Subtype-specific responses of hKv7.4 and hKv7.5 channels to polyunsaturated fatty acids reveal an unconventional modulatory site and mechanism

The KV7.4 and KV7.5 subtypes of voltage-gated potassium channels play a role in important physiological processes such as sound amplification in the cochlea and adjusting vascular smooth muscle tone. Therefore, the mechanisms that regulate KV7.4 and KV7.5 channel function are of interest. Here, we study the effect of polyunsaturated fatty acids (PUFAs) on human KV7.4 and KV7.5 channels expressed in Xenopus oocytes. We report that PUFAs facilitate activation of hKV7.5 by shifting the V50 of the conductance versus voltage (G(V)) curve toward more negative voltages. This response depends on the head group charge, as an uncharged PUFA analogue has no effect and a positively charged PUFA analogue induces positive V50 shifts. In contrast, PUFAs inhibit activation of hKV7.4 by shifting V50 toward more positive voltages. No effect on V50 of hKV7.4 is observed by an uncharged or a positively charged PUFA analogue. Thus, the hKV7.5 channel’s response to PUFAs is analogous to the one previously observed in hKV7.1–7.3 channels, whereas the hKV7.4 channel response is opposite, revealing subtype-specific responses to PUFAs. We identify a unique inner PUFA interaction site in the voltage-sensing domain of hKV7.4 underlying the PUFA response, revealing an unconventional mechanism of modulation of hKV7.4 by PUFAs.

Introduction K V 7 voltage-gated potassium channels are expressed in several tissues, where they serve to attenuate excitability by conducting an outward K + current. The five members of the family, K V 7.1-K V 7.5 (encoded by KCNQ1-KCNQ5 genes), are important for human physiology, which is often emphasized in diseased states caused by dysfunctional channels. For instance, K V 7.1 is predominantly expressed in cardiomyocytes, and mutations to this channel are a known risk factor for developing cardiac arrhythmia (Barhanin et al., 1996;Brewer et al., 2020;Sanguinetti et al., 1996;Tester and Ackerman, 2014). K V 7.2 and K V 7.3 are broadly expressed in neurons, where they form heterotetrameric K V 7.2/7.3 channels, and mutations to these channels may give rise to epilepsy or chronic pain (Biervert et al., 1998;Sanders, 2017). This PUFA-induced facilitation of activation is mediated through a lipoelectric mechanism in which the PUFA tail inserts into the outer leaflet of the lipid bilayer adjacent to the channel, whereupon the negatively charged carboxyl head group of the PUFA interacts electrostatically with positively charged arginines in the upper half of the voltage-sensing domain (VSD) of the channel Yazdi et al., 2021). This electrostatic interaction facilitates the outward movement of the S4 helix, causing a shifted voltage dependence of channel opening toward more negative voltages. However, the effect of PUFAs on K V 7.4 and K V 7.5 remains unstudied. In this study, we therefore aimed to characterize the response of human K V 7.4 and K V 7.5 channels (henceforth referred to as hK V 7.5 or hK V 7.4) to PUFAs, in order to expand our understanding of how the K V 7 family of channels responds to these lipids.
We report that PUFAs facilitate activation of hK V 7.5 by shifting the voltage dependence of channel opening toward more negative voltages. Surprisingly, we find that PUFAs inhibit activation of the hK V 7.4 channel by shifting the voltage dependence of channel opening toward more positive voltages. Thus, the hK V 7.5 channel's response to PUFAs is largely in line with the responses that have previously been observed in hK V 7.1 and hK V 7.2/7.3, whereas the hK V 7.4 channel response is not. Providing a mechanistic explanation for this observation, we identify an unconventional inner PUFA site in the VSD of hK V 7.4 underlying PUFA-induced inhibition of the activation of hK V 7.4. Our study expands our understanding of how members of the hK V 7 family respond to PUFAs and reveal subtype specific responses and sites to these lipids.

Results
The PUFA docosahexaenoic acid facilitates activation of hK V 7.5, but inhibits activation of hK V 7.4 We began with investigating the effects of the physiologically abundant (Kim et al., 2014) PUFA DHA (molecular structure shown in Figure 1A) on homotetrameric hK V 7.5 or hK V 7.4 channels expressed in Xenopus oocytes. Activation of the hK V 7.5 channel was facilitated by 70 μM DHA, as was evident by the significant shift in the midpoint of the voltage dependence of channel opening (V 50 ) toward more negative voltages ( Figure 1B, average shift of -21.5 ± 1.9 mV, p = <0.0001). This allows hK V 7.5 channels to open and conduct a K + current at more negative voltages in the presence of DHA. 70 μM DHA did not cause a consistent change in the maximum conductance (G max ) of the hK V 7.5 channel (average relative ΔG max was 1.06 ± 0.11, p = 0.59).
In clear contrast to the facilitated activation observed for hK V 7.5, the activation of the hK V 7.4 channel was inhibited by 70 μM DHA, as was seen by the significant shift in the V 50 towards more positive voltages ( Figure 1C, average shift of +11.8 ± 1.4 mV, p = <0.0001). Furthermore, the application of DHA led to a more shallow slope of the G(V) curve (average slope factors were 13.1 ± 0.2 mV and 17.6 ± 0.5 mV in the absence and presence of 70 μM DHA, respectively). 70 μM DHA did not cause a consistent change in G max of the hK V 7.4 channel (average relative ΔG max was 1.09 ± 0.08, p = 0.28). Because 70 μM DHA induced significant shifts in V 50 of both hK V 7.5 and hK V 7.4, although in opposite directions, without affecting G max we will throughout the remainder of this study focus our analysis of PUFA effects on V 50 , and G(V) curves shown in figures reporting on PUFA effects will be normalized to visually emphasize V 50 shifts.
The DHA-evoked shift in the V 50 of hK V 7.5 was significant at concentrations as low as 7 μM ( Figure 1D, ΔV 50 = -8.5 ± 1.3 mV, p = 0.0002). The concentration-response curve for the DHA effect on hK V 7.5 predicts a maximum V 50 shift of -25.3 mV, with 12 μM required for 50% of the maximum shift (EC 50 = 12 μM). The DHA effect on the V 50 of hK V 7.4 was also significant at 7 μM ( Figure 1E, ΔV 50 = +4.8 ± 1.6 mV, p = 0.01). The concentration-response curve for the DHA effect on hK V 7.4 predicts a maximum V 50 shift of +13.8 mV, with an EC 50 of 14 μM. The onset of the DHA effect was relatively fast for both the hK V 7.5 and hK V 7.4 channels, reaching a stable level within about 6 min of application ( Figure 1-figure supplement 1). The DHA effect proved difficult to wash out or reverse, as re-perfusion with control solution or control solution supplemented with 100 mg/mL bovine serum albumin (BSA) only partially restored baseline current amplitude for hK V 7.5 and hK V 7.4 ( Figure 1figure supplement 1). Altogether, the PUFA DHA induces a concentration-dependent facilitation of activation of the hK V 7.5 channel, and a concentration-dependent inhibition of activation of the hK V 7.4 channel. (B) Representative current family with corresponding G(V) curve of hK V 7.5 in the absence (left) and presence (middle) of 70 µM DHA. Currents generated by the voltage protocol shown as inset. Red traces denote current generated by a test voltage to -40 mV. The G(V) curves (right) have been normalized between 0 and 1, as described in Materials and methods, to better visualize shifts in V 50 . Curves represent Boltzmann fits (see Materials and methods for details). V 50 for this specific cell: V 50,ctrl = -41.7 mV, V 50,DHA = -59 mV. (C) Same as in B but for hK V 7.4. Blue traces denote current generated by a test voltage to 0 mV. V 50 for this specific cell: V 50,ctrl = -12.6 mV, V 50,DHA = -2.0 mV. (D-E) Concentration-response curve of the DHA effect on V 50 of hK V 7.5 (D) and hK V 7.4 (E). Curves represent concentration-response fits (see Materials and methods for details). Best fits: ΔV 50,max is -25.3 mV for hK V 7.5

Figure 1 continued on next page
The hK V 7.5 response, but not the hK V 7.4 response, changes direction in an electrostatic manner To further understand the molecular basis of the DHA response, we examined the importance of the charge of the DHA head group. Several previous studies identify electrostatic interactions between positively charged arginines in the upper half of S4 of the VSD and the negatively charged head group on PUFAs (and their analogues) as fundamental for facilitating the activation of hK V 7.1, hK V 7.2/7.3 and some other K V channels (Börjesson and Elinder, 2011;Liin et al., 2015;Liin et al., 2016;Liin et al., 2018;Martín et al., 2021). This is seen as a shift in V 50 toward more negative voltages. The electrostatic PUFA effect on V 50 can be tuned, from inducing negative to positive shifts in V 50 , by altering the charge of the PUFA head group (Börjesson et al., 2010;Liin et al., 2015). To investigate if the  Figure 1B), the lack of effect of DHA-me (middle, V 50,ctrl = -43.9 mV; V 50,DHA-me = -44.6 mV), and the positive shift in V 50 induced by DHA+ (right, V 50,ctrl = -42.0 mV; V 50,DHA+ = -37.0 mV). Molecular structure of head groups are shown as insets. (B) Summary of responses to indicated DHA compound. Data shown as mean ± SEM. n = 6-12. Statistics denote one sample t test against a hypothetical mean of 0 mV. ns denotes not significant, ** denotes p ≤ 0.01, **** denotes p ≤ 0.0001. (C-D) same as in A-B but for hK V 7.4. (C) Representative examples of the depolarizing shift in V 50 induced by DHA (left, same data as shown in Figure 1C), the lack of effect of DHA-me (middle, V 50,ctrl = -14.6 mV; V 50,DHA-me = -11.1 mV), and the lack of effect of DHA+ (right, V 50,ctrl = -9.1 mV; V 50,DHA+ = -6.6 mV). (D) Summary of responses to indicated DHA compound. Data shown as mean ± SEM. n = 8-15. Statistics denote one sample t test against a hypothetical mean of 0 mV. ns denotes not significant, **** denotes p ≤ 0.0001. See also Figure 2-source data 1.
The online version of this article includes the following source data for figure 2: Source data 1. Numerical data for Figure 2. and +13.8 mV for hK V 7.4. EC 50 is 12 µM for hK V 7.5 and 14 µM for hK V 7.4. Data shown as mean ± SEM. n = 3-15. See also Figure 1-figure supplement 1, Figure 1-source data 1.
The online version of this article includes the following source data and figure supplement(s) for figure 1: Source data 1. Numerical data for Figure 1.
Figure supplement 1. The DHA response has a rapid onset for both hK V 7.5 and hK V 7.4.

Figure 1 continued
same electrostatic mechanism is at play in the responses of hK V 7.5 and hK V 7.4, we compared the DHA response of the channels with: (1) a DHA analogue with an uncharged methyl ester head group (DHAme), and (2) a DHA analogue with a positively charged amine head group (DHA+). 70 μM of DHA-me did not significantly shift V 50 of hK V 7.5 ( Figure 2A, ΔV 50 = -0.2 ± 0.5 mV, p = 0.67). In contrast, 70 μM DHA+ brought on a small, but significant positive shift in V 50 of hK V 7.5 ( Figure 2A, ΔV 50 = +4.5 ± 0.9 mV, p = 0.001). Thus, for hK V 7.5 the negatively charged DHA facilitates activation, the uncharged DHA-me has no effect, and the positively charged DHA+ inhibits channel activation (effects summarized in bar graph of Figure 2B). This is in line with the lipoelectric mechanism that has been proposed to explain PUFA effects on the hK V 7.1 and hK V 7.2/7.3 channels (Liin et al., 2015;Liin et al., 2016).
Neither 70 μM of DHA-me nor 70 μM of DHA+ had significant effects on V 50 of hK V 7.4 ( Figure 2C, ΔV 50 = +1.1 ± 1.3 mV, p = 0.42 for DHA-me; ΔV 50 = +2.1 ± 1.0 mV, p = 0.068 for DHA+). Thus, while the negatively charged DHA caused a positive shift in V 50 of hK V 7.4, neither the uncharged DHA-me nor the positively charged DHA+ altered the V 50 of hK V 7.4 (effects summarized in bar graph of Figure 2D). Even though a negative charge of the head group seems to be important for the effect, the responses are not in line with the lipoelectric mechanism, making hK V 7.4 unique among the hK V 7 channels. Both hK V 7.5 and hK V 7.4 respond broadly to PUFAs, although with varied magnitudes of responses Next, we studied if the effects on hK V 7.5 and hK V 7.4 are specific to DHA or if they are shared among different PUFAs by testing a series of PUFAs, listed in Table 1. These PUFAs vary in their molecular properties, such as the length of the tail and the number and position of double bonds in the tail. Figure 3A shows a plot of the magnitude of ΔV 50 following exposure of hK V 7.5 to PUFAs (at 70 μM) against the number of carbon atoms in each respective PUFA tail. All PUFAs, regardless of length (14-22 carbon atoms) significantly shifted V 50 of hK V 7.5 toward more negative voltages, although to different extents. The magnitude of the PUFA-induced ΔV 50 of hK V 7.5 followed a pattern of TTA < LA = AA < HTA ≤ EPA < DHA. One observation is that the ω-6 PUFAs LA and AA induced smaller shifts than the ω-3 PUFAs HTA, EPA and DHA. Both AA and EPA are 20 carbon atoms long. However, while AA is an ω-6 PUFA, EPA is an ω-3 PUFA ( Table 1). Figure 3B shows representative G(V) curves for hK V 7.5 that highlight the larger shift in V 50 evoked by 70 μM of EPA (left) than by 70 μM of AA (right). On average, EPA induced a shift of -17.9 ± 2.4 mV, which was significantly larger than the shift induced by AA (ΔV 50 = -9.3 ± 1.7 mV; Figure 3C; p = 0.015). These results suggest that in PUFAs of equal length, the ω-number has an impact on the magnitude of the PUFA response of hK V 7.5. We also compared the positively charged amine analogue DHA+ to an amine analogue of AA (AA+). The positive shift in V 50 of hK V 7.5 by 70 μM AA+ was slightly smaller than that observed with DHA+, and did not differ significantly from a hypothetical shift of 0 mV ( Figure 3C, AA+ ΔV 50 = +1.9 ± 1.4 mV, p = 0.2). However, there was also no statistically significant difference between the effects induced by DHA+ and AA+ (p = 0.13), which indicates that the importance of the ω-number for the amine The online version of this article includes the following source data for figure 3: Source data 1. Numerical data for Figure 3.
analogues should be interpreted with caution. Altogether, these experiments show that many PUFAs activate the hK V 7.5 channel and suggest that the hK V 7.5 channel shows a preference toward ω-3 over ω-6 PUFAs. Figure 3D shows a plot of the magnitude of ΔV 50 following exposure to PUFAs (at 70 μM) against the number of carbon atoms in each respective PUFA tail for the hK V 7.4 channel. The shortest PUFA, TTA, did not evoke a significant shift in V 50 (ΔV 50 = +3.1 ± 1.3 mV, p = 0.051). All remaining PUFAs (at a concentration of 70 μM), however, significantly shifted V 50 of hK V 7.4 toward positive voltages. The magnitude of the PUFA-induced ΔV 50 of hK V 7.4 followed a pattern of TTA <AA < LA ≤ HTA < DHA < EPA with no clear pattern in magnitude of effect based on the ω-number of the PUFA tail. Although the ω-3 PUFA EPA induced a larger shift than the ω-6 PUFA AA ( Figure 3E, F, EPA ΔV 50 = +13.9 ± 1.9 mV; AA ΔV 50 = +3.1 ± 1.0 mV, p = <0.001), AA+ caused a larger shift in V 50 compared to DHA+ ( Figure 3F, DHA+ ΔV 50 = +2.1 ± 1.0 mV; AA+ ΔV 50 = +6.3 ± 1.8 mV, p = 0.049). Moreover, the ω-6 PUFA LA induced comparable shifts in V 50 to those of the ω-3 PUFA HTA. Altogether, these experiments show that many PUFAs inhibit the hK V 7.4 channel, with no obvious pattern of which PUFAs hK V 7.4 shows a preference toward. hK V 7.4/7.5 co-expression and disease-associated hK V 7.4 mutations, but not hKCNE4 co-expression, influence the response to DHA We next characterized the DHA response of oocytes co-injected with cRNAs encoding hK V 7.4 and hK V 7.5 (referred to as hK V 7.4/7.5) to allow for the potential formation of heteromeric channels containing hK V 7.4 and hK V 7.5 subunits. Oocytes co-injected with hK V 7.4 and hK V 7.5 generated currents with biophysical properties that fell between those of each homomeric channel complex (hK V 7.4 V 50 = -11.1 ± 1.4 mV; hK V 7.5 V 50 = -42.2 ± 0.7 mV; hK V 7.4/7.5 V 50 = −22.9 ± 0.6 mV, Table 2). This is in agreement with previous studies showing a V 50 of co-expressed hK V 7.4/7.5 channels intermediate to that of homomeric channels (Brueggemann et al., 2011). In addition, 70 μM of DHA induced an effect on hK V 7.4/7.5 intermediate to that of homomeric hK V 7.4 and hK V 7.5 channels, with no change in V 50 ( Figure 4A, C, ΔV 50 = -0.4 ± 0.6 mV, p = 0.53). K V 7.4 has been shown to co-assemble with the KCNE4 subunit in, for instance, vascular smooth muscle tissue (Jepps et al., 2015). We therefore characterized the DHA response of oocytes co-injected with cRNAs encoding hK V 7.4 and hKCNE4 (referred to as hK V 7.4/KCNE4). In agreement with previous studies (Vanoye et al., 2009;Jepps et al., 2015), this generated currents with biophysical properties fairly similar to those of hK V 7.4 alone ( Table 2). 70 μM of DHA shifted V 50 of hK V 7.4/KCNE4 toward more positive voltages ( Figure 4B-C, ΔV 50 = +8.9 ± 1.3 mV, p = 0.0005). This effect was  comparable to that on hK V 7.4 alone (p = 0.23), suggesting the hKCNE4 subunit does not alter the DHA response of the hK V 7.4 channel. Several mutations in the KCNQ4 gene (encoding hK V 7.4) have been identified in humans and are often linked to DFNA2 non-syndromic hearing loss (Jung et al., 2019;Rim et al., 2021), although the possible contribution to pathology remains to be determined for many mutations. Notably, the F182L and S273A missense mutations in hK V 7.4 ( Figure 4D-E), which are suspected to be linked to impaired hearing (Jung et al., 2019;Kim et al., 2011), exchanges the native hK V 7.4 residue for the hK V 7.5 counterpart. By means of site-directed mutagenesis, we substituted these residues in hK V 7.4, one by one. Under control conditions, we found both mutants to behave fairly similar to wild-type hK V 7.4 ( Table 2). The S273A mutant had a V 50 comparable to wild-type hK V 7.4, whereas the V 50 of the F182L mutant was shifted about 10 mV toward more negative voltages compared to wild-type hK V 7.4 ( Table 2). However, both mutations impaired the inhibitory hK V 7.4 response to DHA. 70 µM of traces denote current generated by a test voltage to 0 mV. Curves represent Boltzmann fits. V 50 for this specific cell: V 50,ctrl = -10.7 mV; V 50,DHA = -3.5 mV. (C) Summary of response of hK V 7.4/7.5 and hK V 7.4/KCNE4 to 70 µM DHA, with responses of hK V 7.4 and hK V 7.5 for reference. Data shown as mean ± SEM. n = 7-15. Statistics denote one sample t test against a hypothetical mean of 0 mV. ns denotes not significant, *** denotes p ≤ 0.001, **** denotes p ≤ 0.0001. (D-E) Impact of hK V 7.4 mutations F182L (D) and S273A (E) on the response to DHA. Structural model (PDB ID -7BYL, Li et al., 2021) of hK V 7.4 with position of F182 and S273 marked. Representative G(V) curve of indicated hK V 7.4 mutants in the absence and presence of 70 µM DHA. Curves represent Boltzmann fits. For these specific cells: F182L: V 50,ctrl = -18.7 mV; V 50,DHA = -12.4 mV. For S273A: V 50,ctrl = -15.8 mV; V 50,DHA = -10.8 mV.
(F) Summary of response of WT hK V 7.4 and indicated mutants to 70 µM DHA. Data shown as mean ± SEM. n = 5-15. Statistics denote one-way ANOVA followed by Dunnett's multiple comparisons test to compare the response of mutants to that of the wild-type. * denotes p ≤ 0.05. See also Figure 4source data 1.
The online version of this article includes the following source data for figure 4: Source data 1. Numerical data for Figure 4.  Liin et al., 2016;Liin et al., 2015. Data for hK V 7.4 and hK V 7.5 from the present study. Data shown as mean ± SEM. n = 3-15. Statistics denote one sample t test against a hypothetical mean of 0 mV. * denotes p ≤ 0.05, ** denotes p ≤ 0.01, **** denotes p ≤ 0.0001. (B-C) Altering extracellular pH tunes the shift in V 50 of hK V 7.5 induced by 70 µM DHA, with a greater response observed at more alkaline pH. Note that in all experiments, the pH of the control solution and DHA-supplemented control solution was identical. (B) Representative example of the DHA response at pH 8.2 (V 50,ctrl = -47.6 mV; V 50,DHA = -92.4 mV). (C) pH-response curve for the DHA effect on V 50 . Data shown as mean ± SEM. n = 3-9. Curves represent pH-response fits (see Materials and methods for details). Best fit for hK V 7.5: ΔV 50,max is -57.9 mV. Data for hK V 7.1 as reported in Liin et al., 2015 is included for comparison (Best fit for hK V 7.1: ΔV 50,max is -26.3 mV.) Dashed line denotes the pH-response curve for hK V 7.5 normalized to ΔV 50,max for hK V 7.1 to illustrate the comparable apparent pKa. Inset shows apparent pKa with 95% CI: hK V 7.1 apparent pKa = 7.68 with 95% CI [7.26, 7.89]; hK V 7.5 apparent pKa = 7.67 with 95% CI [7.52, 7.90]. Note that inconsistent behavior of hK V 7.5 under control conditions at pH higher than 8.2 prevented us from determining the DHA effect on hK V 7.5 at pH 9 and 10. See also  DHA shifted V 50 of the F182L mutant by only +5.5 ± 2.2 mV and the S273A mutant by +4.7 ± 0.8 mV ( Figure 4D-F).
Altogether, these experiments suggest that co-expression of hK V 7.4 and hK V 7.5 subunits, or substitution of specific residues in hK V 7.4 to the hK V 7.5 counterpart, alter the DHA response of hK V 7.4 to approach that of hK V 7.5. However, hK V 7.4 co-expression with the hKCNE4 auxiliary subunit did not alter the DHA response.
The DHA response of hK V 7.5 is greater than that of other hK V 7 subtypes We have in previous studies shown that 70 μM of DHA shifts the V 50 of the hK V 7.1 and hK V 7.2/7.3 channels by about -9 mV ( Figure 5A; hK V 7.1 ΔV 50 = -9.3 ± 0.9 mV, as reported in Liin et al., 2015; hK V 7.2/7.3 ΔV 50 = -9.3 ± 1.7 mV, as reported in Liin et al., 2016). Thus, the extent of the hK V 7.5 shift induced by 70 μM DHA is greater than that of hK V 7.1 or hK V 7.2/7.3 ( Figure 5A). Our experiments indicated the necessity of a negatively charged DHA head group to elicit the activating effect on hK V 7.5 (see Figure 2A). A negatively charged head group is promoted by alkaline pH, which triggers proton dissociation (Börjesson et al., 2008;Hamilton, 1998). In a previous study, the apparent pKa of DHA when near hK V 7.1 (i.e. the pH at which 50% of the maximal DHA effect is seen, interpreted as the pH at which 50% of the DHA molecules in a lipid environment are negatively charged) was determined to be pH 7.7 (Liin et al., 2015). One possible underlying cause of the larger DHA effect on hK V 7.5 is that the local pH environment at hK V 7.5 promotes DHA deprotonation (i.e. inducing a lower apparent pKa of DHA), thus rendering a greater fraction of the DHA molecules negatively charged and capable of activating the channel. To test this, we assessed the effect of 70 μM DHA on hK V 7.5 with the extracellular pH adjusted to either more alkaline (pH = 8.2) or acidic (pH = 6.5, or 7.0) values. Adjusting the extracellular pH had only minor effects on V 50 under control conditions ( Figure 5-figure supplement  1). In contrast, the magnitude of the DHA-induced shift varied with pH. At pH 8.2, at which a majority of DHA molecules are expected to be deprotonated, DHA shifts of V 50 were almost two-fold greater than at physiological pH (pH 7.4 ΔV 50 = -21.5 ± 1.9 mV; pH 8.2 ΔV 50 = -44.4 ± 3.2 mV, p < 0.0001, Student's t test, representative example in Figure 5B). The magnitude of the DHA effect was reduced as the pH was gradually titrated toward more acidic pH ( Figure 5C; Figure 5-figure supplement 1), with no shift in V 50 by 70 μM DHA observed at pH 6.5 (ΔV 50 = +0.1 ± 1.0 mV, p = 0.96). The pH dependence of ΔV 50 induced by DHA for the hK V 7.1 channel is also plotted in Figure 5C, to allow for comparison between the two hK V 7 subtypes. While there is a clear difference in the extrapolated maximum shifts for the two channels (hK V 7.1 ΔV 50,max = -26.3 mV with 95% CI [-18.4,-34.3]; hK V 7.5 ΔV 50,max = -57.9 mV with 95% CI [-47.6,-68.2]), the pH required to induce 50% of the maximum ΔV 50 (i.e. the apparent pKa values) are similar (inset of Figure 5C, hK V 7.1 apparent pKa = 7.68 with 95% CI [7.26, 7.89]; hK V 7.5 apparent pKa = 7.67 with 95% CI [7.52, 7.90]). This indicates that the difference in the DHA effect between hK V 7.1 and hK V 7.5 is a question of increased magnitude, rather than a matter of different apparent pKa of DHA at the channels.
Molecular dynamics simulations predict an unconventional inner VSD PUFA site in hK V 7.4 We next turned our attention toward structural components of the hK V 7.4 channel that may contribute to its unusual PUFA response. In order to determine the interaction sites for DHA on hK V 7.4, we performed coarse-grained molecular dynamics simulations of the channel, solved in an intermediate activation state, embedded in a membrane consisting of 75% POPC and 25% negatively charged DHA randomly distributed in both membrane leaflets ( Figure 6A). Analysis of the residence time of DHA showed overall prolonged interactions with residues on S1 (L91, Y101, and I105), on S2 (V142, F143, and D146), on S4 (R213) and at the top of the pore region ( Figure 6B and Figure 6-figure supplement 1A). Out of the 13 interaction sites characterized by density-based clustering ( Figure 6-figure  supplement 1B), the site at which DHA interacted with the channel the longest (2.156 µs, cluster #3) was comprised of residues located on the lower half of S4. This site was also occupied ~87% of the time ( Figure 6C, Figure 6-figure supplement 1B), suggesting favorable interactions between the side chains of R213, R216, and R219 and the head group of DHA, as well as interactions with the backbone of M217. In the representative binding pose of that site, the head group of DHA is bound in the lumen of the lower portion of the VSD, close to the S4 helix ( Figure 6D). We will refer to this as an unconventional 'inner VSD site', in contrast to the conventional 'outer VSD site' described previously for K V 7.1 (Yazdi et al., 2021). Because of its negative charge, DHA in the inner VSD site presumably prevents the outward motion of the positively charged S4 helix, rationalizing its inhibitory effect. The wedging of the tail between S1 and S2 suggests that the DHA head group accesses this interaction    site via this lipid-exposed interface. Repeating these simulations with uncharged DHA changes the interaction pattern drastically. Indeed, neutral DHA shows a high propensity to localize close to the lower half of the S5 and S6 helix, at a considerable distance from the VSD (Figure 6-figure supplement 2). This site presumably has little functional effect, providing an explanation for the lack of effect of the uncharged DHA-me on hK V 7.4 activation.
Arginine residues in the lower half of S4 of hK V 7.4 contribute to the PUFA response The molecular dynamics simulations were performed using a hK V 7.4 model based on the cryo-EM structure of hK V 7.4 (PDB ID -7BYL), in which S4 is presumably in an intermediate state (Li et al., 2021). S4 is anticipated to move further downward when adopting a resting state and further upward when adopting an activated state. The lack of hK V 7.4 structure with S4 in a resting state prevents us from using simulations to assess DHA interaction with the inner VSD site in this conformational state. Therefore, to functionally assess the importance of the interactions predicted by the molecular dynamics simulations, we substituted each of the innermost arginines (R213, R216, and R219), one by one, with an uncharged glutamine. This generated mutant hK V 7.4 channels with V 50 values shifted toward more positive (R213Q and R219Q) or negative (R216Q) voltages relative to the wild-type channel ( Table 2). However, in contrast to the positive shift in V 50 normally seen in wild-type hK V 7.4 by DHA, each of these arginine mutants responded to 70 μM DHA with a negative shift in V 50 ( Figure 7A-D, R213Q ΔV 50 = -8.8 ± 0.9 mV; R216Q ΔV 50 = -5.7 ± 0.6 mV; R219Q ΔV 50 = -7.3 ± 0.6 mV). Thus, neutralization of any of these three arginines in the lower half of S4 endowed the hK V 7.4 channel with a DHA response typical of other hK V 7 channels (that is, PUFA-mediated facilitation of channel activation), confirming the interaction site predictions from the molecular dynamics simulations. With this typical DHA response, the arginine mutants showed a response more in line with the lipoelectric mechanism, with a negative shift in V 50 induced by negatively charged PUFAs, no or minimal shift in V 50 induced by uncharged PUFA analogues, and a positive shift in V 50 induced by positively charged PUFA analogues (Figure 7-figure supplement 1). This response pattern was clearest for R213Q, which showed the largest negative shift induced by DHA, for which DHA-me had minimal effects on V 50 (ΔV 50 = +0.3 ± 0.1 mV, p < 0.05) whereas DHA+ and AA+ shifted V 50 toward positive voltages (ΔV 50 = +2.6 ± 0.5 mV, p < 0.005 for DHA+; ΔV 50 = +3.1 ± 0.4 mV, p < 0.001 for AA+). The response pattern showed a similar trend for R216Q and R219Q, however, with less robust effects induced by the positively charged compounds (For R216Q: ΔV 50 = +1.6 ± 0.8 mV, p = 0.11 for DHA+; ΔV 50 = +5.6 ± 2.4 mV, p = 0.06 for AA+; for R219Q: ΔV 50 = +2.2 ± 1.0 mV, p = 0.08 for DHA+; ΔV 50 = +4.7 ± 0.7 mV, p < 0.005 for AA+). The positive shifts induced by AA+ on the arginine mutants were larger than for hK V 7.5 ( Figure 3C) and those previously reported for the Shaker K V channel (Börjesson et al., 2010), comparable to the effect on WT hK V 7.4 ( Figure 3F), and slightly smaller than those previously reported for hK V 7.1 (Liin et al., 2015).

Discussion
This study finds that the activity of hK V 7.4 and hK V 7.5 channels is modulated by PUFAs. As such, our study expands upon the understanding of the modulation of the hK V 7 family of ion channels by this class of lipids. Importantly, we find that both hK V 7.4 and hK V 7.5 show surprising responses to PUFAs, compared to previously reported effects on other hK V 7 subtypes. hK V 7.4 is the only hK V 7 subtype for which PUFAs induce channel inhibition, whereas ω-3 PUFAs induce unexpectedly large hK V 7.5 channel activation. Experiments on co-expressed hK V 7.4/7.5 channels and hK V 7.4 channels carrying hK V 7.5 mimetic disease-associated mutations indicate responses to PUFAs that are intermediate of the hK V 7.4 and hK V 7.5 homomers. In contrast, hKCNE4 co-expression did not alter the PUFA response of hK V 7.4. The diverse responses of hK V 7 subtypes to PUFAs demonstrate the importance of carrying out functional experiments on each of the channel subtypes to allow for an understanding of subtype variability in channel modulation.
What mechanistic basis may underlie PUFA-induced facilitation of hK V 7.5 activation? We find that the pattern of how hK V 7.5 responds to PUFAs conforms to the lipoelectric mechanism that has previously been described for the hK V 7.1 and hK V 7.2/7.3 channels, with a hyperpolarizing shift in V 50 induced by negatively charged PUFAs, a depolarizing shift in V 50 induced by a positively charged PUFA analogue, and no effect of an uncharged PUFA analogue. Therefore, we find it likely that PUFAs also facilitate activation of hK V 7.5 through electrostatic interactions with positively charged gating charges in the upper half of the VSD (schematically illustrated in Figure 7E left). However, ω-6 PUFAs are less effective than ω-3 PUFAs at activating the hK V 7.5 channel. This is different from the hK V 7.1/ KCNE1 channel, for which ω-6 or ω-9 PUFAs were identified as the best modulators of channel activity (Bohannon et al., 2019). Although the role of the ω-numbering in determining the extent of PUFA effects on hK V 7.5 should be interpreted with caution (for instance, because the PUFAs also vary in their number of double bonds, see Table 1), it is clear that hK V 7.5 and hK V 7.1/KCNE1 do not follow the same response pattern to different PUFAs. Work on hK V 7.1/KCNE1 has proposed that the KCNE1 subunit impairs the DHA response of hK V 7.1 by rearranging one of the external loops in hK V 7.1, which alters the pH at the outer VSD site of the channel to promote DHA protonation Liin et al., 2015). This led us to test if the greater DHA response of the hK V 7.5 channel was caused by a lower apparent pKa of DHA at hK V 7.5 compared to hK V 7.1-7.3, which might promote DHA deprotonation. However, the similar apparent pKa values of DHA for hK V 7.1 and hK V 7.5 discards this hypothesis. Instead, we find a larger magnitude of the DHA effect on hK V 7.5 at all pH values compared to hK V 7.1. Interestingly, a residue at the top of S5 of hK V 7.1, Y278, which was recently identified as an important 'anchor point' for binding the ω-6 PUFA LA near the VSD to allow for efficient shifts in V 50 (Yazdi et al., 2021), is instead a phenylalanine (F282) in hK V 7.5. A phenylalanine mutation of this 'anchor point' in hK V 7.1 (Y278F) led to a drastic decrease in apparent affinity of LA for hK V 7.1, possibly by removing hydrogen bonding between the head group of LA and the hydroxyl group of the tyrosine side chain (Yazdi et al., 2021). However, because the Y278F mutation of hK V 7.1 also reduced the apparent affinity of DHA for hK V 7.1 (Yazdi et al., 2021), and DHA was the most potent of the PUFAs we tested on hK V 7.5 in this study, sequence variability at this specific position is not likely to underlie the relatively larger effect of DHA on hK V 7.5 compared to hK V 7.1. Thus, we conclude the nature of PUFA binding to hK V 7 channels is more complex and there may be other stabilizing residues in the PUFA interaction site of the hK V 7.5 channel.
We find that the activity of the hK V 7.4 channel is inhibited by PUFAs, and that the direction of the effect is not reversed by reversing the charge of the PUFA head group. Therefore, the PUFA effect on hK V 7.4 does not conform to the lipoelectric mechanism proposed for other K V 7 channels. Additionally, we did not find any relationship between PUFA tail properties and the response of hK V 7.4 to PUFAs, other than that all PUFAs with significant effects shifted V 50 toward more positive voltages. A cryo-EM structure for hK V 7.4 was recently reported (PDB ID -7BYL; Li et al., 2021), which reveals structural differences when comparing to the hK V 7.1 structure (PDB ID -6UZZ; Sun and MacKinnon, 2020). For instance, the entire VSD of hK V 7.4 is rotated 15° clockwise relative to the pore domain of the channel. In molecular dynamics simulations, we also identified an unconventional predicted DHA site in hK V 7.4. Although the lack of hK V 7.4 structures in different conformational states limits the ability to simulate the state dependence of DHA interaction with different sites, our experiments validated the functional relevance of the unconventional predicted site. Taken together, our simulation and experimental data suggest that hK V 7.4 is unique among hK V 7 channels in harboring a functional PUFA site in the inner half of the VSD. The DHA-induced shift in V 50 toward positive voltages in wild-type hK V 7.4 is compatible with DHA stabilizing a resting or intermediate state of S4 through interaction = -35.3 mV. (C) Same as in A but for hK V 7.4 R219Q. Blue traces denote current generated by a test voltage to -10 mV. V 50 for this specific cell: V 50,ctrl = 8.6 mV; V 50,DHA = +0.6 mV. (D) Summary of the response of hK V 7.4 gating charge mutants to 70 μM DHA. Data for wild-type hK V 7.4 has been added for reference. Dashed line indicates the mean ΔV 50 observed in hK V 7.1 and hK V 7.2/7.3 channels exposed to 70 μM DHA (same data as in Figure 5A). Data shown as mean ± SEM. n = 6-15. Statistics denote one sample t test against a hypothetical mean of 0 mV. *** denotes p ≤ 0.001, **** denotes p ≤ 0.0001. (E) Cartoon illustrating facilitation of channel opening from the conventional outer VSD site (left, in red) and inhibition of channel opening from the unconventional inner VSD site (right, in blue) mediated by PUFAs (gray). In the outer VSD site, PUFAs interact with the outermost positively charged gating charges of S4, which facilitates channel activation and channel opening. In the inner VSD site, PUFAs interact with the innermost gating charges of S4, stabilizing an intermediate or resting conformation that impairs channel opening. See also Figure 7-figure supplement 1, Figure 7-source data 1.
The online version of this article includes the following source data and figure supplement(s) for figure 7: Source data 1. Numerical data for Figure 7.  with arginines in the lower half of S4 (schematically illustrated in Figure 7E right). Disruption of DHA interaction with this inner VSD site, through substitution of the innermost S4 arginines, endowed hK V 7.4 with the typical PUFA response of a shift in V 50 toward negative voltages. We therefore suggest that hK V 7.4 also harbors a conventional PUFA site at the outer half of the S4, the functional effect of which is unmasked upon disruption of the functionally dominant inner VSD site. In agreement with this, weak DHA interaction with the conventional outer VSD site is indeed observed in our simulations (corresponding to cluster #9, see Figure 6-figure supplement 1B). Furthermore, our experiments using PUFA analogues indicate that it is possible to charge-tune the PUFA response of the hK V 7.4 arginine mutants from a negative to positive shift in V 50 by altering the charge of the PUFA head group, a feature which is in line with observations on other K V channels electrostatically modulated by PUFAs (Börjesson et al., 2010;Liin et al., 2015). A structural alignment between the VSDs of hK V 7.1 (PDB ID -6UZZ) and hK V 7.4 (PDB ID -7BYL) revealed two possible molecular level reasons for the differential binding of DHA to these two channels: in hK V 7.4 the binding site we uncovered indeed appears both larger, with the S4 helix pushed inwards in hK V 7.1 relative to hK V 7.4, and more accessible, as the cleft between S1 and S2 via which DHA appears to enter the cavity features a bulky, obstructing phenylalanine residue (F166) in hK V 7.1 compared to a smaller valine residue (V142) in hK V 7.4 ( Figure 6-figure  supplement 3).
What potential physiological implications may our findings have? Increased consumption of foods rich in PUFAs has long been associated with improved cardiovascular health (Bang et al., 1980;Kagawa et al., 1982;Saravanan et al., 2010) and multiple mechanisms have been proposed, including those mediated by the immune system, the endothelium and vascular smooth muscle tissue (Massaro et al., 2008). Several ion channels in both endothelial cells and VSMCs have been studied as the molecular correlates of PUFA-mediated vasodilatory mechanisms (Bercea et al., 2021). Of note, PUFA-induced inhibition of L-type Ca V channels (Engler et al., 2000) and PUFA-induced activation of large conductance Ca 2+ -activated K + (BK) channels contribute to the vasodilation induced by PUFAs (Hoshi et al., 2013;Limbu et al., 2018). Intriguingly, a recent study by Limbu et al., 2018 observed an endothelium-independent residual relaxation of rodent arteries induced by ω-3 PUFAs despite pre-treatment with several K + channel inhibitors, suggesting there may be another mechanism underlying PUFA-mediated vasorelaxation that does not involve BK channels. Our finding that PUFAs facilitate activation of hK V 7.5 raises the possibility that K V 7.5 subunits contribute to the residual vasorelaxation observed by Limbu and colleagues.
Besides cardiovascular effects, an increased intake of ω-3 PUFAs has also indicated a potential protective role against hearing loss (Curhan et al., 2014;Dullemeijer et al., 2010;Gopinath et al., 2010). While these studies propose the protective mechanism of PUFAs on hearing stem from vascular effects that improve cochlear perfusion, no direct mechanism of PUFAs on the hearing organ was investigated. Based on our findings, it would be unlikely that the decreased risk of hearing loss is due to direct actions of PUFAs on K V 7.4 channels expressed in OHCs, given that we find the channel activity to be inhibited by PUFAs. However, PUFA-induced facilitation of the activity of other K V 7 channels (e.g. K V 7.1 and K V 7.5) in VSMCs could possibly contribute to improved cochlear perfusion. Of note, the two DFNA2-associated missense mutations in hK V 7.4 we examined, F182L and S273A, showed intrinsic biophysical behavior comparable to that of wild-type hK V 7.4 (see Table 2). This is in overall agreement with previous studies of these mutants performed in other expression systems showing voltage dependence of channel opening approximate to that of the wild-type channel (Jung et al., 2019;Kim et al., 2011). Jung and colleagues found that the S273A mutant reduces the average whole-cell current densities to less than 50% of the wild-type (Jung et al., 2019), which indicates that S273A may be a risk factor for development of hearing loss through its limited ability to generate K + currents. However, whether F182L acts as a risk factor in DFNA2 hearing loss or should rather be considered a benign missense mutation (Kim et al., 2011) remains to be determined. We find that both the F182L and S273A mutants display an impaired response to DHA, which, if anything, is expected to preserve channel function in the presence of PUFAs.
To conclude, we find that hK V 7.4 and hK V 7.5 respond in opposing manners to PUFAs. The hK V 7.5 channel's response is largely in line with the responses that have previously been observed in other hK V 7 channels, whereas the hK V 7.4 channel response is not. Altogether, our study expands our understanding of how the activity of different members of the hK V 7 family are modulated by PUFAs and demonstrates different responses explained by different PUFA sites in different hK V 7 subtypes. Further studies are needed to evaluate putative physiological importance of the effects described in this study in more complex experimental systems.

Molecular biology
Human KCNQ4 (GenBank accession no. NM_004700), human KCNQ5 (GenBank accession no. NM_001160133) and human KCNE4 (GenBank accession no. NM_080671) were used in this study. cRNA for injection was prepared from DNA using a T7 mMessage mMachine transcription kit (Invitrogen, Stockholm, Sweden), and cRNA concentrations were determined by means of spectrophotometry (NanoDrop 2000c, Thermo Scientific, Stockholm, Sweden). Mutations were introduced through site-directed mutagenesis (QuikChange II XL, with 10 XL Gold cells; Agilent Technologies, Kista, Sweden) and confirmed by sequencing at the Linköping University Core Facility.

Two-electrode voltage clamp experiments on Xenopus oocytes
Individual oocytes from Xenopus laevis frogs were acquired either through surgical removal followed by enzymatic digestion at Linköping University, or purchased from Ecocyte Bioscience (Dortmund, Germany). The use of animals, including the performed surgery, was reviewed and approved by the regional board of ethics in Linköping, Sweden (Case no. 1941). Oocytes at developmental stages V-VI were selected for experiments and injected with 50 nL of cRNA. Each oocyte received either 2.5 ng of hK V 7.4 RNA or 5 ng of hK V 7.5 RNA. Co-injected oocytes received a 1:1 mix of hK V 7.4 cRNA (2.5 ng) and hK V 7.5 cRNA (2.5 ng) for the expression of hK V 7.4/7.5, or a mix of hK V 7.4 cRNA (2.5 ng) and hKCNE4 cRNA (1.25 ng) for the expression of hK V 7.4/KCNE4. Injected oocytes were incubated at 8 °C or 16 °C for 2-4 days prior to electrophysiological experiments. Two-electrode voltage clamp recordings were performed with a Dagan CA-1B amplifier system (Dagan, MN, USA). Whole-cell K + currents were sampled using Clampex (Molecular devices, San Jose, CA, USA) at 5 kHz and filtered at 500 Hz. For most experiments, the holding potential was set to -80 mV. If experimental conditions allowed for channel opening at -80 mV, the holding potential was set to -100 mV. Current/voltage relationships were recorded using voltage-step protocols prior to and after the application of test compounds. Activation pulses were generated in incremental depolarizing steps of 10 mV, from -100 mV to +60 mV for hK V 7.4 and hK V 7.4/KCNE4, from -90 mV to 0 mV for hK V 7.5, and from -120 mV to +60 mV for hK V 7.4/hK V 7.5. The duration of the activation pulse was 2 s for hK V 7.4, hK V 7.4/KCNE4 and hK V 7.4/7.5, and 3 s for hK V 7.5. The tail voltage was set to -30 mV for all protocols and lasted for 1 s. All experiments were carried out at room temperature (approx. 20 °C). The extracellular recording solution consisted of (in mM): 88 NaCl, 1 KCl, 0.4 CaCl 2 , 0.8 MgCl 2 , and 15 HEPES. pH was adjusted to 7.4 by addition of NaOH. When experiments were conducted at a lower or higher pH, the pH was adjusted the same day as experiments by the addition of HCl or NaOH. Note that the pH of the extracellular recording solution was identical in the control recording and DHA recording of each oocyte. Recording solution containing test compounds was applied extracellularly to the recording chamber during an application protocol (comprised of repeated depolarizing steps every 10 s to 0 mV for hK V 7.4, hK V 7.4/KCNE4 or hK V 7.4/hK V 7.5 co-injected cells, or to -30 mV for hK V 7.5) until steady-state effects were observed. Solutions containing PUFAs or PUFA analogues were applied directly and manually to the recording chamber via a syringe. A minimum volume of 2 mL was applied to guarantee the replacement of the preceding solution in the recording chamber. The recording chamber was thoroughly cleaned between cells with 99.5% ethanol.

Electrophysiological analysis
GraphPad Prism 8 software (GraphPad Software Inc, Ca, USA) was used for data analysis. The voltagedependence of hK V 7 channels was approximated by plotting the immediate tail currents (recorded upon stepping to the tail voltage) against the preceding test voltages. Data were fitted with a Boltzmann function, generating a G(V) (conductance versus voltage) curve: where G min is the minimum conductance, G max is the maximum conductance, V 50 is the midpoint of the curve (i.e, the voltage determined by the fit required to reach half of G max ) and s is the slope of the curve. The difference between V 50 under control settings and under test settings for each oocyte (the ΔV 50 ) was used to quantify shifts in the voltage-dependence of channel opening evoked by test compounds. The relative difference between G max under control settings and under test settings for each oocyte (the ΔG max ) was used to quantify changes in the maximum conductance evoked by test compounds. Note that representative G(V) curves have been normalized in figures to allow for better visualization of V 50 shifts.
To determine the concentration dependence or the pH dependence of ΔV 50 , the following concentration-response function was used: where ΔV 50,max is the maximum shift in V 50 , C is the concentration of the test compound, EC 50 is the concentration of a given test compound or the concentration of H + required to reach 50% of the maximum effect, and N is the Hill coefficient (set to 1 or -1). For studying pH dependence, the values of C were determined with asymptotic 95% confidence intervals in GraphPad Prism 8 and subsequently log-transformed to acquire the apparent pKa values.

Coarse-grained molecular dynamics simulations
The full-length cryo-EM structure of hK V 7.4 channel in an intermediate activation state (PDB ID -7BYL) (Li et al., 2021) was prepared by building in missing residues using the MODLOOP webserver (Fiser et al., 2008). The channel was embedded in a heterogenous bilayer consisting of 480 1-palmi toyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) molecules and 120 docosahexaenoic acid (DHA) molecules (placed at random and distributed equally in both the leaflets) using the CHARMM-GUI MARTINI (Bilayer system) maker (Qi et al., 2015). In one system, DHA with a negatively charged head group was used while in the other, DHA with a neutral head group was considered. The systems were then solvated by adding a~45 Å layer of water to each side of the membrane. Lastly, systems were ionized to reach a 150 mM KCl concentration. Martini 2.2 force field combined with ElNeDyn (Elastic network in dynamics) was used. The systems were minimized and equilibrated following the default CHARMM-GUI protocol. During equilibration, pressure was maintained at 1 bar through Berendsen pressure coupling; temperature was maintained at 300 K through velocity rescaling thermostat (Bussi et al., 2007) with the protein, membrane and solvent coupled. During production simulation, pressure was maintained at 1 bar through Parinello-Rahman pressure coupling (Parrinello and Rahman, 1981). A time step of 20 fs was used. Finally, 19 µs production simulations were carried out for the system with neutral DHA, while eight 5 µs simulations were run for the system with negatively charged DHA. Simulations were performed using GROMACS version 2020.4 (Abraham et al., 2015). The analysis of the coarse-grained simulations was carried out using PyLipid (Song et al., 2022). Frames extracted every 20 ns were used for the analysis. A dual cut-off scheme (lower limit: -0.5 nm, upper limit: -0.7 nm) was used to determine the interactions between DHA and the protein. Interactions were averaged across all subunits of the protein. Binding regions were determined by identifying clusters of at least four residues that interact with the same DHA molecule at the same time. These residues were found by community analysis of residue interaction networks. The most representative bound pose for each binding site was extracted by a scoring of the bound poses based on a density based scoring function, where the probability density is calculated from the simulation trajectories.

Illustrations
For the structural images shown in Figure 4, the cryo-EM resolved structure of the hK V 7.4 channel (PDB ID -7BYL [Li et al., 2021]) was visualized using The Protein Imager (Tomasello et al., 2020). Visualizations of the MD simulation systems were done using Visual Molecular Dynamics (VMD) (Humphrey et al., 1996).

Statistical analysis
Average values are expressed as mean ± SEM. When comparing two groups, a Student's t test was performed. One sample t test was used to compare an effect to a hypothetical effect of 0 (for ΔV 50 ) and 1 for (ΔG max ). When comparing multiple groups, a one-way ANOVA was performed, followed by Dunnett's multiple comparison test when comparing to a single reference group. A p-value <0.05 was considered statistically significant. All statistical analyses were carried out in GraphPad Prism 8.

Additional files
Supplementary files • Transparent reporting form

Data availability
Numerical data is provided in figures and/or corresponding figure legends, Table 2, and the source data files. Source data for molecular dynamics simulations is accessible at https://osf.io/6fuqs/.
The following dataset was generated: