -80 mV o 0.8 - o 0.6- "O 0) m 0 .4 -

0.2 -

0.0 -

-80 -60 -40 -20 0 20 40 60 membrane potential (mV) Figure 5-4 149

NG108-15 current dissection

As indicated in Chapter 4, the two constituents of the whole NG108-15 M-like current can be isolated in two ways. The KCNQ-like portion appears as the fastest of three exponential relaxation components, or as the residual current in 10 pM WAY-123,398. Figure 5.5 combines a recording of KCNQ2+KCNQ3 (as in Figure 5.1) and of the NG108-15 current recorded in the presence of 10 pM WAY-123,398 (from Figure 4.2) for comparison. The use of a subtraction method for obtaining the slow current unfortunately results in all my NG108- 15 fast currents being recorded on 6-second steps, so it should be borne in mind that the time-scale of this NG108-15 fast current record is only half that of the KCNQ2+KGNQ3 example. Nevertheless, the resemblance of the current profiles is evident both in the early completion of the relaxation and in the spacing of the currents at the end of the steps.

Figure 5.6 shows the similarity of the current/voltage relationships for KCNQ2+3 in OHO celis and for the fast current revealed by WAY-123,398 treatment of NG108-15 ceiis. These data were obtained using the M-like current protocoi (deactivations from -2 0 mV for 1 second on the KGNQ2+KCNQ3 current and 6 seconds for the NG108-15 current). Instantaneous currents are from taii-currents back-fitted with an appropriate number of exponential decay components. The end-of-step values are mean amplitudes, as before.

A single exponential fit was visually satisfactory for NG108-15 deactivations in 10 pM WAY- 123,398, but did not coincide with the initial part of KGNQ2+KGNQ3 current relaxations in GHO ceiis, as illustrated for a single trace from each current type in Figure 5.7a. The time constants from these data are plotted in Figure 5.7b, and show similar magnitudes for the two current types, with the single NG108-15 time constants being intermediate in size between the double KGNQ2+KGNQ3 ones. Figure 5.8 illustrates the averaged reciprocals of the time constants for deactivation from -2 0 mV. These show a decline towards more positive potentials, rather than the U-shaped curve seen in Robbins et al. (1992). The bottom plot in Figure 5.8 is derived from single-exponential fits to relaxations from +50 mV (using the protocol shown at the bottom of Figure 5.1) in five KGNQ2+KGNQ3 currents. These currents were large at +50 mV, ranging from 1.4 to 3.5 nA and averaging 2.4 nA. There is consequently some shortfall in the membrane potential achieved, since series resistance was not compensated. Notwithstanding, a U-shaped curve does appear here, with a minimum at about -3 0 mV, close to the V% of -2 0 mV derived from conductance measurements based on the maximum amplitudes of deactivating currents at -7 0 mV (Figure 5.4). 150

Figure 5.5: KCNQ2+KCNQ3 and the NG108-15 fast current compared (Above) Voltage protocol and KCNQ2+KCNQ3 current deactivating from -20 mV.

(Below) Voltage protocol and WAY-123,398-resistant current in NG108-15, deactivating from -2 0 mV. Note that the scale of the time axis is half that of the upper record. -20 mV 151

-110 mV

200 pA

0.5 s

KCNQ2+KCNQ3 in CHO hm1

20 mV

-100 mV

WAY-123,398 resistant 200 pA current 1 s in NG108-15

Figure 5-5 152

Figure 5.6: Partial current/voitage relationships for KCNQ2+KCNQ3 and the NG108-15 fast current

(Top) Back-fitted instantaneous and end-of-step (at ~1 second) current amplitudes, + s.e.m., for KCNQ2+KCNQ3 currents in eight CHO cells.

(Bottom) Back-fitted instantaneous and end-of-step (at ~6 seconds) current amplitudes, ± s.e.m., for the fast (WAY-123,398-insensitive) current in nineteen NG108-15 cells. 153

KCNQ2+KCNQ3, n=8 — end-of-step ♦ instantaneous 0.5 -

0.0 -

Æ'

-120 -100 -80 -60 -40 -20 0) CO membrane potential (mV) "8 E o c 0 NG108-15 fast, n=19 ■D 3 — end-of-step 0.5 - # instantaneous o_ E CO -t—> C 0 i_ m :

-0.5 -

- 1 2 0 - 1 0 0 -80 -60 -40 -20 membrane potential (mV)

Figuie 5-6 154

Figure 5.7: Deactivation kinetics of KCNQ2+KCNQ3 and NG108-15 fast currents

Figure 5.7a: (Top) Initial region of a relaxation at -5 0 mV (from -2 0 mV) for a KCNQ2+KCNQ3 current in a CHO hml cell, with one- and two-exponential fits superimposed. A one-component fit is inadequate for this current, while a two-component fit matches it well.

(Bottom) Initial region of a relaxation at -5 0 mV (from -2 0 mV) for an NG108-15 fast current, in the presence of 10 pM WAY-123,398, with a one-exponential fit superimposed.

Figure 5.7b: (Top) x + s.e.m. versus voltage from a two-exponential fit to KGNQ2+KCNQ3 current deactivations from -2 0 mV in eight CHO cells.

(Bottom) X + s.e.m. versus voltage from a single-exponential fit to NG108-15 fast (WAY- 123,398-resistant) current deactivations from -2 0 mV in nineteen cells. 155

700 -I

600 KCNQ2+KCNQ3 500 - 2-exponential fit: 400 - TS about 30 < o_ and 100 ms 3 0 0 -

200 1-exponential fit: 3 Ü T about 60 ms

100 -

0 -

-100 0 50 100 150 200 250 300 time (ms)

350 -I

300 - NG108-15 fast current 250 - in WAY-123,398 200 < Q. 1-exponential fit: 150 - T about 60 ms c 2 100 -

50 —

0 -

-50 0 100 200 300 400 500 time (ms) Figure 5-7a 400 -I 156 KCNQ2+KCNQ3, n=8 slow X 3 5 0 - ■ fast X

3 0 0 -

2 5 0 -

w 200 -

1 5 0 -

1 0 0 -

5 0 -

-100 -80-60 -40 -20-120

4 0 0 -, membrane potential (mV)

3 5 0 -

3 0 0 -

NG108-15 fast, n=19 2 5 0 - ■ mean x

2 200 -

1 5 0 -

1 0 0 -

5 0 -

-1 2 0 -1 0 0 -80 -60 -40 -20 membrane potential (mV)

Figure 5-7b 157

Figure 5.8: Deactivation as 1/x for KCNQ2+KCNQ3 and NG108-15 fast currents

(Top) 1/x + s.e.m. versus voltage from a single-exponential fit to nineteen NG108-15 fast (WAY-123,398-resistant) current deactivations from -2 0 mV.

(Centre) 1/x + s.e.m. versus voltage from a two-exponential fit to KCNQ2+KCNQ3 current deactivations from -2 0 mV in eight CHO cells.

(Bottom) 1/x + s.e.m. versus voltage from a single-exponential fit to KCNQ2+KCNQ3 current deactivations from a nominal command voltage of +50 mV in five CHO cells (note that the more positive command voltages will not have been achieved: see Discussion). 100-1 158

8 0 -

6 0 - NG108-15 fast current relaxations from -20 mV CO 4 0 -

2 0 -

—I----- 1----- 1----- 1----- 1----- 1----- 1----- 1----- 1----- 1----- 1----- 1----- 1----- 1----- 1----- 1----- 1 -120 -100 -80 -60 -40 -20 0 20 40

100-,

8 0 - KCNQ2+KCNQ3 relaxations from -20 mV 6 0 -

(/) 4 0 -

2 0 - Î-Z- m m

-120 -100 -80 -60 -40 -20 0 20 40

80 —

KCNQ2+KCNQ3 relaxations 6 0 - from 4-50 mV

—■— mean 1/x

2 0 -

-120 -100 -80 -60 -40 -20 0 20 40 60

Figure 5-8 membrane potential (mV) 159

Pharmacology

The slow portion of the NG108-15 M-like current was blocked with 10 pM WAY-123,398, and the drugs of interest were applied cumulatively in doses following a 1,3,10,30... pattern, giving intervals of about 0.5 log unit. The relaxation at -5 0 mV (during a 6-second step from a holding potential of -2 0 mV) was assessed for a single control trace, and for a trace showing a stable current amplitude at each concentration of the blocker. Using Clampfit, one exponential was back-fitted to NG108-15 fast currents, and two to KCNQ currents in CHO cells (see Methods). From the mathematically-corrected relaxation amplitudes, current inhibition was calculated, plotted, and fitted as described for m Erg la.

Figure 5.9 demonstrates the selectivity of linopirdine, at low concentrations, for

KCNQ2+KCNQ3 currents and the NG108-15 fast current. Doses from 3 to 30 pM linopirdine cause a substantial, concentration-dependent block of these currents. For comparison, linopirdine up to 30 pM had only a modest effect on relaxations of mErgla. The record shown for mKvl .2 is a ramp current, which is only about half-inhibited by 100 pM linopirdine dissolved in dimethylsulphoxide. Similar results were obtained in a further 3 cells with this protocol. [Brief tests suggested that linopirdine was slightly more potent against the NG108- 15 fast current (n=3, not shown) when dissolved in DMSO rather than in 0.1 N HOI. DMSO alone at equivalent concentrations caused no inhibition. I did not test aqueous acidic linopirdine stocks against mKvl .2, but an equal or lesser effect seems likely on this basis].

Figure 5.10 shows concentration/response data for linopirdine, azimilide and tetraethylammonium on the KCNQ2+KCNQ3 current expressed in CHO hml cells. I have added the data for TEA on KCNQ2 and KCNQ3 separately, for comparison. Figure 5.11 gives the equivalent data for the NG108-15 fast current. Table 3 summarises the block of the NG108-15 fast current and of KONG currents by linopirdine, azimilide and TEA. Linopirdine and azimilide are similar in their effects on KCNQ2, KCNQ3 and KCNQ2+KCNQ3 currents; only TEA distinguishes between the different KCNQ channels. The TEA sensitivity of the NG 108-15 fast current is compatible with a composition of KCNQ2+KCNQ3 heteromultimers, perhaps with a relatively high contribution of KCNQ2; the shallow slope for the NG 108-15 fast current may suggest a variable subunit stoichiometry.

I also fitted curves to the drug inhibition data for individual cells, and used a two-tailed Student’s t-test (in Excel) for comparison between the KCNQ channel currents and the NG108-15 fast current, obtaining the P values shown in the lower part of Table 3. No P values of less than 0.05 (which would suggest a significant difference between data for 160

KCNQ currents and the NG108-15 fast current) were obtained. Cross-comparison between KCNQ channel current and NG108-15 slow current inhibition by azimilide and TEA also gave no significant differences. However, see also comments (p.122) about Table 2, and (p.177) in Chapter 6.

As shown in Figure 4.12 and discussed in Chapter 4, firing of NG108-15 cells in response to an applied depolarisation at room temperature was prolonged by linopirdine at a KCNQ channel-blocking dose of 30 pM. By contrast, the ERG channel blocker WAY-123,398 had only a marginal effect on firing rate under these conditions. 161

Figure 5.9: Selectivity of linopirdine The effect of linopirdine on examples of four currents, with the relevant voltage protocol shown above each one.

(Above left). KCNQ2+KCNQ3. Decrease in the holding current at -2 0 mV and the relaxation at -5 0 mV, by 3 pM linopirdine. Most of the standing current and relaxation removed by 30 pM linopirdine.

(Above right) NG108-15 fast current, in the presence of 10 pM WAY-123,398. Big decrease in the holding current at -2 0 mV and the relaxation at -5 0 mV, by 3 pM linopirdine. Most of the standing current and relaxation removed by 30 pM linopirdine.

(Below left) mErgla expressed in a CHO hml cell. Small effect of doses from 3 to 30 pM linopirdine on the relaxation at -5 0 mV; negligible effect on the standing current at 0 mV. Similar effect on relaxations in a further 2 cells.

(Below right) mKvl .2 in an MK2 cell. -50% inhibition of mKvl .2 current (evoked by a ramp from -7 8 to +52 mV) by a large dose of 100 pM linopirdine dissolved in dimethylsulphoxide. Similar result in a further 3 cells. -20 mV -20 mV 1 6 2 n

-50 mV -50 mV

KCNQ2+KCNQ3 NG108-15 fast current

100 pA

100 pA 0.25 s

------control ------control 3 nM linopirdine ...... 3 jiM linopirdine 10 pM linopirdine 10 nM linopirdine ------30 nM linopirdine ------30 nM linopirdine

0 mV +52 mV

-50 mV -78 mV m Ergla mKv1.2

25 pA

2 nA

0.5 s

------control ------control 3 nM linopirdine ...... 100 nM linopirdine 10 nM linopirdine dissolved in DMSO ------30 )iM linopirdine

F igure 5-9 163

Figure 5.10: Concentration/response curves on KCNQ channels All data are plotted as mean inhibition + s.e.m.

(Top) Linopirdine (n=4-5) on KCNQ2+KCNQ3 Hill equation fit parameters:

Y m a x = 1 0 0 %

IC50 = 5.10 + 0.61 pM nH - 0.95 + 0.10

(Centre) Azimilide (n=4) on KCNQ2+KCNQ3 Hill equation fit parameters:

Y m a x = 1 0 0 %

IC50 = 73.55 ± 8.94 pM nn = 0.68 + 0.06

(Bottom) TEA (n=4) on KCNQ2 alone Hill equation fit parameters:

Y m a x = 1 0 0 %

IC50 = 0.29 ± 0.02 mM nH = 0.93 + 0.06

TEA (n=4) on KCNQ2+KCNQ3 Hill equation fit parameters:

Y m a x = 1 0 0 %

IO50 = 3.82 + 0.20 mM nH = 0.76 + 0.03 % inhibition of control current % inhibition of control current % inhibition of control current T1 cq'

O

Q. Ü1 Q. 3 O TD Q. 9- o - 3 CD

O ~ o ~

o

7s

S D S CO O) 165

Figure 5.11: Concentration/response curves on the NG108-15 fast current All data are plotted as mean Inhibition + s.e.m.

(Top) Linopirdine (n=4) Hill equation fit parameters:

Ymax=100% iCso = 1.07 ± 0.23 pM nH = 1.21 ± 0 .2 8

(Centre) Azimilide (n=3) Hill equation fit parameters:

Ymax=100%

ICso = 35.53 ± 3.81 pM nn = 1.04 ± 0.12

(Bottom) TEA (n=3) Hill equation fit parameters:

Ymax=100% ICso = 1.16 + 0.30 mM nn = 0.47 + 0.07 % inhibition of control current % inhibition of control current % inhibition of control current

—^ -‘ MOJ^OlO'slOOtOO-^ o o 0 o o o o o o ooooooooooooo ■ I ■ I ...... 1 ■ I ■ * ‘ ■ I ■ I ■ * o o

CO Q. Q.

3 O m T3 > Q. o CD 1

8 o

8 o

O) CD 167

Table 3: Block of the NG108-15 fast current and of KCNQ currents by linopirdine, azimilide and TEA All ICso values are quoted + s.e.m., firstly, as calculated on averaged inhibition data: Drug NG 108-15 fast KCNQ2 alone KCNQ2+KCNQ3 KCNQ3 alone Linopirdine

ICso zh s.e.m. 1.1 + 0 .2 pM 3.7 ±0.1 pM 5.1 ±0.6 pM 4.8 ± 0.6 pM

Hh 1.2 1.2 1.0 1.3 n (cells) 4 4 4-5 4 Azimilide

ICso 35.5 ± 3.8 pM 70.6 ± 4.8 pM 73.6 ± 8.9 pM 12.7 ±1.8 pM nH 1.0 1.2 0.7 1.0 n (cells) 3 4 4 3 TEA

ICso 1.2±0.3mM 0.3 ± 0.02 mM 3.8 ± 0.2 mM -142 mM nn 0.5 0.9 0.8 -0.7 n (cells) 3 4 4 4 and secondly from fitting data individually for each cell, averaging and applying a t-test: Drug NG108-15 KCNQ2 KCNQ2+ P value, ICso P value, ICso fast alone KCNQ3 KCNQ vs. KCNQ2+3 NG fast vs. NG slow Linopirdine ICso + s.e.m. 1.1 ±0.38 5.88 ± 1.56 0.05 1.2 1.2 nn 3 4 n (cells) Azimilide ICso 42.67 ± 23.67 82.5 ± 32.69 0.40 0.06 0.9 0.95 Hh 3 4 n (cells) TEA ICso 1.62 ± 0.65 0.32 ± 0.09 6.98 ±4.39 0.35 (KCNQ 0.30 0.7 0.95 0.85 2+3): nn 3 4 4 0.07 (KCNQ2) n (cells) 168

5.3: Discussion

The NG108-15 M-like current is now seen to comprise, in addition to a mErgla-like current, another current which resembles that generated by KCNQ2+KCNQ3 heteromultimeric channels expressed in CHO cells.

As previously mentioned, diagnostic criteria for the identities of native currents are: (1 ) the presence of mRNA transcripts and protein; (2) biophysical properties, including various kinetic measurements; (3) sensitivity to blockers; and (4) modulation by agonist action (Selyanko et al. 1999).

Regarding KGNQ2+KCNQ3 as a basis for the NG108-15 fast current, the expression of the channeis (point (1)) is partially covered in Selyanko et al. (1999). Collaborators in the Wellcome Laboratory for Molecular Pharmacology confirmed by RT-PCR the presence of KCNQ2 and KCNQ3 mRNAs in both NG108-15 ceiis and mouse SCG neurones. Immunostaining for the KCNQ channel proteins is currently under study. Modulation by an agonist of Mi receptors (point (4)) has been shown for the NG108-15 fast current and the mouse SCG M-current (Selyanko et al. 1999), and for KCNQ channels in CHO cells (Selyanko et al. 2000). In a CHO cell environment, all the KCNQ channel types from KCNQ1 to KCNQ4 were readily inhibited by the Mi receptor agonist oxotremorine methiodide.

My own data, discussed here, address points (2) and (3): the kinetics and blocker sensitivities of KCNQ2, KCNQ3, KCNQ2+KCNQ3 coexpressed currents and the NG108-15 fast current. As coilaborative data show in Selyanko et al. (1999), the fast current in NG108-15 cells corresponds, in kinetics, pharmacology and capacity for muscarinic moduiation, to the mouse sympathetic ganglion M-current. I show here that the behaviour of the NG108-15 fast current is aiso very similar to that of a heteromultimer of KCNQ2+KCNQ3 subunits.

I found that about 30% of the NG108-15 M-like current was produced by channels with macroscopic properties resembling those of KCNQ2+KCNQ3 expressed in CHO ceils. The fast current contribution to relaxations at -5 0 mV was 33% in collaborative results for 86 cells (Selyanko et al. 1999). Other figures can be calculated from the Discussion section of Chapter 4, where the mErgla-like current contribution is quantified. The consistent kinetic properties of the WAY-123,398-resistant current, and the possibility of blocking it fully with 30 pM linopirdine, confirm that only two current types are present, so the fast current contribution is 100% minus the slow current contribution. 169

Kinetics

The properties of deactivation tail currents compare well for KCNQ2+KCNQ3 and the WAY- 123,398-resistant current in NG108-15 cells (examples in Figure 5.5 and averaged current/voltage data in Figure 5.6). The deactivation time constants plotted in Figure 5.7 are similar in magnitude for the cloned and the native current. The need for a two-exponential fit for KCNQ24-KCNQ3 currents in CHO cells, while a single exponential sufficed for the NG108- 15 fast current, may relate in part to the different length of the deactivation steps. I commonly fitted only the first part of the 6-second step used on the NG108-15 M-like current. Still, the longer sampling interval, and the need to avoid NG108-15 transients while placing fitting cursors, may have concealed a fast component of the relaxation such that a two-exponential deactivation was approximated by a one-exponential fit.

I found the activation curve of the KCNQ2+KCNQ3 channel to be positively shifted relative to literature values for the sympathetic M-current and the NG108-15 M-like current. However, the results of Constanti & Brown (1981) and Robbins et al. (1992), with V% at -4 4 or -4 5 mV, were obtained at raised temperatures. It appears that the V% for the fast M-like current may occur at more positive potentials at room temperature.

Even though I have not been able to study the fast M-like current at strongly positive potentials, there is provisional evidence from deactivation time constants that its activation midpoint V% is positive to -3 0 mV at room temperature. This is because the reciprocal (mono-exponential) time-constants of Figure 5.8, piotted against step potential, do not show an upturn at the positive end of the voltage range used.

A U-shaped relationship of 1/x to step potential, as found for the whole NG108-15 M-like current in Robbins et al. (1992), and the sympathetic M-current in Constanti & Brown (1981), is an expected property of a channel whose opening and closing is determined by a single conformational transition. The closino rate of the channels falls off as voltage becomes more positive, and the opening rate increases rapidly, according to the exponential relationships: a = a(o)exp(+ze/2kT)(V-V%) and p = p(o)exp(-ze/2kT)(V-V%), wherea and p are the opening and closing rate constants, respectively, e is the eiementary charge, and z represents the charge transfer within the channel that leads to opening or closing (Adams et al. 1982).

Further, the time constant x for a current relaxation is the reciprocal of (a+ P) (Hille 1984, p.47), so 1/x follows the U-shaped curve generated by averaging a and p. Since a and p are functions of V% with opposite signs, the point where a = p, and the sum of the two curves is minimal, falls at V%. 170

Figure 5.8 shows reciprocal plots of the time constants obtained for relaxations from -2 0 mV for both fast M-like and KCNQ2+KCNQ3 channels. Neither of these data-sets yields a U- shape over the range of potentials used. However, data from a single-exponential fit of

KCNQ2+KCNQ3 current relaxations from +50 mV shows that 1/x starts to increase with voltage positive to about -2 0 mV. This potential corresponds to the value of V% obtained from the deactivation-derived conductance/voltage curve in Figure 5.2. The U-shape is not well-maintained above 0 mV, probably because uncompensated series resistance has prevented the command voltages from being achieved. However, even if the step preceding deactivation reached only +20 mV rather than +50 mV, it was consistent in each cell, and high enough to indicate the trend in the deactivation rate up to about 0 mV. The NG108-15 fast current has such a low x at -3 0 mV that, on the basis of its general similarity to KCNQ2+KCNQ3 currents, this is likely to be close to the minimum. Thus, I have not been able to measure directly the V% for the NG 108-15 fast current, but it is probably in the range of -3 0 to-20 mV.

However, the assumption of a single opening/closing transition determining fast M-like current behaviour is probably an oversimplification. At best, it is an approximation to the behaviour of native currents which may not be reproduced in a non-neuronal cell type. Activation of the KCNQ currents in CHO cells (centre section of Figure 5.1) reproduces a common observation also visible in oocyte recordings (Wang et al. 1998a; Yang et al. 1998).

KCNQ2 , KCNQ3 and KCNQ2+KCNQ3 currents all show an initial sigmoidal phase of activation. Native M-currents reconstituted from channel recordings in sympathetic neurones by Selyanko et al. (1992) and in NG108-15 cells by Selyanko et al. (1995) show exponential activation. This suggests that some extra complexity may exist in the gating process of KCNQ channels when expressed heterologously. Qnly further experiment will determine whether it is connected with the make-up of heteromultimers, phosphorylation state of the channel, accessory subunits, or other post-translational modifications of the channels.

Blockers

Linopirdine block of the NG108-15 fast current (Table 3), with IC50 1.1 pM, was comparable with the 4-5 pM obtained for KCNQ2, KCNQ3 and their mixtures. Linopirdine blocks

KCNQ2+KCNQ3 currents in Xenopus oocytes with IC 50 4.0 pM (Wang et al. 1998), compared with 3.4 pM (Lamas et al. 1997) or 7.0 pM (Costa & Brown 1997) on the SCG M-current. My fitted Hill slopes for linopirdine block are close to unity, suggesting a simple channel/blocker interaction. Linopirdine is selective enough to discriminate KCNQ-like channels from other 171

clones: at 30 [xM it had little effect on m Ercla. and even 100 linopirdine inhibited mKvl .2 by no more than half.

The ICso of 35.5 pM for azimilide on the fast NG108-15 current falls within the 13-75 pM range found for KCNQ2, KGNQ3, and combinations in CHO cells (Table 3). The low Hill slope (0.7) on KCNQ2+3 might be interpreted as evidence for a mixed channel population, but the data-points do not look biphasic. In any case, this is not observed with the NG108-15 fast current, where the Hill slope is unity.

As stated in Chapter 4, the ERG blocking drug WAY-123,398 had no inhibitory effect on

KCNQ2+KCNQ3 currents (n=3) at 10 pM.

TEA and stoichiometry

Channel block by tetraethylammonium is interesting in relation to the precise molecular identity of M-type currents in various cells. KCNQ2 is particularly sensitive to TEA, KCNQ3 particularly insensitive, and the KCNQ2+KCNQ3 current intermediate in sensitivity. The high sensitivity of KCNQ2 is conferred by the presence of a tyrosine residue just downstream of the potassium-selective pore signature sequence GYG, as in TEA-sensitive channels in the

Shaker group (Hadley et al. 2000). The TEA IC50 of 1.2 + 0.3 mM on the fast NG108-15 current is slightly lower than the value of 3.8 ± 0.2 mM for KCNQ2+KCNQ3 channels (Table 3). The presence of KCNQ2 homomultimers in NG108-15 cells, or a higher proportion of KCNQ2 relative to KCNQ3 in the heteromultimer assemblies, is possible. The shallow Hill slope of 0.5 fitted to the NG108-15 fast current data is compatible with a mixed channel population, but I do not propose to draw conclusions from such a small sample.

Other investigators have expressed KCNQ2 and/or KCNQ3 subunits in another cell type, the tsA-201 strain of human embryonic kidney cells (Shapiro et al. 2000). They fitted Hill equations to KCNQ2 and (very incomplete) KCNQ3 data, obtaining Hill slopes close to 1. For KCNQ2+KCNQ3-cotransfected cells, Shapiro et al. (2000) obtained a Hill slope of 0.56 when they fitted a single Hill equation, suggesting non-homogeneous channel types. They preferred a fit of five Hill equations, to allow for all possible subunit combinations in the channel tetramer.

My own TEA data in CHO cells give Hill slopes 0.8 (KCNQ2+KCNQ3), and 0.9 (KCNQ2), a bit low to be taken as unity. My own fit to incomplete KCNQ3 data would suggest a Hill slope of 0.7. Data obtained by A.A. Selyanko and myself show Hill slopes markedly less than unity 172

for TEA on both NG108-15 current constituents as well as on m Ergla and on the mouse ganglion M-current (Selyanko et al. 1999). The Hill slopes for TEA inhibition of KCNQ1 and KCNQ4 are close to 2, which is as yet unexplained in molecular terms (work by A.A. Selyanko in Hadley et al. 2000). Perhaps the block of KCNQ and ERG channels by TEA involves something other than a simple one-to-one binding reaction. No experimenters have used high enough TEA concentrations to show convincingly that KCNQ3 conforms to the one-subunit, unity-Hill-slope “rule”. I do not think that any of the data presently available for TEA block justify any conclusions about the composition of channels in a mixed transfection.

Overall, the sensitivity of KCNQ2+KCNQ3 to linopirdine, azimilide and TEA matches that of the NG108-15 fast current, strongly supporting the identity of the two currents. 173

Chapter 6: General discussion and conclusions

In summary, I have used the patch-clamp technique to dissect the M-like current in neuroblastoma-glioma hybrid cells, making particular use of kinetic measurements and channel-blocking agents. I conciude that channel-selective blockers, when backed up by kinetic and expression data, provided a particularly valuable tool for Identifying M-like current constituents. My findings are that the M-like current consists partly of a current sensitive to WAY-123,398 and resembling that produced by mErgla channels, and partly of a current sensitive to linopirdine and matching the properties of coexpressed KCNQ2+KCNQ3 subunits. These two blockers at low concentrations eiiminate the whole M-like current, indicating that Kvi .2 and other channels are unlikely to contribute.

What has aiso emerged is that, while the sympathetic M-current has been commonly described as “non-inactivating”, this term cannot be considered valid for the NG108-15 M-like current. Certainly, KCNQ2, KCNQ2+KCNQ3 and the fast component of the M-like current show no signs of inactivation at normal M-current recording potentials. However, the slow WAY-123,398-inhibited current, separated by data subtraction, shows unmistakable inactivation behaviour. It is nevertheless significant that mErgla and the mErgla-like NG108-15 current undergo rapid inactivation and de-inactivation, revealed only by paradoxical changes In current amplitude. This does not invalidate the slow Inactivation seen in Kvi .2 as evidence against its participation in the M-like current.

To recapitulate, four classes of criteria have been specified as important in identifying the moiecular counterparts of native currents (Seiyanko et ai. 1999): (1) the presence of mRNA transcripts and protein; (2) biophysicai properties; (3) sensitivity to blockers; and (4) modulation by agonist action. The molecular determinants of these behaviours are localised to varying extents in the channel structure. The intracellular termini of channel subunits contain distinct sites which affect expression of the protein, some kinetic properties, and moduiation, through membrane targeting, association with accessory subunits and the cytoskeleton, and phosphorylation. Most of these involve specific short peptide sequences. Channel gating depends on the S4 transmembrane voitage sensor and its interactions with the surrounding regions. Milier (1995) makes the point that drastically altered sensitivity to a particular blocker can be caused by variations in channel protein sequence having no effects on channel gating. Biocker sensitivity characteristicaliy relates to amino acid residues in the extraceilular mouth of the channel, which means that it can vary independentiy of kinetic properties. This is valuable for channel identification, because kinetic and pharmacoiogical data provide two effectively separate lines of evidence. 174

I would also like to consider how closely particular types of property need to match between native currents and heterologously-expressed channels, for a firm identification to be made. I have certain reservations in respect of modulation. From a chemist’s point of view, direct molecular interactions are more readily detected than indirect ones, so the most persuasive evidence of channel identity comes from properly-controlled experiments in which one substance binds with high affinity to another.

Molecular biology techniques can rigorously demonstrate the presence of a known channel. Messenger RNA can be visualised in a cell (by in-situ hybridisation) or after extraction (as a Northern blot) by allowing it to bind to a radioactively-labelled nucleotide sequence. This approach can identify candidate channels. Immunocytochemistry techniques bind an antibody stably to the protein of interest. A second antibody is commonly used to improve visualisation. In this case, three molecules are stably bound to each other with high affinity. A further use of antibodies is for co-immunoprecipitation, as used to identify the complexes of alpha- and beta-subunits forming dendrotoxin-binding Kvi channels in the brain (Rhodes et al. 1995). An occasional drawback is that these techniques require prior knowledge of the channel sought: at least a partial gene- or protein sequence is needed to create the oligonucleotide probe for the mRNA, or the antigen to raise the antibodies.

Biophvsical properties such as kinetics and single-channel behaviour may conceivably be altered by intracellular interactions with cytoskeletal molecules and unidentified accessory subunits. This field is still in the empirical stage of development. However, kinetic and single-channel properties of cloned channels can provide a guide to the correct subunit combination for a native channel. For example, minK cotransfection with KvLQTI reproduced the behaviour of the cardiac slow delayed rectifier (Sanguinetti et al. 1996; Barhanin et al. 1996). In the work reported here on mErgla and KCNQ2+KCNQ3 as candidate channels for the NG108-15 M-like current, the kinetic correspondences are clear, and are supported by pharmacology and molecular biology. Any influence of post- translational modification or accessory subunits is minimal, or is parallelled in the CHO cell line used to express the cloned channels. It seems there is a simple reason for Kvi .2 not looking like an M-current: it isn’t one.

Potassium channel modulation involves multiple steps. The most direct known mechanism involves an agonist acting on a receptor, the receptor activating a G-protein, and the G- protein beta-gamma complex binding to the channel to alter its activity, as happens with cardiac inward-rectifying channels. In the case of M-currents, there is evidence for a 175 diffusible messenger acting on the channel at a substantial distance from the receptor (Selyanko et al. 1992), implying a fourth or further stage. Individually, these interactions are transient and may be subject to desensitisation, making it difficult to dissect the mechanism of modulation and prove that the mechanism for a heterologously-expressed channel duplicates that for a native counterpart. There is also no guarantee that all the machinery of neuronal transduction will be available in a non-neuronal cell line used to express the receptor and the channel.

Blocker oharmacoloov is a well-established and valuable tool. Small organic molecules like tetraethylammonium and 4-aminopyridine have long been used for a loose classification of potassium channels. Apamln (for calcium-activated potassium channels) and dendrotoxin (for the Kvi subfamily) tightened up the standard of current-matching between different cells. It became apparent early on that sensitivity to block by agents like dendrotoxin and tetraethylammonium correlated with the amino-acid sequence of the channel protein (Stühmer et al. 1989; MacKinnon & Yellen 1990; Kavanaugh et al. 1991). The demands of medicine identified blockers such as linopirdine and WAY-123,398 by their effects in specific tissues, before the molecular nature of their targets was known. Molecular biology is now playing an increasing role in channel identification, but blockers remain useful for dissecting mixed currents and for studying function.

It is possible to exploit the direct interaction of a blocker molecule and a potassium channel protein in several ways. A labelled blocker can be used to stain tissue sections, or the blocker can be bound to a chromatographic column to extract the channel from the tissue (e.g. Scott et al. 1990). Electrophysiology monitors the interaction by measuring the blocker’s obstruction of the electrical current created by potassium ions flowing through the channel.

There are few theoretical objections to the use of blockers to identify channels. Glycosylation of the extra-membrane region of the channel protein, occurring in some cell types but not others, might conceivably interfere with drug access to a channel mouth. Heteromultimers of two or more types of channel subunit typically show reduced sensitivity to a blocker specific to one of the subunits (as with TEA block of KCNQ2+ KCNQ3 heteromultimers). However, only false necative results in channel matching are likely consequences of these circumstances. The creation of a blocker binding site by heteromultimerisation is possible, but as yet unreported. 176

Selectivity of drugs

Phylogenetic trees depleting relatedness of potassium channel genes (e.g. Coetzee et al. 1999; Derst & Karschin 1998) emphasise the distinctness of the eag, erg and e//r channels. The ShakeMamWy channels, Kvi, etc., cluster with the KCNQ channels, while the eag group appear on a separate branch which diverged farther back in evolutionary time. The genetic isolation of the ether-à-go-go group of channels might lead one to expect that they would have a distinctive set of blockers (as well as other behaviours) not shared with channels of other families. This was, implicitly, one of the justifications for Stansfeld et al.'s (1997) proposal that the eag family might harbour the channels responsible for the M-current. As it turns out, the KCNQ channels underlying the sympathetic M-current are closely related genetically to a previously-known group. Indeed, for Coetzee et al. (1999): “the question may arise whether the KOT family should be considered as part of the Kv family.” Yet the pharmacological profile of the sympathetic M-current did not betray its closeness to the Kv group.

Are there any grounds for the assumption that binding sites for blockers are conserved within channel families? It depends partly on the nature of the blocker binding site, as I shall explain. For this purpose, it is easier to classify types of blocker (which are smaller molecules) than types of channel.

A number of channel-blocking oeotides result from the evolution of animals that make venoms to immobilise prey and defend themselves against predators. They are large molecules, often with high specificity. Alpha-dendrotoxin has 59 amino acids (Harvey 1997) and three identified sites of interaction (Hurst et al. 1991) which confine its blocking effect to some members of the Kvi channel subfamily. Margatoxin, a Kvi .3-specific blocker from scorpion venom, interacts with multiple sites on its target channel (Aiyar et al. 1995). The more distributed the high-affinity interaction sites, the more specific the blocker Is likely to be. The opposite situation arises when sensitivity to a blocker depends on the presence of a single amino acid residue in the channel, as with the tyrosine residue controlling tetraethylammonium block, which may differ among channels in the same family (MacKinnon & Yellen 1990; Kavanaugh et al. 1991 ; Hadley et al. 2000).

The pharmaceutical industry has created many organic molecules with channel-blocking activity. These molecules need to be complex enough to dock into a specific site, and to be capable of legal protection for their inventors under the patent system. Simultaneously, they must be small enough to be easily administered and to remain dissolved in bodily fluids until they reach their site of action. Reconciling these requirements gives most such drugs 177

molecular weights in the range of 400-500. Examples are WAY-123,398 (molecular weight 488), azimilide (MW 458 as base, 531 as hydrochloride), THA (MW 235), and linopirdine (MW 391.5). Specificity varies among molecules of this class. Thus, I found WAY-123,398 impressively selective for one constituent of the NG108-15 M-like current. By contrast, azimilide blocks both ERG channels and members of the KCNQ family. Linopirdine at low concentrations is (among potassium channels) selective for KCNQ- and M-channels, but it also acts on ionotropic receptors, and it can block other potassium channels at higher doses

(Lamas et al. (1997), and my finding of about 50% block of Kvi .2 by 100 pM linopirdine).

Inorganic cations, the smallest channel blockers, may be physiologically relevant but have little diagnostic use. Calcium, magnesium and zinc have all been implicated as endogenous modulators of various ion channels and receptors. Block by barium is almost universal in potassium channels, so I have not focussed on barium inhibition of Kvi .2, m Ergla or KCNQ channels. The substituted cation tetraethvlammonium (TEA) is more useful, showing a range of channel-blocking potencies because of the random occurrence of the channel residue that influences its binding. Consequently, it has proved useful for discriminating KCNQ subunits, as well as showing some differential activity on the two constituents of the NG108-15 M-like current.

Depending on the mixture of currents present, a blocker may point directly to a channel identity (WAY-123,398 on ERG) or a very narrow range of channels (dendrotoxin on the Kvi subfamily). A combination of blockers may be a better fingerprint for any individual channel, for example linopirdine and TEA for KCNQ2. A lack of sensitivity to a particular blocker can eliminate some candidates, but cannot provide a positive identification.

To conclude, there are some good reasons, rooted in biochemistry, why large and medium­ sized channel-blocking molecules can help in the identification of native currents. This will continue to be a good approach to such tasks.

Functional relevance of KCNQ and ERG currents

The M-current in sympathetic ganglion cells has an evident physiological function as a neurotransmitter-sensitive brake on action potential generation (Adams et al. 1982). Nerve cells are designed for versatility: their signalling functions may require a burst of action potentials, a single spike, or none at all for several seconds or minutes. The non-inactivating neuronal M-current builds up over several spikes to halt further firing, but can be switched off by muscarinic receptor activity, allowing both phasic and tonic firing by cells expressing M- 178 currents. This itself indicates that the continual availability of the M-current is not necessary for the moment-to-moment survival of cells expressing it.

It is however likely that potassium flow through KCNQ channels plays an important part in establishing and maintaining functional neuronal circuitry. Homozygous KCNQ2 knockout mice die just after birth, an effect ascribed tentatively to lung dysfunction caused by continuous depolarisation of adrenal chromaffin cells (Watanabe et al. 2000). The KGNQ2 and KCNQ3 mutations reported in humans do not have a dominant-negative effect, and reduce rather than eliminate channel function (Weinreich & Jentsch 2000). The effect - a short-lived neonatal seizure condition - is moderate and has not been reported to involve sympathetic nervous system dysfunction. Developmental plasticity might compensate for the reduced function of these channels. The recent discoveries of not just KCNQ4, but KGNQ5 (Lerche et al. 2000; Schroeder et al. 2000) raise the possibility that the KGNO channel family contains enough redundancy to diminish the effects of neuronal channel lesions.

A current resembling the KGNQ2+KGNQ3 heteromultimer is commonly found in NG108-15 cells, but some cells appear able to survive without it (Ghapter 4, Discussion). While differentiated NG108-15 and similar cells undoubtedly form synaptic connections with each other, fire action potentials and release neurotransmitters, they can undergo little selection pressure with regard to firing behaviour unless they become fatally flooded with calcium. What differentiated NG108-15 cells do show consistently Is a current resembling mErgla, the mouse homologue of the cardiac fast “delayed rectifier”.

In the heart, the rapidly-inactivating ERG current forms part of a cardiac ion channel complement which allows repeated depolarisations over a defined range of frequencies. If the ERG current is missing, failure of the heart muscle to maintain a steady firing rate results in premature death of the organism. The question remains what (if any) the survival function of ERG currents may be in NG108-15 or other cell lines.

Bianchi et al. (1998) propose that ERG-type “inward rectifier” channels, normally only expressed early in cell differentiation, are “frozen” by neoplastic transformation. These channels then afford tumour cells a selective advantage by maintaining a shallow membrane potential that promotes continual cell division. Their effect on membrane potentials is demonstrated by the fact that the cells used by this group can only be induced to show transient currents (e.g. Arcangeli et al. 1995). Such cells cannot respond to moderate depolarisations with a sustained potassium outflow that would restore a negative potential. Gonversely, the “classical” so-called anomalous-rectifier currents from the Kjr family are 179 characteristic of mature differentiated cells. These channels stay open above and below the potassium equilibrium potential Ek, allowing ion flow to set the membrane potential close to

Ek.

ERG currents and neuronal differentiation

Healthy neurones are terminally differentiated - that is, they have stable structural and neurochemical properties and no, or very little, capacity for replacement of damaged tissue. In a neural tumour, or neuroblastoma, neuronal cells have regressed by re-acquiring the ability to divide. NG108-15 cells continue to proliferate when cultured in a high-serum medium without chemical interference, but various treatments cause them to adopt an apparently differentiated state. Some researchers use serum starvation (Arcangeli et al. 1995; Pancrazio et al. 1999). Arcangeli et al. (1995) also used retinoic acid to lock the cells in stages G0-G1 of the cell cycle, in effect differentiating them [see Watson et al. (1992), p.370 for a summary of the cell cycle]. Another standard method of differentiating NG108-15 cells is by raising intracellular cAMP. This (via protein kinase A phosphorylation) activates transcription factors which bind at CREs (cAMP-response elements) to promote transcription of genes (Watson et al. 1992, p329), setting off chemical and morphological changes. My observation is that not all chemically treated cells become morphologically differentiated, and some continue to proliferate, while some untreated cells produce miniature M-like currents. Krystosek (1985) too found a probabilistic relationship between treatment and differentiation.

ERG currents and the cell cycle

Several proposals have been made for the function of ERG-like currents in cell lines. An electrical role for HERG-like currents in tumour cell proliferation (Bianchi et al. 1998) is referred to above. A volume-controlling effect of potassium channels may also sustain proliferative activity. Block of potassium channels with TEA, 4-aminopyridine and caesium (effective against HERG at 1 mM: Trudeau et al. 1995), like chloride channel block and other volume-changing stimuli, decreased NG108-15 proliferation. (Rouzaire-Dubois & Dubois 1998). ERG currents may additionally be implicated in the morphological aspects of differentiation: neurite extension in NG108-15 cells was inhibited by 5 mM caesium, but not by the delayed-rectifier blockers 4-aminopyridine and TEA (Pancrazio et al. 1999)

Conversely, the cell cycle appears to affect the properties of ERG-type currents in neuroblastoma (hybrid) cells. Arcangeli et al. (1995) found that in differentiated SH-SY5Y (human neuroblastoma) cells the “inward rectifier” set a more negative resting potential than in proliferating cells. “Inward rectifier” current density increased 40-50 hours after the start of 180

retinoic acid treatment (Arcangeli et al. 1995). Pancrazio et al. (1999) found that differentiated NG108-15 cells had higher HERG-like current density. ERG mRNA also increased on differentiation, and the voltage dependence of the ERG current shifted by about -1 5 mV. Meyer & Heinemann (1998) found opposite results in SH-SY5Y cells after 2-3 days treatment with retinoic acid, when differentiated cells showed a positive shift in the voltage- dependence of the HERG-like current.

The activation of HERG is shifted positively, by up to 27 mV, by protein kinase A (Kiehn et al. 1998) acting at four consensus phosphorylation sites (Thomas et al. 1999). These sites in HERG, S283, S890, T895 and S1137, are all conserved in mErgla (London et al. 1997). Thus, PKA phosphorylation may also cause an activation shift in mErgla. This could explain what is happening during differentiation: increased PKA activity reduces mErgla current activity, reducing the membrane potential. Decreased potassium outflow promotes an increase in cell volume, adding to the influences inhibiting cell division. A later rebound is caused either by a down-regulation of kinase activity, or by an overwhelming increase in channel expression. The nature of the ERG currents observed in differentiating/differentiated neuroblastoma cells and their hybrids may therefore be highly sensitive to the amount of time elapsed after the application or removal of differentiating agents, explaining the contradictory results above. The effects of protein kinase A and its inhibition on mErgla and NG108-15 slow currents would be worth exploring.

With the identification of the slow M-like current in NG108-15 (Meves et al. 1999; Selyanko 1999), it has become clear that an ERG current persists in the NG108-15 line long after differentiation. A minimal conclusion accommodating Bianchi et al.'s (1998) argument Is that the ERG-type current Is permissive with regard to tumour-like behaviour, but has other roles in non-proliferating cells. Given that ERG channel mRNAs are expressed in the mammalian brain (Shi et al. 1997; Wymore et al. 1997), a role in mature neurones seems likely. The lack of neurological correlates of HERG-related long QT syndrome may be explained by the compensatory effect of other family members (Shi et al. 1997).

Stoichiometry, assembiy and membrane compiexes of KCNQ channeis

In my work, I transfected equal amounts of KGNQ2 and KGNQ3 plasmids into GHO cells, hoping that these cells would process the subunits in the same way as NG108-15 cells. GHO cells yielded a presumed heteromultimer with high expression, and intermediate block by TEA, compared with KGNQ2 or KGNQ3 alone. My proposal is that these consistent properties may point to KGNQ2 and KGNQ3 coassemblinc by preference rather than by random association. If the two subunits have a higher affinity for each other than for their 181 own kind, the most likely assembly is a tetramer in which they are linked 2-3-2-3, giving a two-plus-two ratio of subunits. An alternative possibility is that subunits assemble randomly, but that heteromultimeric channels are more efficiently trafficked to the membrane. Schwake et al. (2 0 0 0 ) showed increased - and quantitatively similar - surface expression of epitope- tagged KCNQ2 and KCNQ3 subunits co-expressed in oocytes, and suggested an analogy with ATP-sensitive potassium channels, where coassembly of different subunits conceals a peptide motif normally responsible for protein retention in the endoplasmic reticulum.

The findings of Shapiro et al. (2000) differ from mine, and from reports on KCNQ channels in oocytes, in that their HEK cell line expressed KGN02 less readily than KCNQ3. Perhaps HEK cells have different trafficking mechanisms from GHO cells and oocytes. Alternatively, there may be differences in the interaction of channels with the intracellular scaffolding. Cooper et al. (2000) demonstrate that human brain KGNQ2+KGNQ3 channels are associated with a sub-membrane complex of tubulin, actin, protein kinase A and the anchoring protein AKAP79, with KGNQ2 rather than KGNQ3 containing the region responsible for cytoskeletal interaction. It will be interesting to see whether human embryonic kidney cells (and other cell lines) also have a mechanism for pinning KGNQ2 channels to the cell surface.

Future directions

Several approaches are under way to corroborate the presence of functional ERG- and KGNQ-family channels in the NG108-15 line, including staining with anti-KGNQ2 and anti- KGNQ3 antibodies, and transfection of dominant-negative mutations of KGNQ2 and KGNQ3 (with a G-»S pore mutation preventing potassium ion flow) into NG108-15 cells. To confirm the expression of such channels, an epitope tag such as Myc or FLAG on the mutated channel subunit enables the use of antibodies that specifically visualise the mutated channels. A suitably-tagged dominant-negative mutant mErgla will be needed to apply the same approach to the NG108-15 slow current.

The single channels so far recorded from NG108-15 cells (Selyanko et al. 1995) appear to correspond to the KGNQ-like constituent of the current, which matches the classic SGG M- current. Where are the ERG channels? The single-channel behaviour of mErgla in a mammalian cell environment needs to be examined, but the sparseness of mErgla expression in transfected GHO cells has been discouraging.

Discrepancies in the reported effect of protein kinase A phosphorylation on the M-current warrant further investigation. Schroeder et al. (1998) reported a PKA-mediated increase in 182

KCNQ2+KCNQ3 currents in Xenopus oocytes and HEK cells, but other researchers found no effect of membrane-permeable cAMP analogues on the M-current (Brown et al. 1989) or PKA activators on KCNQ2 (Tinel et al. 1998). Work by Kiehn et al. (1998) and Thomas et al. (1999) shows inhibition of HERG by PKA acting at sites conserved in mErgla. In NG108-15 cells at least, kinase effects may have been missed in the past because of opposite effects on the two channel types contributing to the M-like current. Combining selective block of one component with PKA activation is now a possibility.

Taking a broader view, M-currents in autonomic neurones and M-like currents in their tumour­ cell relatives are only the tip of the iceberg. The new frontier is, of course, the brain. Apart from the hippocampal M-current, M-like currents in brain neurones have been little studied as yet. Noda et al. (1998) found an M-like current in rat cortical pyramidal neurones which had a relatively high linopirdine IC 50 of 35.8 pM, giving grounds to suspect that an ERG or other linopirdine-insensitive channel type was present. With our present knowledge of blockers such as WAY-123,398, pharmacological dissection of brain M-like currents, as they emerge, should be a fruitful approach. 183

References

Abbott, GW, Sesti, F, Splawski, I, Buck, ME, Lehmann, MH, Timothy, KW, Keating, MT & Goldstein, SAN (1999). MiRPI forms iKr potassium channels with HERG and is associated with cardiac arrhythmia. Cell 97, 175-187.

Abitbol, I, Peretz, A, Lerche, C, Busch, AE & Attali, B (1999). Stilbenes and fenamates

rescue the loss of Iks channel function induced by an LQT5 mutation and other IsK mutants. E/WBOJ. 18, 4137-4148.

Accili, EA, Kiehn, J, Yang, Q, Wang, Z, Brown, AM & Wible, BA (1997). Separable Kvp subunit domains alter expression and gating of potassium channels. J. Biol. Chem. 272, 25824-25831.

Adams, PR, Brown, DA & Constanti, A (1982). M-currents and other potassium currents in bullfrog sympathetic neurones. J. Physiol. (Lond.) 330, 537-572.

Adda, S, Fleischmann, BK, Freedman, BD, Yu, M-F, Hay, DWP & Kotlikoff, Ml (1996). Expression and function of voltage-dependent potassium channel genes in human airway smooth muscle. J. Biol. Chem. 271,13239-13243.

Aiken, SP, Zaczek, R & Brown, BS (1996). Pharmacology of the neurotransmitter release enhancer linopirdine (DuP 996), and insights into its mechanism of action. Adv. Pharmacol. 35, 349-384.

Aiyar, J, Withka, JM, Rizzi, JP, Singleton, DH, Andrews, GO, Lin, W, Boyd, J, Hanson, DC, Simon, M, Dethlefs, B, Lee, C-L, Hall, JE, Gutman, GA & Chandy, KG (1995). Topology of the pore-region of a K^ channel revealed by the NMR-derived structures of scorpion toxins. Neuron 15,1169-1181

Amano, T, Richelsen, E & Nirenberg, M (1972). Neurotransmitter synthesis by neuroblastoma clones. Proa. Natl. Acad. Sci. USA 69, 258-263

Amano, T, Hamprecht, B & Kemper, W (1974). High activity of choline acetyltransferase induced in neuroblastoma x glia hybrid cells. Exp. Cell Res. 85, 399-408. 184

Arcangeli, A, Bianchi, L, Becchetti, A, Faravelli, L, Coronnello, M, Mini, E, Oiivotto, M & Wanke, E (1995). A novel inward-rectifying current with a cell-cycle dependence governs the resting potential of mammalian neuroblastoma cells. J. Physiol. (Lond.) 489, 455-471.

Attali, B, Wang, N, Kolot, A, Sobko, A, Cherepanov, V & Soliven, B (1997). Characterization of delayed rectifier Kv channels in oligodendrocytes and progenitor cells. J. Neurosci. 17, 8234-8245.

Barhanin, J, Lesage, F, Guillemare, E, Fink, M, Lazdunski, M & Romey, G (1996). KvLQTI and IsK (minK) proteins associate to form the Iks cardiac potassium current.Nature 384, 78-80.

Barry, DM, Trimmer, JS, Meriie, JP & Nerbonne, JM (1995). Differential expression of voitage-gated K^ channel subunits in adult rat heart: relation to functional K^ channels? Circ. Res. 77, 361 -369.

Bauer, CK, Meyerhof, W & Schwartz, JR (1990). An inward-rectifying K^ current in clonal rat pituitary cell and its modulation by thyrotropin-releasing hormone. J. Physiol. (Lond.) 429, 169-189.

Bauer, CK, Engeland, B, Wulfsen, I, Ludwig, J, Pongs, O & Schwartz, JR (1998). RERG is a molecular correlate of the inward-rectifying K current in clonal rat pituitary cells. Receptors Channeis 6,19-29.

Beech, DJ, Bernheim, L, Mathie, A & Hille, B (1991). Intracellular Ca^^ buffers disrupt muscarinic suppression of Ca^^ current and M current in rat sympathetic neurons. Proa.

Natl. Acad. Sci. USA 8 8 , 652-656.

Béhé, P & Meves,H (1992). Arachidonic acid has a biphasic effect on the M-current of NG108-15 neuroblastoma x glioma hybrid cells. J. Physiol. (Lond.) 446, 551 P.

Behe, P, Sandmeier, K & Meves, H (1992). The effect of arachidonic acid on the M-current of NG108-15 neuroblastoma x glioma hybrid ceiis. Pflügers Arch. 422,120-128.

Bianchi, L, Wible, B, Arcangeli, A, Tagiialateia, M, Morra, F, Castaldo, P, Crociani, O, Rosati, B, Faravelii, L, Oiivotto, M & Wanke, E (1998). /jergf encodes a K'^ current highly 185

conserved in tumors of different histogenesis: a selective advantage for cancer cells? Cancer Res. 58, 815-822.

Biervert, C, Schroeder, BC, Kubisch, C, Berkovic, SF, Propping, P, Jentsch, TJ & Steinlein, OK (1998). A potassium channel mutation in neonatal human epilepsy. Science 279, 403-406.

Bleich, M, Briei, M, Busch, AE, Lang, HJ, Gerlach, U, Gogeiein, H, Greger, R & Kunzelmann,

K (1997). KvLOT channels are inhibited by the channel blocker 293B. Pflügers Arch. 434, 499-501.

Bofill-Cardona, E, Vartian, N, Nanoff, C, Freissmuth, M & Boehm, 8 (2000). Two different signalling mechanisms involved in the excitation of rat sympathetic neurons by uridine nucleotides. Mol. Pharmacol. 57,1165-1172.

Brahmajothi, MV, Morales, MJ, Liu, 8 , Rasmussen, RL, Campbell, DL & Strauss, HO (1996). In situ hybridization reveals extensive diversity of channel mRNA in isolated ferret cardiac myocytes. Giro. Res. 78,1083-1089.

Brenman, JE, Chao, DS, Gee, SH, McGee, AW, Craven, SE, Santillano, DR, Wu, Z, Huang, F, Xia, H, Peters, MF, Froehner, SC & Bredt, DS (1996). Interaction of nitric oxide synthase with the postsynaptic density protein PSD-95 and al-syntrophin mediated by PDZ domains. Ce//84, 757-767.

Brown, DA & Adams, PR (1980). Muscarinic suppression of a novel voltage-sensitive \C current in a vertebrate neurone. Nature 283, 673-676.

Brown, DA & Selyanko, AA (1985). Membrane currents underlying the cholinergic slow excitatory post-synaptic potential in the rat sympathetic ganglion. J. Physiol. (Lond.) 365, 365-387.

Brown, DA & Higashida, H (1988a). Voltage- and calcium-activated potassium currents in mouse neuroblastoma x rat glioma hybrid cells. J. Physiol. (Lond.) 397, 149-165.

Brown, DA & Higashida, H (1988b). Membrane current responses of NG108-15 mouse neuroblastoma x rat glioma hybrid cells to bradykinin. J. Physiol. (Lond.) 397,167-184. 186

Brown, DA & Higashida, H (1988c). Inositol 1,4,5-trisphosphate and diacylglycerol mimic bradykinin effects on mouse neuroblastoma x rat glioma hybrid cells. J. Physiol. (Lond.) 397, 185-207.

Brown, DA, Marrion, NV & Smart, TG (1989). On the transduction mechanism for muscarine- induced inhibition of M-current in cultured rat sympathetic neurones. J. Physiol. (Lond.) 413, 469-488.

Brown, DA, Higashida, H, Noda, M, Ishizaka, N, Hashii, M, Hoshi, N, Yokoyama, S, Fukuda, K, Katayama, M, Nukada, T, Kameyama, K, Robbins, J, Marsh, SJ & Selyanko, AA (1993). Coupling of muscarinic receptor subtypes to ion channels: experiments on neuroblastoma hybrid cells. Ann. NY Acad. Sci. 7, 237-258.

Busch, AE, Herzer, T, Wagner, CA, Schmidt, F, Raber, G, Waldegger, S & Lang, F (1994a). Positive regulation by chloride channel blockers of Uk channels expressed inXenopus oocytes. Mol. Pharmacol. 46, 750-753.

Busch, AE, Malloy, K, Groh, WJ, Varnum, MD, Adelman, JP & Maylie, J (1994b). The novel class III antiarrhythmics NE-10064 and NE-10133 inhibit Isk channels expressed in Xenopus oocytes and Iks in guinea pig cardiac myocytes. Biochem. Biophys. Res.

Commun. 2 0 2 , 265-270.

Busch, AE, Suessbrich, H, Waldegger, S, Sailer, E, Greger, R, Lang, H-J, Lang, F, Gibson, KJ & Maylie, JG (1996). Inhibition of Iks in guinea pig cardiac myocytes and guinea pig IsK channels by the chromanol 293B. Pflügers Arch. 432,1094-1096.

Busch, AE, Busch, GL, Ford, E, Suessbrich, H, Lang, H-J, Greger, R, Kunzelmann, K, Attali, B & Stühmer, W (1997a). The role of the Uk protein in the specific pharmacological properties of the Iks channel complex. British Journal of Pharmacology 122,187-189.

Busch, AE, Suessbrich, H, Waldegger, S, Sailer, E, Greger, R, Lang, H-J, Lang, F, Gibson, KJ & Maylie, JG (1997b). Inhibition of Iks in guinea pig cardiac myocytes and guinea pig IsK channels by the chromanol 293B. Biophysical Journal 72, abstract M-Pos141.

Busch, AE, Eigenburger, B, Jurkiewicz, NK, Salata, JJ, Pica, A, Suessbrich, H & Lang, F (1998). Blockade of HERG channels by the class III antiarrhythmic azimilide: mode of action. Brit. J. Pharmacol. 123, 23-30. 187

Charlier, C, Singh, NA, Ryan, SG, Lewis, TB, Reus, BE, Leach, RJ & Leppert, M (1998). A pore mutation in a novei KQT-like potassium channei gene in an idiopathic epilepsy family. Nature Genet. 18, 53-55.

Chen, M-L, Hoshi, T & Wu, G-F (1996). Heteromultimeric interactions among K"' channei subunits from Shaker ar\6 eag families in Xenopus oocytes. Neuron 17, 535-542.

Chiesa, N, Rosati, B, Arcangeli, A, Oiivotto, M & Wanke, E (1997). A novei role for HERG channeis: spike-frequency adaptation.J. Physiol. (Lond.) 501, 313-318.

Coetzee, WA, Amarillo, Y, Chiu, J, Chow, A, Lau, D, McCormack, T, Moreno, H, Nadai, MS, Ozaita, A, Pountney, D, Saganich, M, Vega-Saenz de Miera, E & Rudy, B (1999).

Molecular diversity of channels. Ann. N Y Acad. Sci. 8 6 8 , 233-285.

Conforti, L & Miiihorn, DE (1997). Selective inhibition of a slow-inactivating voltage- dependent channel in rat PCI 2 ceils by hypoxia. J. Physiol. (Lond.) 502, 293-305.

Conforti, L, Bodi, I, Nisbet, JW & Miiihorn, D (2000). 0 2 -sensitive K"" channels: role of the Kvi .2 a-subunit in mediating the hypoxic response. J. Physiol. (Lond.) 524, 783-793.

Constanti, A & Brown, DA (1981). M-currents in voltage-clamped mammalian sympathetic neurones. Neurosci. Lett. 24, 289-294.

Cooper, EC, Aidape, KD, Abosch, A, Barbare, NM, Berger, MS, Peacock, WS, Jan, YN & Jan, LY (2000). Colocaiization and coassembiy of two human brain M-type potassium channel subunits that are mutated in epilepsy. Proc. Natl. Acad. Sci. USA 97, 4914- 1919.

Costa, AMN & Brown, BS (1997). Inhibition of M-current in cultured rat superior cervical ganglia by linopirdine: mechanism of action studies. NeuropharmacoiogyZ6, 1747-1753.

Cotton, J, Crest, M, Bouet, F, Aiessandri, N, Goia, M, Forest, E, Karisson, E, Castaneda, O, Harvey, AL, Vita, C & Menez, A (1997). A potassium channei from the sea anemone Bunodosoma granuiifera, an inhibitor for Kvi channels: Revision of the amino acid sequence, disulfide-bridge assignment, chemical synthesis, and biological activity. Eur. J. Biochem. 244,192-202. 188

Creighton, TE (1993) Proteins: Structures and molecular properties. 2^ edition. New York: Freeman.

Gruzbianca, H, Koh, D-S & Hiile, B (1998). Bradykinin inhibits M current via phospholipase 0 and Ca^^ release from IPa-sensiI Pa-sensitive Ca^^ stores in rat sympathetic neurons. Proc. Natl. Acad. Sci. USA 95, 7151-7156.

Curran, ME, Splawski, I, Timothy, KW, Vincent, GM, Green, ED and Keating, MT (1995). A molecular basis for cardiac arrhythmia: HERG mutations cause long QT syndrome. Cell 80, 795-803.

Daniels, MP & Hamprecht, B (1974). The ultrastructure of neuroblastoma glioma somatic cell hybrids. J Cell Biol. 63, 691 -699

Deal, KK, Lovinger, DM & Tamkun, MM (1994). The brain Kvl.1 potassium channel: In vitro and In vivo studies on subunit assembly and posttranslational processing. J. Neurosci. 14, 1666-1676 del Rio, E, Bevilacqua, JA, Marsh, SJ, Hailey, P & Caulfield, MP (1999). Muscarinic Ml receptors activate phosphoinositide turnover and Ca^^ mobilisation in rat sympathetic neurones, but this signalling pathway does not mediate M-current inhibition. J. Physiol. (Lond.) 520, 101-111.

Derst, C & Karschin, A (1998). Evolutionary link between prokaryotic and eukaryotic \C channels. J. Exp. Biol. 201, 2791-2799.

Dixon, JE & McKinnon, D (1996). Potassium channel mRNA expression in prevertebral and paravertebral sympathetic neurons. Eur. J. Neurosci. 8,183-191.

Ellingboe, JW, Spinelli, W, Winkley, MW, Nguyen, TT, Parsons, RW, Moubarak, IF, Kitzen, JM, Von Engen, D & Bagli, JF (1992). Class III antiarrhythmic activity of novel substituted 4-[(methylsulfonyl)amino]benzamides and sulfonamides. J. Med. Chem. 35, 705-716. 189

Faravelli, L, Arcangeli, A, Oiivotto, M & Wanke, E (1996). A HERG-like K"" channel in rat F- 11 DRG cell line: pharmacological identification and biophysical characterization. J. Physiol. (Land.) 496,13-23.

Filippov, AK, Selyanko, AA, Robbins, J & Brown, DA (1994). Activation of nucleotide receptors inhibits M-type K current [Ik(m)] in neuroblastoma x glioma hybrid cells. Pflügers Arch. 429, 223-230.

Filippov, AK, Kozlov, SA, Pluzhnikov, KA, Grishin, EV & Brown, DA (1996). M-type K+ current inhibition by a toxin from the scorpion Buthus eupeus. FEBS Lett. 384, 277-280.

Folander, K, Smith, JS, Antanavage, J, Bennett, C, Stein, RB & Swanson, R (1990). Cloning and expression of the delayed rectifier Uk channel from neonatal rat heart and diethylstilbestrol-primed rat uterus. Proc. Natl. Acad. Sci. USA 87, 2975-2979.

Gendeh, GS, Young, LG, de Medeiros, CL, Jeyaseelan, K, Harvey, AL & Chung, MC (1997). A new potassium channel toxin from the sea anemone Heteractis magnifica: isolation, cDNA cloning, and functional expression. Biochemistry 36,11461 -11471.

Gomperts, SN (1996). Clustering membrane proteins: it’s all coming together with the PSD- 95/SAP90 protein family. Ce//84, 659-662.

Greger, R (1990). Chloride channel blockers. Methods Enzymol. 191, 793-810.

Gribkoff, VK, Lum-Ragan, JT, Boissard, CG, Post-Munson, DJ, Meanwell, NA, Starrett, JE, Kozlowski, ES, Romine, JL, Trojnacki, JT, McKay, MC, Zhong, J & Dworetzky, SI (1996). Effects of channel modulators on cloned large-conductance calcium-activated potassium channels. Mol. Pharmacol. 50, 206-217.

Grissmer, S, Nguyen, AN, Aiyar, J, Hanson, DC, Mather, RJ, Gutman, GA, Karmilowicz, MJ, Auperin, DD & Chandy, KG (1994). Pharmacological characterization of five cloned voltage-gated K^ channels, types Kvi .1 ,1 .2 ,1 .3 ,1 .5 , and 3.1, stably expressed in mammalian cell lines. Mol. Pharmacol. 45: 1227-1234.

Guillemare, E, Honoré, E, Pradier, L, Lesage, F, Schweitz, H, Attali, B, Barhanin, J & Lazdunski, M (1992). Effects of the level of mRNA expression on biophysical properties. 190

sensitivity to neurotoxins, and regulation of the brain delayed-rectifier \C channel Kvi .2. Biochemistry 31,12463-12468.

Gurrola, GB, Rosati, B, Rocchetti, M, Pimienta, G, Zaza, A, Arcangeli, A, Oiivotto, M, Possani, LD & Wanke, E (1999). A toxin to nervous, cardiac, and endocrine ERG channels isolated from Centruroides noxius scorpion venom. FASEB J. 13, 953-962.

Hadley, JK, Noda, M, Selyanko, AA, Wood, 1C, Abogadie, FC & Brown, DA (2 0 0 0 ). Differential tetraethylammonium sensitivity of KGNQ1-4 potassium channels. Brit. J. Pharmacol. 129, 413-415.

Hamprecht, B & Schultz, J (1973). Stimulation by prostaglandin Ei of adenosine 3’: 5’-cyclic monophosphate formation in neuroblastoma cells in the presence of phosphodiesterase inhibitors. FEBS Lett. 34, 85-89.

Hamprecht, B (1977). Structural, electrophysiological, biochemical, and pharmacological properties of neuroblastoma-glioma cell hybrids in cell culture, int. Rev. Cytoi. 49, 99-170

Hart, PJ, Overturf, KE, Russell, SN, Carl, A, Hume, JR, Sanders, KM & Horowitz, B (1993). Cloning and expression of a Kyi .2 class delayed rectifier K^ channel from canine coionic smooth muscle. Proc. Nati. Acad. Sci. USA 90, 9659-9663.

Harvey, AL (1997). Recent studies on dendrotoxins and potassium ion channels. Gen. Pharmacol. 28, 7-12.

Higashida, H (1988). Acetylcholine release by bradykinin, inositol 1,4,5-trisphosphate and phorbol dibutyrate in rodent neuroblastoma cells. J. Physiol. (Lond.) 377, 209-222.

Hille, B (1984). Ionic channels of excitable membranes. Sunderland, Massachusetts: Sinauer Associates Inc.

Hodgkin, AL, Huxley, AF & Katz, B (1952). Measurement of current-voltage relations in the membrane of the giant axon of Loiigo. J. Physiol. (Lond.)^^6, 424-448.

Hodgkin, AL & Huxley, AF (1952a). Currents carried by sodium and potassium ions through - the membrane of the giant axon of Loiigo. J. Physiol. (Lond.)^^6, 449-472. 191

Hodgkin, AL & Huxley, AF (1952b). The components of membrane conductance in the giant axon of Loiigo. J. Physiol. fLonc/.j 116, 473-496.

Hodgkin, AL & Huxley, AF (1952c). A quantitative description of membrane current and its application to conduction and excitation in nerve. J. Physiol. (Lond.) AM, 500-544.

Horn, R & Marty, A (1988). Muscarinic activation of ionic currents measured by a new whole­ cell recording method. J. Gen. Physiol. 92,145-159.

Hu, Q, & Shi, YL (1997). Characterization of an inward-rectifying potassium current in NG108-15 neuroblastoma x glioma cells. Pflügers Arch. 433, 617-625.

Huang, X-Y, Morielli, AD & Peralta, EG (1993). Tyrosine kinase-dependent suppression of a potassium channel by the G protein-coupled ml muscarinic acetylcholine receptor. Cell 75,1145-1156.

Huang, X-Y, Morielli, AD & Peralta, EG (1994). Molecular basis of cardiac potassium channel stimulation by protein kinase A. Proc. Natl. Acad. Sci. USA 91, 624-628.

Hurst, RS, Busch, AE, Kavanaugh, MP, Osborne, PB, North, RA & Adelman, JP (1991). Identification of amino acid residues involved in dendrotoxin block of rat voltage- dependent potassium channels. Mol. Pharmacol. 40, 572-576.

Ito, Y, Yokoyama, S & Higashida, H (1992). Potassium channels cloned from neuroblastoma cells display slowly inactivating outward currents in Xenopus oocytes. Proc. R. Soc.

Lond. 8 248,95-101.

Kavanaugh, MP, Varnum, MD, Osborne, PB, Christie, MJ, Busch, AE, Adelman, JP & North, RA (1991). Interaction between tetraethylammonium and amino acid residues in the pore of cloned voltage-dependent potassium channels. J. Biol. Chem. 266, 7583-7587.

Kharrat, R, Mabrouk, K, Crest, M, Darbon, H, Cughideni, R, Martin-Eauclaire, M-F, Jacquet, G, El Ayeb, M, van Rietschoten, J, Rochat, H & Sabatier, J-M (1996). Chemical synthesis and characterization of maurotoxin, a short scorpion toxin with four disulfide bridges that acts on K^ channels. Eur. J. Biochem. 242, 491-498. 192

Kharrat, R, Mansuelle, P, Sampieri, F, Crest, M, Oughideni, R, van Rietschoten, J, Martin- Eauclaire, M-F, Rochat, H & El Ayeb, M (1997). Maurotoxin, a four disulfide bridge toxin from Scorpio maurus venom: purification, structure and action on potassium channels. FEBS Lett. 406, 284-290.

Kiehn, J, Karle, C, Thomas, D, Yao, X, Brachmann, J & Kübler, W (1998). HERG potassium channel activation is shifted by phorbol esters via protein kinase A-dependent pathways. J. Biol. Chem. 273, 25285-25291.

Kim, E, Niethammer, M, Rothschild, A, Jan, YN & Sheng, M (1995). Clustering of Shaker- type K^ channels by interaction with a family of membrane-associated guanylate kinases. Nature 378, 85-88.

Kim, E, Cho, K-0, Rothschild, A & Sheng, M (1996). Heteromultimerization and NMDA receptor-clustering activity of chapsyn-110, a member of the PSD-95 family of proteins. Neuron 17, 103-113.

Kim, E & Sheng, M (1996). Differential K^ channel clustering activity of PSD-95 and SAP97, two related membrane-associated putative guanylate kinases. Neuropharmacology 35, 993-1000.

Klee, WA & Nirenberg, M (1974). A neuroblastoma x glioma hybrid cell line with morphine receptors. Proc. Natl. Acad. Sci. USA 71, 3474-3477.

Krystosek, A (1985). Neurite formation by neuroblastoma-glioma hybrid cells (NG108-15) in defined medium: stochastic initiation with persistence of differentiated functions.J. Cell Physiol. 125, 319-329.

Kubisch, C, Schroeder, BC, Friedrich, I , Lütjohann, El-Amraoui, A, Marlin, S, Petit, C & Jentsch, TJ (1999). KCNQ4, a novel potassium channel expressed in sensory outer hair cells, is mutated in dominant deafness. Cell 96, 437-446.

Lamas, JA, Selyanko, AA & Brown, DA (1997). Effects of a cognition-enhancer, linopirdine (DuP 996) on M-type potassium currents (Ik(m)) and some other voltage- and ligand-gated membrane currents in rat sympathetic neurons. Eur. J Neurosci. 9, 605-616. 193

Lammers, CH, Schweitzer, P, Facchinetti, P, Arrang, JM, Madamba, SG, Siggins, GR & Piomelli, D (1996). Arachidonate 5-lipoxygenase and its activating protein: prominent

hippocampal expression and role in somatostatin signaling. J. Neurochem. 6 6 , 147-52.

Laube, G, Roper, J, Pitt, JG, Sewing, S, Kistner, U, Garner, GO, Pongs, O & Veh, RW (1996). Ultrastructural localization of Shaker-re\a\e6 potassium channel subunits and synapse- associated protein 90 to septate-like junctions in rat cerebellar pinceaux.Mol. Brain Res. 42, 51-61.

Lees-Miller, JP, Kondo, G, Wang, L & Duff, HJ (1997). Electrophysiological characterization of an alternatively processed ERG channel in mouse and human hearts. Circ. Res. 81,719-726.

Lerche, G, Scherer, GR, Seebohm, G, Derst, G, Wei, AD, Busch, AE & Steinmeyer, K (2000). Molecular cloning and functional expression of KGNQ5, a potassium channel subunit that may contribute to neuronal M-current diversity. J Bid. Chem., published electronically 27 April 2000.

Li, D, Takimoto, K & Levitan, ES (2000). Surface expression of Kvi channels is governed by a G-terminal motif. J. Bid. Chem. 275,11597-11602.

London, B, Trudeau, MG, Newton, KP, Beyer, AK, Gopeland, NG, Gilbert, DJ, Jenkins, NA, Satler, GA & Robertson, GA (1997). Two isoforms of the mouse Ether-a-go-go-re\ate6 gene coassemble to form channels with properties similar to the rapidly activating component of the cardiac delayed rectifier K^ current. Circ. Res. 81, 870-878.

Loussouarn, G, Gharpentier, F, Mohammad-Panah, R, Kunzelmann, K, Baro, I & Escande, D (1997). KvLQTI potassium channel but not Uk is the molecular target for trans-6-cyano- 4-(A/-ethylsulfonyl-A/-methylamino)-3-hydroxy-2,2-dimethyl-chromane. Md. Pharmacd. 52,1131-1136.

Ludwig, J, Terlau, H, Wunder, F, Brüggemann, A, Pardo, LA, Marquardt, A, Stühmer, W & Pongs, O (1994). Functional expression of a rat homologue of the voltage gated ether à go-go potassium channel reveals differences in selectivity and activation kinetics between the Drosophila channel and its mammalian counterpart. EMBOJ. 13, 4451- 4458. 194

MacKinnon, R & Yellen, G (1990). Mutations affecting TEA blockade and Ion permeation In voltage-activated \C channels. Science 250, 276-279.

Marrion, NV (1997). Does r-EAG contribute to the M-current? Trends Neurosci. 20, 243.

Marsh, SJ, Hubbard, A & Brown, DA (1990). Some actions of 9-amlno-1,2,3,4- tetrahydroacrldlne (THA) on cholinergic transmission and membrane currents In rat sympathetic ganglia. Eur. J. Neurosci. 2,1127-1134.

Marsh, SJ & Brown, DA (1991). Potassium currents contributing to action potential repolarlzatlon In dissociated cultured rat superior cervical sympathetic neurones. Neurosci. Lett. 133, 298-302.

Marty, A & Neher, E (1995). TIght-seal whole-cell recording. In Singie-channei recording, 2 nd edition, eds. Sakmann, B & Neher, E, pp. 31-52. New York: Plenum Press.

Mathie, A & Watkins, CS (1997). Is EAG the answer to the M-current? Trends Neurosci. 20, 14.

McGee, R, Simpson, P, Christian, 0, Mata, M, Nelson, P & Nirenberg, M (1978). Regulation of acetylcholine release from neuroblastoma x glioma hybrid cells. Proc. Nati. Acad. Sci. USA 75, 1314-1318.

McKinnon, D (1989). Isolation of a cDNA clone coding for a putative second potassium channel Indicates the existence of a gene family. J. Bid. Chem. 264, 8230-8236.

Meves, H, Schwartz, JR & Wulfsen, I (1999). Separation of M-llke current and ERG current In NG108-15 cells. Brit. J. Pharmacd. 127,1213-1223.

Meyer, R & Heinemann, SH (1998). Characterization of an eag-llke potassium channel In human neuroblastoma cells. J. Physioi. (Lond.) 508, 48-56.

Miller, C (1995). The charybdotoxin family of K^ channel-blocking peptides. Neuron 15, 5- 10.

Miller, C (1996). The Inconstancy of the human heart. Nature 379, 767-768. 195

Morielli, AD & Peralta, EG (1995). Suppression of a potassium channel by G-protein coupled

receptors. Life Sci. 56, pi 035, abstract 67.

Mullaney, I, Dodd, MW, Buckley, N & Milligan, G (1993). Agonist activation of transfected human M l muscarinic acetylcholine receptors in GHO cells results in down-regulation of

both the receptor and the a subunit of the G-protein Gq. Biochem. J. 2 8 9 , 125-131.

Nakamura, K & lijima, T (1994). Postnatal changes in mRNA expression of the channel in rat cardiac ventricles. Jpn. J. Pharmacol. 66, 489-492.

Noda, M, Ghana, M & Akaike, N (1998). Inhibition of M-type current by linopirdine, a neurotransmitter-release enhancer, in NG108-15 neuronal cells and rat cerebral neurons

in culture. Brain Res. 794, 274-280.

Numaguchi, H, Johnson, JP, Petersen, 01 & Baiser, JR (2000). A sensitive mechanism for

cation modulation of potassium current. Nature Neuroscience 3, 429-430.

Pan, X, Ikeda, SR & Lewis, DL (1996). Rat brain cannabinoid receptor modulates N-type Ca^"" channels in a neuronal expression system. Mol. Pharmacol. 49, 707-714.

Pancrazio, JJ, Ma, W, Grant, GM, Shaffer, KM, Kao, WY, Liu, Q-Y, Manos, P, Barker, JL & Stenger, DA (1999). A role for inwardly rectifying K"^ channels in differentiation of NG108-

15 neuroblastoma x glioma cells. J. Neurobiol. 38, 466-474.

PaulmichI, M, Nasmith, P, Hellmiss, R, Reed, K, Boyle, WA, Nerbonne, JM, Peralta, EG & Glapham, DE (1991). Gloning and expression of a rat cardiac delayed rectifier potassium channel. Proc. Nati. Acad. Sci. USA 88, 7892-7895.

Pennington, MW, Mahnir, VM, Krafte, DS, Zaydenberg, I, Byrnes, ME, Khaytin, I, Growley, K & Kem, WR (1996a). Identification of three separate binding sites on SHK toxin, a potent inhibitor of voltage-dependent potassium channels in human T-lymphocytes and rat

brain. Biochem. Biophys. Res. Commun. 219, 696-701.

Pennington, MW, Mahnir, VM, Khaytin, I, Zaydenburg, I, Byrnes, ME & Kem, WR (1996b). An essential binding surface for ShK toxin interaction with rat brain potassium channels.

Biochemistry 35, 16407-16411. 196

Po, S, Roberds, S, Snyders, DJ, Tamkun, MM & Bennett, PB (1993). Heteromultimeric assembly of human potassium channels: molecular basis of a transient outward current? Circ. Res. 72,1326-1336

Poling, JS, Vicini, S, Rogawski, MA & Salem, N (1996a). Docosahexaenoic acid block of neuronal voltage-gated channels: subunit selective antagonism by zinc. Neuropharmacology 35, 969-982.

Poling, JS, Rogawski, MA, Salem, N & Vicini, S (1996b). Anandamide, an endogenous cannabinoid, inhibits Shaker-re\aXed voltage-gated channels. Neuropharmacology 35, 983-991.

Pond, AL, Scheve, BK, Benedict, AT, Petrecca, K, Van Wagoner, DR, Shrier, A & Nerbonne, JM (2000). Expression of distinct ERG proteins in rat, mouse, and human heart. J. Biol. Chem. 275, 5997-6006.

Pongs, O (1992). Molecular biology of voltage-dependent potassium channels. Physiol. Rev. 72, S69-S88.

Rae, J, Cooper, K, Gates, P & Watsky, M (1991). Low access resistance perforated patch recordings using amphotericin B. J. Neurosci. Methods, 3 715-26. ,

Rehm, H, Pelzer, S, Cochet, C, Chambaz, E, Tempel, BL, Trautwein, W, Pelzer, D & Lazdunski, M (1989). Dendrotoxin-binding brain membrane protein displays a \C channel activity that is stimulated by both cAMP-dependent and endogenous phosphorylation. Biochemistry 23, 6455-6460.

Rhodes, KJ, Keilbaugh, SA, Barrezueta, NX, Lopez, KL & Trimmer, JS (1995). Association and colocalization of IC channel a- and p-subunit polypeptides in rat brain. J. Neurosci. 15, 5360-5371.

Robbins, J & Sim, JA (1990). A transient outward current in NG108-15 neuroblastoma hybrid cells. Pflügers Arch. 416,130-137.

Robbins, J, Trouslard, J, Marsh, SJ & Brown, DA (1992). Kinetic and pharmacological properties of the M-current in rodent neuroblastoma x glioma hybrid cells. J. Physiol. (Lond.) 451, 159-185. 197

Robbins, J, Marsh, SJ & Brown, DA (1993). On the mechanism of M-current inhibition by muscarinic ml receptors in DNA-transfected rodent neuroblastoma x glioma cells. J. Physiol. (Lond.) 469, 153-178.

Rogowski, RS, Collins, JH, O’Neill, TJ, Gustafson, TA, Workman, TR, Rogawski, MA, Tenenholz, TO, Weber, DJ & Blaustein, MP (1996). Three new toxins from the scorpion pandinus imperator selectively block certain voltage-gated\C channels. Mol. Pharmacol. 50:1167-1177.

Rouzaire-Dubois, B & Dubois, JM (1998). K'" channel block-induced mammalian neuroblastoma cell swelling: a possible mechanism to influence proliferation.J. Physiol. (Lond.) 510, 93-102.

Rudy, B (1988). Diversity and ubiquity of K channels. Neuroscience 25, 729-749.

Rundfeldt, 0 & Netzer, R (2000). The novel anticonvulsant retigabine activates M-currents in Chinese hamster ovary-cells transfected with human KCNQ2/3 subunits. Neurosci. Lett. 282, 73-76.

Russell, SN, Publicover, NG, Hart, PJ, Carl, A, Hume, JR, Sanders, KM & Horowitz (1994). Block by 4-aminopyridine of a Kvi .2 delayed rectifier \C current expressed in Xenopus oocytes. J. Physioi. (Lond.) ASA, 571-584.

Salata, JJ, Jurkiewicz, NK, Wang, J, Evans, BE, Orme, HT & Sanguinetti, MC (1998). A novel benzodiazepine that activates cardiac slow delayed rectifier \C currents. Moi. Pharmacol. 53, 220-230.

Sanguinetti, MC & Jurkiewicz, NK (1990). Two components of cardiac delayed rectifier K^ current: differential sensitivity to block by class III antiarrhythmic agents. J. Gen. Physioi. 96,195-215.

Sanguinetti, MC, Jiang, C, Curran, ME & Keating, MT (1995). A mechanistic link between an inherited and an acquired cardiac arrhythmia: HERG encodes the Ixr potassium channel. Ce//81, 299-307. 198

Sanguinetti, MC, Curran, ME, Zou, A, Shen, J, Specter, PS, Atkinson, DL & Keating, MT (1996). Coassembly of KvLQTI and minK (IsK) proteins to form cardiac Iks potassium channel. Nature 384, 80-83.

Sankaranarayanan, S & Simasko, SM (1996). Characterization of an M-like current modulated by thyrotropin-releasing hormone in normal rat lactotrophs. J. Neurosci. 16, 1668-1678

Schauf, CL, Moffett, DF & Moffett, SB (1990). Human physiology: foundations and frontiers. St. Louis, Missouri: Times Mirror/Mosby College Publishing.

Schmitt, H & Meves, H (1993). Protein kinase C as mediator of arachidonic acid-induced decrease of neuronal M current. Pflügers Arch. 425,134-139.

Schonherr, R, Rosati, B, Hehl, S, Rao, VG, Arcangeli, A, Oiivotto, M, Heinemann, SH & Wanke, E (1999). Functional role of the slow activation property of ERG K"" channels. Eur. J. Neurosci. 11, 753-760.

Schroeder, BC, Kubisch, C, Stein, V & Jentsch, TJ (1998). Moderate loss of function of cyclic-AMP-modulated KCNQ2/KCNQ3 K^ channels causes epilepsy. Nature 396, 687- 690.

Schroeder, BC, Hechenberger, M, Weinreich, F, Kubisch, C & Jentsch, TJ (2000). KCN05, a novel potassium channel broadly expressed in brain, mediates M-type currents. J. Biol. Chem., published electronically 17 May 2000.

Schwake, M, Pusch, M, Kharkovets, T & Jentsch, TJ (2000). Surface expression and single channel properties of KCNQ2/KCNQ3, M-type K^ channels involved in epilepsy. J. Biol. Chem. 275,13343-13348.

Schwartz, JR & Bauer, CK (1999). The ether-à-go-go-re\a\e6 gene \C current: functions of a strange inward rectifier. News Physiol. Sci. 14,135-142.

Schweitz, H, Bruhn, T, Guillemare, E, Moinier, D, Lancelin, J-M, Béress, L & Lazdunski, M (1995). Kalicludines and kaliseptine. J. Biol. Chem. 270, 25121-25126. 199

Schweitzer, P, Madamba, 8 & Siggins, GR (1990). Arachidonic acid metabolites as mediators of somatostatin-induced increase of neuronal M-current. Nature 346, 464-467.

Schweitzer, P, Madamba, S, Champagnat, J & Siggins, GR (1993). Somatostatin inhibition of hippocampal GA1 pyramidal neurons: mediation by arachidonic acid and its metabolites. J. Neurosci. 13, 2033-2049.

Schweitzer, P (2000). Cannabinoids decrease the M-current in hippocampal GA1 neurons. J. Neurosci. 20, 51-58.

Scott, VES, Parce], DN, Keen, JN, Findlay, JBG & Dolly, JO (1990). a-dendrotoxin acceptor from bovine brain is a channel protein. J. Bid. Chem. 265, 20094-20097.

Scott, VES, Muniz, ZM, Sewing, S, Lichtinghagen, R, Parce], DN, Pongs, O & Dolly, JO (1994a). Antibodies specific for distinct Kv subunits unveil a heterooligomeric basis for subtypes of a-dendrotoxin-sensitive K^ channels in bovine brain. Biochemistry 3 31617- , 1623

Scott, VE, Rettig, J, Parce], DN, Keen, JN, Findlay, JB, Pongs, O & Dolly, O (1994b). Primary structure of a p subunit of a-dendrotoxin-sensitive K^ channels from bovine brain. Proc. Nati. Acad. Sci. USA 91,1637-1641.

Selyanko, AA, Stansfeld, GE & Brown, DA (1992). Glosure of potassium M-channels by muscarinic acetylcholine-receptor stimulants requires a diffusible messenger. Proc. R. Soc. Lond. B 250,119-125.

Selyanko, AA, Robbins, J & Brown, DA (1995). Putative M-type potassium channels in neuroblastoma-glioma hybrid cells: inhibition by muscarine and bradykinin. Receptors Channeis 3,147-159.

Selyanko, AA & Brown, DA (1996). Intracellular calcium directly inhibits potassium m channels in excised membrane patches from rat sympathetic neurons. Neuron 16,151- 162.

Selyanko, AA, Hadley, JK, Wood, IG, Abogadie, FG, Delmas, P, Buckley, NJ, London, B & Brown, DA (1999). Two types of K^ channel subunit, Ergi and KGNQ2/3, contribute to the M-like current in a mammalian neuronal cell. J. Neurosci. 19, 7742-7756. 200

Selyanko, AA, Hadley, JK, Wood, 1C, Abogadie, FG, Jentsch, TJ & Brown, DA (2000). Inhibition of KCNQ1-4 potassium channels expressed in mammalian cells via Mi muscarinic acetylcholine receptors. J. Physiol. (Land) 522, 349-355.

Sewing, 8, Roeper, J & Pongs, O (1996). Kvpi subunit binding specific for Shaker-re\aXe6

potassium channel a subunits. Neuron 16, 455-463.

Shamotienko, OG, Parce], DN & Dolly, JO (1997). Subunit combinations defined for channel Kvi subtypes in synaptic membranes from bovine brain. Biochemistry 36, 8195- 8201.

Shahidullah, M, Hoshi, N, Yokoyama, S & Higashida, H (1995a). Microheterogeneity in heteromultimeric assemblies formed by Shaker{K\/^) and S/?aw(Kv3) subfamilies of voltage-gated channels. Proa. R. Soc. Lond. 6261, 309-317.

Shahidullah, M, Hoshi, T, Yokoyama, S, Kawamura, T & Higashida, H (1995b). Slow inactivation conserved in heteromultimeric voltage-dependent K"" channels between Shaker (Kvi ) and Shaw (Kv3) subfamilies. FEBS Lett. 371, 307-310

Shapiro, MS, Roche, JP, Kaftan, EJ, Cruzblanca, H, Mackie, K & Hille, B (2000). reconstitution of muscarinic modulation of the KCNQ2/KGNQ3 IC channels that underlie the neuronal M current. J. Neurosci. 20,1710-1721.

Shen, NV & Pfaffinger, PJ (1995). Molecular recognition and assembly sequences involved in the subfamily-specific assembly of voltage-gated IC channel subunit proteins. Neuron 14, 625-633.

Sheng, M, Liao, YJ, Jan, YN & Jan, LY (1993). Presynaptic A-current based on heteromultimeric K^ channels detectedin vivo. Nature 365, 72-75

Sheng, M, Tsaur, M-L, Jan, YN & Jan, LY (1994). Gontrasting subcellular localization of the Kvi .2 K^ channel subunit in different neurons of rat brain. J. Neurosci. 14, 2408-2417.

Shi, G, Nakahira, K, Hammond, S, Rhodes, KJ, Schechter, LE & Trimmer, JS (1996). p Subunits promote K^ channel surface expression through effects early in biosynthesis. Neuron 16, 843-852. 201

Shi, W, Wymore, RS, Wang, H-S, Pan, Z, Cohen, IS, McKinnon, D & Dixon, JE (1997). Identification of two nervous system-specific members of the erg potassium channel gene family. J. Neurosci. 17, 9423-9432.

Shi, W, Wang, H-S, Pan, Z, Wymore, RS, Cohen, IS, McKinnon, D & Dixon, JE (1998). Cloning of a mammalian e//c potassium channel gene and EAG mRNA distribution in rat sympathetic ganglia. J. Physiol. (Lond.)5^^, 675-682

Simmons, ML & Chavkin, C (1996). K-opioid receptor activation of a dendrotoxin-sensitive potassium channel mediates presynaptic inhibition of mossy fiber neurotransmitter release. Mol. Pharmacoi. 50, 80-85.

Singh, NA, Charlier, C, Stauffer, D, DuPont, BR, Leach, RJ, Melis, R, Ronen, GM, Bjerre, I, Quattlebaum, T, Murphy, JV, McHarg, ML, Gagnon, D, Rosales, TO, Peiffer, A, Anderson, VE & Leppert, M (1998). A novel potassium channel gene, KCNQ2, is mutated in an inherited epilepsy of newborns. Nature Genet. 18, 25-29.

Southan, AP & Robertson, B (1998). Patch-clamp recordings from cerebellar basket cell bodies and their presynaptic terminals reveal an asymmetric distribution of voltage-gated potassium channels. J. Neurosci. 18 , 948-955.

Spinelli, W, Moubarak, IF, Parsons, RW & Colatsky, TJ (1993). Cellular electrophysiology of WAY-123,398, a new class III antiarrhythmic agent: specificity of Ik block and lack of reverse use dependence in cat ventricular myocytes. Cardiovasc. Res. 27, 1580-1591.

Sprunger, LK, Stewig, NJ & O’Grady, SM (1996). Effects of charybdotoxin on K"" channel (Kvi .2) deactivation and inactivation kinetics. Eur. J. Pharmacol. 314, 357-364.

Stansfeld, CE, Marsh, SJ, Gibb, AJ & Brown, DA (1993). Identification of M-channeis in outside-out patches excised from sympathetic ganglion cells. Neuron 10, 639-654.

Stansfeld, C, Ludwig, J, Roeper, J, Weseloh, R, Brown, D & Pongs, O (1997). A physiological role for ether-a-go-go K^ channels? Trends Neurosci. 20,13-14. 202

Steinlein, OK, Stoodt, J, Biervert, C, Janz, D & Sander, T (1999). The voltage gated potassium channel KCNQ2 and idiopathic generalized epilepsy. Neuroreport AO, 1163- 1166.

Stoffel, M & Jan, LY (1998). Epilepsy genes: excitement traced to potassium channels. Nature Genet. 18, 6-8.

Stühmer, W, Ruppersberg, JP, Schroter, KH, Sakmann, B, Stocker, M, Giese, KP, Perschke, A, Baumann, A & Pongs, O (1989). Molecular basis of functional diversity of voltage- gated potassium channels in mammalian brain. EMBO J. 8, 3235-3244.

Taglialatela, M, Castaldo, P, Pannaccione, A, Giorgio, G & Annunziato, L (1998). Human ether-a-gogo related gene (HERO) K"^ channels as pharmacological targets. Biochem. Pharmacol. 55,1741-1746.

Tang, C-Y, Schulteis, CT, Jimenez, RM & Papazian, DM (1998). Shaker and ether-a-go-go channel subunits fail to coassemble in Xenopus oocytes. Biophys. J. 75, 1263-1270.

Thomas, D, Zhang, W, Karle, CA, Kathofer, S, Schols, W, Kübler, W & Kiehn, J (1999). Deletion of protein kinase A phosphorylation sites in the HERG potassium channel inhibits activation shift by protein kinase A. J. Biol. Chem. 274, 27457-27462.

Tiffany, AM, Manganas, LN, Kim, E, Hsueh, Y-P, Sheng, M & Trimmer, JS (2000). PSD-95 and SAP97 exhibit different mechanisms for regulating \C channel surface expression and clustering. J. Cell Biol. 148, 147-157.

Tinel, N, Lauritzen, I, Chouabe, 0, Lazdunski, M & Borsotto, M (1998). The KGNQ2 potassium channel: splice variants, functional and developmental expression. Brain localization and comparison with KGNQ3. FEBS Lett. 438,171-176.

Titus, SA, Warmke, JW & Ganetzky, B (1997). The Drosophila erg \C channel polypeptide is encoded by the seizure locus. J. Neurosci. 17, 875-881.

Traber, J, Fischer, K, Latzin, S & Hamprecht, B (1975a). Morphine antagonises action of prostaglandin in neuroblastoma and neuroblastoma x glioma hybrid cells. Nature 253,

120- 122. 203

Traber, J, Fischer, K, Buchen, C & Hamprecht, B (1975b). Muscarinic response to acetylcholine in neuroblastoma x glioma hybrid cells. Nature 255, 558-560.).

Trudeau, MC, Warmke, JW, Ganetzky, B & Robertson, GA (1995). HERG, a human inward rectifier in the voltage-gated potassium channel family. Science 269, 92-95.

Tsaur, M-L, Sheng, M, Lowenstein, DH, Jan, YN & Jan, LY (1992). Differential expression of channel mRNAs in the rat brain and down-regulation in the hippocampus following seizures. Neuron 8,1055-1067.

Tytgat, J, Debont, T, Carmeliet, E & Daenens, P (1995). The a-dendrotoxin footprint on a mammalian potassium channel. J. Biol. Chem. 270, 24776-24781.

Uebele, VN, England, SK, Chaudhary, A, Tamkun, MM & Snyders, DJ (1996). Functional differences in Kvi .5 currents expressed in mammalian cell lines are due to the presence of endogenous Kvp2.1 subunits. J. Bioi. Chem. 271, 2406-2412.

Veh, RW, Lichtinghagen, R, Sewing, S, Wunder, F, G rum bach, IM & Pongs, O (1995). Immunohistochemical localization of five members of the Kvi channel subunits: contrasting subcellular locations and neuron-specific co-localizations in rat brain. Eur. J. Neurosci. 7, 2189-2205.

Wang, H, Kunkel, DD, Martin, TM, Schwartzkroin, PA & Tempel, BL (1993). Heteromultimeric K^ channels in terminal and juxtaparanodal regions of neurons. Nature 365, 75-79

Wang, H, Kunkel, DD, Schwartzkroin, PA & Tempel, BL (1994). Localization of Kvl.1 and Kvi .2, two K channel proteins, to synaptic terminals, somata, and dendrites in the mouse brain. J. Neurosci. 14, 4588-4599.

Wang, Q, Curran, ME, Splawski, I, Burn, TO, Millholland, JM, VanRaay, TJ, Shen, J, Timothy, KW, Vincent, GM, de Jager, T, Schwartz, PJ, Towbin, JA, Moss, AJ, Atkinson, DL, Landes, GM, Connors, TD & Keating, MT (1996a). Positional cloning of a novel potassium channel gene: KVLQT1 mutations cause cardiac arrhythmias. Nature Genet. 12,17-23. 204

Wang, S, Morales, MJ, Liu, S, Strauss, HC & Rasmussen, RL (1996b). Time, voltage and ionic concentration dependence of rectification of h-erg expressed inXenopus oocytes. FEBS Lett. 3S9, 167-173.

Wang, Z, Kiehn, J, Yang, Q, Brown, AM & Wible, BA (1996c). Comparison of binding and

block produced by alternatively spliced Kvpi subunits. J. Biol. Chem. 271, 28311-28317.

Wang, S, Liu, 8, Morales, MJ, Strauss, HO & Rasmussen, RL (1997a). A quantitative analysis of the activation and inactivation kinetics of HERG expressed in Xenopus oocytes. J. Physiol. (Lond.) 502, 45-60.

Wang, XJ, Reynolds, ER, Déak, P & Hall, LM (1997b). The seizure locus encodes the Drosophila homolog of the HERG potassium channel. J. Neurosci. 17, 882-890.

Wang, H-S, Pan, Z, Shi, W, Brown, BS, Wymore, RS, Cohen, IS, Dixon, JE & McKinnon, D (1998a). KCNQ2 and KCNQ3 potassium channel subunits: molecular correlates of the M-channel. Science 282,1890-1893.

Wang, H-S, Shi, W, Brown, BS, Costa, AMN, Zhan, J, Wymore, RS, Cohen, IS, Dixon, JE & McKinnon, D (1998b). The KQT2 channel is a molecular correlate of the M-channel in sympathetic neurons. Society for Neuroscience Abstracts, pi 984, 792.1.

Wang, H, Shi, H & Wang, Z (1999). Inactivation block of the HERG human cardiac K^ channels by RP58866. Brit. J. Pharmacoi. 127, 1899-1907.

Warmke, JW & Ganetzky, B (1994). A family of potassium channel genes related to eag in Drosophila and mammals. Proc. Natl. Acad. Sci. USA 91, 3438-3442.

Watanabe, H, Nagata, E, Kosakai, A, Nakamura, M, Yokoyama, M, Tanaka, K & Sasai, H (2000). Disruption of the epilepsy KCNQ2 gene results in neural hyperexcitability. J. Neurochem. 75, 28-33.

Watson, JD, Gilman, M, Witkowski, J & Zoller, M (1992). Recombinant DMA, 2"'* edition. New York: Scientific American Books.

Wei, A, Jegla, T & Salkoff, L (1996). Eight potassium channel families revealed by the C. e/egans genome project. Neuropharmacoiogy 35, 805-829 205

Weinreich, F & Jentsch, TJ (2000). Neurological diseases caused by ion-channel mutations. Curr. Opin. Neurobiol. 10, 409-415.

Weinsberg, F, Bauer, CK & Schwartz, JR (1997). The class III antiarrhythmic agent E-4031 selectively blocks the inactivating inward-rectifying potassium current in rat anterior pituitary tumor cells (GHs/Bg cells). Pflügers Arch. 434,1 -10.

Werkman, TR, Kawamura, T, Yokoyama, 8, Higashida, H & Rogawski, MA (1992). Charybdotoxin, dendrotoxin and mast cell degranulating peptide block the voltage- activated current of fibroblast cells stably transfected with NGK1 (Kv1.2) channel complementary DMA. Neuroscience 50, 935-946.

Werkman, TR, Gustafson, TA, Rogowski, RS, Blaustein, MP & Rogawski, MA (1993). Tityustoxin-Ka, a structurally novel and highly potent channel peptide toxin, interacts

with the a-dendrotoxin binding site on the cloned Kvi .2 [C channel. Mol. Pharmacol. 44: 430-436.

Wymore, RS, Gintant, GA, Wymore, RT, Dixon, JE, McKinnon, D & Cohen, IS (1997). Tissue and species distribution of mRNA for the iKr-like K^ channel,erg. Circ. Res. 80, 261-268.

Xu, J, Yu, W, Jan, YN, Jan, LY & Li, M (1995). Assembly of voltage-gated potassium channels. J. Biol. Chem. 270, 24761 -24768

Xu, H, Dixon, JE, Barry, DM, Trimmer, JS, Merlie, JP, McKinnon, D & Nerbonne, JM (1996).

Developmental analysis reveals mismatches in the expression of\C channel a subunits and voltage-gated K^ channel currents in rat ventricular myocytes. J. Gen. Physiol. 108, 405-419.

Yamagishi, T, Ishii, K & Taira, N (1993). Absence of effects of class III antiarrhythmic agents on cloned cardiac K channels. Jpn. J. Pharmacol. 61, 371-373.

Yang, W-P, Levesque, PC, Little, WA, Conder, ML, Shalaby, FY & Blanar, MA (1997). KvLQTI, a voltage-gated potassium channel responsible for human cardiac arrhythmias. Proc. Natl. Acad. Sci. USA 94, 4017-4021. 206

Yang, W-P, Levesque, PC, Little, WA, Conder, ML, Ramakrishnan, P, Neubauer, MG & Blanar, MA (1998). Functional expression of two KvLQTI-related potassium channels responsible for an inherited idiopathic epilepsy. J. Bioi Chem. 273,19419-19423.

Yokoyama, S, Imoto, K, Kawamura, T, Higashida, H, Iwabe, N, Miyata, T & Numa, S (1989). Potassium channels from NG108-15 neuroblastoma-glioma hybrid cells. FEBS Lett. 259, 37-42.

Yokoyama, S, Kawamura, I , Ito, Y, Hoshi, N, Enomoto, K-l & Higashida, H (1993). Potassium channels cloned from NG108-15 neuroblastoma-glioma hybrid cells. Ann. NY Acad. Sci 707, 60-73).

Yu, W, Xu, J & Li, M (1996). NAB domain is essential for the subunit assembly of both a-a

and a-p complexes of Shaker-like potassium channels.Neuron 16, 441-453.

Zaczek, R, Chorvat, RJ, Saye, JA, Pierdomenico, ME, Maciag, CM, Logue, AR, Fisher, BN, Rominger, DH & Earl, RA (1998). Two new potent neurotransmitter release enhancers, 10,10-bis(4-pyridinylmethyl)-9(1 OH)-anthracenone and 10,10-bis(2-fluoro-4- pyridinylmethyl)-9(10H)-anthracenone: comparison to linopirdine. J. Pharmacoi Exp. Ther. 285, 724-730.

Zhou, B-Y, Ma, W & Huang, X-Y (1998). Specific antibodies to the external vestibule of voltage-gated potassium channels block current.J. Gen. Physiol. Ill, 555-563.

Zou, A, Curran, ME, Keating, MT & Sanguinetti, MC (1997). Single HERG delayed rectifier K^ channels expressed in Xenopus oocytes. Am. J. Physioi 272, H1309-H1314. 0259 Journal o f Physiology (2000), 522.3, j?p.349-355 3 4 9

Rapid Report

Inhibition of KCNQl-4 potassium channels expressed in mammalian cells via muscarinic acetylcholine receptors

A. A. Selyanko, J. K. Hadley, I. C. Wood* R C. Abogadie* T. J. Jentschf and D. A. Brown

Department o f Pharmacology, * Wellcome Laboratory fo r Molecular Pharmacology, University College London, Gower Street, London WCIE 6BT, U K and ^Zentrum f u r Molekulare Neurobiologie Hamburg {ZMNH), Universitat Hamburg, Martinistrasse 85, D -20246 Hamburg, Germany

(Received 22 October 1999; accepted after revision 6 December 1999)

1. KCNQl-4 potassium channels were expressed in mammalian Chinese hamster ovary (CHO) cells stably transfected with muscarinic acetylcholine receptors and currents were recorded using the whole-cell perforated patch technique and cell-attached patch recording.

2. Stimulation of receptors by 10 /fM oxotremorine-M (Oxo-M) strongly reduced (to 0-10% ) currents produced by KCNQl-4 subunits expressed individually and also those produced by KCNQ2+KCNQ3 and KCNQl+KCNEl heteromers, which are thought to generate neuronal M-currents (/k ,m) cardiac slow delayed rectifier currents (fg j, respectively.

3. The activity of KCNQ2-I-KCNQ3, KCNQ2 and KCNQ3 channels recorded with cell-attached pipettes was strongly and reversibly reduced by Oxo-M applied to the extra-patch membrane.

4. It is concluded that Mj receptors couple to all known KCNQ subunits and that inhibition of KCNQ2+KCNQ3 channels, like that of native M-channels, requires a diffusible second messenger.

KCNQl-4 are potassium channels abundantly expressed in KCNQ4 current has recently been identified in mouse outer various tissues. KCNQl has a high level of expression in the hair cells (Marcotti & Kros, 1999). However, sensitivity of heart where it co-assembles with minK (KCNEl) channels KCNQ4 currents to muscarinic receptor stimulation is to give a slow delayed rectifier type current Barhanin unknown. et al. 1996; Sanguinetti et al. 1996; Yang et al. 1997). The present experiments were aimed at establishing whether Mutations in KCNQl contribute to the long QT syndrome the muscarinic inhibition of KCNQ2+KCNQ3 channels which causes cardiac arrhythmia and sudden death (i) involves one of these subunits or both; (ii) is membrane- (Sanguinetti, 1999). KCNQ2 and KCNQ3 are expressed delimited or requires a diffusible second messenger; and exclusively in the nervous system. Mutations in either (iii) can be extended to KCNQl and KCNQ4 subunits which KCNQ2 or KCNQ3 result in a juvenile epilepsy (Biervertet have shorter carboxy termini and, therefore, a reduced a l 1998; Charlier et a l 1998; Singh et a l 1998; Lerche et a l r. r j. j.- ^ ^ j. 1999). KCNQ4 ia expressed in cochlea and brain and its mutations cause deafness (Kubisch et a l 1999).

Heteromeric KCNQ2+KCNQ3 channels probably constitute METHODS the M-channel in sympathetic ganglia (Wang et a l 1998). culture and transfection M-channels control neuronal excitability (Brown, 1988; The procedure for Chinese hamster ovary (CHO) hm l cell culture Marrion, 1997) and can be inhibited via Mj muscarinic transfection has been described previously (Selyanko et a l aceWlcholine receptors (Marrion cl al. 1989; Bemheim cl al. ™ «

heteromultimers equal amounts of KCNQ2 and KCNQ3 cDNAs intrapipette anions are permeant through the pores and all intra­ were used. Cells expressing channels were identified by adding cellular anions impermeant, the Donnan potential would be

CD8 -binding Dynabeads (Dynal A/S) before recording. < 1 1 mV (see Horn & Marty, 1988). Pipette resistance was 2-3 M il and the series resistance was compensated (60-90%). Whole-cell perforated patch recording Recordings were made at room temperature (20-22 °C) 1-4 days Cell-attached single-channel recording after transfection (longer times were required to obtain sufficiently Single-channel recordings were made 2-4 days after transfection, large KCNQ2 and KCNQ3 currents). Cells were bath perfused with when the density of channels was enhanced. The methods of

the following solution (mn): 144 NaCl, 2-5 KCl, 2 CaClg, 0-5 MgClj, recording and analysis were similar to those previously employed 5 Hepes, 10 glucose, pH 7*4 with Tris base. Pipettes were filled for studying the M-channels (Selyanko & Brown, 1999). with an internal solution containing (mM): 80 potassium acetate, 30 Intrapipette solution had the same composition as the bath solution

KCl, 40 Hepes, 3 MgClj, 3 EGTA, 1 CaClj, pH 7'4 with KOH. The used for whole-cell recording. In cell-attached experiments bath liquid junction potential was small (4 mV; Selyanko et al. 1995) and solutions had elevated (25 mM) K^ to pre-depolarise the cell at so no correction was made. Amphotericin B was used to perforate 30 mV. Records were sampled at 4-46 kHz and filtered at the patch (Rae et al. 1991). Perforated patch recording leads to an O'5-l kHz. Pipette resistance was 2-3 Mfl. additional undischarged Donnan potential due to the impermeability In patches containing single channels, the effects of muscarinic of amphotericin pores to large intracellular anions. Assuming all stimulation on single-channel activity were deduced from changes

KCNQ1

200 ms

KCNQ2

m s a sm

200 ms 20 ms

KCNQ3

400 ms 40 ms

KCNQ2+KCNQ3

|500pA

200 ms I

20ms

KCNQ4

200 ms

20 ms

Figure 1 . Activation (left panel, onsets in boxes) and deactivation (right panel) of homomeric (K C N Q l-4) and heteromeric (KONQ2-j-KCNQ3) channels expressed in CHO cells Currents were activated by long (1-2 s; 0'06 Hz) depolarising pulses from a holding potential of —70 mV, and deactivated by 1-2 s hyperpolarising pulses from the full activation potential of -1-50 mV (see top records for the voltage protocols). J. Physiol. 622.^ Modulation o f KC N Q l-4 channels 351 in the open probability while the multi-channel activity was Drugs and chemicals measured as the mean current through these channels. cDNAs for KCNQl, K C N E l, KCNQ2 and KCNQ4 were human and Data acquisition and analysis KCNQ3 cDNA was from the rat. Oxo-M (oxotremorine methiodide) was obtained from R B I (Natick, MA, USA). All other drugs and Data were acquired and analysed using pCLAMP software (version chemicals were obtained from Sigma or BDH (Poole, UK). 6.0). Currents were recorded using an Axopatch 200A (or 200) patch-clamp amplifier, filtered at 1 kHz and digitised at 4 -8 kHz. Activation curves were fitted by the Boltzmann equation: RESULTS 7/7(50) = 1/(1 4- exp(Fo.s - V J/k), Activation of KCNQl-4 channels where 7 is the tail current recorded at —70 mV following a step to Figure 1 shows examples of currents produced by different membrane potential 7(50) is current following a step to KCNQ subunits expressed in CHO cells. All currents had 4-50 mV, and Vq.^ is the membrane potential at which 7 is equal to low thresholds ('^-60 mV), slow and sustained activations 0*57(50). Current size and its inhibition by a muscarinic stimulant and slow deactivations. Activation curves for homomeric was measured from the amplitude of the deactivation tail recorded (KCNQl, KCNQ2, KCNQ3, KCNQ4) and heteromeric at —50 mV. The program ‘Origin’ (version 5.0, Microcal Software Inc.) was used for fitting activation curves and for creating the (KCNQ24-KCNQ3) channels (Fig. 2) were fitted by the figures. Data are shown as means ±s .e .m . unless otherwise stated. Boltzmann equation at l^.g and k, respectively, equal to

KCNQ1 KCNQ4

o o to LO

0.5 0.5

0.0 0.0 -50 0 50 -50 0 50 (mV) FJmV)

KCNQ2 KCNQ3

§

0.5 0.5

0.0 -50 0 50 -50 0 50 Kn (mV) V'JmV)

KCNQ2+KCNQ3

olO

0.5

0.0 ■50 0 50 FJmV)

Figure 2. Activation curves for KCNQl-4 and KCNQ24-KCNQ3 channels The activation curves were obtained as described in Methods and fitted by the Boltzmann equation with

parameters indicated in Results. Error bars indicates .e .m . 352 A, A. Selyanko and others J. Physiol.522.^

-7-8 ± 1-2 mV and 10-0 ± M mV, KCNQl (number of Ml receptor inhibition of KCNQl-4 channels cells, n = 6); -13-8 ± 0*4 mV and 12-1 ± 0-4 mV, KCNQ2 Figure 3+ shows that stimulation of M^ receptors by oxo­ (n = 5); -36-8 ± 1-3 mV and 5-5 ± 1-1 mV, KCNQ3 (n = 3); tremorine-M (Oxo-M) strongly reduced the KCNQ2+KCNQ3 -18-6 ± 0-3 mV and 9-8 ± 0-2 mV, KCNQ4 (n = 7); and current (on average, to 11-6 + 11-4%, w = 3). When -17-7 ± 0-6 mV and 11-9 +0-5 mV, KCNQ2+KCNQ3 recorded in cell-attached patches, the activity of (n = 6). Previously reported half-activation voltages for KCNQ2+KCNQ3 channels was reduced by adding Oxo-M KCNQl (-24 mV, Chouabe of. 1997), KCNQ2 (-37 mV, to the bath (see Fig. average inhibition to 6-4 + 2-8%; Biervertet al. 1998), KCNQ4 (-10 mV, Kubisch et a l 1999) 8 patches). and KCNQ2+KCNQ3 (-40 mV, Wang et a i 1998) channels differed somewhat from those obtained in the present Figure 4 shows that KCNQl-4 subunits expressed experiments. The uncorrected junction potential of ~4 mV individually can also be inhibited by Oxo-M, to 3-0 + 2-9% and (uncertain) Donnan potential in our perforated-patch (n = 5), 9-8+ 1-1% (%=3), 0-0 ±0-0% (n = 6) and recording (see Methods) might contribute to some of the more 3-9 ±3-9% {n = 5), in subunits 1-4, respectively. When positive half-activation voltages we have obtained, but this recorded in cell-attached patches, the activities of KCNQ2 would not explain all the differences. These may therefore and KCNQ3 channels were inhibited by Oxo-M added to the be associated with the different expression systems used. bath, to 2-2 ± 2-0% (5 patches) and 5-3 ± 4-5% (3 patches), respectively (not shown). The current onsets showed sigmoidal activation (in KCNQ4 only at very positive potentials) and all except KCNQ4 Compared with homomeric KCNQl channels, heteromeric channels showed evidence of some inactivation at positive (‘cardiac’) KCNQl+KCNEl channels produced larger membrane potentials: they either slowly declined during currents which had a lower threshold and slower activation continuous membrane depolarisation (KCNQ2 and KCNQ3) kinetics (see also Barhanin et a l 1996; Sanguinetti et a l or ‘saturated’ with increased depolarization (KCNQl and 1996; Yang et a l 1997). These KCNQl+KCNEl currents KCNQ2+KCNQ3). Hyperpolarising voltage steps from the were also abolished by Oxo-M (ri = 3; see Fig. 4). full activation potential of +50 mV removed inactivation All effects of Oxo-M on KCNQl-4 channels were sustained and increased the currents. and slowly reversible.

KCNQ2+KCNQ3 J Whole-cell (perforated patch) +OX0-M (10/iM) ■“oL ±Oxo-M

B

\2 Ca^A Cell- ■attached V5K7 W 25iK* +OXO-M (10 /

h pA

Figure 3. Inhibition of heteromeric KGNQ2+K0NQ3 channels expressed in CHO hml cells via stimulation of Mj muscarinic acetylcholine receptors A, macroscopic KCNQ2-I-KCNQ3 currents were recorded in response to a deactivation voltage protocol (top) before (left), during (centre) and 40 min after (right) addition of the muscarinic stimulant oxotremorine-M (Oxo-M, 10/^m) to the bath. Dashed lines denote zero-current level. B, multi-channel activities were recorded with a cell-attached pipette in response to the equivalent deactivation voltage protocol (top) before (left), during (centre) and 40 min after (right) addition of the muscarinic stimulant oxotremorine-M (Oxo-M, 10 /im) to the bath. The cell was pre-depolarised by using 25 mM bath solution, and the membrane potential in the patch (F„, = —10 mV; Is step to —30 mV) was calculated assuming the cellwas ‘voltage- clamped’ at ~ -3 0 mV. J. Physiol. 6 2 2 .i Modulation o f KC N Q l-4 channels 353

DISCUSSION Oxo-M (88%, present experiments) was not different from The idea that KCNQ2+KCNQ3 channels form M-channels is that for the M-currents in mammalian sympathetic neurones based on the overlapping distributions of KCNQ2-, KCNQ3- (~90%; Hamilton et al. 1997; Haley et a l 1998). and M-channels in the nervous system and the similarities The present experiments showed that homomeric KCNQ2 of the biophysical and pharmacological properties of their and KCNQ3 channels can also be inhibited via activation of macroscopic currents (Wang et al. 1998). The present Ml muscarinic receptors, and also through a diffusible experiments give additional new support to this idea. Thus, second messenger. This suggests that both subunits forming KCNQ2-1-KCNQ3 channels were inhibited via activation of the M-channel couple to Mi muscarinic receptors, probably Mj receptors (as previously reported in oocytes by Wang et through the same mechanism. al. 1998); and furthermore, cell-attached recording showed Two other members of the subfamily, KCNQl and KCNQ4, that, like native M-channels (Selyanko et al. 1992), this have not previously been considered as contributing to inhibition required a ditfusible second messenger. Since in M-channels largely because one of them, KCNQl, is seen as CHO cells Ml receptors couple to the G protein Gq (Mullaney a ‘cardiac’ channel, with low expression in the brain (Yang et et al. 1993), Mi-induced inhibition of KCNQ2-I-KCNQ3 a l 1997). The other, KCNQ4, has not yet been extensively channels, like that of native ganglionic M-channels (Haley et studied, and its distribution in the brain remains to be al. 1998), may be mediated by the subunit. Interestingly, determined (Kubisch et a l 1999). The results obtained in the average inhibition of KCNQ2-1-KCNQ3 currents by 10 /tM present experiments with expressed KCNQl and KCNQ4

I l l ' L l +Oxo-M (10/tM) KCNQl

20 pA

KCNQl+KCNE1

1100 pA

KCNQ2

20 pA

KCNQ3

KCNQ4

|20pA

Figure 4. Inhibition of KCNQl-4 channels via stimulation of Mi muscarinic acetylcholine receptors Experimental protocol as in Fig. 3 i. Responses on recovery were recorded 10-15 min after washing out Oxo-M. 354 A, A, Selyanko and others J. Physiol. 522,3

channels are interesting in three respects. First, when H a m ilt o n , S. E., L oose, M. D., Qi , M., L e v e y , A. I., H il l e , B., recorded with the standard deactivation M-current protocol, M c K n ig h t , G. s ., I d zer d a , R. L. & N athanson , N. M. (1997). Disruption of the m l receptor gene ablates muscarinic receptor- KCNQl and KCNQ4 currents had kinetics strikingly similar dependent M current regulation and seizure activity in mice. to those of the M-current, and these channels too were Proceedings o f the National Academy o f Sciences o f the USA 94, inhibited via stimulation of Mj muscarinic receptors. Thus, 13311-13316. all members of the KCNQ family reported so far can generate H a r v ey , R. D. & H u m e, J. H. (1989). Autonomic regulation of ‘M-channels’ as defined kinetically and pharmacologically. delayed rectifier K'*’ current in mammalian heart involves Second, acetylcholine is a transmitter in both the heart and G proteins. American Journal o f Physiology 267, H818-823. the inner ear, so if M^ receptors and corresponding KCNQ H orn , R. & M a r ty , A. (1988). Muscarinic activation of ionic currents subunits are co-expressed in the same cells, their coupling measured by a new whole-cell recording method. Journal o f General Physiology 9 2 , 145-159. may influence cardiac function and hearing. (Acetylcholine has been reported to reduce ^ g when enhanced by Kub isc h , C., Schroeder , B. C., F r ie d r ic h , T., L u t jo h a n n , B., El - A m r a o u i , a ., M a r l in , S., Pe t it , C. & J entsch , T. J. (1999). isoprenaline (Harvey & Hume, 1989; Yazawa & Kemeyama, KCNQ4, a novel potassium channel expressed in sensory outer hair 1990), but this was probably due to inhibition of adenylate cells, is mutated in dominant deafness.Cell 96,437-446. cyclase, and so was probably mediated by Mg rather than Mj L er c h e , H ., B ie r v e r t , G, A lek o v, A. K ., Sc h l e it h o ff , L., receptors. Also, Freeman & Kass (1995) reported that g in L in d n e r , M., K l in g e r , W., Bretschneider , F., M it r o v ic , N., guinea-pig sino-atrial node cells was inhibited by cholinergic Ju r k a t -R ott, K ., Bode , H ., L e h m a n n -H orn , F. & S t e in l e in , agonists but via a direct interaction with /^g channels.) O.K. (1999). A reduced K^ current due to a novel mutation in KCNQ2 causes neonatal convulsions. Annals o f Neurology 46, Third, because KCNQl and KCNQ4 subunits have short 305-312. carboxy termini (compared with those of KCNQ2+KCNQ3 M a r c o tti , W. & K ros , G J. (1999). Developmental expression of the subunits; see Wang et a l 1998; Kubisch et a l 1999) this potassium current ^ „ contributes to maturation of mouse outer allows us to eliminate substantial parts of KCNQ2+KCNQ3 hair cells. Journal o f Physiology 5 2 0 ,663-660. C-termini as obligatory sites involved in channel inhibition. M a r r io n , N. V. (1997). Control of M-current. Annual Review o f Thus, although the mechanism of muscarinic inhibition of Physiology 69,483-504. M- (and KCNQ-) channels is still unknown, it is clearly M a r r io n , N V., Sm a r t , T. G., M ar s h , S. J, & Bro w n , D. A. (1989). conserved among all four members of the KCNQ family. Muscarinic suppression of the M-current in the rat sympathetic ganglion is mediated by receptors of the M^-subtype. British Journal o f Pharmacology 98, 557-573.

Mu l l a n e y ,-L.rDoDDrM.-W., B u c k ley, N. & M il l ig a n , G. (1993). Agonist activation of transfected human M l muscarinic acetylcholine receptors in CHO cells results in downregulation of Ba r h a n in , J., L esage , F., Gu il l e m a r e , E., F in k , M., L a z d u n s k i , M. both the receptor and the a-subunit of the G-protein Gq. & R omey, G. (1996). KvLQTI and IsK (minK) proteins associate to Biochemical Journal 125-131. form the cardiac potassium current, Nature 384,78-80. 289,

R a e , j .. Cooper , K ., Gates , P. & Watsky, M. (1991). Low access Be r n h e im , L., M a t h ie , A. & H il l e , B. (1992). Characterization of resistance perforated patch recordings using amphotericin B. muscarinic receptor subtypes inhibiting Ca^'*’ current and M current Journal of Neuroscience Methods 15-26. in rat sympathetic neurons. Proceedings o f the National Academy o f 37,

Sciences o f the USA 89,9544-9548. Sa n g u in e t t i , M. C. (1999). Dysfunction of delayed rectifier potassium channels in an inherited cardiac arrhythmia. Annals o f Bie r v e r t , C., Schroeder , B. C., Ku b isc h , G, Be r k o v ic , S. F., New York Academy o f Sciences 8 6 8 ,406-413. P ropping , P., Jentsch , T. & St e in l e in , 0. K. (1998). A potassium channel mutation in neonatal human epilepsy. Science 279, Sa n g u in e t t i , M. G, Cu r r a n , M. E., Z ou, A., Sh e n , J., Spector , 403-406. P. S., At k in s o n , D. L. & K ea t in g , M. T. (1996). Coassembly of KvLQ TI and minK (IsK) proteins to form cardiac I(Ks) potassium Br o w n , D. A. (1988). M-currents. In Ion Channels, vol. 1, ed. channel. Nature 384,80-83. N a r a h a s h i ,T., pp. 55-94. Plenum Publishing, New York.

Selyanko , A. A. & Br o w n , D. A. (1999). M-channel gating and Ch a r l ie r , C., Sin g h , N. A., R y a n , S. G., L e w is , T. B., R eu s, B. E., simulation. Biophysical Journal 77,701-713. L each , R. J. & L eppert , M. (1998). A pore mutation in a novel

KQT-like potassium channel gene in an idiopathic epilepsy family. Selyanko , A. A., H a d l e y , J. K., W ood , I. G , A bogadie , F. G, Naiure Genetics 18,53-55. D elm as , P., Buc k l e y, N. J., L ondon , B. & Bro w n , D. A. (1999). Two types of K'*' channel subunit, ergl and KCNQ2/3, contribute to Chouabe , C., N e y ro ud , N., Gu ic h e n e y , P., L a z d u n s k i , M., the M-like current in a mammalian neuronal cell. Journal o f R omey, G. & B a r h a n in , J. (1997). Properties of K vLQ TI K'*' Neuroscience 7742-7756. channel mutations in Romano-Ward and Jervell and Lange-Nielsen 19, inherited cardiac arrhythmias. EMBO Journal 16,5472-5479. Selyanko , A. A,, R obbins , J. & B ro w n , D. A. (1995). Putative M-type potassium channels in neuroblastoma-glioma hybrid cells: F r e e m a n , L. C. & K ass , R. S. (1995). Cholinergic inhibition of slow inhibition by muscarine and bradykinin. Receptors and Channels 3, delayed-rectifier K^ current in guinea pig sino-atrial node is not 147-159. mediated by muscarinic receptors. Molecular Pharmacology 47,

1248-1254. Selyanko , A. A., Stan sfeld , C. E. & Br o w n , D. A. (1992). Closure of potassium M-channels by muscarinic acetylcholine-receptor H a le y , J. E., A bogadie , F. C., D elm as , P., D a y r e l l , M., Va l l is , Y., stimulants requires a diffusible messenger.Proceedings o f the Royal M il l ig a n , G., Ca u l f ie l d , M. P., Br o w n , D. A. & Bu c k le y, N. J. Society B 260,119-125. (1998). The a subunit of Gq contributes to muscarinic inhibition of the M-type potassium current in ^mpathetic neurons. Journal o f Neuroscience 18,4521-4531. J Physiol 522.^ Modulation o f KC N Q l-4 channels 355

Sin g h , N. A., Ch a r l ie r , C., Sta u ffe r , D., D u Pont , B. R., L each , R .J., M e l is , R., R o n en , G. M., Bj e r r e , I , Quattlebaum , T., Mu r p h y , J. V., M cH a r g , M. L , Gagnon , D., R osales , T., P e if f e r , A., A nderson , V. E. & L eppert , M. (1998). A novel potassium channel gene, KCNQ2, is mutated in an inherited epilepsy of newborns. Nature Gemtics, 18,25-29.

Wan g , H.-S., Pa n , Z., Bro w n , B. S., W ym o r e , R. S., Co h en , I. S., D ix o n , J. E. & Mc K in n o n , D. (1998). KCNQ2 and KCNQ3 potassium channel subunits: molecular correlates of the M-channel. Science 282,1890-1893.

Y ung, W -P , L evesq ue, P. 0., L it t l e , W. A., Condor , M. L., Sh alaby , F. Y. & Bl a n a r , M. A. (1997). K vLQ TI, a voltage-gated potassium channel responsible for human cardiac arrhythmias. Proceedings o f the Natiom l Academy o f Sciences o f the USA 94, 4017-4021.

Y vzawa , K . & K am eya m a , M. (1990). Mechanism of receptor- mediated modulation of the delayed outward potassium current in guinea-pig ventricular myocytes. Journal o f Physiology 421, 135-150.

Acknowledgements This work was supported by the MRC and the Wellcome Trust. We thank Dr D. McKinnon for KCNQ2 and KCNQ3 cDNAs and Dr M. T. Keating for KCNQl cDNA.

Corresponding author A. A. Selyanko: Department of Pharmacology, University College

London, Gower Street, London W C lE 6 BT, U K .

Email: [email protected]

Author’s present address I. C. Wood: School of Biochemistry and Molecular Biology, University of Leeds, Leeds LS2 9JT, U K . British Journal of Pharmacology (2000) 129, 413-415 © 2000 Macmillan Publishers Ltd All rights reserved 0 00 7-1188/00 $15.00 ( Î ) www.nature.com/bjp SPECIAL REPORT Differential tetraethylammonium sensitivity of K C NQ l-4 potassium channels

* U.K. Hadley, ^M. Noda, A.A. Selyanko, Wood, E C. Abogadie & D.A. Brown

'Department of Pharmacology, University College London, Gower St., London W C IE 6 BT; ^Wellcome Laboratory for Molecular

Pharmacology, University College London, Gower St., London W C IE 6 BT and ^Laboratory of Pathophysiology, Graduate School of Pharmaceutical Science, Kyushu University, Japan

In Shaker-gvovLp potassium channels the presence of a tyrosine residue, just downstream of the pore signature sequence G Y G , determines sensitivity to tetraethylammonium (TEA). The K C N Q family of channels has a variety of amino acid residues in the equivalent position. We studied the effect of T EA on currents generated by K C N Q homomers and heteromers expressed in CHO cells. We used wild-type K C N Q l- 4 channels and heteromeric KC N Q 2/3 channels incorporating either wild-type K C N Q 3 subunits or a mutated K C N Q 3 in which tyrosine replaced threonine at position 323

(mutant T323Y). IC 50 values were (mM): K C N Q l, 5.0; K C N Q 2, 0.3; K C N Q 3, >30; K C N Q 4, 3.0; K C N Q 2 + K C N Q 3, 3.8; and K C N Q 2 + KCNQ 3(T323Y), 0.5. While the high T E A sensitivity of K C N Q 2 may be conferred by a tyrosine residue lacking in the other channels, the intermediate T E A sensitivity of K C N Q l and K C N Q 4 implies that other residues are also important in determining T E A block of the K C N Q channels. British Journal of Pharmacology (2000) 129, 413-415 Keywords: TEA; K v L Q T I; K C N Q channels; M-current; Chinese hamster ovary (CHO ) cell Abbreviations: C H O (hm l), Chinese hamster ovary cell (stably expressing human M , muscarinic receptors); EG TA , ethylene glycol-bis(jS-aminoethyl ether)N,N,N',N'-tetraacetic acid; HEPES, N-2-hydroxyethylpiperazine-N'-2-ethanesul- phonic acid; IC50, drug concentration giving 50% block of current; M-channel/M-current, muscarinic-inhibited potassium channel/current; NG108-15, neuroblastoma-glioma hybrid cell line; Uy, H ill slope; pICso, negative log

of IC 5 0 ; TEA, tetraethylammonium

Introduction The K C N Q family of channels includes the ing a mutated K C N Q 3 subunit in which tyrosine replaced cardiac slow delayed rectifier K C N Q l (K v L Q T I) (Barhanin et threonine at position 323. a l, 1996; Sanguinetti et a l, 1996), the neural-specific channels K C N Q 2 and K C N Q 3 (Biervert et a l, 1998; Charlier et a l, Methods Methods for culture and transfection of C H O hm l 1998; Yang et a l, 1998), which have been proposed as cells have been described previously in Selyanko et a l (1999). components of the M-channel (Wang et a l, 1998), and C HO cells (a derivative of the C H O -K l line; Mullaney et a l,

K C N Q 4, whose m R N A is expressed in the cochlea (Kubisch 1993) were grown at 37°C and 5% CO 2 in « -M E M supplemented et a l, 1999). The T E A sensitivity of K C N Q 2 is higher than with 10% foetal calf serum, 1% L-glutamine and 1% penicillin/ that of the M-channel, that of KC N Q 3 is lower, and that of the streptomycin. For recording, cells were plated in 35 mm plastic KCNQ 2/3 heteromultimer closely matches that of the M - dishes, and transfected 1 -2 days after plating using Lipofect current (Yang et a l, 1998; Wang et a l, 1998). Hence, Amine Plus (Life Technologies) as suggested by the manufac­ differential T E A sensitivity can provide helpful information turer. Plasmids (driven by C M V promoter) containing cDNAs

regarding the contribution of different members of this gene for K C N Q channels, and c D N A for C D 8 as a marker, were co­ family to native K ^ channels. transfected at a channel : marker ratio of 10:1. Cells for patch-

The high T E A sensitivity of K C N Q 2 might result from the clamping were identified using C D8 -binding Dynabeads (Dynal presence of a tyrosine residue in the pore loop of the channel (U .K .) Ltd) 1 day after transfection. (see Kavanaugh et a l, 1991). Thus, the Shaker potassium The perfusing solution contained (mM): N aC l 144, K C l 2.5, channel, which is weakly blocked by T E A (IC50 27 m M ), has a CaCl] 2, MgCIi 0.5, N-2-hydroxyethylpiperazine-N'-2-ethane- threonine at position 499 (just downstream of the G Y G pore sulphonic acid (HEPES) 5, and glucose 10 plus Tris base to pH sequence at positions 444-446), but can be made much more 7.4. The pipette solution contained (mM): K acetate 80, K C l sensitive to T E A (0.59 m M ) by replacing T449 with a tyrosine 30, HEPES 40, MgCU 3, ethylene glycol-bis(j?-aminoethyl (MacKinnon & Yellen, 1990). K C N Q 2 also has a tyrosine in ether)N,N,N',N'-tetraacetic acid (EG TA ) 3, CaCli 1 and the corresponding position, whereas K C N Q 3 has a threonine N aO H to pH 7.4. Amphotericin B was used to perforate the (Wang et a l, 1998) and K C N Q l and K C N Q 4 have a valine patch (Rae et a l, 1991). TEA was applied cumulatively, using and a threonine, respectively (Kubisch et a l, 1999). 0.2 M and 2 M aqueous T E A solutions added to the reservoir Using a C H O cell expression system and the perforated- of perfusing fluid. patch configuration of the patch-clamp, we quantified the Data were acquired and analysed using pClamp software effect of T E A on currents generated by homomeric wild-type (version 6.0.3). Currents were recorded using an Axopatch K C N Q l- 4 channels, on heteromeric wild-type KC N Q 2/3 200A (or 200) patch-clamp amplifier, filtered at 1 kHz, and channels and on heteromeric K C N Q 2/3 channels incorporat­ digitized at 1 - 4 kHz. One second steps to —50 m V were applied from a holding potential of —20 mV and the total deactivation relaxation amplitude was measured at — 50 m V in * Author for correspondence; E-mail: [email protected] ''Current address: School of Biochemistry and Molecular Biology, control and for each dose of TEA . Data were normalized and University of Leeds, Leeds, LS2 9JT plotted in Origin (version 5.0, Microcal Software), and curves 414 J.K. Hadley e t al Special Report were fitted with the equation: y = l/(l+ (x /x o )'’) where y is the or with the mutated K C N Q 3 subunit KCNQ 3(T323Y). As is relative current amplitude; x is the concentration of the clear from Figure 2B and Table 1, the potency for the blocker; Xo, theIC50, is the concentration at which y is 0.5; and heteromeric wild-type K C N Q 2 and K C N Q 3 was intermediate p is the power (equivalent to the H ill slope, nn). Additionally, between those of homomeric K C N Q 2 and K C N Q 3. The data were plotted against log concentrations and fitted with the potency for heteromers containing the mutated Boltzman equation: y = 1 / ( 1 where xO is the pIC^o K C N Q 3(T323Y) was increased, approaching that for the (negative log of the IC50), as a check on the true (geometric) wild-type KCNQ2. spread of the standard error. T E A was from Lancaster Synthesis. A ll other chemicals Discussion Our results for TEA inhibition of KCNQ2 and were obtained from Sigma or B D H Chemicals. Human KC N Q 2/3 currents in C H O cells give IC50 values of KCNQl cDNA was kindly provided by Dr M.T. Keating; 0.3 + 0.02 m M and 3.8 ± 0 .2 m M (pIC^o values 3.53 ±0.03 and human K C N Q 2 and rat K C N Q 3 (as studied in Wang et a i, 2.42 ±0.02), respectively. These values accord well with 1998) by D r D . McKinnon; and human K C N Q 4 (as in previously reported IQ values for these channels expressed in Kubisch et a i, 1999) by D r T.J. Jentsch. We thank D r T.J. Xenopus oocytes, which were 0.16 ±0.02 m M and Jentsch and Tatjana Kharkovets for the mutated 3.5 ± 0 .7 m M, respectively (Wang et a i, 1998). The sensitivity KCNQ 3(T323Y), which was based on the human sequence. of KC N Q 3 to T E A blockade has not been quantified, but 5 m M T E A had a negligible effect on K C N Q 3 currents in Results We find that the T E A sensitivity of K C N Q 2 is high, oocytes (Yang et a i, 1998). Even the highest concentration of that of KCNQ3 is low, and that of KCNQl and KCNQ4 is T E A tested in the present experiments, 30m M, which could intermediate. Figure lA a - d shows representative data records completely block K C N Q l, 2 and 4, had a minimal effect on for T E A inhibition of homomeric K C N Q channel currents. KCNQ3 (Figure 2A). Figure 2A and Table 1 show the potencies for these channels. T E A block does not appear to have been tested before on Figure lBa,b shows data records for the heteromeric channels K C N Q l and K C N Q 4 channels, so it was of interest to produced by co-transfecting K C N Q 2 with wild-type K C N Q 3 compare T E A sensitivity of these channels with that of the

Table 1 Half-maximal blocking concentrations of TEA (IC50) in mM, pICjo (negative log of the IC50 in M, fitted separately), and Hill slopes (nn) for block of K C N Q channels expressed in CHO cells; « = number of cells tested

KCNQ2.3 KCNQl KCNQ2 KCNQ3 KCNQ4 KCNQ2,3 (T323Y)

ICso + s.e.mean (mM) 5.0+0.2 0.3 + 0.02 >30 3.0+ 0.3 3.8 + 0.2 0.5 + 0.05 pIQo + s.e.mean 2.30+0.02 3.53 + 0.03 2.53 + 0.04 2.42+0.02 3.29+0.05 Uh 1.8+G.l 0.9+0.06 2.0+0.4 0.8 + 0.03 1.0+6.09 n 4 4 3 4 3

-201 -50 L 1 s

KCNQ1 KCNQ2 KCNQ3 KCNQ4

30*

0.3 < lOOpA lOOpA 100 pA 100 pA

B KCNQ2 + KCNQ3 KCNQ2 + KCNQ3(T323Y) Ol

^ 0.3* / 1100 pA , V lOOpA 30 V r _ _ Ss------/%..._......

Figure 1 Differential TEA sensitivities of K C N Q l-4 (A a -d ) and heteromeric KCNQ2/KCNQ3 and KCNQ2/KCNQ3(T323Y) currents (Ba,b), expressed in CHO cells. Deactivations were recorded in response to 1 s steps from the holding level of -2 0 to — 50 mV, in the absence and presence of different concentrations of TEA as indicated (in mM) on the left side of each trace.

Brltlsli Journal of Pharmacology, vol 129 (3) J.K. Hadley et a l Special Report 415

M-channel components K C N Q l and KC N Q 3. Figure 2 and Table 1 indicate that the slope of the inhibition curve is considerably steeper for K C N Q l and K C N Q 4, with Hill coefficients of around 2, compared with values of about 1 for K C N Q l, K C N Q 3 and their heteromers. Thus, the nature of c the molecular interaction between T E A and the channel g protein may differ between different K C N Q channels. Our 3 Hill coefficient of 0.8 for T E A inhibition of K C N Q l/3 heteromers matches the previously reported value of 0.8 for these channels (Yang et a l, 1998), and is comparable with ▼ KCNQl 0) values of 0.6 for both the mouse superior cervical ganglion M - A KCNQ2 Œ current and the fast component of the neuroblastoma-glioma — X -K C N Q 3 hybrid (NG108-15) M-like current (Selyanko et a l, 1999). ■ KCNQ4 0.0 We find that substituting tyrosine for threonine at position 100 323 in K C N Q 3 apparently increases the T E A sensitivity of the heteromeric KC N Q 2/3 channel, as predicted from previous work on Shaker channels. This implies that homomeric K CNQ 3(T323Y) currents should also be considerably more sensitive to T E A than wild-type KC N Q 3 channels, but unfortunately we were unable to obtain homomeric c K CNQ 3(T323Y) currents of sufficient amplitude to test their T E A sensitivity directly. However, the fact that K C N Q l and 3 KC N Q 4, lacking a tyrosine in this position, also have a higher Ü o 0 5 T E A sensitivity than KCNQ 3, suggests that other residues are important in determining T E A block of K C N Q channels. 0) Finally, it is worth noting that both KCNQl and KCNQ4 Œ yield currents with a similar appearance to M-currents (Figure • KCNQ2 + KCNQ3 > 1). Since their sensitivity to TEA (and to the M-channel blocker, ■ KCNQ2 + KCNQ3(T323Y)' 0.0 linopirdine: Kubisch et al., 1999; A. A. Selyanko, unpublished 0.01 0.1 1 10 100 observations) is similar to that of native M-currents, these may [TEA], mM also be regarded as representing species of ‘M-channels’. The characteristics of TE A blockade described here may be useful in Figure 2 Concentration dependences of inhibition of K C N Q l-4 establishing whether K C N Q l and K C N Q 4 participate in the currents (A), and coexpressed KCNQ2/KCNQ3 and KCNQ2/ formation of any native M-like channels. KCNQ3(T323Y) channels (B) in CHO cells. All values are mean± s.e.mean. Smooth lines are least squares fits to the equation y = l / (1 +(x/xo)'’), superimposed for all data sets except KCNQ3, where the This work was supported by the M .R .C . and The Wellcome Trust. points have been joined by straight lines for clarity. We thank D r T.J. Jentsch for helpful reading of the manuscript.

References

BARHANIN, J., LESAGE. P.. GUILLEMARE, E„ FINK, M., LAZDUNS- RAE, J., COOPER, K., GATES, P. & WATSKY, M. (1991). Low access KI, M. & ROMEY, G. (1996). K yL Q T l and IsK (minK) proteins resistance perforated patch recordings using amphotericin B.J. associate to form the Iks cardiac potassium current.Nature, 384, Neurosci. Methods, 37, 15-26. 78-80. SANGUINETTI, M.C., CURRAN, M.E., ZOU. A., SHEN, J., SPECTOR, BIERVERT, C., SCHROEDER, B.C.. KUBISCH, C , BERKOVIC, S.F., P.S., ATKINSON, D.L. & KEATING, M.T. (1996). Coassembly of PROPPING, P., JENTSCH, T.J. & STEINLEIN, O.K. (1998). A K yL Q T l and m inK (IsK) proteins to form cardiac Iks potassium potassium channel mutation in neonatal human epilepsy. channel. Nature, 384, 80-83. Science, 279, 403-406. SELYANKO, A.A., HADLEY, J.K., WOOD, I.C., ABOGADIE, F.C., CHARLIER, C., SINGH, N.A., RYAN, S.G., LEWIS, T.B., REUS, B E., DELMAS, P., BUCKLEY, N.J., LONDON, B. & BROWN, D.A. LEACH, R.J. & LEPPERT. M. (1998). A pore mutation in a novel (1999). Two types of K ^ channel subunit, Ergl and KCNQ2/3, KQT-like potassium channel gene in an idiopathic epilepsy contribute to the M-like current in a mammalian neuronal cell.J. family. Nature Genetics, 18, 53-55. Neurosci., 19, 7742-7756. KAVANAUGH, M.P., VARNUM, M.D., OSBORNE, P.B., CHRISTIE, WANG, H.-S., PAN, Z., SHI, W„ BROWN, B.S., WYMORE, R.S., M.J.. BUSCH, A.E.. ADELMAN. J.P. & NORTH, R.A. (1991). COHEN, I.S., DIXON, J.E. & MCKINNON, D. (1998). KCNQ2 Interaction between tetraethylammonium and amino acid and KCNQ3 potassium channel subunits: molecular correlates of residues in the pore of cloned voltage-dependent potassium the M-channel. Science, 282, 1890-1893. channels. J. Biol. Chem., 266, 7583-7587. YANG, W.-P., LEVESQUE, P.C., LITTLE, W.A., CONDER, M L., KUBISCH, C , SCHROEDER, B.C., FRIEDRICH, T., LUTJOHANN, EL- RAMAKRISHNAN, P., NEUBAUER, M.G. & BLANAR, M.A. AMRAOUI, A., MARLIN, S., PETIT, C. & JENTSCH, T.J. (1999). (1998). Functional expression of two KvLQTI-related potassium KCNQ 4, a novel potassium channel expressed in sensory outer channels responsible for an inherited idiopathic epilepsy. J. Biol. hair cells, is mutated in dominant deafness. Cell, 96, 437-446. Chem.,213, 19419-19423. MACKINNON, R. & YELLEN, G. (1990). Mutations affecting TEA blockade and ion permeation in voltage-activated K ^ channels. (Received October 21, 1999 Science, 250, 276-279. Revised November 9, 1999 MULLANEY, L, DODD, M.W., BUCKLEY, N. & MILLIGAN, G. (1993). Accepted November 10, 1999) Agonist activation of transfected human M l muscarinic acet­ ylcholine receptors in C H O cells results in down-regulation of both the receptor and the a subunit of the G-protein Gq.Biochem. J., 289, 125-131.

British Journal of Pharmacology, vol 129 (3) The Journal of Neuroscience, September 15,1999, /9(18);7742-7756

Two Types of Channel Subunit, Erg1 and KCNQ2/3, Contribute to the M-Like Current in a Mammaiian Neuronai Cell

A. A. Selyanko,1 J. K. Hadley,^ I. C. Wood ,2 F. C. Abogadie,^ P. Delmas,2 N. J. Buckley,2 B. London,^ and D. A. Brown^

^Department of Pharmacology, ^Wellcome Laboratory for Molecular Pharmacology University College London, London, WC1E6BT, United Kingdom, and ^Cardiovascular Institute, University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania 15213

The potassium M current was originally identified in sympa­ pharmacologically. We provide evidence from PCR analysis and thetic ganglion cells, and analogous currents have been re­ expression studies to indicate that these two components are ported in some central neurons and also in some neural cell mediated by two distinct molecular species of channel: the lines. It has recently been suggested that the M channel in fast component resembles that in sympathetic ganglia and is sympathetic neurons comprises a heteromultimer of KCNQ2 probably carried by KCNQ2/3 channels, whereas the slow and KCNQ3 (Wang et al., 1998) but it is unclear whether all component appears to be carried by mergla channels. Thus, other M-like currents are generated by these channels. Here we the channels generating M-like currents in different cells may be report that the M-like current previously described in NG108-15 heterogeneous in molecular composition. mouse neuroblastoma x rat glioma cells has two components, Key words: potassium channels; neuroblastoma x glioma “fast” and “slow”, that may be differentiated kinetically and hybrid cells; M current; sympathetic neuron; Ergl; KCNQ

The M current (/k(m>) is a low-threshold, slowly activating po­ whether members of other channel gene families might con­ tassium current that exerts an inhibitory control over neuronal tribute to the generation of M-like currents. excitability; this inhibition can be relieved by neurotransmitters In the present experiments, we have attempted to identify the acting on G-protein-coupled receptors, leading to enhanced ex­ molecular species of channels that generate the M-like cur­ citability and reduced spike-frequeney adaptation (Brown, 1988; rent (/K(M,ng)) NG108-15 mouse neuroblastoma x rat glioma Marrion, 1997). The current was originally described in sympa­ cells. These currents have been particularly well characterized thetic neurons (Brown and Adams, 1980; Constanti and Brown, (Higashida and Brown, 1986; Brown and Higashida, 1988a,b; 1981), and analogous currents have subsequently been identified Fukuda et al., 1988; Schafer et al., 1991; Robbins et al., 1992, in a variety of other neuronal and non-neuronal cells. Because the 1993; Selyanko et al., 1995). Like the channels in sympathetic precise kinetic and pharmacological properties of the current vary neurons, they are inhibited by transmitters acting on G-protein- somewhat in different cell types, the name “M-like” is often linked receptors coupled to phospholipase C (e.g., bradykinin and applied to this current family. Ml and M3 muscarinic receptors) (Higashida and Brown, 1986; Recently, evidence has been provided to indicate that the Fukuda et al., 1988), with similar consequences for cell firing channels that generate the M current in rat sympathetic neurons (Robbins et al, 1993). On the other hand, the kinetics of /K(M,ng> are composed of a heteromeric assembly of KCNQ2 and KCNQ3 appear more complex than those of the ganglionic M current subunits (Wang et al., 1998; see also Yang et al., 1998). These are (Robbins et al, 1992), and the two currents differ in their sensi­ two homologs of the KCNQl (KvLQTI) channel, mutations of tivities to 9-aminotetrahydroacridine (cf. Marsh et al, 1990; Rob­ which are responsible for one form of the cardiac “long QT” bins et al, 1992) and linopirdine (cf. Aiken et al, 1995; Lamas et syndrome (Yang et al., 1997). In contrast, KCNQ2 and KCNQ3 al, 1997; Noda et al, 1998). It has previously been suggested that are restricted to the nervous system, and mutations in these Shaker-typQ Kvi.2 channels, cloned from NG108-15 cells channels are associated with a form of infant epilepsy termed (Yokoyama et al, 1989), may contribute to /K(M,ng) (Morielli and “benign familial neonatal convulsions” (Biervert et al., 1998; Peralta, 1995). However, the insensitivity of /K(M,ng) to dendro­ Charlier et al., 1998; Schroeder et al., 1998; Singh et al., 1998). toxin (Selyanko et al., 1995) makes this unlikely. Instead, we However, it is not yet known whether all M-like channels are provide evidence to indicate that two different types of composed of these two subunits (or homologs thereof), or channel contribute to the M-like current in NG108-15 cells: the mouse eîker-a-go-go-ichteâ gene (mergla), also expressed in the brain (London et al, 1997), and KCNQ2/KCNQ3, the proposed

Received May 18,1999; revised June 28,1999; accepted July 2,1999. substrate for the ganglionic current (Wang et al, 1998). B.L was supported by a Grant-In-Aid from the American Heart Association, and the other authors were supported by the United Kingdom Medical Research Council MATERIALS AND METHODS and the Wellcome Trust. We thank Misbah Malik-Hall, Brenda Browning, and Mariza Dayrell for tissue culture and Svjetlana Miocinovic (Biology Program, Cell cultures. NG108-15 mouse neuroblastoma x rat glioma hybrid cells, CalTech, Pasadena, CA) for participation in some experiments. subclone BM8 (PMl), transfected to express pig brain M j muscarinic Correspondence should be addressed to Dr. A. A. Selyanko, Department of receptor (Fukuda et al, 1988), were cultured and differentiated as de­ Pharmacology, University College London, Gower Street, London WCIE 6 BT, UK. scribed previously (Robbins et al, 1992). Chinese hamster ovary (CHO) Copyright © 1999 Society for Neuroscience 0270-6474/99/197742-15$05.00/0 cells stably transfected with cDNA encoding human muscarinic Selyanko et al. • M-Like Current in a Neuronal Cell J. Neurosci., September 15,1999, ï9(18):7742-7756 7743

Table 1. Fast and slow kinetic components of/g/Mne) and their comparison with the kinetics of ^K(M) ^nd /jviergl

T, contribution

Current Fast Slow

^K(M,ng) Total (control) 117 ± 5 msec 658 ± 29 msec 3434 ±159 msec 33.1 ± 2.1% (86/101) 25.7 ± 1.7% (82/101) 41.2 ± 2.2% (86/101) 333 ± 126 msec 2356 ± 247 msec 42.5 ± 5.1% (14/101) 57.6 ± 5.1% (15/101) Fast (WAY-insensitive) 104 ± 7 msec (37) Slow (WAY-sensitive) 456 ± 29 msec 2718 ± 156 msec 41.7 ± 3.5% (34/36) 58.3 ± 3.5% (36/36) ^K(M) 93 ± 5.6 msec (n = 15) ^mergla 183 ± 10.6 msec 1012 ± 85 msec 64.6 ± 2.4% (22/23) 35.4 ± 2.4% (23/23) ^merglb 37.1 ± 19.4 msec 112 ± 42.5 msec 70.8 ± 15.8% (3/3) 29.2 ± 15.8% (3/3)

^K(M,ng)> 7x(M), ^meigia> 3"^ 7n,ergib werc recordcd from NG108-15, mouse sympathetic, and CHO cells, respectively. Deactivation was recorded in response to a voltage step to -50 mV from a holding potential of -20 mV (/K(M,ng) or 0 mV (/nergia ^ud fmergib)- Mcans ± SEM of T and contributions were obtained from deactivation relaxations recorded at -50 mV. Number of cells in parentheses; see Results.

Vn 0) ()^G108-15 mouse neuroblastoma x rat glioma) mV -2 0 -j -50 L 6s

-20 Cell -50 ■bW AY 123,398 WAY-sens. Control -i-Linopircline ^ (10 [xM) WAY-lnsens. / (3 ^M) Linop.-sens. 50 pA

100 pA

CellB

l5 0 p A Control Linop.-insens. / Lin.-sens. WAY-sens. -bWAY 123,398 (10 ^xM) 100 pA

Figure 1. Pharmacological separation of two components of the M-like (/K(M,ng)) current in NG108-15 mouse neuroblastoma x rat glioma cells. M-like currents recorded in two cells {A, B) as deactivating tail currents produced by voltage steps from -2 0 mV (holding potential) to -5 0 mV. In each cell, the control current had two different components, fast and slow, which could be blocked by WAY 123,398 and linopirdine, respectively. Insets on the right show the difference (blocker-sensitive) currents. The currentin A consisted predominantly of the slow component (blocked by WAY 123,398), whereas that in B was predominantly fast and blocked by linopirdine. 7744 J. Neurosci., September 15,1999, 79(18):7742-7756 Selyanko et al. • M-Llke Current in a Neuronal Cell

^K(Mng) (NG108-15 mouse neuroblastoma x rat glioma) KCNQ

-20

control 'fast' T = 76 ms 'slow' T = 340 ms T = 2124 ms

-H 10^M WAY 123,398 Figure 3. cDNA from mouse SCG, rat SCG, and chemieally differenti­ ated NG108-15 cells, was amplified using primers to the erg family of potassium channel genes or primers recognizingKCNQ2 and KCNQ3 'fast' potassium channel genes. Amplified products were obtained from all cell t = 75 ms types but not from a negative control containing no template, indicating 100 pA that members of these families are expressed by these cells. Sequence analysis revealed that the amplified product obtained from NG108-15 cDNA with primers to the erg genes was predominantly, if not exclusively, mergl. Analysis of the amplified product obtained using the KCNQ primers in both rat SCG and NG108-15 has shown these cells express both KCNQ2 and KCNQ3. M-1 kb ladder DNA size standards. difference

elamping were identified by adding CD8-binding Dynabeads (Dynal, 'slow' Great Neck, N Y ) the day after transfection. For immunocytochemistry, ... t = 315 ms a plasmid containing cDNA for jellyfish green fluorescent protein (GPP) . T = 1972 ms was used as a marker for transfection. Reverse transcription PCR. RNA was extracted from cell lines and superior cervical ganglia (SCG) using RNAzol B (Biogenesis Ltd.) and reverse-transcribed using oligo-dT and mouse murine leukemia virus reverse transcriptase (Promega, Madison, W I). The oligonucleotides used to amplify the erg gene family were: erg-s 5' C C CYTTCAAG G C- Figure 2. Kinetic analysis of the M-like current in a NG108-15 mouse MGTGTGGG and erg-a 5' CTG G TH A G R C TG C TG AA G G T. Primers neuroblastoma x rat glioma cell. Total was activated by holding at were designed such that the amplified product spanned at least one intron -2 0 mV and then deactivated by a 6 sec step {top record) to -5 0 mV to ensure amplification products were not derived from contaminating before [A] and after (B) addition of the og-ehannel blocker WAY 123,398 genomic DNA. The primers used forKCNQ PCR were KCNQs 5'ACCT- (10 jixM). In control (A) the deactivation tail was fitted {smooth line, GGARGCTBCTGGCTC and KCNQa 5'CCKCTYTTCTCAAAGT- superimposed) by the sum of fast ( t = 76 msec, 254 pA) and two slow (x = GCTTCTG. These primers were designed to amplify both KCNQ2 and 340 msec, 37 pA and 2124 msec, 70 pA) exponential curves. WAY 123,398 KCNQ3 sequences. Cycling conditions were 95°C for 5 min and then 30 abolished both slow components without affecting the fast component{B, cycles of 95°C for 30 sec; 60°C for 1 min and 7 2 T for 1 min followed by T = 76 msec). C, Slow component obtained by subtracting the record a final step of 72°C for 10 min. Aliquots of the reaction mixture were shown in B from that shown in A was fitted {smooth line, superimposed) visualized on a 2% (w/v) Metaphor agarose (PMC BioProducts, Rock­ hy the sum of two kinetic components (x = 315 and 1972 msec). All land, M E ). PCR products were cloned using the pGEM-T vector (Pro­ records were obtained from the same cell. Dashed lines denote zero mega) and recombinant plasmids sequenced usingTaq polymerase, fluo- current levels. resceinated dye terminators and an Applied Biosystems 377 automated DNA sequencer. Immunocytochemistry. This was performed using antibodies raised receptors were maintained in culture as described in Mullaney et al. against synthetie peptides corresponding to the last 14 amino acids of the (1993). Superior cervical ganglion neurons were prepared as described C-terminal fragment of mergl (mergl-CT) and the first 17 amino acids previously (Owen et al., 1990) from 6 -week-old C57 mice and 14-d-old of merglb (mergl-NT), respectively. NG108-15 cells were seeded onto Sprague Dawley rats, and used after 1-2 d in culture. Recordings from all polyornithine-coated glass coverslips to allow immunocytoehemistry. three types of cell were made at room temperature (20-22°C), under NG108-15 cells were washed in TBS (5.5 mM Tris, pH 7.4, and 137 mM identical experimental conditions (solutions, pipettes, etc.). NaCl) and fixed in acetone for 20 min at room temperature. The fixed Culture and transfection of CHO hml cells. CHO hml cells are cells were then treated with normal swine serum (1:10) in TBS for 30 CHO-Kl cells, previously transfected with the human M j receptor (M ul­ min. Once excess serum was removed, the mergl-CT primary antibody laney et al., 1993). Cells were grown in 50 ml flasks at 37“C and 5% C O2. was applied at 1:1000 dilution for 1 hr at room temperature. Different The culture medium was a-M E M supplemented with 10% fetal calf concentrations of the primary antibody were tested to optimize immu­ serum, 1% L-glutamine, and 1% penicillin/streptomycin. Cells were split nochemical labeling and minimize nonspecific staining of the tissue twice weekly when confluent, plated in 35 mm dishes, and transfected background. Bound antibodies were then detected using alkaline 1-2 d after plating using “LipofectAmine Plus” (Life Technologies, phosphatase-eonjugated secondary antibodies (1:500; Dako, Carpinteria, Gaithersburg, M D ) according to the manufacturer’s recommendations. CA), largely as described in Abogadie et al. (1997). Transfected mam­ Plasmids containing mergla and CD8 cDNAs, both driven by cytomeg­ malian CHO cells were briefly washed with PBS and fixed for 20-30 min alovirus promoter, were cotransfected in a ratio of 10:1. Cells for patch in PBS eontaining 4% paraformaldehyde. Fixed eells were then rinsed Selyanko et al. • M-Like Current in a Neuronal Cell J. Neurosci., September 15,1999, ?9(18):7742-7756 7745

with PBS, blocked for 10 min with ESA, and permeabilized with 0.1% Triton X-100 for 5 min. Incubation with the anti-mergl antibody and A Differentiated NG108-15 cells labeling were carried out as described for the NG108-15 cells. Perforated-patch whole-cell recording. Cells were bath-perfused with the solution of the following composition (in mM): 144 NaCl, 2.5 KCl, 2 CaCli, 0.5 MgClj, 5 HEPES, and 10 glucose, pH 7.4, with Iris base. Pipettes were filled with the “internal” solution containing 90 mM K acetate, 20 mM KCl, 40 HEPES, 3 MgClj, 3 mM EGTA, and 1 mM CaCU. The pH was adjusted to 7.4 with NaOH. Amphotericin B was used to perforate the patch (Rae et al., 1991). The series resistance was not compensated because the error introduced was reasonably small. Thus, with the electrodes used (2-3 MO), the series resistance was6 -8 MO, and most of the currents were <0.5 nA, so the voltage error would be <5 mV. As confirmation that the voltage error was small, no corre­ lation was found between deactivation time constants and initial current amplitude. Data accpiisition and analysis. Data were acquired and analyzed using pClamp software (version 6.0.3). Currents were recorded using an Axo­ patch 200A (or 200) patch-clamp amplifier, filtered at 1 kHz, and digi­ tized at 1-4 kHz. In eurrent-elamp experiments, currents were injected, and membrane potential was recorded using an Axoclamp-2 amplifier. Activation curves were fitted by the Boltzmann equation: ///(50) = 1/(1 exp(K,/2 - y)lk), where / is current at the test potential (estimated from the amplitudes of exponentials baekfitted to the beginning of the test step), 7(50) is current at +50 mV, ¥^,2 is the membrane potential, V, B Undifferentiated NG108-15 cells at which I is equal to Vi 7(50). Inhibition of the current was measured from the change in the amplitude of the deactivation tail recorded at -5 0 mV. Each tail was fitted by one or more exponentials, and the tail amplitude was taken as the sum of the amplitudes of all components contributing to it after backfitting them to the beginning of the hyper- polarizing pulse. In cells that had theerg-type component, backfitting was necessary to exclude not only the (relatively fast) capacity transient, but also the brief rising phase of the tail caused by the deinactivation of erg-type channels. To include the fast component (when present) in the fit, we had to position the fitting cursor earlier in the trace than was

appropriate for the pure erg current. This tends to skew thet values obtained for the erg components to larger values. This may explain the difference in slow rvalues between the total and WAY 123,398-sensitive currents (Table 1). Inhibition curves were fitted by the Hill equation: Y = ^max * + IQ o ""), where y^ax is the maximum inhibition, % is the blocker concentration, n^ is the slope (Hill coefficient), and IC50 is the concentration corresponding to the half-maximal inhibition. Individual currents were measured and fitted using the Clampfit software, whereas the program “Origin” (version 5.0, Microcal Software) was used for fitting activation and inhibition curves and for creating the figures. Drugs and chemicals. Linopirdine (DuP 996) was obtained from Re­ search Bioehemieals (Natick, M A ). WAY 123,398 and azimilide were kindly provided by Wyeth-Ayerst Research (Princeton, NJ) and Dr. A. Busch (DG Cardiovascular, Frankfurt, Germany), respectively. All other Mouse SCG drugs and chemicals were obtained from Sigma or BDH Chemicals (Poole, UK).

RESULTS Fast and slow M-like (/K(M,ng)) currents in NG108-15 mouse neuroblastoma x rat glioma cells V w m We recorded M-like currents (/K(M,ng)) from 101 chemically dif­ ferentiated NG108-15 cells using perforated-patch electrodes. Cells were carefully selected for their “neuron-like” appearance, ; + i,e., large size and well-developed neuropil (Robbins et al,, 1992), When studied with the conventional M-current voltage protocol, that is, by stepped hyperpolarization after predepolarization to m approximately -2 0 mV (see Materials and Methods), currents showed characteristic M-like deactivation tails. However, the time course of these tail currents varied considerably from one cell to another. Figure 1 illustrates two extreme examples of this variation. Thus, in Figure 1+1, deactivation during a 6 sec hyper- Figure 4. Immunocytochemicai detection of mergl protein in differen­ tiated NG108 cells. Immunostaining for the C terminus of mergl in polarizing step was very slow, with an apparent “time-constant” chemically differentiated (A) and undifferentiated (B) NG108 cells, and of ~ 2 sec, whereas in Figure \B, deactivation was complete in dissociated mouse SCG cells (C). Note thatin A, a large NG108-15 cell within 2 sec. (arrow) showed labeling of strong intensity, whereas adjacent smaller and Because a number of tumor cells (including neuroblastoma bipolar eells were not stained. Micrographs were obtained using bright- cells) have been reported to expressHERG-\ike currents with field optics. Seale bars, 20 /am. 7746 J. Neurosci., September 15, 1999, 79(18):7742-7756 Selyanko et al. • M-Like Current in a Neuronal Cell

U a (CHO) i.,

-50^ -100*

250 pA a a

2 20

-10 -100 -50 0 50 -100 -50 0 50 ^JmV)

Figure 5. Characteristics of the mergla current (^mergla) cxpressed in CHO cells, /mergia was aeti- 1000 vated by long (8 see) depolarizing voltage steps 1000 from the holding level of -8 0 mV (Aa) and deac­ tivated by hyperpolarizing steps after its full aeti- &100 vation by a prepulse to +50 mV (Ba). Leak- subtracted steady-state 1-V relationships obtained at the end of the depolarizing and hyperpolarizing pulses, respeetively, are shown in Ab and Bb (open 100 -100 -50 0 50 -100 -50 0 50 circles), and an “instantaneous” I-V relationship ^rn(mV) obtained at the beginning of the hyperpolarizing pulses for current deactivation is shown in Bb (filled circles). Activation (Ac) and deactivation (Be) time constants were plotted semilogarithmi- 1.0 cally, and t-V relationships were fitted by straight = -5.9 ± 0.6 mV lines with t at 0 mV and the slope equal to 1218 ± k= 12.2 ± 0.5 mV 1 msec and -0.013 ± 0.001 m V “ ^ in Æ and 403 ± 1 msec and 0.012 ± 0.0006 m V (filled circles) and 8 0.5 3437 ± 1 msec and 0.014 ± 0.002 m V " ' (open circles) in Be. C, Activation curve fitted by the Boltzmann equation at ~ -5 .9 ± 0.6 mV and k = 12.2 ± 0.5 mV. Reeords in Aa and Ba were from the same eell. In Ab, Ac, Bb, Be, and C the 0.0 -100 -50 mean data are shown (vertical lines indieate SEMs) 0 50 obtained from six and nine cells, respectively.

relatively slow rates of deactivation (Bianchi et al., 1998), we Because 10 p.M W AY 123,398 produces a complete block oferg wondered whether these might contribute to the long deactivation channels without affecting other K ^ channels such as sympathetic tails. We tested this pharmacologically, using the HERG channel- neuron M channels (see below), we analyzed the fast and slow blocking drug W AY 123,398 (Spinelli et al., 1993; Faravelli et al., components of the deactivation tails in more detail by recording 1996). As shown in Figure 1+1,10 p-M W AY 123,398 blocked most currents in the absence and presence of W AY 123,398. Figure 2 of the long deactivation, leaving a residual fast component which exemplifies the results obtained in 86 of 101 cells so examined. was then eliminated by the M channel-blocking drug linopirdine Here, the control current recorded in the absence of W AY (Aiken et al., 1995; Lamas et al., 1997). In contrast, the fast- 123.398 (Fig. 2+1) showed three components, a fast component deactivating current in Figure 1B was strongly blocked by linopir­ with a time constant of 76 msec, and a slower, biexponential dine, leaving a slower component that was blocked in turn by component with time constants of 340 msec and 2.1 sec. W AY W AY 123,398. Thus, comparison of the linopirdine- and W AY- 123.398 (Fig. 2 3 ) eliminated the slower component, leaving only

sensitive currents (Fig. 1,inserts) showed that in fact each cell had the fast component ( t , 75 msec), whereas the difference (W AY- two components to the deactivation currents, fast and slow, and sensitive) current (Fig. 2C) showed only the biexponential slow

that the overall time course of current deactivation was deter­ component ( t , 315 msec and 2.0 sec). Thus, the time constants of mined by their proportion. Furthermore, both components con­ the residual current recorded after application of W AY 123,398 tributed to the sustained current recorded at -2 0 mV. and of the subtracted (WAY-sensitive) current accurately repro- Selyanko et al. • M-Like Current in a Neuronal Cell J. Neurosci., September 15, 1999, 19(18):7742-7756 7747

-100

Control WAY 123,398 (10 nM)

500 pA

A B UJCHO) (WAY-seiîisitive)

Figure 6. Comparison of the slow M-like (^K(M.ng)) current in NG108-15 mouse neuro­ blastoma X rat glioma cells (A) and mouse-ergla current (/^ergia) expressed in CHO cells (5 ). The slowly deactivating component of /K(M,ng) (Aa) was recordcd in response to long hyperpo­ larizing steps (holding level, -2 0 mV; step du­ 200 pA ration, 6 sec; interval, 60 sec; increment, -1 0 mV). These slow /K(M.ng) currents were obtained by subtracting the currents recorded in the pres­ ence of 10 p-M WAY 123,398 {box, right) from those in the absence of the blocker{box, left). For comparison {Ba), /^crgia is shown, obtained with a similar voltage protocol (note the holding potential of 0 mV in Ba). Deactivation tails were 6- fitted by double-exponential curves {smooth 4- lines, superimposed). Leak-subtracted steady- 2 state {filled circles) and instantaneous {open cir­ 0 cles) I-V relationships for /K(M.ng) and /mergla wcrc obtained by measuring the -2 current at the beginning and end of the voltage ■4 pulse. Fast {filled circles) and slow {open circles) -6 time constants for deactivation of /K(M,ng) i^c) -120 -100 60 -40 -20 •120 -100 -80 -60 -40 -20 0 and /merçia i^c) wcrc plotted semilogarithmi- 10000 10000 cally against membrane potential, and t-V rela­ tionships were fitted by straight lines with t at 0 mV and the slope equal to 2134 ± 1 msec and 1000- 1000 0.015 ± 0.0009 m V-^ {filled circles) and 15737 ± 1 msec and 0.016 ± 0.002 mV"^ {open w 100 T 100- circles) in Ac and 680 ± 1 msec and 0.011 ± 0.0008 mV"^ {filled circles) and 5432 ± 1 msec and 0.012 ± 0.0007 mV {open circles) in Be. In Ab, Ac, Bb, and Be the mean data are shown •120 -100 -80 -60 -40 -20 0 -120 -100 -80 -60 -40 -20 0 {vertical lines indicate SEMs) obtained from 26 Membrane potential (mV) Membrane potential (mV) and 29 cells, respectively. duced the fast and slow components of the composite initial gested that these might be composite currents, resulting from the current. In this cell, the fast and slow components contributed 70 deactivation of two different species of channel: one com­ and 30%, respectively of the total tail current, and both contrib­ posed of linopirdine-sensitive KCNQ2I3 subunits, or homologs uted to the steady outward current at the holding potential as thereof (Wang et al., 1998), and the other comprising a member judged from the effect of W AY 123,398 on the holding current. (or members) of the erg family. We therefore sought evidence for On average, in the 86 cells expressing both currents, the fast- and the presence of transcripts of these channels by RT-PCR (see slow-deactivating components contributed -3 3 and 67%, respec­ Materials and Methods). Transcripts for both erg and KCNQ2/3 tively, to the total tail current (Table 1). In the other 15 cells, the were detected (Fig. 3). tail current showed only the slowly deactivating component and Sequence analysis of 10 independent clones from the erg PCR was fully suppressed by W AY 123,398. showed that each clone contained ergl D N A sequence (data not mRNAs for mergl and KCNQ2 and KCNQ3 in shown), suggesting that these cells express predominantly, if not NG108-15 cells exclusively, mergl transcript (London et al., 1997). The KCNQ2I3 The clearly distinguishable effects of linopirdine and W AY transcript contained m R NA for both KCNQ2 and KCNQ3. It may 123,398 on the tail currents illustrated in Figures 1 and 2 sug­ be noted in Fig. 3 that these transcripts were also present in 7748 J. Neurosci., September 15,1999, 19(18):7742-7756 Selyanko et ai. • M-Like Current in a Neuronal Ceil

mV underlying glial cell layer. This latter (negative) finding accords 0- with the absence of any effect of WAY 123,398 on membrane -50 currents recorded from these neurons (see below). Also, no immunostaining of differentiated NG108-15 cells was detected slow'V„„(NG108-15) (CHO) after exposure to an antibody raised against an N-terminal se­ mergla quence unique to the short form of mergl (merglb), the expres­ sion of which is restricted to cardiac cells (London et al., 1997). This suggests that the protein tagged by the C-terminal antibody is the product of the long-form transcript mergla (London et al., 5 0.3- 1997; see also below). Strong immunoreactivity for mergl-CT antibody was also de­ tected in CHO cells transfected with mergla cDNA. No staining 200 pA was observed in untransfected cells, in cells transfected only with the GFP plasmid, or in cells treated with preabsorbed antibody (data not shown).

Slow /K(M,nq) 's mimlckecl by mergla current (/mergla) expressed in CHO cells The above results suggested that the slow component of the C 100 M-like current in NG108-15 cells might well be carried by mergla channels. We tested this further by expressing mergla cDNA in mammalian CHO cells. Figure 5 illustrates the resultant membrane currents, /mergla was activated by membrane depolar­

c ization to equal or positive to -40 mV, but showed substantial o mergla K(M,ng) inactivation during the (long) depolarizing command at poten­

.c tials positive to 0 mV (Fig.5Aa). As a result, the “steady-state” c current-voltage curve was “bell-shaped” (Fig.5Ab). The time course of this composite activation (accompanied by inactivation) could be described by two exponentials, accelerating strongly 0.10.01 10 with depolarization (Fig. 5Ac). When the cell was hyperpolarized [WAY 123,398], pM after a depolarizing prepulse to +50 mV, there was a large transient enhancement of the current, caused by removal of Figure 7. The slow component of the NG108-15 M-like current and the channel inactivation, followed by a slower deactivation (Fig. mergla current are equally sensitive to WAY 123,398. Records in A and 5Ba,b). Two deactivation components were detected that were B show the slow component of /K(M,ng) and /^ergia deactivation tail strongly shortened by membrane hyperpolarization (Fig. 5Bc). currents recorded on stepping from 0 to -5 0 mV in the presence of increasing concentrations of WAY 123,398 (0,0.3,1, and 10 im ). Plots in The mean activation curve deduced from tail currents followed a C show mean percent inhibition of the tail currents{open circles, /K(M,ng)i Boltzmann equation with V y j = -5 9 ± 0.6 mV and k - 12.2 ± n = A; filled circles, /^ergiai " ^ 6). See Table 2 for fitted parameters. 0.5 mV (Fig. 5C). However, when individual curves were fitted, they showed a great variation in Vy2 (range, between -2 7 and 13 mV; n = 9) and small variation ink (range, between 6.9 and 10). dissociated neurons from mouse and rat SCG (see also Shi et al., These results accord well with previous observations on mergla 1997; Wang et al., 1998). currents in oocytes (London et al., 1997). Mergl protein expression in chemically differentiated We next compared the properties of /^ergia deactivation more NG108-15 cells closely with those of the WAY-sensitive slow component of I^- The presence of mRNA transcripts does not necessarily correlate (M ,ng), using the “standard” M current protocol, that is, currents with translated protein products. We therefore tested for the were preactivated by holding at the depolarized potential of 0 mV expression of mergla protein in NG108 cells by immunocyto­ (-2 0 mV in the case o f /^(M .n g )? to avoid contamination by other, chemistry using a specific antibody raised against the C-terminal primarily C a -dependent, currents) and then deactivated by fragment of mergl (mergl-CT; see Materials and Methods). As a 6 sec step commands to various negative potentials. As shown in control for specificity, the m ergl-CT primary antibody was pre­ Figure 6, there was a close correspondence between the two. incubated overnight with a 10-fold molar excess of the immuno­ There were three main differences. First, deactivation of /^ergia genic peptide. Figure A, A and B, compare mergl immunolabeling was preceded by a larger transient reactivation: this presumably in chemically differentiated and nondifferentiated NG108-15 reflected the greater steady-state inactivation of /^ergia 0 mV cells, respectively. Strong labeling was observed in both cell than that of slow /K(M,ng) -2 0 mV. Second, the threshold for bodies and neuropil of most large chemically differentiated activation of /^ergia was ~10 mV more positive than that for slow NG1G8-15 cells of the type we normally selected for electrophys* ^K(M,ng) (Fig- ^AbJSb): the reason for this is not known but may iological recording, whereas smaller cells with less well-developed simply relate to different cell types. Third, whereas both showed processes showed weak or no immunoreactivity. No staining was a biexponential deactivation, the time constants for the two observed in cells treated with preabsorbed antibody. Nondiffer­ components of /mergla deactivation measured at -50 mV were entiated cells, which normally expressed very small /R(M,ng)» -40% of those for deactivation of the slow /K(M,ng> measured at showed moderate or no staining. Interestingly, no staining was the same potential (Table 1). The time constants showed a com­ observed in dissociated mouse SCG neurons (Fig.AC), or over the parable voltage dependence (Fig. 6,compare Ac, Be), so this Selyanko et al. • M-Like Current in a Neuronal Cell J. Neurosci., September 15,1999,19(18):7742-7756 7749

Table 2. Effects of channel blockers on fast and slow /K(M,ng)> ^mergia» andK(M) /,

Blocker Slow /,K(M,ng) 'mergla Fast /,K(M,ng) 'K(M)

WAY 123,398

IC50 (/^m) 0.4 ± 0.07 0.3 ± 0.02 no effect (10 /am ) no effect (10 /am )

Hh 1.0 ± 0.2 0.9 ± 0.07 (4) (6) (37) (9 ) Azimilide

IC50 (/am) 6.5 ± 0.7 6.4 ± 0.3 31.1 ± 1.8 12.9 ± 0.7

" h 0.9 ± 0.08 0.8 ± 0.03 0.9 ± 0.05 1.3 ± 0.09 (5)(4) (4) (3) THA

IC50 (/AM) 11.5 ± 1.5 35.7 ± 9.7 1542 ± 200 1279 ± 75

Dh 0.6 ± 0.05 0.6 ± 0.1 0.7 ± 0.07 0.8 ± 0.03 (3) (7) (3) (4) Linopirdine

IQ o (/AM) » 3 0 >30 1.2 ± 0.02 3.5 ± 0.2 % 1.2 ± 0.03 0.8 ± 0.04 (4) (6) (6) (6) TEA

IC50 (mM) 16.7 ± 3.4 24.1 ± 3.6 1.1 ± 0.09 10.9 ± 0.7

" h 0.8 ± 0.1 0.6 ± 0.1 0.6 ± 0.03 0.6 ± 0.02 (6) (3) (5) (4)

IC50 (/am) 134 ± 6 492 ± 57 248 ± 26 393 ± 36

" h 1.0 ± 0.04 1.0 ± 0.1 0.9 ± 0.07 0.8 ± 0.05 (3) (5) (3) (3)

Mean ± SEM. Number of cells in parentheses. Inhibition constants (IC50) were determined from dose-response curves of the type illustrated in Figures 7 and 9. Effects on fast and slow currents in NG108-15 cells were determined from kinetic analysis of deactivation tails as illustrated in Figure 2, supplemented (where appropriate) by tests after inhibiting the slow component with WAY 123,398, as in Figure 9.

difference may be explained by the different aetivation thresholds M channel- and KCNQ2I3 channel-blocking agent linopirdine at and/or the different size of the voltage step (in four CHO eells concentrations up to 30 /am [Noda et al. (1998) reported an IC50 using a prepulse protocol, the first and the second slow deactiva­ of 36 /AM against the NG108-15 current, but this was measured tion T values for /mergla were slower after a prepulse to -20 mV from the depression of the composite current, and the Hill slope compared with a prepulse to 0 mV, by 41 and 57%, respectively). was rather shallow, so probably reflected its primary action on the For comparison, we also examined the properties of currents fast current component, see below]. Also, both currents were very generated by the short “eardiac” isoform merglb expressed in insensitive to tetraethylammonium (TEA), with IC50 values of 17 CHO cells (/mergibl « = 3; data not shown). Whereas the voltage and 24 mM (Table 2). Thus, /mergla provides a good match for the dependence of /mergib activation and deactivation were very sim­ slowly deactivating component of ^KfM.ng)? both kinetically and ilar to that of the slow /^^ ^g) and of/^^^^gia, both aetivation and pharmacologically. deactivation of /^ergib were several times faster (Table 1), as

previously reported in oocytes by London et al. (1997) (see also Fast /K(M,ng) IS mlmlckecI by the M current (/k(m)) in Lees-Miller et al, 1997). Hence, and in accordance with the lack mouse sympathetic neurons of antibody staining mentioned above and the absenee of merglb As noted above (Fig. 1), the fast component of /K(M,ng) was mRNA expression in the nervous system (London et al., 1997), it readily suppressed by 10 /a m linopirdine. Since linopirdine blocks is unlikely that merglb channels contribute to slow /K(M,ng)- M currents in sympathetic ganglia (Lamas et al., 1997; Wang et

Slow /K (M ,ng) 3"^ / ^ e r g i a sMow Similar pharmacology al., 1998), this suggested that fast/K(M,ng) might correspond to the We next compared the sensitivity of the slowly deactivating “true” (ganglionic-type) M current. We assessed this by compar­ component of the NG108-15 current /K(M,ng) with that of CHO- ing fast WAY-insensitive /K(M,ng) with the M current recorded expressed mergla currents to some blocking drugs. As shown in from dissociated mouse SCG neurons. (We used mouse neurons Figure 7, both tail currents were blocked by the anti-arrhythmic because (1) the parent neuroblastoma to the NG108-15 hybrid drug WAY 123,398 with equal facility (IC50 values, 0.4 and 0.3 ju,M, cell line is derived from mouse neural crest; and (2) sequence respectively; Table 2). They were also equally sensitive to another analysis of the PCR products obtained with the KCNQ primers

anti-arrhythmic drug, azimilide (IC50 values, 6.4 and 6.5 jum, showed that NG108-15 cells expressed both the mouse KCNQ2 respectively; Table 2; Busch et al., 1998). /„,ergia was also inhibited and the mouse KCNQS. Because we have successfully used these

by 9-aminotetrahydroacridine (THA; IC50 value, 36 /a m ; Table 2), primers to amplify the rat KCNQ2 and KCNQ3 genes, we con­ a compound that had previously proved unexpectedly potent in clude that NG108-15 cells express predominantly, if not exclu­ inhibiting the M-like current in NG108-15 cells (Robbins et al, sively, the mouse KCNQ genes.) Figure 8 shows families of fast 1992). In contrast, neither current was inhibited by the ganglionic /K(M,ng) and /k(m) (Ba) activated by membrane depolariza- 7750 J. Neurosci., September 15,1999,79(18):7742-7756 Selyanko et al. • M-Llke Current in a Neuronal Cell

^KiM.naiK (M ,n g) (NG108-15 mouse neuroblastoma x rat glioma) •'‘’nL5Qi e WAY-sensitive ' 100 pA WAY 123,: r (10 (iM) 200 pA

A B 'last'JNG108-15) K (M ,n g) 'm (rnouse SCG) (+WAY 123,398)

K(A)

200 pA Figure 8. Characteristics of the fast M-like (/K(M,ng)) current in a NG108-15 mouse neuro­ blastoma X rat glioma cell (v4) and the M cur­ rent (/k(m)) in a mouse SCG neuron(B). The fast-deactivating component of /K(M,ng) was recorded in the presence of 10 /am WAY 123,398 at different membrane potentials (hold­ ing level, -2 0 mV; step duration, 1 sec; interval, 10 sec; increment, -10 mV).Box, Inhibition of the slow, WAY-sensitive component in this cell: currents before and after application of 10 /am 1.0 WAY 123,398, kft, and the difference current, right. For comparison (Ba), 7^^) is shown, ob­ tained with the same voltage protocol. Deacti­ o 0,0 c\t vation tails were fitted by single exponential 0.0 curves {smooth lines, superimposed). Leak- subtracted steady-state (filled circles) and in­

stantaneous (open circles) I- V relationships for -0.5 ■120 -100 -80 -60 -40 -20 ^K(M,ng) (Ab) and 7^^) i^a) were obtained by -120 -100 -80 -60 -40 -20 measuring the current at the beginning and end C of the voltage pulse. When the time constants for deactivation off^^.^g) (^c) and 7^ ^ ) (Be) 100 were plotted semilogarithmically against mem­ 100- brane potential, t-V relationships were fitted by E straight fines with t at 0 mV and the slope equal to 271 ± 1 msec and 0.009 ± 0.0004 mV inAc and 399 ± 1 msec and 0.013 ± 0.0009 m V “^ in Be. In Ab, Ac, Bb, and Be the mean data are 10 -120 -100 -80 -60 -40 -20 shown (vertical lines indicate SEM) obtained -120 -100 -80 -60 -40 -20 Membrane potential (mV) from 30 and 15 cells, respectively. Membrane potential (mV)

tion to -2 0 mV and deactivated by 1 sec hyperpolarizations at sitive to this compound. Furthermore, both fast /K(M,ng) and

-30 to -100 mV. The slow component of/K(M,ng) was eliminated mouse 7k(m ) were 1.5-2 orders of magnitude less sensitive to using 10 /AM WAY 123,398, as shown in the box in Figure 8. Fast THA than were slow /K(M,ng) or /^ergia (IQo values, 1.5 and 1.3 J^K(M,ng) and 7k(m) had similar voltage dependences (Fig.SA,B) to mM; Table 2). This accords with the relative insensitivity of rat

each other, and also to slow/g^M.ng) and/^ergia (compare Fig. 6). SCG 7k(m ) to THA reported previously (Marsh et al., 1990). However, unlike slow /R(M,ng)» deactivation tails of fast /^^.ng) Figure 9 shows the responses of fast /K(M,ng) and mouse 7^^) to and mouse /^(M) were fitted with single exponential curves: these linopirdine and TEA. Whereas both currents were considerably had time constants similar to each other but much shorter than more sensitive than slow /K(M,ng) or /^ergia to linopirdine, the those in the slow /K(M,ng) and ^^rgia (Table 1). neuroblastoma-glioma current was clearly more sensitive than

Pharmacological comparison of fast /K(M,ng) and the mouse SCG current (IC 5 0 values, 1.2 and 3.5 /a m , respectively; mouse /k,m) Table 2), Likewise, fast /K(M,ng) was more readily blocked than

Like fast /K (M ,n g )> mouse / ^ ( m ) was unaffected by 10 jlam WAY the mouse / k(m ) by TEA (see Discussion). Nevertheless, although 123,398. 7k(m) recorded from 5 rat SCG neurons was also insen­ not completely identical pharmacologically, fast /K(M,ng) and Selyanko et al. • M-Llke Current in a Neuronal Cell J. Neurosci., September 15,1999,19(18):7742-7756 7751

'fast'V„„(NG108-15) 'k(m) (mouse SCG) (+10 nM WAY 123,398)

- 20-1 -50 ' b 0.3- 2 0.3V

200 pA I 50 pA

B a

30" < LU 00 pA I- |100 pA 30“

Linopirdine TEA

100 100

Figure 9. Differential sensitivities of fast /K(M,ng) ^nd

'K(M ) /k(m) to linopirdine and TEA. The fast (WAY-insensitive) .2 50 'K(M) K(M,ng) component o ff^M ^g) and /^(M) was recorded in response to 1 sec steps from the holding level of -2 0 to -5 0 mV, in the absence and presence of different concentrations of linopirdine {Aa, Ab) and TEA {Ba, Bb). C, Concentration dependences of inhibition of the two currents by linopir­ 0.01 100 0.01 100 dine [a) and TEA {b). Smooth lines are the fits by the Hill [blocker], nM [blocker], mM equation. For the parameters of the fit see Table 2.

mouse SCG /^(m) are clearly similar and together show an obvi­ shortened by 25.4 ± 10.0% and 34.6 ± 9.0% {n = 6) and 36.5 ± ous difference from slow /K(M,ng) and /^ergia- 4.8% and 27.8 ± 10.1% (n = 4), respectively. Figure 12 shows examples of inhibitions of fast /K(M,ng) and Both fast and slow fK(M,ng)> and Lergia, are inhibited through activation of M1 muscarinic receptors mouse by Oxo-M. Such inhibitions were observed in four of four NG108-15 cells (mean inhibition, 72.6 ± 13.8%; n = 4). As Activation of muscarinic receptors inhibits both the total expected (Hamilton et al., 1997), similar inhibition was consis­ (composite) M-like current (/K(M,ng)) NG108-15 cells (Fukuda tently observed in sympathetic neurons. et al., 1988; Robbins et al., 1991,1993), and the M current /^(m) in rat (Marrion et al., 1989; Bernheim et al., 1992) and mouse Fast and slow /K(M,ng) control firing in NG108-15 cells (Hamilton et al., 1997) sympathetic neurons. We therefore exam­ The function of the M current is to act as a brake on repetitive ined the effects of a muscarinic stimulant, oxotremorine-M firing. Thus, inhibition of the M current in sympathetic neurons, (Oxo-M; 10 iiu ), on each component of/K(M,ng)» as well as/j„ergiaj either by muscarinic agonists (Brown and Selyanko, 1985) or by in Ml muscarinic receptor-transformed NG108-15 and CHO an M channel-blocking agent (Wang et al., 1998), is associated cells. Figure 10 shows that Oxo-M inhibited the slow /^^i ^g) (^) with increased repetitive firing during depolarizing current injec­ and /mergia (^)- Such inhibitions were observed in seven of eight tions. A similar effect has also been reported in M^-transformed NG108-15 cells (mean inhibition, 42.3 ± 13.8%) and six of six NG108-15 cells after application of a muscarinic agonist (Rob­ mergla-expressing CHO cells (mean inhibition, 50.7 ± 10.8%). bins et al., 1993). However, in the latter case, it is not clear Inhibition of both slow /K(M,ng) and /^ergia was accompanied by a whether this results from inhibition of the fast or slow /K(M,ng), or significant acceleration of their deactivation kinetics (Fig. 11): on both. We tested this by injecting long (7 sec) depolarizing cur­ average, the two components in slow /K(M,ng) and /^ergia were rents into NG108-15 cells and then observing the effects of 7752 J. Neurosci., September 15, 1999, 79(18):7742-7756 Selyanko et ai. • M-üke Current in a Neuronal Cell

W g) (NG108-15 mouse neuroblastoma x rat glioma)

a -20/-50m V Oxo-M (10 LiM) .02 Hz 600

400

200

04

- 20- -60 Control

Oxo-M

100 pA

B mergla (CHO) a 0 / -50 mV Oxo-M(IOpM) 2001 0.025 Hz mm PA

1 0 0 -

Figure 10. Muscarinic inhibition of slow /RtM.ng) in a NG108-15 mouse neuroblastoma x rat glioma cell (/4) and ^mergia Gxprcssed in a CHO cell (B). Currents (Aa, Ba) were produced by holding at -2 0 mV (A) or 0 mV (B) and giving repeated steps (at 0.02 Hzin A and 0.025 Hz in B) to -5 0 mV control for 6 sec. Both steady-state currents at the holding potentials and deactivation currents at the test potential were reduced control Oxo-M by bath-application of oxotrcmorine-M (Oxo-M] 10 /x m ) . In an NG108-I5 cell, inhibition of /K(M.ng) was preceded by a % transient activation of a C a-activated current (Aa). Families of currents in b and c were obtained in response to incremental (-1 0 mV) hyperpolarizing voltage steps before Oxo-M (h) and during (c) action of Oxo-M. The insert in B shows 50 pA superimposed current produced by voltage steps from 0 to 1 = 0 -5 0 mV before and during the action of Oxo-M.

selectively inhibiting fast and slow /^(M.ng) with linopirdine and KCNQ2 and KCNQ3 are analogs of KCNQl, which coassembles W AY 123,398, respectively. The depolarizing currents produced with KCNE (minK) subunits to give the cardiac current /^s (the a short burst of repetitive firing in 20 of 22 NG108-15 cells tested slow component of the cardiac “delayed rectifier”), and mutation and single action potentials in the remaining two cells. Figure 13 of which causes one form of the cardiac long Q T syndrome (Yang shows that 30 /x m linopirdine (which blocked the fast current et al., 1997). KCNQ2 and KCNQ3 have so far been detected only completely and inhibited the slow /K(M,ng) by only 33%) produced in brain and ganglia, and are implicated in a form of juvenile a strong reduction in spike adaptation, whereas 10/ x m WAY epilepsy (Biervert et al., 1998; Charlier et al., 1998; Schroeder et 123.398 had a much weaker effect. Neither linopirdine nor W AY al,, 1998; Singh et al., 1998; Yang et al., 1998). M ergla is one 123.398 had any effect on spike repolarization. isoform of the mouse homolog oferg, originally cloned from a rat DISCUSSION brain hippocampal cDNA library (Warmke and Ganetzky, 1994); m RNA for erg is found mainly in heart and brain (London et al., The main point emerging from this work is that the M-like 1997; Wymore et al., 1997). Mutations of the human homolog current in NG108-15 cells (/K(M.ng)) is a composite current gen­ erated by at least two channel types: a fast-deactivating set of HERG give rise to a cardiac long Q T syndrome (Curran et al., channels similar (but not quite identical) to those carrying the M 1995; Sanguinetti et al., 1996), whereas mutations in Drosophila current in mouse sympathetic neurons and tentatively identified erg are responsible for the seizure phenotype associated with as KCNQ2IKCNQ3 (Wang et al., 1998), and a slower-deactivating hyperactivity in the flight motor pathway (Titus et al., 1997; Wang current probably formed from mergla (London et al., 1997). et al., 1997). Thus, both these channel types have been implicated Selyanko et al. • M-üke Current in a Neuronal Cell J. Neurosci., September 15,1999, 19(18):7742-7756 7753

'slow'V„^,(NG108-15)

A -20 - 2 0 - -5QL -50

Con

Oxo + WAY 123,398 (10 nM)

100 pA

50 pA = 0

■fast'/,,,„^,(NG108-15) 'k(m) Onouse SCG)

(+10 nM WAY 123,398) (CHO) mergla A B B -20—I - 20- ] -50 L -50 L 1 s 1 s -50

Control, WAY 123,398 (10 jiM)

20 pA +Oxo-M (10|j,M)

100 pA 100 pA Figure 11. Muscarinic inhibition of slow /K(M,ng) in a NG108-15 mouse neuroblastoma x rat glioma cell (A) and f^ergia expressed in a CHO cell (B) is accompanied by acceleration of their deactivation kinetics. Super­ Figure 12. Muscarinic inhibition of fast /R(M,ng) in nn NG108-15 mouse imposed are deactivation tails obtained by stepping to -5 0 mV from -2 0 neuroblastoma x rat glioma cell (A) and /^(M) in a mouse sympathetic (B). (A) mV (A) or 0 mV (B) in control (Con) and in the presence of 10 /xm neuron Fast-deactivating7K(M,ng) currents are shown during 1 sec oxotremorine-M (Oxo). Smooth lines are double-exponential fits with time of hyperpolarization from -2 0 to -5 0 mV, before and during the action constants (indicated by arrows) equal to 276 and 1271 msec (Con) and 176 of oxotremorine-M (Oxo-M-, 10 /xm). Both currents were obtained in the and 708 msec (Oxo) in A and 92 and 398 msec (Con) and 71 and 290 msec presence of 10 /xM WAY 123,398. (The effect of WAY 123,398 on the total (Oxo) in B. ^K(M,ng) recorded in this cell with a longer, 6 sec pulse, is shown in the box.) Currents in a mouse sympathetic neuron were obtained before and after addition of WAY 123,398 and WAY 123,398 + Oxo-M. in the control of excitability in vivo. This is the first example of these two channels forming overlapping and functionally similar components of membrane current. both fast and slow components of /K(M,ng) are inhibited by stim­ Our conclusion that two different channels are involved is ulating Ml muscarinic acetylcholine receptors and that the nature based on: (1) the presence of mRNA transcripts and protein; (2) of the inhibition of these two components matches that for inhi­ biophysical properties (voltage threshold and deactivation param­ bition of mouse /^(m) and /mergia^ respectively. eters); (3) sensitivity to potassium channel blockers; and (4) 7k(m) in rat ganglion cells has been ascribed to current through modulation of the channel by an agonist. Thus, we demonstrate channels composed of heteromultimeric assemblies of expressed the presence of mRNA and protein for m e rg l, and mRNA for KCNQ2 and KCNQ3 subunits (Wang et al., 1998). This may also KCNQ2 and KCNQ3, in NG108-15 cells. We show that the be true for in mouse ganglion cells and for the fast compo­ kinetics of mergla heterologously expressed in mammalian cells nent of /K(M,ng) ill NG108-15 cells, because we have found that correspond closely to those of the slow/K(M,ng)- We also show that both cell types show transcripts for these subunits. However, as the two are equally sensitive to the merg-stlecXivQ blocking agents yet, we cannot exclude a contribution from other, homologous, WAY 123,398 and azimilide, and insensitive to the ganglionic M KCNQ subunits. Furthermore, fast /K(M,ng) was distinctly more current and KCNQ2/3 channel-blocking agent linopirdine. On the sensitive to TEA and to linopirdine than was mouse ganglion other hand, we find that the fast /K(M,ng) kinetically matches the /k(m)> suggesting that the subunit composition of the presumed- mouse SCG that these two are pharmacologically similar KCNQ channels in these two cell types might differ. (although not quite identical), and that they can be distinguished The presence of functional mergla chaimels in these cells is from both slow /K(M,ng) &nd /^ergia by their greater sensitivity to not, in itself, particularly surprising, because mRNA transcripts linopirdine and insensitivity to WAY 123,398. We also show that have been identified in a number of neuroblastoma-derived cells 77 5 4 J. Neurosci., September 15,1999, /9(18):7742-7756 Selyanko et al. • M-Like Current in a Neuronal Cell

NG108-15 Cell A Cell B

7s 7s +0.23 nA .22 nA t= -0.07 nA i -0.03 nA

Control Control

0 mV

5 spikes 5 spikes

" -90 mV

+ 30 ^iM linopirdine + 10pM\NAY123J98

II b 16 spikes 6 spikes

+ 10pM W AY 123,398 + 30 |liM linopirdine

Figure 13. Effects of inhibiting fast and slow /K(M,ng) on firing in NG108-15 cells. Records show action potential trains in two NG108-15 cells {A, B) produced by long (7 sec) depolarizing pulses {top records) from the holding 20 mV potential of -9 0 mV in the absence of drugs {Aa, Ba), in 19 spikes 23 spikes the presence of 30 pM linopirdine {Ab), 10 p,M WAY 123,398 {Bb), or in the presence of both linopirdine and WAY 123,398 {Ac, Be).

(including NG108-15) using probes to the human homolog neuroblastoma-derived cell line, not dissimilar to the role of HERG (Bianchi et al., 1998), and an “inwardly rectifying” cur­ ganglionic M channels (Jones and Adams, 1987; Brown, 1988). In rent, retrospectively similar to an erg current, was reported in the present experiments, it appeared that inhibition of the fast NG108-15 cells by Hu and Shi (1997) (see also Bianchi et al., (M) channels had more effect on the response of NG108-15 cells 1998). However, in previous experiments, the properties of this to a sustained current injection than did inhibition of the slow current were mostly studied using solutions containing a raised (merg) channels (Fig. 13). However, the relative contribution of concentration, so that its relation to the M-like current was these two currents to spike frequency adaptation may depend on difficult to discern. It is clear from the present experiments (using the nature of the testing pulse protocol, because merg currents normal external concentrations) that the activation range of activate and deactivate more slowly thanKCNQ currents, and /mergla overlaps that of the true but that deactivation of hence accumulate during repetitive depolarization (Schonherr et /mergla contrffiutes a distinctive slow component to composite al., 1999). current deactivation; and furthermore, that the proportional con­ The transduction mechanism for Mj-mediated inhibition of tributions of /mergla ^nd /^(m) to the totul M-like current vary KCNQ2I3 and mergl is still unknown. Inhibition of /mergia and appreciably from cell to cell. slow /K(M,ng) was accompanied by accelerated deactivation, which The mergla and (presumed) KCNQ currents also overlap func­ may indicate the involvement of protein kinase C (PKC); in tionally. Thus, Chiesa et al. (1997) have provided evidence that HERG currents expressed in Xenopus ooeytes, similar accelera­ erg channels play a role in spike frequency adaptation in another tion produced by thyrotropin-releasing hormone receptor activa- Selyanko et al. • M-Llke Current in a Neuronal Cell J. Neurosci., September 15, 1999, 79(18):7742-7756 7755 tion was mediated by PKC (Barres et al., 1998). This would Bianchi L, Wible B, Arcangeli A, Taglialatela M, Morra F, Castaldo P, accord with an earlier proposal regarding the mechanism of Crociani 0 , Rosati B, Faravelli L, Olivotto M, Wanke E (1998) Herg encodes a current highly conserved in tumours of different histo­ inhibition of the M-like current in NG108-15 cells by bradykinin genesis: a selective advantage for cancer cells? Cancer Res 58: (Higashida and Brown, 1986). Although receptor activation in 815-822. NG108-15 cells produces a strong elevation in intracellular Biervert C, Schroeder BC, Kubisch C, Berkovic SF, Propping P, Jentsch [Ca^"^] (Robbins et al., 1993), it is unlikely that Ca^"^ could be a T, Steinlein OK (1998) A potassium channel mutation in neonatal human epilepsy. Science 279:403-406. messenger for muscarinic inhibition of slow /K(M,ng) or ^mergia Brown DA (1988) M-currents. In: Ion channels, Vol 1, (Narahashi T, because, in CHO cells, the Ca^"^ ionophore ionomycin (5 juM) ed), pp 55-94. New York: Plenum. produced an insignificant reduction in /^ergia (to 89.4 ± 9.7% of Brown DA, Adams PR (1980) Muscarinic suppression of a novel control; n = 3). As a control for the effectiveness of ionomycin in voltage-sensitive K^ current in a vertebrate neurone. Nature 283:673-676. these cells, it caused a complete block of Kvl.2 channels ex­ Brown DA, Higashida H (1988a) Voltage- and calcium-activated potas­ pressed in CHO cells when recorded in perforated-patch or sium currents in mouse neuroblastoma x rat glioma hybrid cells. cell-attached configurations; direct application of 500 UM Ca^^ J Physiol (Lond) 397:149-165. blocked Kvl.2 channels when recorded in the inside-out config­ Brown DA, Higashida H (1988b) Inositol 1,4,5-trisphosphate and diac- ylglycerol mimic bradykinin effects on mouse neuroblastoma x rat uration (A. A. Selyanko and J. K. Hadley, unpublished glioma hybrid cells. J Physiol (Lond) 397:185-207. observations). Brown DA, Selyanko A A (1985) Two components of muscarine- Do products of erg genes contribute to M-like currents in other sensitive membrane current in rat sympathetic neurones. J Physiol neurons? Transcripts for mergl (London et al., 1997), and for the (Lond) 358:335-363. Busch AE, Eigenburger B, Jurkiewicz NK, Salata JJ, Pica H, Suessbrich rat homologs ergl and erg3 (Shi et al., 1997; Wymore et al., 1997) H, Lang F (1998) Blockade of HERG channels by the class III anti- are present in mammalian brain. Furthermore, both expressed arrhythmic azimilide: mode of action. Br J Pharmacol 123:23-30. mergla currents and the slow, presumed-mergla component of Charlier C, Singh NA, Ryan SG, Lewis TB, Reus BE, Leach RJ, Leppert the M-like current in NG108-15 cells were inhibited by stimulat­ M (1998) A pore mutation in a novel KQT-like potassium channel gene in an idiopathic epilepsy family. Nat Genet 18:53-55. ing Ml muscarinic receptors, so they could contribute to Chiesa N, Rosati B, Arcangeli A, Olivotto M, Wanke E (1997) A novel muscarinic-inhibitable M-like currents previously recorded in role for HER G channels: spike-frequency adaptation. J Physiol central neurons. True, the presence of mRNA transcripts may not (Lond) 501:313-318. betoken the assembly of functional channels: thus, no appropriate Constanti A, Brown DA (1981) M-Currents in voltage-clamped mam- mahan sympathetic neurones. Neurosci Lett 24:289-294. erg-like component of membrane current could be recorded from Curran M E, Splawski I, Timothy KW, Vincent GM, Green ED, Keating mouse or rat sympathetic neurons, in spite of the presence of M T (1995) A molecular basis for cardiac arrhythmia: HERG muta­ mRNAs (Shi et al, 1997; see also this paper), nor could we detect tions cause long QT syndrome. Cell 80:795-804. mergl immunoreactivity. However, this may not be the case for Faravelli L, Arcangeli A, Olivotto M, Wanke E (1996) A HERG-like K^ channel in rat F-11 DRG cell line: pharmacological identification other mammalian neurons. For example, the M-like current re­ and biophysical characterization. J Physiol (Lond) 496:13-23. corded from isolated rat cortical neurons has been reported to be Fukuda K, Higashida H, Kubo T, Maeda A, Akiba I, Bujo H, Mishina M, an order of magnitude less sensitive to linopirdine than either the Numa S (1988) Selective coupling with K^ currents of muscarinic ganglionic or hippocampal cell current (Noda et al, 1998; cf. acetylcholine receptor subtypes in NG108-15 cells. Nature 335:355-358. Aiken et al, 1995; Lamas et al, 1997; Schnee and Brown, 1998). Hamilton SE, Loose MD, Qi M, Levey AI, Hille B, McKnight GS, Although other explanations are possible, our findings suggest Idzerda RL, Nathanson N M (1997) Disruption of the m l receptor that this might arise from a contribution by erg channels to the gene ablates muscarinic receptor-dependent M current regulation and cortical neuron current. In view of the significance of M-like seizure activity in mice. Proc Natl Acad Sci USA 94:13311-13316. Higashida H, Brown DA (1986) Two polyphosphatidylinositide metab­ channels as potential targets for cognition-enhancing drugs (Zac- olites control two K^ currents in a neuronal cell. Nature 323:333-335. zek and Saydoff, 1993), further information regarding the degree Hu Q, Shi Y L (1997) Characterization of an inward-rectifying potas­ of heterogeneity in the molecular composition of the channels sium current in NG108-15 neuroblastoma x glioma cells. Pfiügers Arch underlying M-like currents in different neurons would be helpful. 433:617-625. Jones SW, Adams PR (1987) The M-current and other potassium cur­ rents of vertebrate neurons. In: Neuromodulation (Kaczmarek LK, Levitan IB, eds), ppl59-186. New York: Oxford UP. REFERENCES Lamas JA, Selyanko AA, Brown DA (1997) Effects of a cognition- enhancer, linopirdine (DuP 996), on M-type potassium currents (Ik(m )) and some other voltage- and ligand-gated membrane currents in rat Abogadie PC, Vallis Y, Buckley NJ, Caulfield MP (1997) Use of sympathetic neurons. Eur J Neurosci 9:605-616. antisense-generating plasmids to probe the function of signal transduc­ Lees-Miller JP, Kondo C, Wang L, Duff HJ (1997) Electrophysiological tion proteins in primary neurons. In: Methods in molecular biology, Vol characterization of an alternatively processed ERG K^ channel in 83: Receptor signal transduction protocols (Challiss RAJ, ed), pp mouse and human hearts. Circ Res 81:719-726. 217-225. Totowa, NJ: Humana. London B, Trudeau MC, Newton KP, Beyer AK, Copeland NG, Gilbert Aiken SP, Lampe BJ, Murphy PA, Brown BS (1995) Reduction of spike DJ, Jenkins NA, Satler CA, Robertson GA (1997) Two isoforms of frequency adaptation and blockade of M-current in rat CAl pyramidal the mouse ether~a-go-go-Te\a.teà gene coassemble to form channels with neurones by linopirdine (DuP 996), a neurotransmitter release en­ properties similar to the rapidly activating component of the cardiac hancer. Br J Pharmacol 115:1163-1168. delayed rectifier K^ current. Circ Res 81:870-878. Barros F, Gomez-Varela D, Viloria CO, Palomero T, Giraldez T, De-la- Marrion N V (1997) Control of M-current. Annu Rev Physiol Pena P (1998) Modulation of human erg channel gating by acti­ 59:483-504. vation of a G protein-coupled receptor and protein kinase C. J Physiol Marrion N V, Smart TG, Marsh SJ, Brown DA (1989) Muscarinic sup­ (Lond) 511:333-346. pression of the M-current in the rat sympathetic ganglion is mediated Bernheim L, Mathie A, Hille B (1992) Characterization of muscarinic by receptors of the M^-subtype. Br J Pharmacol 98:557-573. receptor subtypes inhibiting Ca^’^ current and M current in rat sym­ Marsh SJ, Hubbard A, Brown DA (1990) Some actions of 9-amino- pathetic neurons. Proc Natl Acad Sci USA 89:9544-9548. I,2,3,4-tetrahydroacridine (THA) on cholinergic transmission and 7756 J. Neurosci., September 15,1999, 79{18):7742-7756 Selyanko et al. • M-Like Current in a Neuronal Ceil

membrane currents in rat sympathetic ganglia. Eur J Neurosci channels in neuroblastoma-glioma hybrid cells: inhibition by muscarine 2:1127-1134. and bradykinin. Receptors Channels 3:147-159. Morielli AD, Peralta EG (1995) Suppression of a potassium channel by Singh NA, Charlier C, Stauffer D, DuPont BR, Leach RJ, Melis R, Ronen G-protein coupled receptors. Life Sci 56:1035. GM, Bjerre I, Quattlebaum T, Murphy JV, McHarg ML, Gagnon D, Mullaney I, Dodd MW, Buckley N, Milligan G (1993) Agonist activation Rosales T, Peiffer A, Anderson VE, Leppert M (1998) A novel po­ of transfected human M l muscarinic acetylcholine receptors in CHO tassium ehannel gene,KCNQ2, is mutated in an inherited epilepsy of cells results in down-regulation of both the receptor and the a-subunit newborns. Nat Genet 18:25-29. of the G-protein Gq. Biochem J 289:125-131. Shi W, Wymore RS, Wang HS, Pan Z, Cohen IS, McKinnon D, Dixon JE Noda M, Obana M, Akaike N (1998) Inhibition of M-type current by (1997) Identification of two nervous system-specific members of the erg linopirdine, a neurotransmitter-release enhancer, in NG108-15 neuro­ potassium channel gene family. J Neurosci 17:9423-9432. nal cells and rat cerebral neurons in culture. Brain Res 794:274-280. Spinelli W, Moubarak IF, Parsons RW, Colatsky TJ (1993) Cellular Owen DG, Marsh SJ, Brown DA (1990) M-current noise and putative eleetrophysiology of WAY-123,398, a new class III anti arrhythmic M-channels in cultured rat sympathetic ganglion cells. J Physiol (Lond) agent: specificity of I^ block and lack of reverse use dependence in cat 431:269-290. ventricular myocytes. Cardiovasc Res 27:1580-1591. Rae J, Cooper K, Gates P, Watsky M (1991) Low access resistance Titus SA, Warmke JW, Ganetzky B (1997) The Drosophila erg perforated patch recordings using amphotericin B. J Neurosci Methods channel polypeptide is encoded by the seizure locus. J Neurosci 37:15-26. 17:675-681. Robbins JA, Caulfield MP, Higashida H, Brown DA (1991) Genotypic Wang H-S, Pan Z, Brown BS, Wymore RS, Cohen IS, Dixon JE, McK­ m3-muscarinic receptors preferentially inhibit M-currents in DNA- innon D (1998) KCNQ2 and KCNQ3 potassium channel subunits: transfected NG108-15 neuroblastoma x glioma hybrid cells. Eur J Neu­ molecular correlates of the M-channel. Science 282:1890-1893. rosci 3:820-824. Wang XJ, Reynolds ER, Deak P, Hall, L M (1997) The seizure locus Robbins J, Trouslard J, Marsh SJ, Brown DA (1992) Kinetic and phar­ encodes the Drosophila homolog of the HERG potassium channel. macological properties of the M-current in rodent neuroblastoma x J Neurosci 17:882-890. glioma hybrid cells. J Physiol (Lond) 451:159-185. Warmke JW, Ganetzky B (1994) A family of potassium channel genes Robbins J, Marsh SJ, Brown DA (1993) On the mechanism of M-current related to eag in Drosophila and mammals. Proc Natl Acad Sci USA inhibition by muscarinic ml receptors in DNA-transfected rodent neu­ 91:3438-3442. roblastoma X glioma cells. J Physiol (Lond) 469:153-178. Wymore RS, Gintant GA, Wymore RT, Dixon JE, McKinnon D, Cohen Sanguinetti MC, Curran M E, Spector PS, Keating M T (1996) Spectrum IS (1997) Tissue and species distribution of mRNAfor the iKr-likeK"^ oiHER G K^ channel dysfunction in an inherited cardiac arrhythmia. Proc Natl Acad Sci USA 93:2208-2212. channel, erg. Circ Res 80:261-268. Schafer S, Behe P, Meves H (1991) Inhibition of the M current in NG Yang W-P, Levesque PC, Little WA, Condor ML, Shalaby FY, Blanar 108-15 neuroblastoma x glioma hybrid cells. Pfiügers Arch M A (1997) KvLQTl, a voltage-gated potassium channel responsible 418:581-591. for human cardiac arrhythmias. Proc Natl Acad Sci USA 94:4017-4021. Schnee M E, Brown BS (1998) Selectivity of linopirdine (DuP 996), a Yang W-P, Levesque PC, Little WA, Conder ML, Ramakrishnan P, neurotransmitter release enhancer, in blocking voltage-dependent and Neubauer MG, Blanar M A (1998) Functional expression of two calcium-activated potassium currents in hippocampal neurons. J Phar­ KvLQTl-related potassium channels responsible for an inherited idio­ macol Exp Ther 286:709-717. pathic epilepsy. J Biol Chem 273:19419-19423. Schonherr R, Rosati B, Hehl S, Rao VG, Arcangeli A, Olivotto M, Yokoyama S, Imoto K, Kawamura T, Higashida H, Iwabe N, Miyata T, Heinemann SH, Wanke E (1999) Functional role of the slow activa­ Numa, S (1989) Potassium channels from NG108-15 neuroblastoma- tion property of ERG K^ channels. Eur J Neurosci 11:753-760. glioma hybrid cells. Primary structure and functional expression from Schroeder BC, Kubisch C, Stein V, Jentsch TJ (1998) Moderate loss of cDNAs. FEBS Lett 259:37-42. function of cyclic-AMP-modulated KCNQ2IKCNQ3 K"^ channels Zaczek R, Saydoff J (1993) Depolarization activated releasers of trans­ causes epilepsy. Nature 396:687-690. mitters as therapeutics for dementia: preclinical characterization of Selyanko AA, Robbins J, Brown DA (1995) Putative M-type potassium linopirdine (DuP 996). Curr Opin Invest Drugs 2:1097-1104.