Alkanols inhibit voltage-gated K+ channels via a distinct gating modifying mechanism that prevents gate opening

Alkanols are small aliphatic compounds that inhibit voltage-gated K+ (Kv) channels through a yet unresolved gating mechanism. Kv channels detect changes in the membrane potential with their voltage-sensing domains (VSDs) that reorient and generate a transient gating current. Both 1-Butanol (1-BuOH) and 1-Hexanol (1-HeOH) inhibited the ionic currents of the Shaker Kv channel in a concentration dependent manner with an IC50 value of approximately 50 mM and 3 mM, respectively. Using the non-conducting Shaker-W434F mutant, we found that both alkanols immobilized approximately 10% of the gating charge and accelerated the deactivating gating currents simultaneously with ionic current inhibition. Thus, alkanols prevent the final VSD movement(s) that is associated with channel gate opening. Applying 1-BuOH and 1-HeOH to the Shaker-P475A mutant, in which the final gating transition is isolated from earlier VSD movements, strengthened that neither alkanol affected the early VSD movements. Drug competition experiments showed that alkanols do not share the binding site of 4-aminopyridine, a drug that exerts a similar effect at the gating current level. Thus, alkanols inhibit Shaker-type Kv channels via a unique gating modifying mechanism that stabilizes the channel in its non-conducting activated state.

1-HeOH had a slightly higher affinity and yielded a concentration-response curve with an IC 50 value of 2.7 ± 0.2 mM (n = 7) and a Hill coefficient of 1.13 ± 0.22 (Fig. 1D). Monitoring the development of I K inhibition and analyzing the remaining steady-state I K amplitude upon application of 50 mM 1-BuOH or 3 mM 1-HeOH (IC 50 concentrations) indicated that: (1) the I K inhibition developed rapidly and was fully reversible upon wash-out of both alkanols, and (2) both alkanols did not induce major alterations in the voltage dependence of channel opening nor the time constants of channel activation (τ I Kac ) and deactivation (τ I Kdeac ) ( Fig. 2 and Table 1). An apparent channel inactivation behavior or rising phase in the deactivating (I Kdeac ) tail current (i.e. a hooked tail), which are typical hallmarks for an open channel blocker, were not observed ( Fig. 1A-C). Thus, 1-BuOH and 1-HeOH inhibited the I K amplitude without affecting the kinetics, and both compounds achieved this through a mechanism most likely different from open channel block, as proposed previously 4 .
1-BuOH and 1-HeOH accelerate VSD deactivation and immobilize approximately 10% of the gating charge. The I K measurements only report on the final opening of the channel gate, which is an end state in the activation pathway from closed to open. From I G analysis it has been reported that the VSD traverses at least one non-conducting activated state before the channel gate opens. Channel gate opening subsequently slows down VSD deactivation 10,15 , which can be visualized by gradually prolonging the duration of the depolarizing pre-pulse (Fig. 3A). Thus, to assess whether 1-BuOH and 1-HeOH affect transitions early in the activation pathway, i.e. before the channel gate opened, we tested the effect of both compounds on the I G recordings of the non-conducting Shaker-IR pore mutant W434F 25 . During wash-in of both 1-BuOH and 1-HeOH we noted a concentration-dependent acceleration of the deactivating (I Gdeac ) gating currents (Fig. 3B,C). Plotting the time constant of VSD deactivation (τI Gdeac , obtained by fitting the decaying phase of I Gdeac ) as a function of 1-BuOH or 1-HeOH concentration yielded concentration-response curves with IC 50 values of 67 ± 1 mM (n = 10) and 3.0 ± 0.4 mM (n = 6), and Hill coefficients of 1.3 ± 0.4 and 1.6 ± 0.3, respectively (Fig. 3D).
To examine whether this acceleration in τ I Gdeac was associated with a reduction in gating charge movement, we integrated the activating (I Gac ) gating currents (elicited during the depolarizing test pulse) after reaching steady-state modification of the τ I Gdeac kinetics. This analysis indicated that there was an alkanol-dependent reduction in gating charge movement concomitantly with the acceleration in τ I Gdeac . The reduction in total gating charge as a function of alkanol concentration yielded for 1-BuOH and 1-HeOH concentration-response curves with IC 50 values of 88 ± 2 mM (n = 10) and 13.8 ± 1.6 mM (n = 6), and Hill coefficients of 1.5 ± 0.2 and 1.5 ± 0.2, respectively (Fig. 3E). Based on these concentration-response curves, the maximal reduction in charge movement amounted to approximately 10% and 12% upon application of 300 mM 1-BuOH and 30 mM 1-HeOH, respectively.
To determine the kinetics and voltage dependence of VSD activation, we applied incremental depolarizing voltage steps starting from a constant hyperpolarized initial voltage (activation protocol, Fig. 4A). To characterize VSD deactivation adequately, a deactivation pulse protocol was used (Fig. 4B). Integrating the I Gac recordings, obtained in control conditions and after steady-state 1-BuOH and 1-HeOH modification, yielded charge vs. voltage QV curves (Fig. 4C). Interestingly, the QV curves determined in presence of 1-BuOH or 1-HeOH displayed V 1/2 and slope factor values similar as in control condition (Table 1). This indicated that neither alkanol affected the voltage dependence of the remaining gating charge movement. As noted during the wash-in protocol (Fig. 3B,C), both alkanols accelerated τ I Gdeac without markedly altering the I Gac kinetics (τ I Gac , Fig. 4D,E). Thus, both alkanols accelerated τ I Gdeac and immobilized approximately 10% of the gating charge movement but did not affect the voltage dependence of the early VSD movements. These observations indicated that in presence of 1-BuOH or 1-HeOH the Shaker channel is able to reach the non-conducting activated state but it cannot pass the subunit-cooperative step leading to channel gate opening. Accordingly, the τ I Gdeac values in presence of saturating alkanol concentrations should corresponded to τ I Gdeac in control conditions when the activating pre-pulse is very short and channels only reach the non-conducting activated state. In control conditions τ I Gdeac amounted at −120 mV to 0.32 ± 0.03 ms (n = 6) upon a brief 0.5 ms depolarization, determined from pulse protocols shown in Fig. 3A. In presence of 300 mM 1-BuOH or 30 mM 1-HeOH τ I Gdeac at −120 mV were 0.48 ± 0.08 ms (n = 4) and 0.53 ± 0.10 ms (n = 4) respectively (Fig. 4D,E), which are indeed similar to the value in control conditions. 1-alkanols and 4-AP have different binding sites but immobilize the same gating charge component. The impact of 1-BuOH and 1-HeOH on the I G recordings of the Shaker-IR-W434F channel was reminiscent of the effect of 4-AP that prevents the channels from passing the late subunit-cooperative step of channel gate opening, resulting in a similar 10% reduction in gating charge movement 14 . To assess if 4-AP and 1-BuOH immobilized the same gating charge component, we determined the reduction in gating charge movement using a mixture of 1 mM 4-AP and 300 mM 1-BuOH, which for both compounds are saturating concentrations. First, we applied 1 mM 4-AP that resulted in an approximately 10% loss of gating charge movement and an acceleration of τ I Gdeac , as has been described before 14,23 . After establishing a steady-state 4-AP effect, we applied 300 mM 1-BuOH in the continued presence of 1 mM 4-AP. The addition of 1-BuOH did not result in an extra reduction of gating charge movement or further acceleration of the I Gdeac kinetics (Fig. 5). This indicated that both compounds affected the same gating charge component and further supported that alkanols stabilize the channel in the non-conducting activated state similar to 4-AP.
Although alkanols and 4-AP exert a similar effect at the gating current level, they may act through different binding sites. Whereas the binding site of 4-AP partially overlaps with that of internal pore   blockers 26,27 , alkanols have been proposed to target the electromechanical coupling that is located outside the K + pore. To test whether 4-AP and 1-BuOH have structurally different binding sites, we performed drug competition experiments using IC 50 concentrations of 4-AP (30 μ M) and 1-BuOH (50 mM). After establishing approximately 50% steady-state I K inhibition with 4-AP, we applied a mixture of 30 μ M 4-AP and 50 mM 1-BuOH. This mixture resulted in 78.7 ± 4.1% (n = 7) inhibition of I K (Fig. 6), thus yielding an additional inhibition of 29% in I K amplitude compared to each compound separately.
To evaluate whether 4-AP and 1-BuOH competed, the expected inhibition of the mixture was calculated using a syntopic (both compounds compete) or an allotopic (no competition) model 28 . Using an allotopic model and the experimentally determined inhibition of each compound separately, the predicted inhibition of the mixture was 81.4 ± 2.4% (n = 7). With a syntopic model the predicted inhibition was 73.4 ± 2.8% (n = 7). Because the experimentally determined inhibition (78.7%) differed only statistically (p < 0.05) from the predicted value of the syntopic model (Fig. 6), our data matched best an allotopic model indicating that there was no competition between both compounds.

1-BuOH and 1-HeOH activate the Shaker-IR-P475A mutant by accelerating channel opening.
A previous study reported that substituting a highly conserved proline residue in the S6 c of the Shaw2 channel (the second proline of a highly conserved PXP motif within the S6 c of K v channels) by a neutral amino acid such as alanine inverted the effect of the alkanols 22 . Thus, instead of inhibiting the channel mutant, application of alkanols potentiated the current amplitude. An alanine substitution for the corresponding proline (P475) in Shaker-IR shifted the threshold for channel opening towards more depolarized potentials by affecting the late step(s) of channel gate opening while leaving earlier VSD transitions unaffected 29 . Consequently, the Shaker-IR-P475A mutant displays slow I Kac kinetics that is only weakly voltage-dependent. Applying 1-BuOH or 1-HeOH to the Shaker-IR-P475A mutant resulted in a concentration-dependent increase in I K and an acceleration of τ I Kac (Fig. 7A,B), which is in agreement with previous data obtained in the Shaw2 channel 22 . With higher concentrations of 1-BuOH or 1-HeOH the typical conduction versus voltage GV curves, which were determined from normalizing the deactivation tail current of activation protocols (Fig. 8A), appeared to become steeper and to shift slightly towards more hyperpolarized potentials (Fig. 8B, Table 1). However, concomitantly with the accelerated τ I Kac kinetics, also the inactivation process became more pronounced and the peak I K amplitude started to decrease at higher alkanol concentrations (Fig. 7A,B). Therefore, the small hyperpolarizing shift and steepening of the GV curves could be an apparent effect due to the accelerated channel inactivation. To test this possibility, we determined the normalized conduction G from the peak outward currents using the Goldman-Hodgkin-Katz current equation. The GV curves obtained with this approach, which should be less sensitive to inactivation, were in presence of alkanols similar to those in control conditions (Fig. 8B). Thus, although both compounds resulted in I K activation, neither 1-BuOH nor 1-HeOH affected the voltage dependence of channel opening substantially. To evaluate if the pronounced channel inactivation behavior reflects in fact open channel block, we examined I Kdeac more closely. In contrast to what is expected with open channel block, the I Kdeac recordings did not cross nor did they display a noticeable hook (Fig. 7A,B). In fact, the τ I Kdeac kinetics accelerated markedly which suggested that also the accelerated channel inactivation was due to gating modification. All these effects were fully reversible upon wash-out of both alkanols.
The I Kac of Shaker-IR-P475A displayed two components and was best approximated with a double exponential function yielding a fast and a slow τ I Kac component 29 . However, the fast component contributed only marginally to the overall I K amplitude and the weighted τI Kac kinetics approximated the value of the slow component in control condition (Fig. 8C). 1-BuOH or 1-HeOH accelerated channel opening markedly but approximating the I Kac currents with a double exponential function indicated that the time constants of both the fast and slow component were similar to those obtained in control condition. However, the contribution of the fast component in the total current amplitude increased as a function of alkanol concentration (Fig. 8C). Consequently, the weighted τ I Kac accelerated with increasing alkanol concentration (Fig. 8D,E). Similar to I Kac , the weighted τ I Kdeac kinetics, obtained from fitting I Kdeac with a double exponential function, accelerated in an alkanol concentration-dependent manner (Fig. 8D,E). Plotting the weighted τ I Kac as a function of 1-BuOH or 1-HeOH concentration yielded concentration-response curves with IC 50 values of 58.8 ± 3.0 mM (n = 5) and 4.6 ± 0.8 mM (n = 4), and Hill coefficients of 1.5 ± 0.4 and 1.3 ± 0.3, respectively (Fig. 8F). This indicated that the alanine substitution for P475 in S6 c did not affect the affinity for alkanols, suggesting that the conformation of the binding site remained intact.

1-BuOH and 1-HeOH did not affect the VSD movements of the P475A mutant. Since the
Shaker-IR-P475A mutant did not affect the early VSD movements, the QV curve was split and displayed two gating charge components whereby the late one corresponded with the voltage dependence of channel gate opening 29 . Analyzing the gating currents of Shaker-IR-W434F-P475A in presence of 300 mM 1-BuOH or 30 mM 1-HeOH indicated that the voltage dependence of neither the early nor the late gating charge component was affected by 1-BuOH or 1-HeOH ( Fig. 9 and Table 1). This was in agreement with the absence of an obvious shift in the threshold of channel opening (Fig. 8B). Also the I Gac time constants, which in Shaker-IR-W434F-P475A report directly on the kinetics of the early VSD movements 29 , were unaffected by 1-BuOH or 1-HeOH. These I G data confirmed that 1-BuOH and 1-HeOH did not affect the voltage-dependent transitions of the Shaker-IR-P475A mutant but facilitated a late largely voltage-independent transition in the activation pathway, a transition that is compromised by the P475A mutation.

Discussion
1-BuOH and 1-HeOH inhibited the Shaker-IR channel in a concentration-dependent manner without displaying the classic hallmarks of an open channel blocker. Therefore, alkanols appear to act as gating modifiers that stabilize the channels in a non-conducting state 4 . To elucidate which state is stabilized by alkanols, we determined the impact of 1-BuOH and 1-HeOH on the I G recordings of Shaker-IR-W434F. Both alkanols caused a concentration-dependent reduction in gating charge movement associated with accelerated VSD deactivation (Fig. 3). This data indicated that alkanols interfere with the transition from the non-conducting activated conformation to full channel gate opening, which occurs in a highly subunit-cooperative (concerted) manner. Consequently, the reduction in either I K or gating charge movement as a function of alkanol concentration yielded concentration-response curves with similar IC 50 values and Hill coefficients (Figs 1,3). Since alkanols have been proposed to operate via the S4-S5 linker 18 , there are 4 potential alkanol binding sites on the channel that appear to operate largely independently (Hill coefficients of approximately 1). As they interfere with a subunit-cooperative transition, binding of a single alkanol molecule (occupying only one out of four binding sites) can be sufficient to prevent channel gate opening and losing about 10% of the gating charge movement. Accordingly, a previous study, which used concatemeric constructs, showed that channels with less than four high affinity binding sites (e.g. only 2) were still inhibited by alkanols 4 . Thus, we propose that alkanols inhibit Shaker-IR currents by preventing the channels of passing the final concerted step in the activation sequence that opens the channel gate. Therefore, the ionic current data analysis represented in figure 2 reports on the channels that were free of alkanols which explains why both the normalized GV curves (Fig. 2C) and the kinetics (Fig. 2D) in presence of alkanols were similar to control conditions. The impact of 1-BuOH and 1-HeOH on Shaker's gating charge movement was reminiscent to that of the well-characterized drug 4-AP 14 , and both compounds stabilize the Shaker K v channel in the non-conducting activated state. However, both compounds achieve this by acting via distinct binding sites (Fig. 6). Whereas the binding determinants for 4-AP, including those for guanidine compounds that possibly work in a similar manner 30 , reside within S6 c 31,32 , alkanols are suggested to distort the coupling between the S4-S5 linker and S6 c 33 . The observation that alkanols rescued partly the kinetics of the Shaker-IR-P475A mutant, favors the idea that alkanols alter the conformation of the S4-S5 linker and/or S6 c without disrupting their communication completely. By altering the conformation of the S4-S5 linker and/or S6 c , the electromechanical coupling is compromised as its operation relies on a correct positioning of both segments with respect to each other 34 .
Mutations that affect the communication between the VSD and the channel gate might therefore alter the alkanol effect, as is the case in Shaker-IR-P475A. Apparently, 1-BuOH and 1-HeOH did not shift the voltage dependence of the late gating charge component in Shaker-IR-P475A (Fig. 9), which is expected if the mutation was only to affect the equilibrium constant of the transition from the non-conducting activated to the open state. Therefore, the structural consequences of the P475A mutation should be more severe and the Shaker-IR-P475A mutant displayed, accordingly, a biphasic current activation that in absence of alkanols is dominated by the slow component. We propose that alkanol binding to the Shaker-IR-P475A channel alters the conformation of the S4-S5 linker and/or its communication with S6 c (as it does in WT Shaker-IR), and in doing so it coincidentally restores the conformation of the S6 c channel gate that is compromised by the mutation. Alkanols then act as activators of the Shaker-IR-P475A mutant by yielding I K current activation that is dominated by the fast component (Fig. 8C-E), thus accelerating a late largely voltage-independent transition of channel gate opening. Notably, the effect of both 1-BuOH and 1-HeOH on the Shaker-IR-P475A mutant was comparable to the behavior of poly-unsaturated-fatty acids (PUFAs): accelerating channel opening followed by more pronounced channel inactivation at higher concentrations (Fig. 7A,B). PUFAs have been shown to alter the kinetics of K v channels leading to current activation or current inhibition, in part by accelerating the inactivation process 35 . At low concentrations several PUFAs act as channel activators but at higher concentrations they result in channel inhibition [36][37][38] . Whereas their activating property is ascribed to their ability to shift the voltage dependence of channel opening towards more hyperpolarized potentials and to facilitate the late subunit-concerted transition of channel opening 39,40 , their molecular mechanism to induce channel inhibition is still debated 41 . Whereas alkanols most likely target the S4-S5 linker of K v channels 33 , PUFAs supposedly exert their effect through the VSD 39,40 , although a role for the S4-S5 linker has been suggested 42 .
Alkanols and 4-AP immobilize the same VSD movement(s) in the Shaker-IR channel (Fig. 5), but both compounds achieve this via distinct drug binding sites (Fig. 6) and a different mechanism of action. This conclusion is further supported by the observation that the mutant Shaker-IR-P475A is activated by alkanols (Figs 7,8) but is insensitive to 4-AP 29 . The presence of other (possibly overlapping) intracellular or lipid-accessible binding sites for gating modifying compounds is supported by: (1) the finding that the gating modifier toxin gambierol occupies a lipid exposed S5-S6 crevice outside the K + pore 43 , a binding site which is most likely shared by psora compounds 44 and (2) the observation that ruthenium complexes uncouple VSD movement from channel gate opening but in contrast to alkanols they immobilize about 50% of the gating charge 45 . Furthermore, the binding site for the volatile anesthetic halothane has been shown to overlap with that of alkanols 21 , and both isoflurane and servoflurane, which belong to the same class of halogenated general anesthetics, potentiate K v channels instead of inhibiting them [46][47][48] .
The intoxicating and sedating effects of exposure to high alkanol concentrations are well described and ion channels (including K v channels) do most likely form one of their molecular targets. We provide a mechanistic basis for understanding their effect on K v channels and show that 1-BuOH and 1-HeOH interfere directly with the gating apparatus of the Shaker-IR K v channel. They inhibit Shaker-IR by stabilizing the non-conducting activated state preventing the channels from passing the final subunit-concerted transition leading to channel gate opening. They achieve this through a unique gating modifying mechanism different from that of 4-AP. Our findings strengthen the idea that there exist different intracellular drug binding sites that via distinct mechanisms of action exert a similar gating modifying effect; this opens new possibilities for designing modulators of K v channels.

Methods
Molecular Biology. The N-terminal deletion Δ 6-46 Shaker clone (Shaker-IR), which removes fast inactivation 49 , was used in this study. The W434F mutation, which yields a non-conducting Shaker-IR-W434F channel 25 , and the P475A mutation were introduced as described previously 29 . All channel constructs were expressed using a pGW1 expression vector. The plasmid that codes for the green fluorescent protein, used to identify transfected cells, was purchased from Clontech (Palo Alto, CA, USA). Plasmid DNA for mammalian expression was obtained by amplification in XL2 Bluescript cells (Stratagene), and afterwards isolated using the endotoxin-free Maxiprep kit (Macherey-Nagel, Düren, Germany). The cDNA concentration was determined by UV absorption.
Cell culture. HEK293 cells were cultured in Modified Eagle's Medium (MEM) supplemented with 10% fetal bovine serum, 1% penicillin/streptomycin and 1% non-essential amino acids (Invitrogen, Carlsbad, CA, USA). Cells were transiently transfected with the appropriate channel DNA plasmids using polyethyleneimine that was purchased from Sigma-Aldrich (St Louis, MO, USA), details of procedure was described previously 29 . Electrophysiology. Whole-cell ionic I K or gating I G current measurements were done at room temperature (20 to 23 °C) using an Axopatch-200B amplifier and the recordings were digitized with a Digidata-1200 A acquisition system (Molecular Devices, Sunnyvale, CA, USA). Both I K and I G recordings were digitized at 10 kHz sampling rate after passing a 5 KHz Bessel low-pass filter. Command voltages and data storage were controlled with pClamp10 software. Patch pipettes were pulled from 1.2 mm quick-fill borosilicate glass capillaries (World Precision Instruments, Sarasota, FL, USA) with a P-2000 puller (Sutter Instrument Co., Novato, CA, USA) and afterwards heat-polished, to have patch pipettes with a resistance of approximately 1.5 MOhm determined with the filled pipette in the bath solution.
For I K measurements the cells were constantly superfused with external bath solution that contained (in mM) NaCl 130, KCl 4, CaCl 2 1.8, MgCl 2 1, HEPES 10, Glucose 10, adjusted to pH 7.35 with NaOH. The patch pipettes were filled with internal solution containing (in mM) KCl 110, K 4 BAPTA 5, K 2 ATP 5, MgCl 2 1, HEPES 10, adjusted to pH 7.2 with KOH. For I G measurements the monovalent cations were replaced with N-methyl-D-glucamine (NMG + ). The bath solution contained (in mM) NMG + 140, HEPES 10, Glucose 10, MgCl 2 1, CaCl 2 1.8, titrated to pH 7.35 with HCl. The pipette solution contained (in mM) NMG + 140, HEPES 10, EGTA 10, MgCl 2 1, titrated to pH 7.2 with HCl. Junction potentials were zeroed with the filled pipette in the bath solution and experiments were excluded from analysis if the voltage error estimate exceeded 5 mV after series resistance compensation. For I G measurements, leak currents and remaining capacitive currents were subtracted online using a −P/6 protocol (using a holding potential of −95 mV). I K recordings were not leak corrected.
Drug solutions. 1-BuOH and 1-HeOH (Sigma-Aldrich, St. Louis, MO, US) were directly dissolved in the external recording solution for either I K or I G measurements. The different test concentrations were daily made as both compounds are volatile lowering the effective concentration upon storage. For the highest concentration of 1-BuOH tested (300 mM), the osmolarity of the extracellular solution for I K or I G recordings increased by approximately 300 mOsm resulting in a total osmolarity of ~640 mOsm. Because of the rapid partitioning of alkanols across the plasma membrane, we expected a minor impact of this increase in osmolarity. Indeed, the cells tolerated remarkably well the perfusion of the 300 mM 1-BuOH solution. This was not the case when the cells were perfused with a 600 mOsm extracellular solution that contained glucose, which does not easily partition across the plasma membrane, to increase osmolarity (data not shown). 4-AP was purchased from Sigma-Aldrich and after dissolving it in the external recording solutions the pH was adjusted to 7.35 using HCl. All compounds were applied to the cells using a pressurized fast perfusion system equipped with a quartz micromanifold (ALA scientific, Farmingdale, NY, USA), allowing rapid exchange of the external solutions.
Scientific RepoRts | 5:17402 | DOI: 10.1038/srep17402 Data analysis. Details of pulse protocols used to elicit I K or I G recordings were adjusted to determine the biophysical properties of each construct adequately and are shown in the figures or described in legends. All the graphs were built using SigmaPlot 11.0 (Systat Software Inc., San Jose, CA, USA). If not mentioned otherwise, the conductance vs. voltage (GV) curves were determined from analyzing normalized tail current amplitudes and the charge vs. voltage (QV) curves by integrating the activating I G currents. The QV and GV curves of Shaker-IR were fitted with a Boltzmann equation: y = 1/{1 + exp[− (V − V 1/2 )/k]}, where V represents the applied voltage, V 1/2 the midpoint potential at which 50% of the total charge has moved or half of the channels have opened, and k the slope factor. For the P475A mutant the GV curve was also approximated with a single Boltzmann equation whereas its QV curve was approximated with the sum of two Boltzmann distributions. Activation I K kinetics (τ I Kac ) were determined by approximating the rise in I Kac with a single or double exponential function. Deactivation I K kinetics (τ I Kdeac ) were obtained from single or double exponential fits to the I Kdeac decay elicited at various repolarizing potentials following a 25 ms depolarizing pre-pulse to + 20 mV that activated the channels. When a double exponential function was used to determine the fast (τ fast ) and slow (τ slow ) component of the τ I Kac and τ I Kdeac kinetics, the weighted time constants (τ W ) were calculated based on the amplitude of each component: τ W = (A fast /(A fast + A slow )) x τ fast + (A slow /(A fast + A slow )) x τ slow , with A fast and A slow the amplitude of the fast and slow component respectively. The I G activation and deactivation kinetics (τ I Gac and τ I Gdeac ) were determined by fitting the decaying part of I Gac and I Gdeac with a single exponential function. All results are expressed as mean ± S.E.M. with n the number of cells analyzed.
Concentration-response curves (both from I K and I G analysis) were fitted in the program OriginPro 8 (OriginLab Corp., Northampton, MA, USA) with a Hill equation: I effect = I min + ({I max − I min }/{1 + ([alkanol]/IC 50 ) Hill coefficient }), where [alkanol] is the concentration of 1-BuOH or 1-HeOH and IC 50 the concentration that induces 50% effect. To test whether 1-BuOH shares a similar and/or overlapping binding site with 4-AP, we performed competition experiments based on a previously described approach 28 . The method is based on comparing the experimental determined inhibition to the expected level of channel inhibition using an allotopic (non-competing) or a syntopic model (competing). Formulas used for calculating the expected inhibition in presence of both compounds (I NX,Y ) according to the allotopic and syntopic model were I NX,Y = (I NX + I NY − I NX I NY ) and I NX,Y = ((I NX + I NY − 2I NX I NY )/(1− I NX I NY )), respectively. I NX and I NY were the experimentally determined level of channel inhibition induced by each compound independently. I.e. I NX was the inhibition induced by 1-BuOH and I NY the level of inhibition induced by 4-AP. A two way analysis of variance (ANOVA) was used to determine the differences of a dual inhibition. A post hoc Dunnett's was used to compare both models.