Modulation of Closed−State Inactivation in Kv2.1/Kv6.4 Heterotetramers as Mechanism for 4−AP Induced Potentiation

The voltage−gated K+ (Kv) channel subunits Kv2.1 and Kv2.2 are expressed in almost every tissue. The diversity of Kv2 current is increased by interacting with the electrically silent Kv (KvS) subunits Kv5−Kv6 and Kv8−Kv9, into functional heterotetrameric Kv2/KvS channels. These Kv2/KvS channels possess unique biophysical properties and display a more tissue-specific expression pattern, making them more desirable pharmacological and therapeutic targets. However, little is known about the pharmacological properties of these heterotetrameric complexes. We demonstrate that Kv5.1, Kv8.1 and Kv9.3 currents were inhibited differently by the channel blocker 4−aminopyridine (4−AP) compared to Kv2.1 homotetramers. In contrast, Kv6.4 currents were potentiated by 4−AP while displaying moderately increased affinities for the channel pore blockers quinidine and flecainide. We found that the 4−AP induced potentiation of Kv6.4 currents was caused by modulation of the Kv6.4−mediated closed−state inactivation: suppression by 4−AP of the Kv2.1/Kv6.4 closed−state inactivation recovered a population of Kv2.1/Kv6.4 channels that was inactivated at resting conditions, i.e. at a holding potential of −80 mV. This modulation also resulted in a slower initiation and faster recovery from closed−state inactivation. Using chimeric substitutions between Kv6.4 and Kv9.3 subunits, we demonstrated that the lower half of the S6 domain (S6c) plays a crucial role in the 4−AP induced potentiation. These results demonstrate that KvS subunits modify the pharmacological response of Kv2 subunits when assembled in heterotetramers and illustrate the potential of KvS subunits to provide unique pharmacological properties to the heterotetramers, as is the case for 4−AP on Kv2.1/Kv6.4 channels.


Introduction
Voltage−gated K + (Kv) channels are K + selective membrane proteins that open, close and/or inactivate in response to changes in the membrane potential [1]. They exist as tetramers of α−subunits each consisting of six transmembrane segments (S1−S6) [2]. The four S5−S6 segments generate the central ion conducting pore while the S1−S4 segments form the voltage−sensing domains (VSDs). The channel pore can be sealed off by the channel gate at the level of the S6 bundle artery vasoconstriction [16,17], mutations in Kv8.2 have been associated with an inherited retinal dystrophy [18] and epilepsy susceptibility [19] and recently the involvement of Kv6.4 in a fast motor neuron biophysical signature has been demonstrated [20].
Because the S6c sequence of KvS subunits differs from that of Kv2 subunits and since certain Kv2/KvS combinations appear to display a more tissue−specific expression than the more ubiquitously expressed Kv2 subunits, these Kv2/KvS heterotetramers have a therapeutic potential as more specific pharmacological targets. In this study, we demonstrate that Kv6.4 subunits confer a distinct pharmacological profile to the Kv2.1/Kv6.4 heterotetramers for the common channel inhibitor 4−aminopyridine (4−AP) in which the Kv6.4 S6c sequence plays an important role. 4−AP is known to inhibit Kv2.1 channels, but surprisingly it potentiated the Kv2.1/ Kv6.4 current. This 4−AP response was unique for the Kv2.1/Kv6.4 heterotetramers and not observed with other KvS subunits suggesting that various Kv2.1/KvS heterotetramers harbor a unique pharmacological profile.

Transient transfection and cell culture
Mouse fibroblast Ltk − cells (American Type Culture Collection CLL. 1.3) were chosen for the electrophysiological experiments because of low levels of endogenous current [23,24]. Ltkcells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% horse serum (purchased from Gibco1, Life Technologies, Gent, Belgium) and 1% penicillin/streptomycin at 37°C and under 5% CO 2 . Ltk − cells were transiently transfected with 0.05 μg-5 μg cDNA of wild type (WT) or mutant channel subunits along with 0.5 μg eGFP as transfection marker using the polyethylenimine reagent (Sigma−Aldrich, St. Louis, MO, USA) according to the manufacturer's instruction. To ensure that only a population of heterotetrameric Kv2.1/KvS channels was formed, the KvS and Kv2.1 subunits were co−transfected in a 10:1 cDNA ratio. Twenty−four hours post transfection the Ltk − cells were trypsinized and used for the electrophysiological experiments.

Electrophysiology
Current recordings were obtained in the whole cell configuration at room temperature (20 −23°C) with an Axopatch−200B amplifier (Axon Instruments, Union City, CA, USA). The current recordings were low−pass filtered through a Bessel filter and sampled at 1 − 10 kHz with a Digidata 1200A data acquisition system (Axon Instruments). Data storage and command voltages were controlled with the pClamp10 software (Axon Instruments). Pipettes were pulled using a P−2000 puller (Sutter Instruments, Novato, CA, USA) from borosilicate glass capillaries (World Precision Instruments, Sarasota, FL, USA) and afterwards heat polished to obtain patch pipettes with a resistance between 1 MO− 2.5 MO. The patch pipettes were filled with an intracellular solution (ICS) containing 110 mM KCl, 5 mM K 2 ATP, 5 mM K 4 BAPTA, 2 mM MgCl 2 , 10 mM HEPES and adjusted to pH 7.2 with KOH. The Ltk − cells were continuously perfused with an extracellular solution (ECS) containing 145 mM NaCl, 4 mM KCl, 1 mM MgCl 2 , 1.8 mM CaCl 2 , 10 mM HEPES, 10 mM glucose and adjusted to pH 7.35 with NaOH. Junction potentials between the ICS and ECS were zeroed with the filled pipette in the bath solution. 4−aminopyridine (4−AP) (Sigma−Aldrich), flecainide (S.A. Meda Pharma N.V., Brussels, Belgium) and quinidine (Eli Lilly & co, Indianapolis, IN, USA) were diluted in ECS solution to obtain a stock solution of 180 mM, 2 mM and 1 mM, respectively. Stock solutions were balanced to a pH of 7.35. The desired working concentrations of 4−AP, flecainide and quinidine were made by diluting these stock solutions in ECS solution and were applied by using a fast perfusion system (ALA Scientific Instruments, Farmingdale, NY, USA).

Data analysis and statistics
Voltage protocols are represented in the figures and were adjusted to the biophysical properties of the different Kv channels. The voltage dependence of activation and inactivation was fitted with a single Boltzmann function according to y = 1 / (1 + exp (−(V-V 1/2 ) / k)), with V representing the voltage applied, V 1/2 the voltage at which 50% of the channels are activated or inactivated, and k the slope factor. Time constants of activation (from + 80 to −10 mV) and deactivation (−20 to −70 mV) were obtained by fitting the raw current traces of either the activation or deactivation (not shown) protocol with a single or double exponential function. Dose−response curves were obtained by plotting y, the fraction of current remaining at +30mV, as a function of drug concentration, [D], and fitted with the

Nomenclature
We reported in 2002 the cloning and properties of three KvS subunits designated Kv6.3, Kv10.1 and Kv11.1, based on the degree of sequence homology in the S1-S6 region [15]. The Kv subunit nomenclature was revised by IUPHAR in 2005 [13]. Kv10.1 became Kv6.3, Kv6.3 was renamed to Kv6.4 and Kv11.1 to Kv8.2. Thus, the Kv6.4 mentioned in this paper is the KvS subunit that displayed the largest hyperpolarizing shift in the voltage-dependence of inactivation in the original report [15].

4−AP potentiates Kv2.1/Kv6.4 currents while inhibiting other Kv2/KvS combinations
A common feature of all the KvS subunits is that they lack the 2 nd proline residue of the PXP motif within the bottom end of the S6 segment (S6c) (Fig 1), a channel region that has been shown to be involved in the interaction with pore blockers [8][9][10]. Substituting the 2 nd proline of the PXP motif by an alanine or glycine drastically reduced the 4−AP affinity of the Shaker Kv channel [25]. This raised the question whether the KvS subunits also affect the pharmacological profile of Kv2 containing heteromers, in addition to altering the biophysical properties. Since the effect of 4−AP on Kv2/KvS heteromeric channels has not been investigated extensively, we screened heterotetramers of Kv2.1 and either Kv5.1, Kv6.4, Kv8.1 or Kv9.3 for their sensitivity to 4−AP (Fig 2). These KvS subunits were chosen as representative members for each of the four KvS subfamilies. Application of 18 mM 4−AP, which is the reported IC 50 value for Kv2.1 inhibition [26,27], blocked 10 ± 2%, 56 ± 2% and 12 ± 1% of Kv5.1, Kv8.1 and Kv9.3 containing currents, respectively (n = 7) (Fig 2A and 2B). Full dose−response relationships for the different heterotetrameric Kv2.1/KvS and for homotetrameric Kv2.1 channels were determined and compared to obtain a relative measurement of the 4−AP sensitivity (Fig 2C). No proper IC 50 value and Hill coefficient (n H ) could be determined for the Kv2.1/Kv5.1 and Kv2.1/Kv9.3 channel due to their weak sensitivity for 4-AP inhibition (Fig 2C). Kv2.1/Kv8.1 channels displayed sufficient sensitivity and were inhibited with an IC 50 of 24.1 ± 2.0 mM and n H = 1.4 ± 0.2 (n = 5). Homotetrameric Kv2.1 channels displayed an IC 50 of 4.5 ± 1.8 mM (n = 6) for inhibition by 4−AP, and Hill coefficient n H of 1.0 ± 0.1 (n = 6), which is a higher affinity compared to the value of 18mM reported previously [27]. However, this difference most likely reflects the difference in experimental conditions namely the used heterologous expression system. Our IC 50 value corresponds well with the reported sensitivity in HEK cells and cultivated neurons/myocytes (0.5mM to 5mM) [28,29] but is lower compared to similar experiments performed in oocytes (18mM) [27]. The dose−response relationships of Kv5.1, Kv8.1 and Kv9.3 currents were all right−shifted towards a higher concentration range in comparison to Kv2.1 currents. Thus, Kv5.1, Kv8.1 and Kv9.3 heterotetramers display an altered sensitivity for 4−AP.
Interestingly, application of 4−AP increased the Kv6.4 current by 16 ± 1% (n = 9). To exclude that this effect was due to a potential change in intracellular pH caused by the high 4−AP concentration, we increased the buffering capacity of the "intracellular" pipette solution from 10 mM to 20 mM HEPES; even with this solution we observed an increase in Kv2.1/ Kv6.4 currents after 4−AP application. This potentiating effect of 4-AP was not conserved within the Kv6 subfamily because heterotetrameric Kv2.1/Kv6.3 channels (before the IUPHAR classification also known as Kv10.1) were inhibited by 4-AP (S1A Fig), although with a reduced sensitivity compared to homotetrameric Kv2.1 channels (S1 Fig) [13,15].
To examine whether this effect was unique to 4−AP, we tested the effect of two other chan-  While not significantly altering the voltage dependence of activation, 4−AP shifted the voltage dependence of inactivation of Kv2.1/Kv6.4 channels slightly towards more depolarized potentials (p = 0.031, Fig 3D and Table 1). To exclude that this was due to a possible depolarizing drift in our recordings, the V 1/2 of inactivation was performed over the course of 25 min. (S4A Fig). The V 1/2 of inactivation displayed a small, non significant drift towards hyperpolarized potentials when comparing the initial recording (t = 5 min.) with the final  inactivation during the 5 second prepulse but also the degree of U−type inactivation. The reduction in channel inactivation was most pronounced at potentials below the threshold for BC gate opening, i.e. more negative than −40 mV. Under control conditions 12 ± 2, 23 ± 3, 46 ± 5 and 73 ± 3% (n = 8) of Kv2.1/Kv6.4 channels are inactivated at −80, −70, −60 and −50 mV, respectively; 4−AP significantly decreased the amount of inactivation at these potentials to roughly 1 (p = 0.033), 12 ± 3 (p = 0.044), 31 ± 6 (p = 0.041) and 55 ± 7% (p = 0.036), respectively (n = 7) (Fig 3D). This suggested that 4−AP prevented (or reduced) the transition from a closed state to a closed−inactivated state in Kv2.1/Kv6.4 channels. The concentration−dependence of the 4−AP induced current increase saturated at concentrations above 10 mM, EC 50 = 2.9 ± 0.1 mM (n = 5), but for consistency, all of the following experiments were performed with the 18 mM 4−AP screening concentration ( Fig 3E).

Effect of 4−AP on the kinetics of closed−state inactivation
4−AP could interfere with the closed to closed−inactivated state transition in Kv2.1/Kv6.4 channels by altering the induction of and/or recovery from closed−state inactivation. To investigate whether 4−AP affects the rate of development of closed−state inactivation, we used a protocol as illustrated in Fig 4A; the initial current amplitude was recorded at a test pulse to +60 mV (P1) followed by a 5s recovering pulse to −100 mV to eliminate any  inactivation that had developed, a pulse of variable duration to −60 mV to develop closed −state inactivation and a second test pulse to +60 mV (P2) to obtain the current amplitude after the initiation of closed−state inactivation. The fraction P2/P1 represents the degree of inactivation and was plotted against the time spent at −60 mV (Fig 4C, normalized plot). The holding potential was kept at −100 mV to avoid development of closed−state inactivation between the different current traces. Under control conditions (upper panel Fig 4A, black circles Fig 4C) approximately 70% of the Kv2.1/Kv6.4 channels were inactivated after 50s at −60 mV and closed−state inactivation developed with a time constant of 5.4 ± 0.2 s (n = 6).
In the presence of 4−AP (lower panel Fig 4A, red circles Fig 4C), the total amount of inactivation was reduced to approximately 40% and the time constant was increased to 9.3 ± 0.9 s (n = 6, p = 0.032). To investigate the rate of recovery from closed−state inactivation, a similar protocol was used ( Fig 4B); the initial +60 mV (P1) and recovery −100 mV pulse remained the same but were now followed by a 10 s pulse to −60 mV to induce closed−state inactivation, followed by a pulse of variable duration to −100 mV to allow the channels to recover from closed−state inactivation and a second test pulse to +60 mV (P2) to obtain the current amplitude after the recovery from closed−state inactivation. The P2/P1 ratio now represents the fraction of Kv2.1/Kv6.4 channels that had recovered from a closed−inactivated state ( Fig  4D). To confirm that 4−AP interferes with the transition from the closed state to the closed−inactivated state in Kv2.1/Kv6.4 channels, a twin−pulse protocol was used: two depolarizing pulses to +60 mV were separated by a 10 s interval at −100 mV allowing the Kv2.1/Kv6.4 channels to recover from closed−state inactivation (τ for recovery from closed−state inactivation was 1.2 ± 0.2 s, Fig 4D). Under control conditions, the current amplitude obtained during the second depolarizing pulse was 16.1 ± 0.9% (n = 7) higher than the current amplitude obtained at the first depolarizing pulse (Fig 4E, black trace). However, in the presence of 4−AP no significant differences in current amplitude were observed between the first and second depolarizing pulse and the difference in current amplitude (%) between the control and 4-AP conditions was found to be statistically significant (p < 0.001, Fig 4E, red trace). Furthermore, the peak amplitude of both pulses corresponded to the current amplitude of the second pulse obtained under control condition. These results indicated that a substantial fraction of the Kv2.1/Kv6.4 channels (~15%) reached the closed−inactivated state at a holding potential of −80 mV and that 4−AP prevented (or reduced) this closed state to closed−inactivated state transition due to a faster recovery from and slower initiation of closed−state inactivation (Fig 4C and 4D). To confirm our hypothesis further we promoted closed−state inactivation in Kv2.1/Kv6.4 channels by applying our two− pulse protocol after a 5 s pre−pulse to −60 mV (Fig 4F). This −60 mV pre−pulse induced more closed−state inactivation under control conditions: the current amplitude of the second depolarizing pulse was 56 ± 6% (n = 6) higher than that of the first depolarizing pulse. In presence of 4−AP the difference between the second and first depolarizing test pulse was reduced to only 18 ± 4% (n = 5), as expected if 4−AP reduced the closed−state to closed−inactivated state transition through an increased recovery rate (at -100 mV) from and slower initiation rate (at -60 mV) to the closed−inactivated state (Fig 4F). The difference in current amplitudes (%) between control and 4−AP conditions proved to be statistically significant (p = 0.002).
Interestingly, some other Kv2.1/KvS and homotetrameric Kv2.1 channels display closed− state inactivation but were inhibited by 4−AP. If the 4-AP induced potentiation and kinetic  (Fig 2). Therefore, the potentiation of Kv2.  (Fig 6A and 6D) (Fig 6E and Table 1). Additionally, the characteristic inactivation from the closed state was preserved in the chimera and closely resembled that of WT Kv2.1/Kv6.4 heterotetramers (Fig 6C and 6E). Application of 18 mM 4−AP caused an inhibition of Kv2.1/Kv6.4_S6c_Kv9.3 currents of 12 ± 3% (n = 7), which is comparable to the 12 ± 1% inhibition of WT Kv2.1/Kv9.3 heterotetramers (p = 0.295) and in marked contrast to the activator effect in WT Kv2.1/Kv6.4 channels (Fig 6A and 6B). 4−AP did not affect the voltage dependence of channel inactivation (p = 0.582). These results showed that exchanging the S6c region from Kv6. Solid lines represent the exponential fit. Note that 18 mM 4−AP decreased the amount of inactivation and slowed the initiation rate of closed−state inactivation (C) while accelerating the recovery from closed−state inactivation (D). (E) The twin pulse protocol (top) reveals that under control conditions (black trace) a fraction of channels can be recovered from their closed−inactivated state by a 10s pulse to −100mV, seen as an increase in peak amplitude from P1 to P2. With 4−AP (red trace), the amplitude at both test pulses were the same as P2 in control, suggesting that 4−AP interferes with the transition from the closed to the closed−inactivated state. The holding potential was kept at −80 mV between different current traces. (F) Inducing more closed−state inactivation by a 5s prepulse to −60 mV increased the difference in peak amplitude between P1 and P2 pulse by approximately 50% in control. 4−AP reduced the current less after the prepulse to −60 mV compared to control conditions and reduced the difference in peak amplitude of the test pulse to approximately 15%, suggesting that 4−AP interfered with the closed to closed−inactivated state transition.  Table 1). Due the total amount of inactivation developing at the V 1/2 of inactivation but did not significantly alter the induction and recovery rate of closed−state inactivation. (E) The twin pulse protocol (top), as described in Fig 4, revealed that under control conditions (black trace) no substantial fraction of Kv2.1/Kv9.3 channels is in a closed−inactivated state as the peak amplitude between P1 and P2 is equal. 4−AP inhibited both P1 and P2 to the same extent. Thus, in contrast to Kv2.1/Kv6.4 channels, no closed−inactivated Kv2.1/Kv9.3 channels can be recovered by 4−AP or the −100 mV recovering pulse. (F) Bar chart of the kinetics of closed−state inactivation before and after application of 4−AP: recovery rate (left), induction rate (middle) and degree of inactivation after a 10s pulse to −45 mV (right). * represent statistical significance. Black and grey bars represent control and 18 mM 4−AP, respectively. doi:10.1371/journal.pone.0141349.g005

4−AP potentiates currents mediated by Kv2.1/Kv6.4 heteromeric channels
The S6c domain has been shown to form a binding site for channel inhibitors such as 4−AP, flecainide, propafenone and quinidine [9][10][11][12]35]. The S6c of most Kv channels contains a conserved PXP motif that, besides being important for channel gating [6,36], has been shown to modulate the channel´s sensitivity to 4−AP and internal pore blockers (e.g. quinidine). Substituting the second proline of the PXP motif by an alanine drastically decreased the affinity for 4 −AP in Shaker [25]. Interestingly, all the members of the Kv7 (KCNQ) subfamily lack a full PXP motif (Fig 1) and neither of them is sensitive to 4−AP [37]. The fact that the KvS subunits also miss the 2 nd proline of the PXP motif (Fig 1) (Fig 2C), which is in agreement with previous findings on Kv9.3 channels [16]. However, the most intriguing change in 4−AP sensitivity was induced by the Kv6.4 subunit: in this case 4−AP potentiated, rather than inhibited, Kv6.4 currents. Interestingly, in mouse bladder smooth muscle cells and gastrointestinal smooth muscle cells it has been demonstrated that 4-AP was capable of potentiating native Kv currents [38,39]. In addition, the expression of several KvS subunits was confirmed in mouse bladder smooth muscle cells [38,40]. Importantly, Thorneloe and Nelson documented both the presence of Kv6.4 (in their study named Kv6.3 following the old nomenclature) in isolated bladder myocytes, and potentiation of the delayed rectifier Kv current by 10 mM 4-AP in these cells, thus providing physiological relevance for our heterologously obtained data [38].
The potentiation was a unique characteristic of 4−AP since the internal pore blockers quinidine and flecainide still blocked Kv2.1/Kv6.4 channels, although with a slightly higher affinity compared to homotetrameric Kv2.1 channels (S2C Fig). A similar increase in quinidine affinity was also observed in the Shaker mutant with a PXA sequence [25]. The potential to convert the 4−AP effect from inhibition to current potentiation appears not to be conserved within all members of the Kv6 family since a previous study reported that Kv2.1/Kv6.2 heterotetramers display a higher degree of current inhibition by 18 mM 4−AP compared to Kv2.1 homotetramers [41]. We observed that Kv2.1/Kv6.3 heterotetramers are also inhibited by 4-AP (S1 Fig

4−AP interferes with closed−state inactivation transition
In heterotetrameric Kv2.1/Kv6.4 channels, the voltage dependence of inactivation is shifted by 40 mV towards more negative potentials without substantially affecting the voltage dependence of activation. Consequently, the Kv2.1/Kv6.4 channels start to inactivate at potentials that are below the apparent threshold for channel activation. Typically, ion channels with preferential closed−state inactivation possess a U− (or J) shaped (steady state) inactivation profile, i.e. at stronger depolarizations the degree of inactivation diminishes, a characteristic strongly present in Kv2.1/Kv6.4 channels [32,[42][43][44]. At −80 mV, approximately 12% of the Kv2.1/Kv6.4 channels populate the closed−inactivated state under control conditions, as evidenced using a twin pulse protocol in Fig 4E, and 4−AP restored this difference in current amplitude suggesting a possible interference with closed−state inactivation. An extra 5 s prepulse to −60 mV in the twin pulse protocol increased the total amount of closed−state inactivation and allowed 4−AP to recover more Kv2.1/Kv6.4 channels from their inactivated state, strengthening this hypothesis ( Fig 4F). If Kv2.1/Kv6.4 channels exhibit a small, non zero open probability (P o ) at −60 mV, it cannot be excluded that, in combination with the duration of the conditioning pulse, 4−AP (partly) acts on open−state inactivation, which has been previously reported for the Shaker channel [45]. However, the observations that 4−AP reduced and slowed induction of inactivation, enhanced the rate of recovery and diminished the degree of U−shaped inactivation all points towards a loss of closed−state inactivation and more inactivation at positive potentials (e.g. +80 mV). The latter observation can be explained by a strengthening of open−state inactivation, as suggested by the macroscopic current kinetics in Fig 3A, which is in contrast to previous findings that 4−AP can reduce open−state inactivation [45]. The exact underlying mechanism of closed−state inactivation has not been resolved yet, but several studies indicate that it involves a dynamic coupling between S6c and the S4−S5 linker, which forms the electromechanical coupling that translates VSD movement into BC gate opening [43,46]. If an uncoupling occurs at this S6c/S4−S5linker interface, BC gate (and channel) opening is prevented and the channel enters a nonconductive, inactivated state. Due to the subunit−cooperative nature of BC gate opening in Shaker−related Kv channels, disrupting the electromechanical coupling of one subunit (e.g. Kv6.4) would be sufficient to enter this nonconductive state. A similar mechanism of uncoupling has been proposed for hyperpolarized cyclic nucleotide−gated (HCN) channels where the uncoupling between the VSD and the channel gate (at hyperpolarized potentials) results in desensitization to voltage, observed as gating charge immobilization [47,48]. Interestingly, such a desensitization to voltage, observed as gating charge immobilization, also occurs in Kv4.2 channels that exhibit prominent closed −state inactivation [49].
If closed−state inactivation involves a collapse of the selectivity filter, as in C−type inactivation, then the process is most likely triggered by rearrangements in S6c (in the vicinity of the BC gate). Indeed, in C−type inactivation the S6c reorientations associated with BC gate opening are transmitted upwards along the S6 helix up to the level of the selectivity filter [50][51][52]. The final collapse of the selectivity filter is probably a process whereby the destabilizing input of 1 subunit is sufficient to trigger inactivation [53,54]. If closed−state inactivation has a similar mechanism, the rearrangements in S6c should occur, and be transmitted upwards, before channel opening. For the Kv2.1/Kv6.4 heteromeric channels we showed that the gating charge movement displays a component at hyperpolarized potentials with a voltage dependence that matches the voltage−dependence of inactivation [55]. This suggests that at −80 mV already a substantial amount of the gating charge moves (originating from the S4 movement of either the Kv6.4 or Kv2.1 subunits) which could potentially be translated in conformational changes of S6c associated with inactivation gating. It is currently not known whether the S6c of Kv6.4 actually moves, but if it does and if this movement occurs at more hyperpolarized potentials, channel inactivation could be triggered. In this model our data would strengthen the hypothesis that one subunit is sufficient to initiate the collapse of the selectivity−filter [50,51].
Here we propose that 4−AP modulates closed−state inactivation in the heteromeric Kv2.1/ Kv6.4 channels either by stabilizing the coupling at the S6c/S4−S5linker interface or by preventing S6c reorientations associated with inactivation gating. In tuberomammillary neurons, it has been demonstrated that 4−AP enhanced the Kv4 A−type current (I A ) at subtreshold potentials [56]. It was suggested that 4−AP−bound channels were unable to inactivate resulting in a potentiation at subthreshold potentials. This observation fits with our finding that 4−AP prevents closed−sate inactivation, a process that is also present in Kv4.2 channels [31].

Molecular determinants of the 4−AP binding site
Location of the 4−AP binding site remains uncertain but both the S4−S5 linker and the cytoplasmic half of S5 and S6 have been shown to be important for 4−AP sensitivity in Kv channels [10,27,30,57]. Molecular docking studies of aminopyridines confirmed this and placed the 4 −AP binding pocket at the S4−S5 linker and S6c interface [58]. However, the location of 4−AP binding site does not allow us to rule out whether closed−state inactivation arises from uncoupling the S6c/S4−S5linker interface or conformational changes that lead to selectivity filter collapse. In both cases, 4−AP unbinding is required in order for closed−state inactivation to occur.
Kv2.1/Kv9.3 heteromers display a large hyperpolarizing shift in the voltage−dependence of inactivation (similar to Kv2.1/Kv6.4 heteromers), but they are inhibited, rather than potentiated by 4−AP. This suggests that the hyperpolarized position of the inactivation curve is not the only determinant for the potentiating effect of 4−AP. Kirsch and Drewe were able to transfer the high affinity of Kv3.1 for 4−AP onto Kv2.1 by exchanging both the S5 and S6 domains [27]. Given the sequence differences in S6c between Kv6. 4 (Fig 6). However, the opposite substitution, i.e. replacing the S6c region of Kv9.3 by the Kv6.4 sequence, did not transfer the 4−AP potentiation effect onto the Kv9.3 currents. Interestingly, the Kv2.1/Kv9.3_S6c_Kv6.4 channels displayed altered biophysical properties, including a depolarizing shift of approximately 20mV in the voltage dependence of inactivation and a reduced degree of U−shaped inactivation. This suggests that the 4−AP effect on Kv2.1/KvS heterotetramers is dependent on both the S6c sequence and the presence or absence of the prominent closed−state inactivation process. It cannot be excluded that other channel regions are involved in creating the 4−AP binding site, especially the S4−S5 linker and the cytoplasmic half of S5 due to their contribution to dynamic coupling in the known mechanism of closed−state inactivation.

General conclusion
Our present study reports that KvS subunits have the ability to alter the pharmacological profile of Kv2.1 channels upon heterotetramerization. Thus, heterotetrameric Kv2/KvS channels not only display unique biophysical properties but also unique pharmacological effects as is the case for 4−AP on Kv2.1/Kv6.4 channels where it interferes with the closed−state inactivation process. It has been demonstrated that heterotetrameric Kv2.1/KvS channels regulate several (patho)physiological processes, e.g. mutations of Kv8.2 contribute to epilepsy susceptibility, as they contribute to several native delayed rectifier currents [19,20,59,60]. Because of their structural differences and unique biophysical properties, different KvS subunits can not only alter the Kv2 pharmacology, but it is also likely that KvS subunits could be pharmacologically specifically targeted, without influencing Kv2 homomers, allowing interventions in physiological processes regulated by unique Kv2/KvS heteromers.