Carboxyl-group compounds activate voltage-gated potassium channels via a distinct mechanism

Voltage-gated potassium (K_V) channels play a crucial role in cardiac and neuronal excitability (Hille, 2001). Inherited or acquired mutations in K_V channel genes can cause cardiovascular, neurological, and metabolic disorders known as “channelopathies” (Ashcroft, 2000), and pharmaceuticals, such as the antiseizure drug retigabine, target a K_V channel (Miceli et al., 2018).

K_V channels consist of two types of domains: a pore domain surrounded by four voltage-sensor domains (VSD) (Fig. 1 A, for clarity only two VSDs are shown). The VSD contains a mobile voltage sensor S4, which responds to alterations in the voltage across the membrane; when the intracellular side becomes more positive, the positively charged voltage sensor moves from a down-state to an up-state to pull the gate open and lets potassium ions flow out from the cell (Fig. 1 A) (Delemotte et al., 2011; Henrion et al., 2012; Jensen et al., 2012; Long et al., 2007). While pharmaceuticals targeting the pore domain to block the ion-conducting pore have been known and used for decades (Hille, 1977; Ragsdale et al., 1994; Strichartz, 1976; Zhou et al., 2001), few drugs activate voltage-gated channels, and no drug targeting the VSD is available in clinical practice.

Previous research has shown that lipophilic and negatively charged small-molecule compounds, such as some resin acids and derivatives thereof, and polyunsaturated fatty acids (PUFAs) are effective activators of several K_V channels and have the potential to suppress neuronal and cardiac excitability (Börjesson et al., 2008; Cui et al., 2016; Imaizumi et al., 2002; Liin et al., 2016; Ottosson et al., 2015, 2021; Salari et al., 2018). One proposed mechanism is that the negatively charged compound binds at the interface between the membrane’s lipid bilayer and the channel’s VSD to electrostatically interact with the positively charged arginines of S4 in the VSD, which drives the voltage sensor S4 from a resting (down) state to an activated (up) state, to open the ion-conducting pore (Site 1 in Fig. 1 A) (Börjesson and Elinder, 2011; Ottosson et al., 2017). We refer to these types of compounds as “lipoelectric,” because the membrane-exposed site requires lipophilicity and the electrostatic pulling requires electrostatic interactions between charges (Börjesson et al., 2008).

In contrast, other types of compounds bind to other parts of the VSD to open the channel via other mechanisms. For instance, biaryl sulfonamides (Fig. 1 B) bind from the extracellular solution to the top of the VSD (Site 2 in Fig. 1 A); they bind in a crevice between the transmembrane segments of the VSD to keep the voltage sensor S4 in an activated (up) state in both Na and K channels (Ahuja et al., 2015; Liin et al., 2018). Recently, a large family of warfarin-like tautomers was reported to bind to the intracellular side of the human K_V1.5 and the Drosophila Shaker K_V channel, to the linker between the VSD and pore domain (Site 3 in Fig. 1 A), and to stabilize this linker in an activated state (Silverå Ejneby et al., 2020). In addition, compounds can bind to the pore domain to activate the channel (4 in Fig. 1 A). A well-known example is the discontinued antiseizure drug retigabine (Wuttke et al, 2005).

A crucial point in the development of pharmaceuticals against specific diseases is to find drugs that selectively act on one type of channel while leaving other channels unaffected. In this study, we searched for new lipoelectric compounds. To do so, we used the Shaker K_V channel supplemented with two additional gating charges at the top of the voltage sensor S4 to increase the electrostatic effect (Ottosson et al., 2014). This channel will be referred to as the 3R Shaker K_V channel because three arginines (R; the native top arginine in addition to the two added ones) are important for the effect. The reason for using this supplemented channel was that we expected it to identify even weak lipoelectric channel activators, which might be activators of other K_V channels too.

Here, we report on a high-throughput screen of 10,096 compounds with follow-up experiments on the site and mechanism of action. We describe a large family of carboxyl (COOH)-group compounds (Fig. 1 B) acting via the lipoelectric mechanism, distinct from several other K_V-channel activating compounds.

The details of the high-throughput screen have been described elsewhere (Liin et al., 2018; Silverå Ejneby et al., 2020; and references therein). In brief, the 3R Shaker K_V channel (Ottosson et al., 2014) was expressed in a CHO-K1 stable cell line and studied with a 384-well IonWorks equipment. The holding voltage was ‒80 mV, and 150 ms voltage-clamp steps were applied to +10 mV, +30 mV, and +50 mV. The cells were kept at ‒80 mV for 300 ms between each step. The current signal was sampled at 10 kHz. No external electronic filter was used. All recordings were made at room temperature (∼21°C). Modulation of the K_V-mediated current was assessed by dividing the post-scan steady-state K_V current (for the last 10 ms of each pulse) by the respective prescan K_V current for each well and corrected for run down. For the quantitative analysis presented in this paper, we used the data from +10 mV. All data between 0.75 and 1.25 (mean ± 3 SD) were considered to have no effect. If multiple experiments were carried out, we adjusted the significance level by dividing the deviation from 1 by n^0.5, where n is the number of experiments.

Compounds were searched for in the following chemical libraries (library name [company or organization]): Acids (Vitas-M Laboratory Ltd.); Consortium set (Specs)*; Drug Like (Enamine)*; Hit Finder (Maybridge)*; Ion Channel Ligand and Nuclear Receptor Ligand (Enzo Life Sciences Inc.); Known Drugs (Prestwick Chemical); LCBKI Primary Screening Set (Chemical Biology Consortium Sweden)*; Natural derivatives (TimTec); NIH Clinical Collection (National Institutes of Health); Peptidomimetics (ChemDiv Inc.)*; Tocriscreen Plus (Tocris Bioscience). Apart from LCBKI, provided by the Chemical Biology Consortium Sweden, and the NIH clinical collection, provided by the Libraries Roadmap Initiative, all compound libraries were commercially available. Whenever a selection was made (denoted by *), a chemist was instructed to search for small, hydrophobic, and partially charged compounds. The final concentration, in the test solution, was 10 µM of the compound. The tested compounds were added to the extracellular solution.

Frequency diagrams (shown in Fig. 3) were constructed by calculating the floating average frequencies of a specific family of compounds among all compounds. All data on the effects of the compounds were binned in intervals of 0.01. The number of compounds per bin is plotted in Fig. 2 A. An effect value of 1 means no effect and the grey area (0.75–1.25) means no significant effect. To calculate the frequency of a specific family of compounds among all compounds for a specific effect, we included all data points (one per compound) within a range of ± 0.05. In the effect range 0.54–1.47, the number of included data for each calculated frequency point was 50–2,394 (Fig. 2 B, red line). Outside this range, the effect window was increased to make sure that at least 50 data/compounds were used for each plotted point (Fig. 2 B black curves). The effect value (the relative current plotted on the x-axis in Fig. 3) was the average effect of all included data/compounds. The original screen included 10,096 compounds (Silverå Ejneby et al., 2020). In this previous investigation, we reported on a small family (259 compounds) of a distinct warfarin-like tautomer motif, which turned out to be a very effective K_V channel activator. In the presented frequency analysis, we excluded this family, which left us with 9,837 compounds.

The Shaker H4 channel (Kamb et al., 1988) was modified to remove the fast N-type inactivation (ShH4IR) (Hoshi et al., 1990). This channel will be referred to as the wild-type (WT) Shaker channel. To enhance the sensitivity to lipophilic and negatively charged compounds, we introduced two positively charged arginines (M356R and A359R) in the extracellular end of the voltage sensor S4; these two arginines in addition to one native arginine (R362) made the channel more sensitive to the PUFA docosahexaenoic acid and resin acids (Ottosson et al., 2014). Because all three arginines are important for the effect, we refer to this channel as the 3R Shaker K channel. The human K_V7.2/7.3 channel was expressed through coinjection of K_V7.2 (Gene bank accession number NM_004518) and hK_V7.3 (Gene bank accession number NM_004519) in a 1:1 ratio (Liin et al., 2016). Point mutations in K_V7.2 were introduced using site-directed mutagenesis (QuikChange II, Agilent) and verified by sequencing at a core facility at Linköping University, Sweden. cRNA (from linearized DNA) was prepared using the mMessage mMachine T7 kit (Invitrogen) (Larsson et al., 2020).

All animal experiments were approved by Linköping’s local Animal Care and Use Committee. The experiments were performed in accordance with relevant guidelines and regulations. African clawed frogs (Xenopus laevis) were anesthetized with 1.4 g/liter ethyl 3-aminobenzoate methanesulfonate salt (tricaine). After an incision through the abdomen, a batch of oocytes was removed. Clusters of oocytes were separated by incubation for ∼1 h in a Ca-free O-R2 solution (in mM: 82.5 NaCl, 2 KCl, 5 HEPES, and 1 MgCl₂; pH adjusted to 7.4 by NaOH) containing Liberase Blendzyme. The oocytes were then incubated at 8°C in a modified Barth’s solution (MBS; in mM: 88 NaCl, 1 KCl, 2.4 NaHCO₃, 15 HEPES, 0.33 Ca(NO₃)₂, 0.41 CaCl₂ and 0.82 MgSO₄; pH adjusted to 7.6 by NaOH) supplemented with penicillin (25 U/ml), streptomycin (25 μg/ml), and sodium pyruvate (2.5 mM) for 2–24 h before injection. 50 nl of cRNA (50 pg) were injected into each oocyte using a Nanoject injector (Drummond Scientific). Injected oocytes were kept at 8°C in MBS until 1 day before electrophysiological recordings, after which they were incubated at 16°C. All chemicals were supplied by Sigma-Aldrich unless stated otherwise.

All manual two-electrode voltage clamp recordings were performed at room temperature (20–23°C) using a GeneClamp 500B amplifier and a Digidata 1440A digitizer with pClamp 10 software (all from Molecular Devices, Inc.). Data were sampled at 10 µs and the signal was filtered at 2 kHz by a Bessel filter. The amplifiers’ leakage subtraction was not used, but digital leakage subtraction was used during the analysis to remove any leakage current between the two most negative test pulses.

Compounds were initially dissolved at 30 mM in 99.5% EtOH and stored at −20°C. Compounds were subsequently diluted to the desired test concentration in an extracellular solution. The oocyte was placed in a bath surrounded by 1 K extracellular solution that contained (in mM): 88 NaCl, 1 KCl, 0.4 CaCl₂, 0.8 MgCl₂, and 15 MES (for pH 5.5 and 6.5), HEPES (for pH 7.4 and 8.0), or CHES (for pH 9.0), pH adjusted by NaOH (reaching a sodium concentration of ∼100 mM). The control solution and the compound solution were added to the bath with a peristaltic pump-driven perfusion system. Two glass microelectrodes were inserted into the oocyte using micromanipulators. The microelectrodes were pulled from borosilicate glass, filled with 3 M KCl, and had a resistance of 0.5–2 MΩ.

For the Shaker potassium channel recordings, all channels were closed when the membrane voltage was clamped to −80 mV, and this voltage was set as the holding voltage. Currents were evoked from the holding voltage of −80 mV by 100-ms long, 5-mV steps ranging from −80 up to +50 mV (WT) or +70 mV (3R). For the WT human K_V7.2/7.3 channel, the holding voltage was set to −100 mV and currents were evoked by 3,000-ms long 10 mV steps ranging from −120 up to +50 mV. This was followed by a 2,000-ms-long step at −30 mV, producing a tail current from which the initial currents were measured and analyzed. For the R198Q K_V7.2 mutant channel, the holding voltage was set to −120 mV and currents were evoked by 6,000-ms long 10 mV steps ranging from −140 up to +10 mV, which was followed by a tail current-producing 3,000-ms long step at −60 mV from which the initial currents were measured and analyzed.

To tune the charge of some compounds, when studying the effect on the WT K_V7.2/7.3 channel we altered the pH in the extracellular solution. The effect of the compound at a certain pH was always compared with a control solution at the same pH and the compound solution. The pH alteration per se had a modest effect on the apparent voltage sensed by the channel, seen as a shift of voltage-dependent parameters along the voltage axis (Prole et al., 2003).

The manual electrophysiological data from oocytes were processed and analyzed with Clampfit 11.2 (Molecular Devices, LLC.) and GraphPad Prism 10 (GraphPad Software, Inc.).

JChem for Excel version 23.5.0.480 (ChemAxon Ltd.) was used for drawing chemical structures and calculating the molecular weight, the acidic dissociation constant (pK_a), and the octanol/water partitioning coefficients (LogP and LogD) for uncharged and charged compounds, respectively, as previously described (Liin et al., 2018; Silverå Ejneby et al., 2020).

Average values are expressed as mean ± SEM. Statistical analyses were done using an unpaired t test when comparing two data points against each other. P < 0.05 was considered statistically significant. Most statistical analyses were performed in GraphPad Prism 10 (GraphPad Software, Inc.). The statistical analyses of the frequency diagrams (Fig. 3) and associated Table 1 were done with skewed binomial distributions calculated in Microsoft Excel 16 (Microsoft Corporation).

To search for K_V channel activators, we screened 10,096 compounds on the 3R Shaker K_V channel. To increase the possibility of finding “lipoelectric” compounds, the channel was supplemented with two extra positively charged residues in the extracellular portion of the voltage-sensor S4, referred to as the 3R Shaker K_V channel (Ottosson et al., 2014). The methodology of the screen has been described before (Liin et al., 2018; Silverå Ejneby et al., 2020) and we have previously found and described two families of compounds acting as K_V-channel activators: (1) one family has a distinct warfarin-like motif (compounds bind to Site 3 in Fig. 1 A [Silverå Ejneby et al., 2020]). In total, 259 compounds belonged to this family, and because data from this family was analyzed in its entirety, we excluded these compounds from the present analysis, leaving us with a total of 9,837 compounds. (2) The other family had a biaryl sulfonamide motif (compounds bind to Site 2 in Fig. 1 A [Liin et al., 2018]). Because we only analyzed data from a small fraction of these compounds (26 compounds in total), compounds from this family will be included in the initial analysis of this study.

The 9,837 analyzed compounds were divided into six classes (Fig. 1 B and Table 1): compounds containing a (1) COOH group (2,298 compounds in total), (2) amide group (2,021 compounds; excluding compounds in [3]), (3) biaryl amide groups (BAA; 733 compounds; an amide group flanked by, in most cases, aromatic rings, but in some cases also other rings), (4) sulfonamide groups (270 compounds, excluding compounds in [5]), (5) biaryl sulfonamides groups (BASA; 605 compounds, a sulfonamide group flanked by, in most cases, aromatic ring, but in some cases also other rings), and (6) none of (1)–(5) (other; 5,012 compounds). Because there is some overlap between the groups, the sum is larger than the total number of compounds. All combinations and numbers are presented in Table 1. To see if any of the groups constitute more activators or inhibitors than expected from a random distribution, we constructed frequency diagrams (Fig. 3). All compounds with an effect E ± 0.05 (x-axis) were included in the frequency analysis of effect E. For effects E < 0.54 and E > 1.47, the range was increased to make sure that data from 50 compounds were included in each data point (see Materials and methods).

The COOH compounds were clearly overrepresented among the clear activators (E ≥ 1.60, Fig. 3 A and Table 1); while constituting only 23% of all compounds, they constitute almost 60% of the best activators, suggesting that the COOH group might be a critical part of a K_V-channel activator. Amides are neither activators nor inhibitors, while BAAs are clearly overrepresented among the activators but not among the inhibitors (Fig. 3 A and Table 1). Sulfonamides are neither activators nor inhibitors, while BASAs are overrepresented among weak activators and weak inhibitors (see analysis below; Fig. 3 A and Table 1). The last group, containing all other compounds, constitutes 51% of all compounds. They are clearly overrepresented among the best inhibitors but underrepresented among the activators (Fig. 3 A and Table 1).

However, while COOHs, BAAs, and BASAs are overrepresented among the activators, we cannot rule out that it is the combination of these motifs that makes them activators. Therefore, we also analyzed the subgroups. COOHs in isolation, in combination with BAAs, and in total were significantly increased among the activators (Fig. 3 B and Table 1). In contrast, BAAs were only increased in combination with COOH and in total, but not in isolation (Fig. 3 C and Table 1). This suggests that it is the COOH group per se that is critical and possibly a COOH group in combination with BAA that makes a fruitful mix. The BASA motif in any combination is neither a significant activator nor an inhibitor if similarly analyzed (Table 1), but if we perform a statistical analysis of BASA occurrence in more narrow windows, we find that BASAs in isolation is a significant weak activator and weak inhibitor (Fig. 3 D); data points outside the vertical dashed lines above the continuous horizontal line denote statistically significant overrepresentation of the motif. In a previous study, we showed that BASAs bind in a groove at the top of the VSD to keep the voltage sensor S4 in an up-state (Liin et al., 2018). The current analysis suggests that BASAs also can be weak inhibitors, possibly by binding to the same site to press down S4. This will not be analyzed further in this article.

Above we described that a COOH group is the most significative functional group of K_V channel activators, possibly in combination with BAA. While most of the 2,298 COOH compounds did not have any effect or were weak inhibitors, 157 of them (6.8% of all COOH compounds) significantly increased the current (see Materials and methods for significance calculation), but most of them were only weak activators. 10 µM of 38 different COOH compounds increased the current of the 3R Shaker K_V channel at +10 mV by at least 60% (Effect E ≥ 1.60; Table 2). These strong activators were slightly heavier than the average COOH compound (384 g/mol (red) versus 348 g/mol (black); Fig. 4 A and Table 2), they were clearly more hydrophobic (LogP = 4.51 versus 2.36, Fig. 4 B and Table 2; LogD = 1.82 versus −0.30, Fig. 4 C and Table 2) and had a marginally higher pK_a value (3.89 versus 3.54, Fig. 4 D and Table 2).

Of the 38 strong activators 24 were run in duplicates in the screen (Table 2), and the mean deviation from the average value for these 24 compounds was 20%, suggesting robust data from the screen. Even though these 38 compounds do not unexpectedly show great variability in structure, some common motifs can be found (Fig. 5). In most of the compounds (29/38), the COOH group is directly attached to an aromatic ring of six or five atoms (columns 1–4, Fig. 5). 15 of these have a characteristic motif with an amide group (in most cases in ortho position) directly connected to the aromatic ring (columns 2–3, Fig. 4). Most of these also have an aromatic or non-aromatic ring directly connected to the amide motif (the BAA compounds). In other cases, there is an adjacent ring to the COOH connected ring (column 4, Fig. 4). In the last group, the COOH group is connected to a linker or a free carbon chain (column 5, Fig. 4). These structures are not surprising; in previous studies, we have reported that other COOH compounds of the types mentioned here can activate the 3R Shaker K_V channel: polyunsaturated fatty acids (Ottosson et al., 2014), resin acids and a derivative thereof (Ottosson et al., 2015, 2017), and resin acid derivatives where the COOH group is attached via a carbon chain (a stalk), optimally of two atoms (Ottosson et al., 2021; Silvera Ejneby et al., 2018). While the shapes of the activators are diverse, the COOH/BAA motif is striking. 12 of the 38 top compounds have this motif (in columns 2–3, Fig. 3). In the remainder of this article, we will focus on a selection of compounds from this top group and study the effects in more detail to decide on the site of action and possible selectivity.

To explore the site and mechanism of action of the COOH compounds and whether there is a common site and mechanism of action, we explored the effect of nine of the compounds with two-electrode voltage clamp on 3R Shaker K_V channels expressed in Xenopus oocytes. Compound #29 at 100 µM increased the current at 0 mV by a factor of 58 (Fig. 6, A and B). This current increase was mainly caused by a shift of the conductance versus voltage, G(V), a curve in the negative direction along the voltage axis, and to a much smaller degree by an increase of the maximum conductance (Fig. 6 B). The shift of the G(V) curve was −45.9 ± 2.1 mV (n = 3). Also, 10 µM of compound #29 had a large effect on the current (Fig. 6 C) and the shift was −14.4 ± 1.5 mV (n = 3). All nine explored compounds’ concentration dependently shifted the G(V) along the voltage axis (Fig. 6 D). To quantify the data in a simple way, we assumed that all compounds shifted the G(V) equally much when fully bound (see Materials and methods). The common maximum shift was −47.0 mV (dotted line in Fig. 6 D). Eight of the compounds had estimated c_½ values in the range of 16–51 µM, while one (#54) had a significantly higher c_½ value of 141 µM.

Because the COOH compounds were identified in the screen with the 3R Shaker K_V channel, our hypothesis was that these compounds act from the lipid bilayer/S4 interface via an electrostatic mechanism to pull the voltage sensor to an activated position. To test this hypothesis, we explored the effect of the compounds on the WT Shaker K_V channel, which lacks the two supplemented arginines at the top of the voltage sensor S4 of the 3R Shaker K_V channel. In previous research on PUFAs and resin-acid derivatives, suggested to electrostatically act on S4, we have found a reduction by a factor of about 2–3 for WT compared with 3R in the concentration range 10–100 µM (Ottosson et al., 2014, 2015, 2017; Silverå Ejneby et al., 2021). 10 µM of compound #29 significantly opened the WT Shaker K_V channel (Fig. 7 A) by shifting the G(V) curve in a negative direction along the voltage axis (Fig. 7 B; −7.7 ± 0.5 mV, n = 3), but the G(V) shift, as well as the increase in maximum conductance, is reduced (Fig. 7 B, compare with Fig. 6 B). For most compounds, the G(V) shift at 10 µM was reduced by a factor of 2–3 (dashed lines in Fig. 7 C). (Note that only eight of the original compounds were studied on the WT channel because the last compound was not available to order for these experiments.) Thus, these data suggest that all studied COOH-group compounds act via an electrostatic mechanism. This contrasts sharply with other chemical motifs, previously published from the screen: the warfarin-like tautomers (Silverå Ejneby et al., 2020) and the biaryl sulfonamides (Liin et al., 2018).

To investigate if the COOH compounds also act on a pharmaceutically more relevant channel, we explored the effects on the human K_V7.2/7.3 channel, which is a validated target for antiseizure drugs (Main et al., 2000; Schenzer et al., 2005). 100 µM of compound #51 clearly increased the current at −60 mV by a factor of about nine (red traces in Fig. 8, A and B). This increase is mainly caused by a shift of the G(V) curve in a negative direction along the voltage axis (Fig. 8 C). In addition, there is a small increase in the maximum conductance (Fig. 8 C). To estimate apparent affinities to the channel, we employed the same strategy as above for the 3R Shaker K_V channel. The common maximum shift was −17.2 mV (Fig. 8 D). Two of the compounds had c_½ values in the range of 30–50 µM (#29 and #51), while the rest had c_½ values >100 µM.

How does the effect on the K_V7.2/7.3 channel compare with the effect on the WT Shaker K_V channel? Four of the compounds (#13, #42, #51, and #54) had almost the same effect at 10 µM on the K_V7.2/7.3 channel and the WT Shaker K_V channel (Fig. 9, A, B, and G), one compound (#29) had half the effect on K_V7.2/7.3 compared with the WT Shaker K_V channel (Fig. 9, C and G), while two compounds (#7 and #27) had almost no effect on the K_V7.2/7.3 channel despite having significant effects on the WT Shaker K_V channel (Fig. 9, E, B, and G). In particular, compound #7 is interesting; it was the most potent compound in the screen of 9,837 compounds (excluding the warfarin-like tautomers) and it was the most potent COOH compound (at 10 µM) in the studies on both the WT Shaker K_V channel and the 3R Shaker channel expressed in oocytes (Fig. 3 F). However, the compound had no effect on the K_V7.2/7.3 channel (at 10 µM, Fig. 8 D). This suggests that it should be possible to reach selectivity between these two channels by modulating the structure of the compounds. However, based on the available data in this study, we think it is premature to speculate about which molecular elements are critical to reach selectivity.

The data on the Shaker K_V channel suggest that compound-induced channel activation is caused by an electrostatic effect. To test if this also is the reason for the effect on the WT K_V7.2/7.3 channel, we performed two types of experiments: (1) an alteration of the charge of the compound and (2) an alteration of the charge of the channel.

First, we altered the charge of the compound by altering the pH in the extracellular solution. Although the COOH compounds have theoretical pK_a values around 4 (Table 1) and should be almost fully charged in a water solution at neutral pH, the proposed site of action in the low dielectric environment in the lipid bilayer pushes the functional pK_a value up to around 7 (Börjesson et al., 2008; Rooney et al., 1983). The G(V) shift induced by 100 µM of compound #51 changed from −19.1 ± 2.6 mV (n = 4) at pH 8.0 to +5.5 ± 1.1 mV (n = 4) at pH 5.5; the apparent pK_a value was 6.8 ± 0.2, the maximum (asymptotic) negative shift, for a fully charged compound at high pH values was −20.2 ± 1.7 mV, and the minimum (asymptotic) shift and for a non-charged compound at low pH values it was +6.7 ± 2.6 mV (Fig. 10 A). This pH dependence supports the hypothesis that the effect is electrostatic. The small shift in positive direction along the voltage axis at low pH suggests that the neutral form of the compound has a slightly higher affinity for the closed, non-activated, channel than for the open, activated, channel. We failed to record reliable currents at pH 9 with 100 µM of compound #51 (the oocytes did not tolerate this treatment), but for 10 µM of compound #51, we succeeded in obtaining data for the complete pH range from 5.5 to 9. The pK_a value was not changed (7.0 ± 0.7 mV) but the (total) amplitude was reduced by 50% (from 26.9 to 13.7 mV; Fig. 10 A). To control for possible compound precipitation at low pH values, we prepared a solution of 100 µM of compound #51 at pH 5.5, which was centrifuged at 16 g for 1 h. The supernatant was extracted and brought to pH 8.0 using 10 M NaOH, after which experiments were conducted according to the standard protocol. The resulting G(V) shifts were −16.6 ± 0.9 mV (n = 5), thus similar to the results from the standard experiments at pH 8.0, suggesting that compound precipitation is not the reason for the reduction of the G(V) shift at low pH values.

Second, we altered the charge of the voltage sensor by neutralizing the top charge of S4 in K_V7.2 (R198Q) and expressed it as a homotetramer (Ottosson et al., 2021). The induced G(V) shift at 100 µM of compound #51 was halved compared with the WT K_V7.2 homotetramer, from −12.9 ± 1.9 (n = 4) to −6.6 ± 0.6 mV (n = 3) (Fig. 10, B and C), suggesting that at least part of the G(V)-shifting effect of compound #51 was caused by electrostatic interaction with R1 in the voltage-sensing domain. There was no difference in effect between K_V7.2/7.3 and K_V7.2 (Fig. 10 C). It is possible that neutralization of the two most extracellularly located gating charges of S4 abolishes channel opening, but this double mutated channel did not express.

In this study, we have shown that a lipophilic compound (4 < LogP < 7) with a COOH group is a common motif among strong activators of the 3R Shaker K_V channel. The mechanism of action is possibly common among the compounds, and the data are consistent with a mechanism where the lipophilic compounds are located at the interface between the lipid bilayer and the voltage sensor (Site 1 in Fig. 1 A). From this position, the compound can electrostatically activate the voltage sensor to an up-state to pull the gate open and thereby open the ion-conducting pore.

In our screen, we only identified COOH compounds activating the channel via the lipoelectric mechanism. It is possible that other channel-activating COOH compounds were missing because the screen was designed to find lipoelectric compounds. However, what speaks against this is that on the screen we in fact identified other types of compounds acting via other sites and mechanisms (Liin et al., 2018; Silverå Ejneby et al., 2020). Thus, we suggest that COOH compounds in general act via the lipoelectric site and mechanism and not via other sites and mechanisms.

While this site and mechanism of action on the Shaker K_V and K_V7.2/7.3 channels clearly overlap with other COOH compounds, such as PUFAs and resin acids (Ottosson et al., 2017; Silverå Ejneby et al., 2021; Yazdi et al., 2021; Yazdi et al., 2016), it is distinct from the sites and mechanisms of action of other channel-activating compounds such as biaryl sulfonamides (Na_V 1.7, [Ahuja et al., 2015]; Shaker K_V, [Liin et al., 2018]; site 2 in Fig. 1 A), warfarin-like tautomers (Shaker K_V and K_V1.5, [Silverå Ejneby et al., 2020]; site 3 in Fig. 1 A), and the antiseizure drug retigabine (K_V7.2–7.5, [Main et al., 2000]; site 4 in Fig. 1 A). Thus, we propose, as a general mechanism, that COOH-containing compounds in most cases activate K_V channel via the lipoelectric mechanism. But, this does not mean that all lipophilic compounds with a COOH group activate all K_V channels. As is evident from the screen, most lipophilic COOH compounds do not activate the 3R Shaker K_V channel (Fig. 3 B), and some channel-activating compounds were selective for specific channels (e.g., Fig. 7 E). Thus, we suggest that refined selectivity between different types of channels will be possible to develop for these lipoelectric compounds.

The data are available from the corresponding author upon reasonable request.

Christopher J. Lingle served as editor.

We thank Per-Eric Lund for running the high-throughput experiments, Gunnar Nordvall for chemistry advice and compound selection, Anders B. Eriksson (Science for Life Laboratory) for help with compound handling, Andreas Nolting for construction of the cell line, Ulrika Yngve (at the Drug Discovery and Development platform at Science for Life Laboratory) for identification of compounds in the screen, and Johan Brask, Damon Frampton, Nina Ottosson, and Hani Abdelrahman for aid, discussions, and insights during the experimental and analytical process. We thank the Chemical Biology Consortium Sweden (CBCS) at the Science for Life Laboratory (Solna, Sweden) for providing the chemical libraries tested on the screen. The NIH Clinical Collection was provided through the National Institutes of Health Molecular Libraries Roadmap Initiative. The LCBKI primary screening set was provided by CBCS. Part of this work was assisted by Karolinska High Throughput Center (KHTC), a core facility at Karolinska Institutet with affiliation to Science for Life Laboratory (https://www.scilifelab.se/facilities/khtc/).

This work was supported by grants from the Swedish Research Council (2020-01019), the Swedish Heart-Lung Foundation (20210596), the Swedish Brain Foundation (2022-0219), and the County of Östergötland.

Author contributions: O. Rönnelid: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Visualization, Writing—original draft, Writing—review and editing, F. Elinder: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing—original draft, Writing—review and editing.