Cannabidiol counters the effects of a dominant-negative pathogenic Kv7.2 variant

A Kv7.2H260 deletion mutation reduces Kv7 currents

To better understand the effects of Kv7.2H260del on brain activity, we conducted a series of electrophysiological and imaging studies at the single cell level to define its effects on homomeric and heteromeric channels that incorporate Kv7.3. It is known that Kv7.3 and Kv7.2 channels expressed in isolation produce only small amplitude currents. Kv7.3 normally increases the membrane expression of other Kv7 channel subunits and can invoke a hyperpolarizing shift in the voltage dependence for activation to increase current amplitude (Schwake et al., 2000; Springer et al., 2021). The Kv7.2H260del mutation is positioned in the external loop of the channel between transmembrane domains 5 and 6 (Figures 1A and 1B). To examine the influence of the H260del mutation on channel biophysical properties, we conducted whole cell patch clamp recordings in tsA-201 cells transiently transfected with wild-type (Kv7.2 wt) or mutant Kv7.2 (Kv7.2mut) cDNA, alone or in combination with Kv7.3. Kv7 current was pharmacologically isolated (see STAR Methods) and activated by a 1.5 s command pulse from a holding potential of −80–40 mV in steps of 10 mV (Figures 1C and 1D). Both Kv7.2 wt and Kv7.2 wt + Kv7.3 channel currents were as previously reported in presenting as a slow activating and non-inactivating current during the command pulse (Figure 1C) (Nappi et al., 2020). Kv7.2 wt current reached a peak amplitude of under ∼2500 pA and Kv7.3 expressed alone produced even less current (451.2 ± 75.7 pA, n = 16) (Figure 1C). Coexpressing Kv7.2 + Kv7.3 cDNA to allow formation of heteromeric channels produced far greater current with peak amplitudes of ∼6000 pA (Figures 1C and 1D). Current from Kv7.2mut expressed alone was dramatically reduced relative to those of Kv7.2 wt expressed alone (Figures 1C and 1D), with the current density (measured at 10 mV) being reduced from 119.2 ± 22.0 pA/pF (n = 10) for Kv7.2 wt to 7.5 ± 1.5 pA/pF (n = 6) for Kv7.2mut (p = 0.0311) (Figure 1E). Coexpressing Kv7.2mut + Kv7.3 also dramatically reduced channel current density from 285.2 ± 31.3 pA/pF (n = 16) for Kv7.2 wt + Kv7.3 to 20.3 ± 4.3 pA/pF for Kv7.2mut + Kv7.3 (n = 22; p < 0.0001) (Figure 1E).

As previous studies have shown that Kv7.2 mutations can shift the channel voltage dependence (Nappi et al., 2020), we examined the influence of H260del on voltage dependence for activation of Kv7.2 channels. We first determined the effects of Kv7.2mut on a heteromeric combination with Kv7.2 wt channels (1:1 ratio). Expressing Kv7.2mut shifted the steady-state activation curve measured for Kv7.2 wt current in the positive direction (Figure 2F), with the half-activation voltage (Va) shifted by 5.6 mV for Kv7.2mut + Kv7.2 wt heteromers relative to Kv7.2 wt homomers (Kv7.2 wt −32.8 ± 1.3, n = 11; Kv7.2mut + Kv7.2 wt −27.2 ± 1.3, n = 10; p = 0.007). A slightly larger rightward shift of 6.4 mV was measured when coexpressing Kv7.2mut + Kv7.2 wt + Kv7.3 (−29.4 ± 1.3, n = 15) relative to Kv7.2 wt + Kv7.3 (−35.8 ± 0.8, n = 17; p = 0.00017) (Figure 1G).

A rightward shift in voltage for activation will contribute to the observed effects of the Kv7.2mut isoform in reducing channel currents. The additional reduction in current density (Figure 1E) could reflect a shift in channel conductance or a change in membrane channel insertion. To address the potential for a change in conductance, we measured the maximum conductance on activation plots. When currents were normalized on conductance plots, the values differed by less than 5%. But comparing the values for absolute conductance showed that Kv7.2mut homomeric channels had a significantly lower maximum membrane channel conductance of 8.3 ± 1.6 nS (n = 10) compared to 21.2 ± 3.7 nS (n = 11) for Kv7.2 wt channels (p = 0.0323) (Figure 1H). Similarly, coexpressing Kv7.2mut + Kv7.3 wt significantly reduced the maximum membrane conductance from 59.3 ± 4.6 nS (n = 17) for Kv7.2 wt + Kv7.3 to 19.2 ± 2.4 nS for Kv7.2mut + Kv7.3 (n = 15; p < 0.0001, post-hoc Tukey’s test following two-way ANOVA) (Figure 1H).

A Kv7.2H260 deletion mutation shifts the balance of Kv7 subunit expression at the membrane

To gain a more specific measure of how the H260del mutation affects membrane expression of Kv7 subunits, we conducted a biotinylation assay on tsA-201 cells expressing either Kv7.2 wt or Kv7.2mut with or without Kv7.3 coexpression. Surface Kv7.2 and Kv7.3 level was calculated as a ratio of biotinylated Kv7 subunits in the total cell lysates and then normalized to the wild-type condition. Semi-quantification of the membrane level of Kv7.2 subunits in the biotinylated membrane samples with Western blot revealed a significant increase in the membrane fraction of Kv7.2mut subunits compared to Kv7.2 wt (Figures 2A and 2B). The ratio of Kv7.2 in membrane fraction lysate to Kv7.2 in total lysate in Kv7.2mut expressing cells was 1.43 ± 0.10 (n = 4) times higher than that for Kv7.2 wt (p = 0.021) (Figures 2A and 2B). However, the membrane level of Kv7.2 in cells that also coexpressed Kv7.2mut and Kv7.3 was not significantly different from the wild-type channels (Figures 2A and 2B). The total Kv7.2 protein level showed a trend of increasing for the homomeric Kv7.2 channels, and was significantly increased in cells coexpressing Kv7.2mut + Kv7.3 channels compared to Kv7.2 wt + Kv7.3 (1.52 ± 0.13, n = 5, p = 0.00729) (Figure 2E). We also quantified the level of Kv7.3 on the membrane for tsA-201 cells coexpressing Kv7.2 wt or Kv7.2mut cDNA. In contrast to measures of Kv7.2 subunits, cells coexpressing Kv7.3 + Kv7.2mut exhibited a significant decrease in the level of Kv7.3 at the membrane (0.52 ± 0.13, n = 4, p = 0.02107) (Figures 2C and 2D) without affecting its total level (Figure 2F).

Together, these results reveal that coexpressing Kv7.mut with Kv7.3 subunits results in a significant increase in the level of membrane-associated Kv7.2mut subunits but a corresponding decrease in surface Kv7.3. The parallel increase in total Kv7.2mut membrane protein but not total Kv7.3 effectively shifts the balance of membrane expression for these two isoforms. The extent to which this corresponds to a shift in the stoichiometry of subunits assembled within heteromeric channels or a change in the extent of heteromeric vs homomeric channel expression cannot be determined from this dataset. However, an increased expression of Kv7.2mut subunits can be expected to contribute to the lower maximal conductance and right-shifted Va of channels detected electrophysiologically (Figures 1D–1F).

Super-resolution imaging supports heteromeric Kv7 channel formation with altered expression patterns

One possible explanation for the effects of an increase in Kv7.2mut expression on Kv7 current would be to interfere with the heteromeric assembly of subunits or their expression at the membrane. To further assess protein interactions, we used direct stochastic optical reconstruction microscopy (dSTORM) to image Kv7 immunolabels and quantify the degree of overlap of label emissions and the minimal distance between neighboring immunolabeled clusters. By combining dSTORM with total internal reflection fluorescence (TIRF) illumination, we could further restrict visualization of Kv7 immunolabels to within 150 nm of the coverslip surface to which cultured cells are attached. Kv7.2 subunits were detected directly with an anti-Kv7.2 antibody and Kv7.3 using an HA-tagged construct. The precision of resolution determined under our conditions reached 20–40 nm in the lateral plane. As such, the threshold for minimal cluster diameter was set at 40 nm, a size we expect reflects a signal from more than one subunit or channel.

Dual labeling of tsA-201 cells coexpressing Kv7.2 and Kv7.3 cDNAs revealed distinct fluorescent clusters either in isolation or positioned with enough proximity to exhibit overlap in the fluorescent signal emissions consistent with formation of heteromeric Kv7.2-Kv7.3 channels (Figures 3A–3D). Magnified regions of interest (ROI) further suggested a more extensive degree of overlap between Kv7.3 and Kv7.2mut compared to Kv7.3 and K7.2 wt subunits (Figures 3B–3D). Calculating the cluster density per unit area revealed a substantial difference in the absolute density of detected clusters, with 151 ± 15 clusters/μm² for Kv7.2 wt (n = 9) and 91 ± 3.4 clusters/μm² for Kv7.3 (n = 9; p = 0.0033) (Figure 3E). A similar difference in the density of clusters was apparent for the Kv7.2mut label (134 ± 14 clusters/μm²; n = 9) compared to Kv7.3 (95.6 ± 4.9 clusters/μm²; n = 9; p = 0.027) (Figure 3E). However, no significant differences were found between the number of Kv7.2 wt vs Kv7.2mut clusters (Kv7.2 wt, 151 ± 15 clusters/μm², n = 9; Kv7.2mut, 134 ± 14 clusters/μm², n = 9; p = 0.43) or between the ratios of Kv7.3 + Kv7.2 wt clusters (1.65 ± 0.13, n = 9) to that of Kv7.3 + Kv7.2mut clusters (1.43 ± 0.14, n = 9; p = 0.288). Quantification of immunolabeled clusters then indicated that both Kv7.2 wt and Kv7.2mut isoforms exhibit a higher level of expression compared to Kv7.3, and with similar relative levels for Kv7.2 wt/Kv7.3 to that of Kv7.2mut/Kv7.3 (Figure 3E).

To gain an estimate of the degree of colocalization, we calculated Manders’s correlation coefficient to quantify the fractional overlap of one immunolabel emission upon the other. This analysis revealed a marked disparity in the overlap of Kv7.3 labels compared to Kv7.2 isoforms. Thus, a smaller percentage of the total intensity of Kv7.2 wt label overlapped with Kv7.3 (M2) compared to a large portion of Kv7.3 labels that overlapped with Kv7.2 wt (M1) (M1wt vs M2wt, n = 20; p = 6.29 × 10⁻⁸) (Figure 3F). A similar difference was detected in the extent of overlap for the Kv7.2mut subunit with Kv7.3 (M2mut), in which a larger portion of Kv7.3 labels overlapped with Kv7.2mut (M1mut) (M1mut vs M2mut, n = 24; p = 2.2 × 10⁻¹⁶). These results are in line with the noted cluster density disparities between Kv7.2mut and Kv7.3 (cf. Figures 4E and 4F). Comparison of either M1 or M2 between the Kv7.2 wt and Kv7.2mut condition showed significantly higher overlap when Kv7.3 was transfected with Kv7.2mut (M1wt vs M1mut, n = 22, p = 0.0039; M2wt vs M2mut, n = 22, p = 0.0017). These calculations suggest a higher incidence of colocalization between Kv7.2mut and Kv7.3 than for Kv7.2 wt and Kv7.3. This finding was further supported by higher Pearson correlation coefficients in cells cotransfected with Kv7.3 + Kv7.2mut than with Kv7.2 wt (n = 22; p = 2.312 × 10⁻⁶) (Figure 3G).

The reasons for a difference in absolute cluster density (Figure 3E) and the extent of isoform overlap (Figures 3F and 3G) are unknown but suggests that the greater number of Kv7.2 clusters provide more than enough subunits to heteromerize with available Kv7.3 clusters. Analysis of the degree of overlap of signals further suggests that Kv7.2mut was more effective at associating with Kv7.3 than was Kv7.2 wt. A higher incidence of this presumed heteromeric formation was also evident visually in the number of overlapping fluorophore emissions (cf. Figures 3B–3D). However, we note that these interactions were not obligatory in that isolated clusters for a given subunit could also be detected.

dSTORM-TIRF imaging allows for calculation of the minimum distance to nearest neighbor cluster centroids. For this, we first coexpressed Kv7.2 wt and Kv7.3 cDNA and plotted histograms of the minimal distance between clusters of each isoform (Figure 3H). The histogram plot of minimal neighbor distance between Kv7.2 wt clusters (Kv7.2 wt – Kv7.2 wt) was moderately right-skewed with a distribution over 100–1600 nm and a peak ∼500 nm (Sk₂ = 0.532; median = 565; K = 77. 2, n = 4). Similar results were found for histogram plots of the minimal distance between Kv7.3 clusters, with a distribution between 125 and 1750 nm and a peak ∼800 nm (Sk₂ = 0.334; median = 737; K = 95.5, n = 4) (Figure 3H). In contrast, the histogram for minimal neighbor distances between clusters of Kv7.2 wt and Kv7.3 was strongly right-skewed with values between 0 and 950 nm and a peak ∼150 nm (Sk₂ = 0.564; median = 300; K = 5.04, n = 4) (Figure 3H). A comparison of the minimal neighbor distances for cells coexpressing Kv7.2mut + Kv7.3 yielded similar results. Specifically, the histogram of minimal neighbor distance between Kv7.2mut clusters was moderately right-skewed over 125–1750 nm and peak at ∼625 nm (Sk₂ = 0.435; median = 630; K = 53.4, n = 7) (Figure 3I). The histogram for minimal distances between Kv7.2mut and Kv7.3 was again strongly right-skewed over 0–975 nm and peak ∼150 nm (Sk₂ = 0.655; median = 263; K = 10.0, n = 7) (cf. Figure 3I). The strong skew in minimal cluster distance histograms for Kv7.2 + Kv7.3 combinations is important in signifying a strong preferred association between these subunits, again as expected for successful heteromeric assembly of channel subunits.

Overall, these data support the electrophysiology on the ability for Kv7.2mut subunits to associate with Kv7.3 subunits at proximities consistent with formation of heteromeric channels. Yet, it is still important to acknowledge that dSTORM images restricted even to 150 nm depth of field through TIRF illumination could still report channels within the most superficial sites of the smooth ER or in transport to the membrane.

Kv7.2mut subunits associate at nanometer distance with Kv7.3 expected for heteromeric channel formation

The heteromeric assembly of specific Kv7 channel subunits into a functional channel can directly affect channel properties and the extent of incorporation at the membrane level (Maljevic et al., 2011; Schroeder et al., 1998; Schwake et al., 2000). Previous studies have shown that Kv7 subunits assemble close enough to promote fluorescence resonance energy transfer (FRET) when fluorescent-tagged channels are coexpressed in HEK293 cells (Bal et al., 2008). To test if Kv7.2mut affected the association between channel subunits, we used spectral-FRET to analyze Kv7 subunit interactions in tsA-201 cells (Wallrabe and Periasamy, 2005). EGFP was used as a donor molecule and mKate2 as an acceptor molecule and assessed using spectral confocal microscopy (Asmara et al., 2017).

Cells in the control and experimental groups were transfected with Kv7.2wt-EGFP + mKate2, and Kv7.2 wt/mut-EGFP + Kv7.3-mKate2, respectively. The vast majority of cells cotransfected with tagged channel subunits exhibited EGFP and mKate2 fluorescence when excited at 488 and 552 nm, respectively (Figure 4A). Spectral images were then taken with 488 nm laser and fluorophore channels were linearly unmixed. ROIs were selected to analyze the average intensity of EGFP before and after photobleaching the mKate2 at 552 nm to estimate the extent of energy transfer between EGFP and mKate2 (Wallrabe and Periasamy, 2005). The intensity of green fluorescence increased in the Kv7.2wt-EGFP + Kv7.2wt-mKate2 group after photobleaching mKate2 (9.8% ± 1.0%, n = 5, p = 1.9 × 10⁻⁵) (Figures 4B and 4C), whereas the control group coexpressing Kv7.2wt-EGFP and mKate2 exhibited a slight decrease in EGFP intensity (4.7% ± 1.4%, n = 6) (Figures 4B and 4C). Together, these data verify the occurrence of FRET between Kv7.2wt-EGFP with Kv7.2wt-mKate2 as a heteromeric assembly of labeled subunits with spatial separations of <10 nm. For cells coexpressing the Kv7.2mut subunit, the intensity of EGFP also increased after photobleaching mKate2 (12.1% ± 1.7%, n = 6, p = 1.7 × 10⁻⁵) (Figure 4C). Finally, a significant increase in fluorescence intensity was observed for coexpression of Kv7.2 wt or Kv7.2mut + Kv7.3 subunits (Kv7.2 wt + Kv7.3 7.9% ± 1.1%, n = 4, p = 2.1 × 10⁻⁴; Kv7.2mut + Kv7.3 7.2% ± 2.0%, n = 5, p = 7.1 × 10⁻⁴) (Figure 4C).

Taken together, these results suggest that Kv7.2mut does not impede the formation of Kv7.2-Kv7.3 heteromeric channels, as also suggested by dSTORM analysis. The lower membrane channel conductance measured with Kv7.2mut expression (Figure 1) thus does not reflect a loss of heteromeric channel assembly.

External loop mutations

Pathogenic variants found in Kv7.2 and Kv7.3 impact multiple possible residues that affect different aspects of channel function and lead variably to BFNS or more severe forms of epilepsy and encephalopathy (Milh et al., 2013; Nappi et al., 2020; Vanoye et al., 2022). The Kv7.2H260del examined here is positioned close to two other reported Kv7.2 mutations. These are also in the extracellular loop between S5 and S6 but at positions 258 (c.773A>G p.Asn258Ser, N258S) and 265 (c.793G>A p.Ala265Thr, A265T) (Cooper et al., 2015). In the case of the Kv7.2 N258S variant, the patient phenotype was BFNS (Yalçin et al., 2007). At the channel level, N258S presented as a dominant-negative factor that reduced current in Kv7.2 N258S + Kv7.3 channels, but with a reduced number of mutant channels at the membrane (Maljevic et al., 2011). No change in voltage dependence for activation was found, leading to a proposed defect in protein folding or trafficking (Maljevic et al., 2011). This differs from the Kv7.2H260del in which we detected a significant depolarizing shift in Va and no change in total Kv7 protein at the membrane. Although we cannot rule out the potential for a disorder in protein folding to account for the decrease in maximal membrane conductance in Kv7.2H260del + Kv7.3 channels, our results would suggest no specific disruption in ability for subunits to heteromerize, or for channels to traffick or insert at the membrane. Instead, the primary loss of function in Kv7.2H260del is reflected in a reduction in current through a shift in voltage for activation and a loss of membrane conductance not reported for the nearby N258S mutation.

The Kv7.2 A265T variant was detected in patients in some of the earliest studies linking Kv7.2 mutations with Ohtahara Syndrome (Milh et al., 2013; Saitsu et al., 2012; Weckhuysen et al., 2012), similar to the epilepsy phenotype described for our proband with Kv7.2H260del. The A265T variant also acted in a dominant-negative manner to suppress Kv7 current but failed to exhibit any change in voltage dependence for activation or altered levels of surface expression (Milh et al., 2013), all effects that are opposite to that found for the Kv7.2H260del variant.

These comparisons support conclusions recently emphasized in a systematic comparison of Kv7.2 mutants using automated patch clamp where the net effect of a Kv7.2 mutation produced variable results (Vanoye et al., 2022). The current work now identifies differences at the channel level for mutations in the extracellular loop between S5 and S6 even when separated by only 2 amino acids, as highlighted by Kv7.2 N258S and H260del that lead to symptoms characteristic of BFNS or NOEE, respectively.

Therapeutic actions of CBD on a Kv7.2 mutation

A great deal of attention has been focused on achieving rescue of Kv7 channel function across the numerous pathogenic variants found in Kv7.2 or Kv7.3 subunits. A number of Kv7 activators have been reported that could provide potential recovery of function in Kv7.2 mutations (Borgini et al., 2021; Incontro et al., 2021; Kanyo et al., 2020; Larsson et al., 2020b; Vanoye et al., 2022), although non-specific actions or side effects have led to removal of some of these drugs from clinical use (Brickel et al., 2020). The proband with Kv7.2H260del had resolution of seizures through use of CBD:THC oil. The ability for CBD to act as an anti-epileptic treatment is well known (Williams and Stephens, 2020). CBD or eCBs were also recently reported to act as agonists for Kv7 channels in CHO, superior cervical ganglion, and hippocampal neurons (Incontro et al., 2021; Zhang et al., 2022). We therefore tested CBD in relation to Kv7 wild-type and Kv7.2mut subunits expressed in tsA-201 cells. Here, we found that 1 μM CBD induced ∼ -5.7 mV hyperpolarizing shift in Va for coexpression of the Kv7.2 wt + Kv7.3 isoforms. While this is less than a shift of up to −17.5 mV reported for 1 μM CBD application in Zhang et al. (2022), it could reflect several differences that include cell lines (CHO vs tsA-201), CBD source, temperature, and drug application procedures. Importantly, we found that CBD also shifted the Va for Kv7.2mut + Kv7.2 + Kv7.3 by ∼ -7 mV to −37.5 ± 1.3 mV compared to controls, a value that was not significantly different from that recorded after CBD application to cells coexpressing Kv7.2 wt + Kv7.3 (p = 0.31). These results are important in representing the first evidence that CBD can offset deficits in channel function caused by a Kv7.2 channel variant relevant to a patient’s experience. Yet since CBD induced a hyperpolarizing shift in the Va of channels expressing either wild-type or mutant subunits (Figures 5A and 5B) (Zhang et al., 2022), we cannot conclude that its effects are specifically exerted on the mutant subunit. We would also interpret this to suggest no change in the binding efficacy of Kv7.2H260del-containing channels for CBD compared to wild-type channels, but this would need to be addressed through modeling (Incontro et al., 2021). Further work will also be needed to assess the potential for CBD to work in a similar fashion on channels incorporating the wide range of Kv7 mutations found in subjects (Cooper et al., 2015; Vanoye et al., 2022).

The mechanism(s) by which CBD offset the effects of Kv7.2mut on channel voltage dependence has not been established. In intact tissue, CBD is known to interact with a host of voltage- and ligand-gated channels that will modulate cell excitability and plasticity (Ghovanloo et al., 2018; Pertwee et al., 2010; Ross et al., 2008; Watkins, 2019). Since our recordings were conducted on principal subunits of Kv7 channels expressed in tsA-201 cells in vitro, many of these potential alternative sites of action will not be relevant here. Any potential actions of CBD on other potassium channels (Watkins, 2019) that might be expressed in tsA-201 cells are minimized given our blockade of a large spectrum of voltage- and calcium-gated potassium channels during recordings. It is known that CBD has a poor affinity for both CB1 and CB2 receptors but negatively modulates CB1 through an allosteric interaction (Laprairie et al., 2015). Our tests with CBD in the presence of the CB1and CB2 receptor blockers would indicate that the effects seen here do not involve CBD actions through these receptors. The ability to record effects by CBD within 20 min of drug application would be consistent with a direct interaction on Kv7 channels, as reported for eCBs and CBD (Incontro et al., 2021; Larsson et al., 2020b; Zhang et al., 2022). However, it is possible that CBD could interact with other receptors or channels in tsA-201 cells that could invoke other factors that modulate Kv7 channels (i.e. PiP2, PKC, calmodulin, arachidonic acid, and ERK1-2) (Borgini et al., 2021; Greene and Hoshi, 2016; Hudson et al., 2019; Larsson et al., 2020a).

The widespread influence of CBD on neural circuit function has identified a range of potential targets to reduce epileptic seizures in subjects (Baculis et al., 2020; Greene and Hoshi, 2016; Williams and Stephens, 2020). It is interesting that in structures such as neocortex and the hippocampus Kv7 channels can further influence factors that range from morphological development through intrinsic excitability and circuit-wide theta frequency resonance (Baculis et al., 2020; Greene and Hoshi, 2016; Incontro et al., 2021; Peters et al., 2005). The current study together with recent reports of direct actions by eCBs and CBD on Kv7 channels (Incontro et al., 2021; Zhang et al., 2022) suggests that part of this influence could come about by modifying Kv7 channel function. Further work will also be required to determine the extent to which CBD acts specifically on Kv7 channels to account for anti-epileptic actions or autistic behaviors in subjects with Kv7 mutations.

Resource availability

Experimental model and subject details

Method details

We gratefully acknowledge J. Forden and F. Visser of the Hotchkiss Brain Institute Molecular facility for technical assistance and H. Asmara for helpful discussions and support. This work was supported by a Canadian Institutes of Health Research Operating Grant (R.W.T.) (PJT156153).