An epilepsy-associated KV1.2 charge-transfer-center mutation impairs KV1.2 and KV1.4 trafficking.

We report on a heterozygous KCNA2 variant in a child with epilepsy. KCNA2 encodes KV1.2 subunits, which form homotetrameric potassium channels and participate in heterotetrameric channel complexes with other KV1-family subunits, regulating neuronal excitability. The mutation causes substitution F233S at the KV1.2 charge transfer center of the voltage-sensing domain. Immunocytochemical trafficking assays showed that KV1.2(F233S) subunits are trafficking deficient and reduce the surface expression of wild-type KV1.2 and KV1.4: a dominant-negative phenotype extending beyond KCNA2, likely profoundly perturbing electrical signaling. Yet some KV1.2(F233S) trafficking was rescued by wild-type KV1.2 and KV1.4 subunits, likely in permissible heterotetrameric stoichiometries: electrophysiological studies utilizing applied transcriptomics and concatemer constructs support that up to one or two KV1.2(F233S) subunits can participate in trafficking-capable heterotetramers with wild-type KV1.2 or KV1.4, respectively, and that both early and late events along the biosynthesis and secretion pathway impair trafficking. These studies suggested that F233S causes a depolarizing shift of ∼48 mV on KV1.2 voltage dependence. Optical tracking of the KV1.2(F233S) voltage-sensing domain (rescued by wild-type KV1.2 or KV1.4) revealed that it operates with modestly perturbed voltage dependence and retains pore coupling, evidenced by off-charge immobilization. The equivalent mutation in the Shaker K+ channel (F290S) was reported to modestly affect trafficking and strongly affect function: an ∼80-mV depolarizing shift, disrupted voltage sensor activation and pore coupling. Our work exposes the multigenic, molecular etiology of a variant associated with epilepsy and reveals that charge-transfer-center disruption has different effects in KV1.2 and Shaker, the archetypes for potassium channel structure and function.

A child patient with epileptic seizures has a de novo, missense, heterozygous mutation c.698T > C in KCNA2. KCNA2 encodes KV1.2 voltage-gated potassium channel subunits (1), which are expressed in central neurons and primarily localized to axon initial segments, juxtaparanodes, and axon preterminals (2). KV1.2 channels form a delayed rectifier conductance that regulates action potential repolarization, contributing to firing frequency and synaptic transmission (2–4). Neuronal firing behavior can be further modulated by KV1.2-subunit interaction with other KV1 subunits: e.g., an A-type conductance is formed by KV1.2 and KV1.4 heteromeric channels (3, 5, 6). A growing number of both gain- and loss-of-function KCNA2 variants are implicated in epileptic encephalopathy (7–10).

The mutation causes substitution F233S at the second (S2) transmembrane helix of KV1.2 (Fig. 1). Located within the voltage-sensing domain (VSD), F233 is part of the charge transfer center critical for voltage sensitivity (Fig. 1) (11, 13, 20). The equivalent mutation (F290S) in the Shaker KV channel from Drosophila impairs voltage-dependent channel opening (11) by disrupting the final VSD activation transition, which couples VSD activation to pore opening (13). However, no investigation of the same mutation has been published in the KV1.2 channel, even though the KV1.2-2.1 channel chimera (which encompasses the KV1.2 F233 site) is the gold standard for potassium channel structure (11, 21). We sought to characterize the effects of mutation F233S in KV1.2, as well as its neuronal molecular partner, KV1.4.

Fig. 1.

Discovery of a KCNA2 mutation in a patient with epilepsy. (A) Membrane topology of the KV1.2 subunit and model of a homotetrameric KV1.2 channel (8). F233S is on helix S2, within the VSD. (B) Model of KV1.2 VSD activation and deactivation (8). The electric field is focused on F233 (magenta, star symbol), which is part of the charge transfer center (11). Upon membrane depolarization (+Vm), positively charged residues on S4 [side chains in dark blue (12)], traverse the membrane electric field, compelling S4 to move outward. Movement of the fourth conserved arginine R4 is thought to couple VSD activation to pore opening (13, 14). Upon membrane repolarization (−Vm), S4 returns to its resting state, inducing pore closure (15–19). (C) Partial EEG of the sleeping patient, showing right posterior epileptiform discharges (red X). (D) Sequencing chromatograms of the patient and his healthy parents. Inset: Magnified view at the mutation site (KCNA2 c.698T > C, heterozygous).

Results

Discovery of a KCNA2 Variant in a Patient with Epilepsy.

The proband is a male infant with onset, at 15 mo, of a febrile generalized convulsive seizure that lasted 5 min. The second seizure (20 mo), febrile status epilepticus, lasted 45 min. The third seizure (23 mo) was prolonged status epilepticus associated with roseola infection and fever, whereupon he was hospitalized with intubation. The initial electroencephalogram (EEG) showed right hemispheric slowing; a follow-up EEG showed right posterior temporal slowing and multifocal spikes (Fig. 1). The proband continued to have occasional seizures, often associated with illness and fever and usually requiring rescue medication. His development has been delayed with some mild autistic features. The patient was screened with an exome-based DNA panel of 224 epilepsy-associated genes (SI Appendix, Table S1). A heterozygous c.698T > C (p.Phe233Ser) missense change in KCNA2 (transcript: NM_004974.3) was identified. The variant was confirmed in the patient by Sanger sequencing. Targeted Sanger sequencing of DNA from parents showed its de novo origin (Fig. 1). This variant has not been recorded in population databases [the genome aggregation database and the exome aggregation consortium (22), ClinVar (23), and the human gene mutation database (24)] and was classified as potentially deleterious by the Sorting Intolerant from Tolerant (SIFT) (25) and MutationTester2 (26) prediction algorithms. No other pathogenic variants were found.

F233S Completely Prevents the Cell-Surface Trafficking of KV1.2 Channels.

Electrophysiological experiments in Xenopus oocytes injected with KV1.2(F233S) complementary ribonucleic acid (cRNA) revealed no detectable current (Fig. 2). To differentiate between a severe functional or trafficking deficiency, we performed trafficking assays in mammalian COS-7 cells, as previously (8). Cells were transfected with wild-type (WT) or F233S KV1.2 constructs (Fig. 2) reporting both total and surface protein (27). These experiments showed that KV1.2(F233S) subunits do not traffic to the cell surface (Fig. 2 ).

Fig. 2.

F233S completely prevents the cell-surface trafficking of KV1.2 channels. (A) Cut-open oocyte Vaseline gap (COVG) experiments showing that Xenopus oocytes injected with KV1.2(F233S) cRNA exhibit no voltage-dependent currents, even upon strong depolarization (red). Vm, membrane potential; Im, membrane current. (B) Fluorescent KV1.2 constructs to investigate trafficking. Top: WT KV1.2 with N-terminally fused enhanced green fluorescent protein (EGFP), reporting total protein, and extracellularly accessible hemagglutinin (HA) tag, reporting surface protein. Bottom: As on top, with F233S. (C) Confocal micrographs of nonpermeabilized COS-7 cells transfected with the constructs in B. (D) Flow cytometry of live COS-7 cells transfected with the constructs in B. Left: Cell-density plots of log(EGFP) (total protein) against log(α-HA) (cell-surface expression). The vertical solid line separates low- and high-EGFP cells (Left and Right). The horizontal dashed line separates high-EGFP cells into α-HA negative (Bottom) and α-HA positive (Top). Right: Percentage of α-HA-positive cells, normalized to cells expressing KV1.2(WT). Errors are SEM.

KV1.2(F233S) Subunits Cause a Dominant-Negative Suppression of KV1.2 Channel Conductance.

As the patient is heterozygous, both WT and F233S-bearing KV1.2 subunits are present in his neurons and could tetramerize during early biogenesis as nascent polypeptides (28). We injected different amounts of WT and F233S cRNA in Xenopus oocytes to emulate homozygous WT (double cRNA dose corresponding to two WT alleles), half-dose WT (single cRNA dose corresponding to one WT allele), heterozygous (single WT and single F233S cRNA doses), and homozygous mutant (double F233S cRNA dose). The cells responded linearly to injected cRNA, exhibiting 0.57- ± 0.07-fold conductance in the half-dose WT condition compared to homozygous WT (1.0 ± 0.098) (Fig. 3 ). The former was not significantly different from a theoretical sample with mean 0.5 and the same σ and n (P = 0.50). Heterozygous cells exhibited ∼20% conductance compared to homozygous-WT cells, significantly less than half-dose WT (P = 1.3e−5), demonstrating dominant-negative loss of KV1.2 function. The current in these conditions had similar voltage dependence (Fig. 3), indicating that the main effect of F233S is its trafficking defect. Homozygous-mutant cells produced no current (Fig. 3 ).

Fig. 3.

F233S subunits cause a dominant-negative suppression of KV1.2 channel conductance. (A) Representative OpusXpress (TEVC) current traces from cells injected with KV1.2 WT and F233S cRNA. 1×RNA = 0.5 ng/oocyte. (B) Macroscopic conductance (Gtotal) relative to 2× (homozygous) WT (1.0 ± 0.098); 1×WT (0.57 ± 0.073); 1×WT + 1×F233S (0.19 ± 0.022). (C) Voltage dependence of 2×WT (V0.5 = −5.8 ± 0.78 mV; zeff = 2.3 ± 0.12 e0); 1×WT (V0.5 = −7.6 ± 0.90 mV; zeff = 2.7 ± 0.15 e0); 1×WT + 1×F233S (V0.5 = −8.3 ± 0.66 mV; zeff = 3.3 ± 0.13 e0). (D) Current evoked by depolarization to 100 mV in homozygous mutant (2×F233S; 300 ± 31 nA) and sham-injected cells (450 ± 95 nA); P = 0.15. Errors are SEM.

KV1.2(F233S) Subunits Sequester, and Are Concomitantly Rescued by, KV1.2(WT).

The dominant-negative loss of KV1.2 conductance is likely due to the association of WT and trafficking-deficient F233S subunits, decreasing the availability of the former on the cell surface. We directly evaluated surface trafficking of WT subunits in the presence of either WT or F233S subunits, and vice versa, by flow cytometry. COS-7 cells were transfected with two KV1.2 constructs in equal proportion, representing two KCNA2 alleles, each bearing a different pair of tags reporting total and surface protein expression (Fig. 4). We evaluated surface expression in live cells positive for both total protein tags (Fig. 4). When a KV1.2(WT) was coexpressed with another KV1.2(WT) construct, it exhibited high surface expression. Crucially, KV1.2(WT) trafficking was significantly reduced in the presence of KV1.2(F233S).

Fig. 4.

KV1.2(F233S) subunits sequester, and are concomitantly rescued by, KV1.2(WT). (A) Constructs used to evaluate KV1.2 cell-surface trafficking. Each construct emulates one allele. KV1.2 with N-terminally fused EGFP, reporting total protein production; an extracellular HA tag (as in Fig. 2) cotransfected with KV1.2 with N-terminally fused mRFP1, reporting total protein production; and an extracellular bungarotoxin (BTX) binding-site (BBS) tag. (B) Flow cytometry experiments on cells transfected with the constructs in A. The cell-density plots show total protein [log(EGFP)] and surface staining [log(α-HA)] in live cells positive for mRFP1. The vertical dashed lines separate negative (Left) and positive (Right) EGFP cells. Percentage of cells with a positive surface (α-HA) signal (normalized to WT*|WT, i.e., 100 ± 0.85%): WT*|F233S: 55 ± 2.0%; F233S*|WT: 22 ± 1.5%; F233S*|F233S: 1.3 ± 0.27%. Signals from mRFP1 and BTX in EGFP-positive cells are in . Errors are SEM. n.s.= not significant.

The previous result suggests that WT and F233S subunits can assemble into trafficking-deficient heterotetramers. Is it possible that a different fraction of WT/F233S heterotetramers can reach the surface? Indeed, a significant amount of F233S subunits were detected on the surface of cells coexpressing WT, indicating rescue (Fig. 4). The same results were obtained using the oppositely labeled constructs, although bungarotoxin signals were weaker than those from hemagglutinin-tagged constructs ().

Only KV1.2(WT/F233S) Heterotetramers with One F233S Subunit Are Likely Trafficking Competent.

The results so far suggest that some KV1.2(WT/F233S) heterotetramers are trafficking deficient, causing WT-subunit sequestration, while others are trafficking competent, causing F233S-subunit rescue. Since the N-terminal tetramerization domain of KV1 subunits facilitates oligomerization early in biosynthesis (28), one would expect the full complement of tetrameric WT/F233S stoichiometries to form. Could it be that only specific stoichiometries are trafficking competent? To direct the distribution of tetrameric stoichiometries, we implemented an "applied transcriptomics" experimental paradigm, varying the molar proportion of KV1.2(F233S) RNA injected in oocytes over a broad range while keeping the amount of WT RNA constant. KV1.2 conductance decreased precipitously in a F233S dose-dependent manner (Fig. 5). In cells with the most F233S cRNA, there was also a ∼5-mV depolarizing shift in voltage dependence ().

Fig. 5.

Only 3WT:1F233S heterotetramers are trafficking capable. (A) Relative conductance in cells injected with KV1.2(WT) cRNA and increasing proportion of KV1.2(F233S) cRNA. The voltage dependence in all conditions is in . Superimposed curves show number of tetramers of specified composition, relative to the 0-F233S condition, generated by a model assuming binomially distributed tetramerization. (B) Representative COVG current traces from cells injected with dimeric KV1.2 concatemer cRNA. Relative macroscopic conductance (C) and voltage dependence (D) of cells injected with KV1.2(WT)-KV1.2(WT) (WT-WT) cRNA (relative Gtotal = 1.0 ± 0.25; V0.5 = −3.3 ± 1.5 mV; zeff = 2.3 ± 0.16e0), WT-F233S (relative Gtotal = 0.053 ± 0.017; V0.5 = 19 ± 0.29 mV; zeff = 2.7 ± 0.083e0) or F233S-WT (relative Gtotal = 0.068 ± 0.014; V0.5 = 22 ± 0.59 mV; zeff = 2.3 ± 0.10e0). (E) No current was observed in cells injected with F233S-F233S up to 200 mV (450 ± 48 nA) compared to uninjected cells (560 ± 150 nA; P = 0.53). Errors are 95% CI (A) or SEM (B–E).

Assuming that F233S does not affect maximal open-probability or single-channel conductance, the data reflect the number of channels on the cell surface. Thus, the data were overlaid with models predicting increasing numbers of F233S subunits participating in trafficking heterotetramers. The model allowing no F233S-containing heterotetramers (black curve) failed to account for the data, predicting too low conductance at high F233S proportions. This is consistent with the observation that F233S subunits are rescued by WT (Fig. 4 and ). The green model, allowing one F233S subunit per heterotetramer, described the data best. Models allowing more F233S subunits (cyan and purple curves) greatly overestimated the observed conductance. There was no evidence of cells being overwhelmed by cRNA at the amounts used (). A thorough discussion of the model assumptions, considering additional simulations and subsequent experimental results, is included in the .

The green model suggested that the 2WT:2F233S stoichiometry was trafficking deficient. To corroborate this premise, we constructed concatenated KV1.2 dimers, which are expected to assemble as pseudotetrameric dimers of dimers (29, 30). Accordingly, KV1.2(WT)-KV1.2(F233S) (WT-F233S) dimers exhibited very low conductance (∼5%) compared to WT-WT, supporting that the 2WT:2F233S stoichiometry is strongly trafficking impaired (Fig. 5 ). This result independently excluded the cyan model in Fig. 5. Despite the very low expression level of WT-F233S channels, it was possible to characterize their voltage dependence: their half-activation potential was shifted (ΔV0.5) by 22 ± 0.34 mV compared to WT-WT (Fig. 5), demonstrating that F233S also impaired KV1.2 voltage dependence in a dominant-negative manner. The inverted construct F233S-WT had similar properties (ΔV0.5: 25 ± 0.69 mV; Fig. 5 ). F233S-F233S–injected cells did not exhibit current distinguishable from uninjected cells (Fig. 5 ).

The KV1.2(F233S) Dominant-Negative Trafficking Deficiency Extends to KV1.4 Subunits.

Paralogous potassium channel subunits were shown to assemble into heteromeric channels soon after their cloning (31, 32). In mammals, KV1.4 subunits are molecular partners of KV1.2, forming heterotetrameric channels that contribute to the A current in neurons (3, 5, 6). We used flow cytometry to directly evaluate KV1.2(F233S) rescue by KV1.4 and whether KV1.4 subunits are sequestered by KV1.2(F233S). COS-7 cells were cotransfected with two constructs: KV1.4 and KV1.2 (WT or F233S), each with a different set of fluorescent tags for total and surface protein expression (Fig. 6). KV1.4 surface trafficking was reduced in the presence of KV1.2(F233S) (Fig. 6). At the same time, KV1.4 assisted the trafficking of KV1.2(F233S) subunits (Fig. 6). This represented partial rescue, as the cells did not achieve the same KV1.2 surface levels as for WT subunits (Fig. 6).

Fig. 6.

KV1.2(F233S) subunits concomitantly suppress and are rescued by KV1.4. (A) Constructs used to evaluate KV1.2 and KV1.4 cell-surface trafficking. KV1.4 with N-terminally fused EGFP, reporting total protein production; an extracellular HA tag cotransfected with KV1.2 with N-terminally fused mRFP1, reporting total protein production; and an extracellular BBS tag. (B) Flow cytometry experiments on cells transfected with constructs in A. The cell-density plots show total protein [log(EGFP)] and surface staining [log(α-HA)] in live cells positive for mRFP1. The vertical dashed line separates negative (Left) and positive (Right) EGFP. Percentage of cells with a positive surface (α-HA) signal [normalized to KV1.4*|KV1.2(WT), i.e., 100 ± 2.4%]: KV1.4*|KV1.2(F233S): 8.5 ± 1.9%. (C) As in B, showing the signals from KV1.2 constructs cotransfected with KV1.4. Plots of log(mRFP1) (total protein) against log(BTX) (cell-surface expression). The vertical dashed line separates negative (Left) and positive (Right) mRFP1 cells among the plotted live, EGFP-positive cells. Percentage of cells with a positive surface (BTX) signal [normalized to KV1.2(WT)*|KV1.4, i.e., 100 ± 2.7%]: KV1.2(F233S)*|KV1.2(F233S): 3.7 ± 0.36%; KV1.2(F233S)*|KV1.4: 16 ± 3.1%. Errors are SEM.

KV1.4 and KV1.2(F233S) Subunits Likely Form 3:1 and 2:2 Heteromeric Channels.

KV1.4 subunits are more adept at surface trafficking, both lacking endoplasmic reticulum (ER) retention signals and possessing a forward trafficking motif (33–35). In fact, they enhance KV1.2 trafficking (36). We hypothesized that KV1.4 is more proficient than KV1.2(WT) in rescuing KV1.2(F233S). Indeed, in cells expressing KV1.4, conductance was significantly decreased only when KV1.2(F233S) cRNA was added in eightfold excess (Fig. 7). Binomial tetramerization models suggested that KV1.4 homotetramers and 3:1 and 2:2 KV1.4:KV1.2(F233S) heterotetramers can traffic (cyan curve), anticipating also some KV1.4 sequestration (no 1:3 heterotetramers). While KV1.4 conductance amplitude did not clearly depend on F233S dosage, its voltage dependence did, exhibiting a progressive shift to more depolarized potentials with increasing F233S (Fig. 7 and ) and supporting that KV1.4/1.2(F233S) heterotetramers can traffic in oocytes. Note also the spontaneous decrease in current during depolarization (Fig. 7) caused by an N-terminal inactivation motif blocking the channel pore (37, 38).

Fig. 7.

KV1.4 and KV1.2(F233S) subunits form 3:1 and 2:2 heteromeric channels. (A) Relative peak conductance in cells injected with KV1.4 cRNA and increasing molar proportion of KV1.2(F233S) cRNA. Superimposed curves show the number of tetramers of specified composition relative to the 0-F233S condition, generated by a model assuming binomially distributed tetramerization. (B) Fitted Boltzmann distributions of peak macroscopic conductance from the same cells as in A. Data, parameters, and individual curves in . The potential of 20% activation (V0.2) best shows the F233S-dependent shift in voltage dependence on the right. Error bars are ±95% CI. (C) Representative TEVC current traces demonstrate altered voltage-dependent properties in heteromeric channels. (D) Representative COVG current traces from cells injected with dimeric KV1.4/KV1.2 concatemer cRNA. Relative macroscopic conductance (E and F) and voltage dependence (G) of cells injected with 1.4-1.2(WT) (relative Gtotal = 1.0 ± 0.29; V0.5 = −4.8 ± 1.3 mV; zeff = 1.6 ± 0.038 e0); 1.4-1.2(F233S) (relative Gtotal = 1.1 ± 0.23; V0.5 = 23 ± 1.2 mV; zeff = 1.2 ± 0.027 e0); 1.2(WT)-1.4 (relative Gtotal = 1.0 ± 0.15; V0.5 = −8.2 ± 0.95 mV; zeff = 1.9 ± 0.12 e0); 1.2(F233S)-1.4 (relative Gtotal = 0.062 ± 0.012; V0.5 = 24 ± 1.2 mV; zeff = 1.8 ± 0.044 e0). Conductance for 1.4-1.2 constructs, which exhibit fast inactivation, was calculated at peak current. Errors are SEM.

As before, we used dimeric concatemers to evaluate the trafficking and biophysical properties of the 2:2 KV1.4:KV1.2 stoichiometry. KV1.4-KV1.2(F233S) constructs exhibited full conductance compared with KV1.4-KV1.2(WT) (P = 0.76; Fig. 7 ), supporting that the 2:2 KV1.4:KV1.2(F233S) stoichiometry is trafficking capable. However, the inverted construct, KV1.2(F233S)-KV1.4, exhibited strongly diminished conductance (Fig. 7 ). The significance of this result for the mechanism of F233S-caused trafficking impairment is discussed below. Both KV1.2(F233S)-containing constructs exhibited right-shifted voltage dependence compared to their WT counterparts (Fig. 7). Finally, our constructs with a C-terminal KV1.4 lacked fast inactivation (Fig. 7). Since the inactivation particle is located at the KV1.4 N terminus (37, 38), perhaps it cannot block the channel when N-terminally tethered.

Operation of the KV1.2(F233S) VSD.

We exploited the rescue of KV1.2(F233S) subunits by WT KV1.2 and KV1.4 to investigate the function of the F233S-bearing VSD. To selectively track the activation of F233S-bearing subunits, we used voltage-clamp fluorometry (VCF) (39, 40). In this approach, a Cys is substituted at a position expected to undergo conformational changes [A291C, at the extracellular flank of KV1.2 S4 (8, 41)]. The Cys is then labeled with a thiol-reactive, environment-sensitive fluorophore, such as MTS-TAMRA. Thus, changes in ensemble fluorescence (ΔF) report VSD activation transitions. KV1.2(A291C) channels showed robust downward ΔF signals (Fig. 8), reporting VSD activation upon membrane depolarization and its rapid deactivation upon repolarization. By only incorporating the A291C substitution in F233S-bearing subunits, only the movements of F233S-containing VSDs are optically reported; the VSDs of rescuing KV1.2 and KV1.4 subunits also undergo voltage-dependent movements, but they are not fluorescently labeled and therefore their movements are not optically reported (). These experiments revealed that the rescued KV1.2(F233S) VSD activates with a modestly affected voltage dependence, as well as near-normal valence and activation kinetics (Fig. 8 ).

Fig. 8.

Operation of the KV1.2(F233S) VSD. (A) VCF experiments on KV1.2 homotetramers, fluorescently labeled outside the S4 helix (A291C) to optically track the voltage-dependent activation of the VSD (Fig. 1) Vm, Im (black), and simultaneously acquired fluorescence deflections (ΔF, red). VSD activation causes fluorophore quenching, reported as negative ΔF (8). (B and C) As in A, but the fluorescence label is in KV1.2(F233S) subunits rescued by WT KV1.2 (B) or KV1.4 (C). Note that VSD deactivation is delayed in C (interpreted below). (D) Normalized macroscopic conductance (G, black) and VSD activation (ΔF, red), fit to Boltzmann distributions. KV1.2(A291C) G: V0.5 = 14 ± 2.8 mV, zeff = 1.6 ± 0.14 e0; ΔF: V0.5 = −50 ± 1.8 mV, zeff = 1.3 ± 0.22 e0; n = 4. KV1.2/1.2(F233S,A291C) G: V0.5 = −0.89 ± 2.4 mV, zeff = 1.6 ± 0.074 e0; ΔF: V0.5 = −64 ± 4.0 mV, zeff = 1.2 ± 0.11 e0; n = 6. KV1.4/1.2(F233S,A291C) G: V0.5 = 16 ± 0.75 mV, zeff = 1. 1 ± 0.039 e0; ΔF: V0.5 = −56 ± 1.0 mV, zeff = 1.6 ± 0.13 e0; n = 6. Errors are SEM. Note that the KV1.2(A291C) conductance (black filled circles) is right shifted by ∼15 mV compared to WT due to the A291C mutation (8); the KV1.2/1.2(F233S,A291C) conductance (black open circles) has a large contribution from WT homotetramers, so it is relatively left shifted. (E and F) Interpretation of data from KV1.4/1.2(F233S,A291C) heterotetramers (C). The diagrams in E show pore and VSD states in a hypothetical KV1.4/1.2(F233S,A291C) heterotetramer (only two subunits shown). Numbered transitions are registered on the exemplary current and fluorescence traces in F. (E) Top Left: The Vm is negative; the VSDs are resting (label in high-fluorescence state), and the pore is closed (no current). Upon depolarization (transition 1) VSDs activate and the channel opens (Top Right), reported as current onset and downward ΔF in F. Next (transition 2), KV1.4 inactivation particles block the channels (Bottom Right), reported as current reduction. Following membrane repolarization, the inactivation particle persists in the pore, preventing its closure (Bottom Left). It takes time for the particle to dissociate (transition 3), delaying VSD deactivation (off-charge immobilization), reported as slow ΔF recovery. In the absence of an inactivation particle, the VSD deactivates with fast kinetics ().

KV1.4 subunits confer a particular advantage to this experimental paradigm, enabling the direct observation of VSD-pore coupling. Their unique (in the KV1 family) N-terminal inactivation particle binds to, and blocks, the open pore (N-type inactivation, Fig. 7) (37). In doing so, the inactivation particle prevents pore closure (42, 43), which in turn impedes the return of the VSD to its resting conformation. This phenomenon, termed off-charge immobilization, persists until the inactivation particle dissociates and the pore closes. It is experimentally observable in gating-current (44, 45) and optical (46) experiments.

Accordingly, in the presence of KV1.4 subunits, and at depolarizations that evoked pore opening (and N-type inactivation), KV1.2(F233S) VSD deactivation was markedly slower (Fig. 8), consistent with off-charge immobilization. This indicates that the rescued KV1.2(F233S) VSDs were functionally coupled to the state of the channel pore (Fig. 8 ). When KV1.2(F233S) subunits were rescued by KV1.4 lacking an inactivation particle, their VSD deactivated quickly ().

Discussion

In this work, we report on a KCNA2 variant discovered in a male child with epilepsy (Fig. 1 ). The protein product of KCNA2, KV1.2 subunits of voltage-gated potassium channels are expressed throughout the brain as homotetrameric channels (Fig. 1) or participating in heterotetrameric complexes with other KV1 channel subunits (47), where they regulate neuronal excitability and synaptic release. The KCNA2 variant studied here results in the substitution of a residue (F233S) at the core of the bundle of transmembrane helices comprising the channel VSD (Fig. 1). This mutation completely abolished the surface trafficking of KV1.2(F233S) subunits (Fig. 2). We also discovered that the mutant subunits are dominant negative, suppressing the surface trafficking of their WT allele (Figs. 3 and 4 and ) but also KV1.4, a known molecular partner of KV1.2 (Fig. 6). Yet the severe trafficking defect of KV1.2(F233S) subunits was ameliorated in the presence of WT KV1.2 and KV1.4 subunits, demonstrated by both surface labeling (Figs. 4 and 6 and ) and electrophysiological evidence (Figs. 5 and 7). This allowed the direct observation of the mutant VSD activation (Fig. 8 and ). Our work classifies this previously unreported KCNA2 variant as a dominant-negative mutation that leads to loss of KV1.2 channel function as a phenotype and could also decrease the function of other KV1-family molecular partners.

How Does F233S Affect KV1.2 Functional Properties?

F233 has an eminent role in voltage sensing as the charge transfer center (Fig. 1) (11). How does its substitution affect channel function? Since KV1.2(F233S) homotetramers could not traffic (Fig. 2), it was not possible to directly determine their voltage dependence. In the Shaker K+ channel, the archetype for functional KV studies, some VSD mutations have been shown to perturb voltage dependence proportionally to the number of affected subunits (29, 48). KV1.2(WT/F233S) concatemers, which represent the 2WT:2F233S stoichiometry, exhibit a depolarizing shift (ΔV0.5) of 24 mV, so we may infer that KV1.2(F233S) homotetramers would exhibit a 48-mV shift. The equivalent mutation in Shaker(F290S) induced a stronger ΔV0.5 of ∼80 mV (11).

The four VSDs in a KV channel are thought to activate independently by membrane depolarization (early transitions) and then undergo a concerted late transition that leads to pore opening (49). In Shaker, mutations that disrupt VSD-pore coupling result in depolarizing shifts of late VSD transitions and pore opening (49), an effect recapitulated by F290S (13). This was not observed in the VSD of KV1.2(F233S) (Fig. 8), while off-charge immobilization (in the presence of KV1.4) strongly indicates functional VSD-pore coupling (Fig. 8 and ). However, the requisite association of KV1.2(F233S) with rescuer subunits may have rectified its late VSD transitions, occurring in concert with the movements of WT VSDs in the heterotetrameric channels.

Our view is that charge-transfer-center disruption has similar effects on the two channels, albeit milder in KV1.2. Previous work indicates that KV1.2 has a thicker hydrophobic septum than Shaker (12), perhaps rendering its VSD more resilient to charge-transfer-center disruption.

Mechanism of F233S-Mediated Trafficking Impairment.

Why does the cellular protein biosynthesis and secretory machinery prevent KV1.2(F233S)-subunit expression? We propose that the F233S mutation hinders trafficking at two distinct stages: "early" and "late". The former likely include processes such as translation, helix formation, and translocation, thought to occur early in channel biosynthesis (50). The latter likely involve the trafficking of translocated subunits and their interaction with ER proteins.

Starting with the KV1.4/KV1.2(F233S) concatemer experiments, the construct with an N-terminal KV1.4 subunit is trafficking capable, whereas the construct with an N-terminal KV1.2(F233S) is trafficking impaired (Fig. 7 ). The only difference between the two is the order in which the two domains are synthesized and introduced to the cell secretory pathway. In related KV1.3 channels, helix S2 functions as the signal for translocation and establishment of N terminus topology (51). KV1.3 S2 helix formation depends on the presence of a WF motif in the middle of the helix (W232 and F233 in KV1.2) that prevents premature S2 helical formation (52). As this motif is highly conserved across KV channel subfamilies, its helical-suppressor function is likely crucial to translocation and could be a major first impediment for the functional expression of N-terminal KV1.2(F233S) peptides, whether they are monomers (Fig. 2) or concatenated with WT KV1.2 (Fig. 5 ) or KV1.4 (Fig. 7 ). Aberrant KV1.2(F233S) translocation may in turn interfere with other processes important for trafficking, such as electrostatic interactions within the VSD (53) or glycosylation at the S1-S2 linker (54, 55).

By contrast, when KV1.2(F233S) is the C-terminal partner of KV1.4, trafficking is not affected (Fig. 7 ). In this construct, the N-terminal KV1.4 likely undergoes a successful translocation first. Why does the C-terminal KV1.2(F233S) not oppose translocation? We suggest that this is evidence of cooperative translocation. It is possible that such a mechanism (a capable N-terminal partner assisting the translocation of downstream domains) is employed by natural concatemers, such as two-pore channels and pseudotetrameric voltage-gated calcium and sodium channels (56).

Why does N-terminal WT KV1.2, then, not facilitate the trafficking of C-terminal KV1.2(F233S) (Fig. 5 )? Likely it does, but trafficking is then halted by later events, unrelated to biosynthesis, folding, and translocation. These may include posttranslational modification and interaction with the ER protein quality-control apparatus. KV1.2 subunits have ER retention motifs which KV1.4 lack; additionally, KV1.4 possess forward trafficking signals (33–35). These elements could underlie the successful trafficking of 3:1 and 2:2 KV1.4/KV1.2(F233S) heterotetramers and KV1.4-KV1.2(F233S) concatemers, but not 2:2 KV1.2(WT)/KV1.2(F233S) heterotetramers or the KV1.2(WT-F233S) concatemers. It appears that inclusion of trafficking-deficient KV1.2(F233S) subunits in heterotetramers imposes a quantal trafficking penalty, abated by trafficking-enhancing elements, as in KV1.4.

In conclusion, we propose this model for KV1.2(F233S)-subunit trafficking: Nascent peptides tetramerize early (28), assembling into homo- and heterotetramers. KV1.2(F233S) homotetramers are immediately excluded, encountering an early translocation impediment and receiving no assistance from WT partner subunits. This early block may also prevent heterotetramers with three KV1.2(F233S) subunits from trafficking; alternatively, these are blocked later in the ER. Heterotetramers with two KV1.2(F233S) translocate successfully but then have different fates, depending on the trafficking adeptness of their partner subunits: those with KV1.2 are blocked (Fig. 5), whereas those with KV1.4 traffic successfully (Fig. 7).

By comparison, Shaker(F290S) channels were reported to have "medium current level" (11) and sufficient expression for the measurement of gating currents (13). We find it interesting that while the effects of F→S are similar in the two channels, their severities are inverted: strong functional impairment and modest trafficking deficiency in Shaker, and vice versa for KV1.2. This disparity indicates that the trafficking and functional defects are not strictly linked. Since F233/F290 is critical for two altogether different functions—S2-helix formation and translocation of the nascent polypeptide (52) and voltage sensing in the mature protein (11, 13)—these could be differentially modified by other parts of the KV1.2 and Shaker channels.

There is still much to discover about the identity of early and late events of KV1.2(F233S) trafficking impairment. We are certain that this variant, serendipitously discovered in an epilepsy patient, can serve as a tool to investigate the fundamental aspects of channel biogenesis, trafficking, and oligomerization.

Neuronal Consequences of a Dominant-Negative KCNA2 Variant.

The ability of KV subunits to form heteromeric channels is a major source of conductance diversity in the brain. However, it can act as a double-edged sword, as trafficking deficiencies in one gene can have multigenic dominant-negative effects. As the two somatic KCNA2 alleles are presumably under the same transcriptional control, our experiments suggest that that the neurons of the heterozygote patient will have ∼20% KV1.2 conductance (Fig. 3 ). Effects on other subunits will depend on their relative expression levels. We posit that epileptogenesis is mainly driven by KV1.2 loss of function, with some KV1.4 decrease in neurons where these genes are coexpressed. KV1.1 are also likely suppressed by KV1.2(F233S), as they traffic less proficiently than KV1.2 or KV1.4 due to their potent ER retention signal (35). Loss of KV1.2 function should prolong action potentials in both excitatory and inhibitory neurons. In the former, this can cause increased synaptic release (4). On the other hand, fast-gating KV3 channels are more abundant in inhibitory neurons (57–59), where they delimit action potential duration (60, 61); thus, some inhibitory synaptic release is protected from loss of KV1.2. The resulting imbalance of excitatory and inhibitory signals could veer neuronal circuits toward epilepsy (62).

Methods

This study was reviewed and deemed exempt by the Children’s Hospital Los Angeles Institutional Review Board. A consent form by Children's Hospital Los Angeles compliant with the Health Insurance Portability and Accountability Act of 1996 (HIPAA) was signed by the patient's parents. DNA was extracted from the peripheral blood of the patient and each of his unaffected parents as comparators using a commercially available kit (Promega Maxwell RSC DNA Extraction Kit). Original trafficking-assay and KV1.4 plasmids were described in (27) and (63), respectively. Confocal microscopy, cut-open oocyte Vaseline gap (64), and VCF (65) experiments were performed as previously described (8). For some flow cytometry experiments, the methodology [previously described in Pantazis et al. (8)] was expanded to four colors to evaluate synthesis and surface trafficking of two constructs simultaneously. Two-electrode voltage-clamp (TEVC) experiments were performed in an OpusXpress 6000A (Axon Instruments) parallel voltage clamp to facilitate testing of a large number of cells.