Article Information

• PDF (313 kb)
• Full Text with Large Figures
• Cited by in Scopus (5)

PubMed

• Articles by Henrik Sindal Jensen
• Articles by Søren-Peter Olesen

Related Articles

• Gating and Flickery Block Differentially Affected by Rubidiu...

  Gating and Flickery Block Differentially Affected by Rubidium in Homomeric KCNQ1 and Heteromeric KCNQ1/KCNE1 Potassium Channels Biophysical Journal, Volume 78, Issue 1, 1 January 2000, Pages 211-226 Michael Pusch, Lara Bertorello and Franco Conti Abstract The voltage-gated potassium channel KCNQ1 associates with the small KCNE1 subunit to form the cardiac IKs delayed rectifier potassium current and mutations in both genes can lead to the long QT syndrome. KCNQ1 can form functional homotetrameric channels, however with drastically different biophysical properties compared to heteromeric KCNQ1/KCNE1 channels. We analyzed gating and conductance of these channels expressed in oocytes using the two-electrode voltage-clamp and the patch-clamp technique and high extracellular potassium (K) and rubidium (Rb) solutions. Inward tail currents of homomeric KCNQ1 channels are increased about threefold upon substitution of 100mM potassium with 100mM rubidium despite a smaller rubidium permeability, suggesting an effect of rubidium on gating. However, the kinetics of tail currents and the steady-state activation curve are only slightly changed in rubidium. Single-channel amplitude at negative voltages was estimated by nonstationary noise analysis, and it was found that rubidium has only a small effect on homomeric channels (1.2-fold increase) when measured at a 5-kHz bandwidth. The apparent single-channel conductance was decreased after filtering the data at lower cutoff frequencies indicative of a relatively fast “flickery/block” process. The relative conductance in rubidium compared to potassium increased at lower cutoff frequencies (about twofold at 10Hz), suggesting that the main effect of rubidium is to decrease the probability of channel blockage leading to an increase of inward currents without large changes in gating properties. Macroscopic inward tail currents of heteromeric KCNQ1/KCNE1 channels in rubidium are reduced by about twofold and show a pronounced sigmoidal time course that develops with a delay similar to the inactivation process of homomeric KCNQ1, and is indicative of the presence of several open states. The single channel amplitude of heteromers is about twofold smaller in rubidium than in potassium at a bandwidth of 5kHz. Filtering at lower cutoff frequencies reduces the apparent single-channel conductance, the ratio of the conductance in rubidium versus potassium is, however, independent of the cutoff frequency. Our results suggest the presence of a relatively rapid process (flicker) that can occur almost independently of the gating state. Occupancy by rubidium at negative voltages favors the flicker-open state and slows the flickering rate in homomeric channels, whereas rubidium does not affect the flickering in heteromeric channels. The effects of KCNE1 on the conduction properties are consistent with an interaction of KCNE1 in the outer vestibule of the channel. Abstract | Full Text | PDF (280 kb)

• MiRP1 Forms IKr Potassium Channels with HERG and Is Associat...

  MiRP1 Forms IKr Potassium Channels with HERG and Is Associated with Cardiac Arrhythmia Cell, Volume 97, Issue 2, 16 April 1999, Pages 175-187 Geoffrey W Abbott, Federico Sesti, Igor Splawski, Marianne E Buck, Michael H Lehmann, Katherine W Timothy, Mark T Keating and Steve A.N Goldstein Summary A novel potassium channel gene has been cloned, characterized, and associated with cardiac arrhythmia. The gene encodes MinK-related peptide 1 (MiRP1), a small integral membrane subunit that assembles with HERG, a pore-forming protein, to alter its function. Unlike channels formed only with HERG, mixed complexes resemble native cardiac I channels in their gating, unitary conductance, regulation by potassium, and distinctive biphasic inhibition by the class III antiarrhythmic E-4031. Three missense mutations associated with long QT syndrome and ventricular fibrillation are identified in the gene for MiRP1. Mutants form channels that open slowly and close rapidly, thereby diminishing potassium currents. One variant, associated with clarithromycin-induced arrhythmia, increases channel blockade by the antibiotic. A mechanism for acquired arrhythmia is revealed: genetically based reduction in potassium currents that remains clinically silent until combined with additional stressors. Summary | Full Text | PDF (257 kb)

• N-Type Calcium Channel Inactivation Probed by Gating-Current...

  N-Type Calcium Channel Inactivation Probed by Gating-Current Analysis Biophysical Journal, Volume 76, Issue 5, 1 May 1999, Pages 2530-2552 Lisa P. Jones, Carla D. DeMaria and David T. Yue Abstract N-type calcium channels inactivate most rapidly in response to moderate, not extreme depolarization. This behavior reflects an inactivation rate that bears a U-shaped dependence on voltage. Despite this apparent similarity to calcium-dependent inactivation, N-type channel inactivation is insensitive to the identity of divalent charge carrier and, in some reports, to the level of internal buffering of divalent cations. Hence, the inactivation of N-type channels fits poorly with the “classic” profile for either voltage-dependent or calcium-dependent inactivation. To investigate this unusual inactivation behavior, we expressed recombinant N-type calcium channels in mammalian HEK 293 cells, permitting in-depth correlation of ionic current inactivation with potential alterations of gating current properties. Such correlative measurements have been particularly useful in distinguishing among various inactivation mechanisms in other voltage-gated channels. Our main results are the following: 1) The degree of gating charge immobilization was unchanged by the block of ionic current and precisely matched by the extent of ionic current inactivation. These results argue for a purely voltage-dependent mechanism of inactivation. 2) The inactivation rate was fastest at a voltage where only ∼ 1/3 of the total gating charge had moved. This unusual experimental finding implies that inactivation occurs most rapidly from intermediate closed conformations along the activation pathway, as we demonstrate with novel analytic arguments applied to coupled-inactivation schemes. These results provide strong, complementary support for a “preferential closed-state” inactivation mechanism, recently proposed on the basis of ionic current measurements of recombinant N-type channels (Patil et al., 1998. . 20:1027–1038). Abstract | Full Text | PDF (418 kb)

• …more

Copyright © 2007 The Biophysical Society. All rights reserved.
Biophysical Journal, Volume 92, Issue 8, 2747-2756, 15 April 2007

doi:10.1529/biophysj.106.101287

Channels, Receptors, and Electrical Signaling

Inactivation as a New Regulatory Mechanism for Neuronal Kv7 Channels

Henrik Sindal Jensen^*, ‡, ,  , Morten Grunnet^*, † and Søren-Peter Olesen^*, †

^* The Danish National Research Foundation Centre for Cardiac Arrhythmia, Department of Medical Physiology, The Panum Institute, University of Copenhagen, DK-2200 Copenhagen N, Denmark
^† NeuroSearch A/S, Ballerup, Denmark
^‡ H. Lundbeck A/S, Copenhagen, Denmark

Address reprint requests to Henrik Sindal Jensen, Dept. of Molecular and Cellular Pharmacology, H. Lundbeck A/S, Ottiliavej 9, DK-2500 Copenhagen, Denmark. Tel.: 45-36-43-26-23; Fax: 45-36-43-82-71.

The Kv7 family of ion channels has received much attention since they have been identified as the molecular correlates of several important human K⁺ conductances. The Kv7 family is unique in the sense that mutations in four of the five genes have been linked to human hereditary diseases 1,2. Within the Kv7 family the most thorough characterization has been performed on the Kv7.1 channel (for a recent review see Jespersen et al. 3). It was cloned in 1996 from patients suffering from the proarrhythmic condition long QT syndrome 4 and later demonstrated to associate with the β-subunits KCNE1 to constitute the cardiac I_Ks current 5,6.

The neuronal M-current was first described in 1980 as slowly activating, noninactivating K⁺ current, which could be inhibited through activation of muscarinic receptors 7. The molecular component behind the M-current was later identified as Kv7.2 and Kv7.3, and it has been demonstrated how mutations in these genes are linked to a neonatal form of epilepsy 8,9,10,11,12. Kv7.4 was cloned in 1999 and mutations in this gene have been associated with deafness 13,14,15. Kv7.5, like Kv7.2–4, is primarily neuronal, although no disease has yet been associated to mutations in this gene 16,17,18.

Due to the tight coupling of Kv7 channels to dysfunction and disease, great effort has been invested to understand the regulation of these channels. Kv7.1 channels are modulated by a number of factors including protein kinase A (PKA) and protein kinase C (PKC), cell volume changes, external acidification, and phosphatidyl inositol phosphate (PIP₂) 19,20,21,22,23,24. Additionally, Kv7.1 channel activity can be regulated by interaction with all members of the KCNE family of β-subunits 5,6,25,26,27,28.

As the molecular correlates of the M-current, Kv7.2 and Kv7.3 can be inhibited via muscarinic M1 receptor activation, although inhibition of the M-current can also be achieved through activation of other G-protein coupled receptors 29,30,31. PIP₂ has been demonstrated to be the link between muscarinic receptors and the M-current 32,33,34,35, and it is necessary for maintaining channel activity. Kv7.4 channels are regulated by factors such as PKA, PKC, intracellular Ca²⁺, changes in cell volume, PIP₂, and phosphorylation by glucocorticoid-inducible kinases 19,36,37. Finally, changes in cell volume, extracellular Zn²⁺, acidification, and muscarinic receptor activation modulate Kv7.5 16,17. In addition to the described regulations, Kv7 channel activation is voltage dependent. A common feature is an initial opening at potentials around −60mV and a slow time constant for both activation and deactivation compared to other voltage-gated K⁺ channels.

One biophysical parameter that differentiates the Kv7 channel subtypes is inactivation. Inactivation has been reported for Kv7.1 channels and has been demonstrated to be time and voltage dependent 5,6. Coassembly with either KCNE1 or KCNE4 β-subunits can eliminate Kv7.1 channel inactivation 38,39,40. In contrast, it is generally believed that Kv7.2-5 channels—along with the M-current—do not inactivate 7,11,13,17,18,39. However, a recent report shows that the activity of Kv7.4 channels can be augmented by hyperpolarizing prepulses, indicating that they may undergo inactivation 37. We have devoted this study to a thorough investigation of the possible inactivation properties of all the Kv7 channels with emphasis on Kv7.4 channels.

cRNA was synthesized by standard T7 in vitro run-off transcription of human Kv7.1, Kv7.2, Kv7.3, and Kv7.4 and murine Kv7.5. A total of 50 nl of ∼0.5μg/μl cRNA was injected into stage V–VI oocytes isolated from Xenopus laevis according to national guidelines. Oocytes were kept at 19°C in Kulori solution (in mM: 90 NaCl, 1 KCl, 1 MgCl₂, 1 CaCl₂, 5 HEPES, pH 7.4) and currents could be detected after 2–3 days.

Conventional two-electrode voltage-clamp recordings were conducted at room temperature in Kulori solution using a Dagan CA-1D amplifier (Minneapolis, MN) and Pulse software (HEKA, Lambrecht/Pfalz, Germany). The borosilicate electrodes were filled with 1M KCl and had a tip resistance of 0.5–2 MΩ. All drugs were dissolved in dimethylsulfoxide (DMSO) and diluted so the final DMSO concentration was <0.3%. At this concentration no vehicle effects were observed. Retigabine, BMS-204352, and XE-991 were synthesized at NeuroSearch A/S.

Data were obtained from more than one batch of oocytes, and analyses were performed in SigmaPlot. Data analyses were done with the Marquardt-Levenberg algorithm, and decaying currents were best fitted to a three-parameter single exponential function: I(x)=A_∞+A×exp(−x/τ). Rising currents were best fitted to a three-parameter single exponential function: I(x)=A₀+A×(1-exp(−x/τ). For both equations I(x) is the current at time or voltage x, A is the amplitudes of the exponential, and τ is the time or voltage constant. For functions describing an exponential decline A_∞ is the horizontal asymptotic value at time ∞, and for those describing an exponential rise A₀ is the current amplitude at time 0. The percentage of control are calculated as the I_x/I_CTRL×100%, where I_x is the normalized current amplitude after retigabine, XE-991, or BMS-204352 application and I_CTRL is the corresponding current amplitude recorded in Kulori solution. Normalization is done to the current amplitude recorded at V_act=+40mV after a V_pre of −120mV. The linear curve fits on the percentage of control data for the pharmacological experiments are fitted to p(x)=αx+β, where p(x) is the percentage of control at potential x, α is the slope, and β the crossover point of x=0. Values are given as mean±SE where n indicates the number of observations, and the mean is calculated as the mean of the particular parameter from several individual fits.

The extent of inactivation of homomeric Kv7.1, 2, 4, and 5 channels was investigated by two-electrode voltage clamp experiments in Xenopus oocytes. After injection of the respective cRNAs, typical Kv7 currents could be activated by standard voltage-step protocols consisting of a 3-s prepulse at −80mV (V_pre), a 2-s activation pulse at gradually more positive potentials from −100 to +60mV in 20-mV increments to activate the channels (V_act), and ended by a 1-s step to −30mV to facilitate recording of the tail current (Figure 1AD). Water-injected control oocytes did not give rise to interfering currents. From such experiments, the inactivation of the Kv7 channels is normally not visible, although indication of an inactivation process can be seen from a small hook on the tail current indicative of a release from inactivation (e.g., Figure 1AD).

Display large version of this figure

Figure 1

Two-electrode voltage-clamp experiments on Kv7.1, 2, 4, and 5 homomeric channels in Xenopus oocytes (A–D). The channels were activated by voltage steps from −80mV to potentials ranging from −100 to +60mV in 20-mV increments (2-s duration), and tail currents were measured at −30mV. Indications of an inactivation process of Kv7.1 and Kv7.5 can be seen by the small hook on the tail current caused by release from inactivation during the initial phase of the 1-s pulse to −30mV as well as the marked difference in activation/deactivation kinetics.

To study this phenomenon in more detail, we designed the inactivation protocol shown in Fig. 2 with an initial 20-s-duration V_pre to clamp the cell membrane at holding potentials of −120mV to +60mV in 20-mV increments. This step was followed by a V_act step to +40mV. During the 5-s test pulse at +40mV the noninactivated channels do activate fully, but since they also to some degree inactivate during the test pulse, we use the peak current as a measure of available channels. Due to the channels inactivating during the upstroke of the test current, the protocol may tend to overestimate the degree of inactivation. Plotting the maximal current amplitudes measured during the V_act step to +40mV as a function of the preceding V_pre holding potential clearly revealed the steady-state inactivation of homomeric Kv7.1, Kv7.4, and Kv7.5 (Figure 2AD). At physiologically relevant potentials, inactivation was most pronounced for Kv7.4 and Kv7.5, which displayed an inactivation of ∼30% at V_pre=−50mV when compared to V_pre=−120mV (Figure 2E). These experiments demonstrate that in a typical resting neuron (V_m ∼ −70mV), significant Kv7 channel inactivation occurs. For the following in-depth investigations, we focused on Kv7.4.

Display large version of this figure

Figure 2

Magnitude of inactivation. (A–D) The inactivation properties of homomeric Kv7 channel complexes were investigated by voltage-clamp experiments in Xenopus oocytes. After a 20-s-long prepulse at potentials ranging from −120mV to +60mV (V_pre), the maximal current amplitudes were measured during a 5-s pulse at +40mV (V_act) for each channel subtype: A. Kv7.1, B. Kv7.2, C. Kv7.4, and D. Kv7.5 (inset shows the voltage protocol). (E) The degree of inactivation of each Kv7 channel complex is revealed by plotting the current amplitude normalized to the level measured after a V_pre of −120mV as a function of the preceding V_act potential. Maximal inactivation of ∼80% is observed for Kv7.5 at +60mV.

The onset of inactivation was studied by a three-step protocol consisting of a 10- or 20-s prepulse clamping the cell membrane at −120mV to recover from inactivation, followed by a step to 0mV of variable duration (0–28s in 2-s increments), allowing different degrees of channel inactivation, and finally a 5-s step to +40mV (Figure 3A). The time-dependent onset of inactivation is shown by plotting the maximal current amplitude recorded at the V_act step to +40mV as a function of the time spent at 0mV (Figure 3B). The results could best be fitted to a three-parameter one-exponential function: I(x)=A_∞+A×exp(−x/τ), which gave a mean decay time constant, τ, of 9.55±1.14s, a noninactivation component, A_∞, of 31%±2.8%, and a relative amplitude of the inactivating current, A, of 72%±3.5%, n=7. To investigate if the 10-s prepulse at −120mV was long enough to facilitate full recovery from inactivation, we repeated the experiment with a V_pre at −120mV of 20s. This resulted in a significantly faster inactivation (τ of 5.06±1.10s, n=3; P<0.05, Student's t-test) and a significantly lower maximal inactivation and amplitude of the exponential (A_∞=48%±4.3%, P<0.01 and A=51%±3.9%, P<0.05), indicating that some inactivation must remain even during a 10-s prepulse at −120mV (Figure 3B).

Display large version of this figure

Figure 3

Time dependence of inactivation. (A) The onset of inactivation of Kv7.4 at 0mV was studied by a two-step protocol designed to allow the channels to inactivate increasingly. The inset in B shows the voltage protocol employed, which clamped the cell membrane at −120mV for 20s to allow for recovery from inactivation (V_pre), followed by a step to 0mV of 0–28-s duration in 2-s increments. The second step was a depolarizing step to +40mV for 5s during which the maximal current amplitude was measured. This experiment was also performed with a 10-s-long V_pre (current traces not shown). (B) Plotting the maximal current amplitude normalized to the amplitude at V_{0mV, 0s} as a function of the time spent at 0mV revealed a single-exponential correlation of the onset of inactivation for both protocols. Fitting the data from the two experiments to I(x)=A_∞+A×exp(−x/τ) gave significantly different time constants: τ_{10s prepulse}=9.55±1.14s (n=7) and τ_{20s prepulse}=5.06±1.10s (n=3); P<0.05, t-test. The curves are generated from the fitted parameters.

The time-dependent recovery from inactivation of Kv7.4 was studied by a two-step protocol composed of a 20-s-long prepulse at −120mV followed by a 5-s activation pulse to +40mV (V1_act). During this first depolarizing pulse the channels inactivate. To quantify the time needed to recover from inactivation, this pulse was followed by a hyperpolarizing step to −120mV ranging from 250ms to 3750ms in 250-ms increments. The recovery from inactivation is reflected in the current amplitude elicited by a second pulse to +40mV (V2_act) (Figure 4A). A clear time-dependent recovery from inactivation was shown by plotting the current amplitude of V2_act in percentage of the preceding V1_act maximal current amplitude as a function of the time spent at −120mV. This could best be fitted to a rising three-parameter one-exponential function with a time constant of 1.47±0.21s, an A₀ of 69.8%±2.6%, and an A of 24.8%±3.0%, n=7 (Figure 4B).

Display large version of this figure

Figure 4

Time-dependent recovery from inactivation of Kv7.4. (A) The time course of recovery from inactivation was studied by a two-step protocol in which two activating pulses follow a 20-s-long hyperpolarizing pulse at −120mV. Each +40-mV pulse (V1_act and V2_act) is of 5-s duration, and the variable parameter is the length of the separating −120mV hyperpolarizing pulse. The duration of the hyperpolarizing pulse ranges from 250ms to 3750ms in 250-ms increments (inset shows the voltage protocol). (B) The maximal current amplitude at V2_act is depicted in percentage of the maximal current amplitude at V1_act subtracting the A₀ (the calculated current amplitude in percentage at time 0) as a function of the release time at −120mV (i.e., I[V2_act]/I[V1_act]×100%-A₀), and this revealed a single-exponential relationship between the release time and the degree of recovery from inactivation. Fitting the data to I(x)=A₀+A×(1-exp(−x/τ)) gave a time constant τ=1.47±0.21s (n=7). The curve is generated from the fitted parameters.

Not only increasing time but also increasing hyperpolarizing potentials would be expected to release the inactivation of Kv7.4. We employed a protocol similar to that shown in Fig. 4, but the time between V1_act and V2_act fixed at 3s. The variable parameter in this experiment was the holding potential of the intermediate hyperpolarizing step that releases the inactivation, which was clamped at potential ranging from −30 to −120mV (Figure 5A). Similar to the time dependence experiments, we plotted the maximal current amplitude of V2_act in percentage of the V1_act maximal current amplitude but now as a function of the holding potential of the intermediate hyperpolarizing step (Figure 5B). Fitting these data to a three-parameter one-exponential decaying function resulted in a voltage constant of 54.9±3.4mV, A_∞=39.7%±3.7%, and A=6.0%±0.88%, n=10 (Figure 5B).

Display large version of this figure

Figure 5

Voltage-dependent recovery from inactivation of Kv7.4. (A) The kinetics of recovery from inactivation was studied by a two-step protocol similar to the one used in the time-dependent recovery experiments, but with the duration of the release pulse fixed at 3s. The variable in this experiment was the voltage employed to clamp the membrane during the release pulse, which ranged from −120mV to −30mV in 10-mV increments (inset shows the protocol). (B) The maximal current amplitude recorded at V2_act in percentage of the corresponding amplitude of V1_act plotted as a function of the holding potential of the release pulse could best be fitted to a single-exponential function: I(x)=A_∞+A×exp(−x/τ), resulting in a voltage-constant τ=54.9±3.4mV (n=10).

As Kv7.3 and Kv7.4 can form functionally distinct heteromers 13, we wanted to investigate the inactivation properties of coexpressed Kv7.3 and Kv7.4, and we did this by a protocol comparable to the one shown in Fig. 2 (Figure 6A). A significant difference in the inactivation properties of homomeric Kv7.4 versus heteromeric Kv7.3+4 channels was evident in the range −40 mV–+60mV when plotting the maximal current amplitude of the V_act step as a function of the V_pre potential (Figure 6B). In contrast to homomeric Kv7.4 channels, which display a flat S-shaped course in the inactivation behavior with no local maximum, heteromeric Kv7.3+4 channels display less inactivation between V_pre of −20mV and +40mV where the maximal amplitude reaches 89.0%±3.5%, n=10 at V_pre=+20mV relative to V_pre=−120mV (homomeric Kv7.4 amplitude at V_pre=20mV is 65.7%±2.8%, n=27 relative to V_pre=−120).

Display large version of this figure

Figure 6

Kv7.3+4 heteromers inactivate differently from Kv7.4 homomers. (A) Cells expressing heteromeric Kv7.3+4 channels were investigated for the inactivation behavior of the respective channels with the same voltage protocol as in Fig. 2 (expanded to include a prepulse step to +80mV). (B) The figure shows the normalized maximal current recorded at a 5-s-long depolarizing pulse to +40mV preceded by a 20-s-long prepulse (V_pre) ranging from −120mV to +80mV in 20-mV increments for both Kv7.4 and Kv7.3+4. Current amplitudes are plotted as a function of the holding potential during the 20-s-long prepulse, and the two channel complexes are significantly different in this respect after a prepulse ranging from −40mV to +60mV. Asterisks indicate the level of significance (*, P<0.05; **, P<0.005, t-test, n=10 and 27, Kv7.3+4 and Kv7.4, respectively).

A handful of compounds has been used to address the function of Kv7 channels (e.g., retigabine, XE-991, and BMS-204352). The positive modulator retigabine presently undergoing clinical phase III studies has especially been the subject of intense study. It has been shown that retigabine enhances the open probability, shifts the activation curve leftward, and increases the macroscopic current of Kv7 channels 16,41,42,43,44. However, none of these studies has investigated the effect on the inactivation of Kv7.4. To address this we employed the standard inactivation voltage protocol, as used in Figure 2 and Figure 6, in the absence and presence of 10μM retigabine (Figure 7AB). The two curves describing the measured currents at each potential normalized to the control current at −120mV and plotted as a function of V_pre were parallel over the range of V_pre tested (−120 mV–+80mV) (Figure 7E). When plotting the retigabine-induced current relative to the control current at each potential the modulation of inactivation by retigabine was also minor (the potentiation ranged from 170%±15% at a V_pre of −120mV to 269%±50% at a V_pre of +80mV, n=7). We quantified this as the slope of the best linear fit to the percentage of control data, which was calculated as α_RTG=0.33%±0.21%/mV, n=7 (Figure 7EG).

Display large version of this figure

Figure 7

Pharmacological modulation of Kv7.4 inactivation. (A–D) Two-electrode voltage-clamp experiments on three different oocytes under control conditions (A) and after application of 10μM retigabine (RTG; B), 10μM BMS-204352 (C), and 100μM XE-991 (D). The voltage protocol consist of a 20-s-long prepulse (V_pre) ranging from −120mV to +80mV in 20-mV increments, followed by a depolarizing step to +40mV (V_act) and ended by a 1-s step to −30mV. The scale bars are for control and RTG traces 3μA and 5s, for the BMS experiment 6μA and 5s, and for XE-991 1μA and 5s. (E) Plot showing the normalized current levels before and after application of RTG, BMS, and XE-991, respectively. (F) Graphs illustrate the best linear fit of the measured current expressed as percentage of control data. Best-fit slope values are (in percentage per mV) α_RTG=0.33±0.21 (n=7), α_XE-991=−0.016±0.017 (n=5), and α_BMS=5.24±1.51 (n=3). (G) Histogram illustrating the differences between the slope values determined for RTG, BMS, and XE-991.

In a similar manner we tested if the inhibitory effect of XE-991 was related to a changed inactivation. Hence, 100μM XE-991 was applied as described above, and the rate of inactivation of Kv7.4 was compared before and after drug application (Figure 7AD). As with retigabine, the maximal current amplitude relative to control conditions over the range of V_pre tested was close to parallel and in the range of 40%–45% (45%±4.3% at V_pre=−120mV to 45%±8.6% at V_pre=+80mV, n=5). The slope of the best linear fit was calculated to α_XE-991=−0.016%±0.017%/mV, n=5 (Figure 7EG).

BMS-204352 was originally developed for the treatment of stroke, and the mechanism was thought to be by opening BK channels 45. However, it has also been shown to be a potent activator of Kv7 channels 41,46,47. A detailed characterization of the mode of action on the Kv7 channels has not yet been described, so we investigated the effect of BMS-204352 on the inactivation properties of Kv7.4. A total of 10μM BMS-204352 potentiated the current to 690%±106%, n=5 of control measured at the end of a depolarizing step during the following repeated voltage protocol: 4s at −80mV, 2s at +40mV, and 1s at −30mV (not shown). In contrast to retigabine and XE-991, a clear effect on inactivation was observed with BMS-204352 (Figure 7AC). The maximal V_act current amplitude increased over the range of V_pre tested from 578%±15% of control at V_pre=−120mV to 1745%±422% at V_pre=+80mV, n=3. The slope of the best linear fit of the percentage of control data reflects abolishment of Kv7.4 inactivation by BMS-204352, as it was calculated to α_BMS=5.24%±1.51%/mV, n=3 (Figure 7EG).

The nonneuronal Kv7.1 potassium channel has been described to inactivate 38,39,48, whereas the general notion has been that the other Kv7 family members do not inactivate. Contrary to this general view on the biophysical properties of the neuronal Kv7 channels, we here show that Kv7.4 and Kv7.5 channels indeed do inactivate. We have performed detailed studies on Kv7.4 showing that the macroscopic inactivation is voltage dependent and is augmented as a function of increased holding potentials, reaching a maximum >60% at +80mV. Kv7.4, as well as Kv7.5 channels, was found to inactivate to a larger extent than Kv7.1.

The kinetics of Kv7.4 channel inactivation is slow, with time constants of 5–10s depending on the duration of the preceding hyperpolarizing pulse (τ_{10s prepulse}=9.55±1.14s and τ_{20s prepulse}=5.06±1.10s). Recovery from inactivation of Kv7.4 is both time and voltage dependent. When inactivation was released by hyperpolarizing pulses at −120mV at increasing time intervals, the recovery from inactivation was clearly time dependent with a time constant of 1.47±0.21s. The inactivation kinetics of Kv7.4 is apparently much slower than that of Kv7.1, as Tristani-Firouzi and Sanguinetti found that Kv7.1 inactivates monoexponentially at +20mV with a time constant of 130ms after an initial delay of ∼75ms 38. The time constants for recovery from inactivation of Kv7.1 was likewise much faster, ranging from 30–40ms and showing very little voltage dependence at voltages ranging from −90 to −30mV 38. However, since the kinetics of inactivation and recovery from inactivation in their study was determined from the hooked tail current, they may only reflect the early phase of both phenomena. The degree of Kv7.1 inactivation in both studies was found to be ∼35% at +40mV with a similar voltage dependence despite the different protocols used to reflect the inactivation.

It has been demonstrated that Kv7.3 and Kv7.4 can form functional heteromers 13, so we investigated if such complexes had an inactivation profile different from Kv7.4 homomers. The quantity of inactivated heteromers was significantly decreased at potentials from −40mV to+60mV. This indicates that more active Kv7.3+Kv7.4 channel complexes will be present at these potentials relative to homomeric Kv7.4 channels.

In a final investigation we addressed if different Kv7 channel modulators had any impact on the inactivation properties of the channels. It was demonstrated that the Kv7.4 channel activator retigabine and the inhibitor XE-991 did not mediate their effects through changed inactivation. In contrast, BMS-204352 abolished the inactivation of Kv7.4 channels. These data demonstrate that the Kv7 openers retigabine and BMS-204352 have distinct modes of action.

It is known from Kv7.1 inactivation that this can be abolished by coexpression with KCNE1 (minK), which is in agreement with the notion that the endogenous I_Ks current does not inactivate 38. It has been suggested that this modification of the channel kinetics is accomplished by the C-terminus of KCNE1 protruding into the Kv7.1 complex and binding to the pore loop 49. The modest effect of Kv7.3 and retigabine as opposed to the strong effect of BMS-204352 on Kv7.4 inactivation may likewise be caused by the physical interaction resulting in an allosteric modulation of the channel complex. However, we do not know the specific binding sites of these molecules to the Kv7.4 complex. It is likely that BMS-204352 binds to a site other than retigabine, since they are structurally quite different molecules. Further, the two compounds modify the Kv7 channel subtypes differentially since BMS-203352 blocks Kv7.1 whereas retigabine shows no effect on this channel 44,50. Likewise, the drugs show differential effects on the neuronal Kv7 channels 51.

One of the ways a drug can activate an ion channel is by removing its inactivation, as is the case for BMS-204352. The effects of BMS-204352 is, however, significantly larger than what can be accounted for by reduced inactivation alone (Fig. 7). The compound also activates the Kv7.4 channel by shifting the voltage-dependent activation curve toward negative values 50 as well as by activating a voltage-independent Kv7.4 current 41. Since the open probability of Kv7.4 channels at positive membrane potentials is only ∼0.18 52, there is room for an increased opening of the channel in the presence of compound, which must also occur for the current to increase by more than fivefold at the noninactivation potential of −120mV.

In this study Kv7.4 and Kv7.5 are found to inactivate to the largest degree, whereas Kv7.2 does not inactivate. Kv7.4 and Kv7.5 exhibit a number of similar characteristics in contrast to Kv7.2: Kv7.4 has a greater amino acid sequence homology with Kv7.5 than with Kv7.2; pharmacologically BMS-204352 activates Kv7.4 and Kv7.5, whereas it does not modify Kv7.2/3 50; biophysically the unitary currents of Kv7.4 and Kv7.5 are 2.1 and 2.2 pS, whereas that of Kv7.2 is 6.2 pS 52,53; the maximal open probability of Kv7.4 and Kv7.5 are 0.18 and 0.17, respectively, whereas it is 0.07 for Kv7.2. This does not in itself explain why Kv7.4 and Kv7.5 inactivate and Kv7.2 does not, and to address this question it is important to study the residues in the Kv7 channels involved in the inactivation process of which we currently have no clear indication.

This study shows that Kv7.4 channel inactivation is different from classical N- and C-type inactivation found in voltage-gated Na⁺ channels and voltage-gated K⁺ channels of the Shaker type 38,48,54,55,56,57. Features that distinguish Kv7.4 inactivation from N- and C-type inactivation are slow onset and voltage dependency. This is also true for Kv7.1 inactivation 38, suggesting that the inactivation mechanism for these two Kv7 channel subtypes are similar.

The inactivation property may have important bearings for the pathophysiology and pharmacology of the neuronal Kv7 channels. Within the family of Kv7 channels, inactivation at physiologically relevant potentials was most prominent for Kv7.4 and Kv7.5 channels. Of the five Kv7 channel subunits, Kv7.4 displays the most restricted expression pattern, being present in the inner ear and the auditory pathways of the brain but also in certain dopaminergic neurones 58,59. Hence, Kv7.4 channels could represent a particularly relevant target for treatment with drugs specifically activating the Kv7.4 channel as efficacy, and side effect profiles of such a compound could prove superior to panreactive compounds affecting all neuronal members of the Kv7 family.

In conclusion we have described inactivation as a new regulatory mechanism for Kv7.4 and Kv7.5 channels. At physiologically relevant resting potentials, the current is reduced by more than 30% due to steady-state inactivation. These results add to a more profound understanding of the biophysical nature of the Kv7 channel family and their physiological impact. Pharmacological channel modulators influence the inactivation properties differentially, so this newly discovered regulatory mechanism is of further importance when developing drugs targeted for these channels. Physiological signaling molecules may also influence the steady-state inactivation of Kv7 channels, and if this proves to be the case it will be an efficient way of modulating neuronal excitability.