Loss-of-function variants in the KCNQ5 gene are implicated in genetic generalized epilepsies

Abstract

Background: De novo missense variants in KCNQ5, encoding the voltage-gated K⁺ channel K_V7.5, have been described to cause developmental and epileptic encephalopathy (DEE) or intellectual disability (ID). We set out to identify disease-related KCNQ5 variants in genetic generalized epilepsy (GGE) and their underlying mechanisms.

Methods: 1292 families with GGE were studied by next-generation sequencing. Whole-cell patch-clamp recordings, biotinylation and phospholipid overlay assays were performed in mammalian cells combined with homology modelling.

Findings: We identified three deleterious heterozygous missense variants, one truncation and one splice site alteration in five independent families with GGE with predominant absence seizures; two variants were also associated with mild to moderate ID. All missense variants displayed a strongly decreased current density indicating a loss-of-function (LOF). When mutant channels were co-expressed with wild-type (WT) K_V7.5 or K_V7.5 and K_V7.3 channels, three variants also revealed a significant dominant-negative effect on WT channels. Other gating parameters were unchanged. Biotinylation assays indicated a normal surface expression of the variants. The R359C variant altered PI(4,5)P₂-interaction.

Interpretation: Our study identified deleterious KCNQ5 variants in GGE, partially combined with mild to moderate ID. The disease mechanism is a LOF partially with dominant-negative effects through functional deficits. LOF of K_V7.5 channels will reduce the M-current, likely resulting in increased excitability of K_V7.5-expressing neurons. Further studies on network level are necessary to understand which circuits are affected and how this induces generalized seizures.

Funding: DFG/FNR Research Unit FOR-2715 (Germany/Luxemburg), BMBF rare disease network Treat-ION (Germany), foundation 'no epilep' (Germany).

Keywords: Exome sequencing; Genetic generalized epilepsy; KCNQ5; Loss-of-function; Patch-clamp.

Figure 1

Variants affecting the K_v7.5 potassium channel. (A) Pedigrees of patients with individuals from Table 1 indicated by number. (B) Schematic of the K_v7.5 subunit of which four assemble to form a channel. Each subunit consists of a voltage-sensor domain (S1-S4) and a pore forming region (S5 and S6) including the pore loop. Point mutations are localized in highly conserved regions of the C-terminus (R359C, L692V, Q735R), while a splice site variant causes a deletion of the S5 segment and parts of the pore forming loop (E265_T306del) and a frame-shift mutation leads to an early stop codon in the pore loop (A301Gfs*64). (C) Amino acid alignment across multiple species (top) and across all human K_V7 family members (bottom) shows evolutionary conservation of R359, L692, Q735 and their surrounding amino acids in K_V7.5. Species from top to bottom: human, mouse, rat, Carolina anole, macaque, pig. ID = intellectual disability, Abs = absence seizure, Myo = myoclonic seizure, GTCS = generalized tonic clonic seizure, FebS = febrile seizure, asymp = asymptomatic.

Figure 2

Functional effects of K_v7.5 WT and mutant channels in Chinese hamster ovarian cells. (A) Representative K⁺ current traces from KCNQ5 WT (black), R359C (green), L692V (orange), Q735R (blue) and untransfected control cells (CTRL, yellow) during voltage steps from -80 mV to +60 mV in 10 mV increments. (B) Peak K+ currents of cells either transfected with WT or one mutated channel subunit were normalized by cell capacitances and plotted versus voltage. All variants result in a significant reduction in current density compared to the WT. WT, n = 11; R359C, n = 10; L692V, n = 10; Q735R, n = 10; CTRL, n = 10. (C) Comparison of maximum peak current density at +60 mV. All variants show a significant reduction compared to the WT. (D) Voltage-dependent activation curves. Lines represent Boltzmann functions fit to the normalized tail current. Currents in the R359C variant were too small to establish such a relationship. (E) Peak K+ currents normalized by cell capacitances and plotted versus voltage of cells either transfected with WT (1 µg) or WT and one mutated channel subunit (1 µg + 1 µg). The significant reduction persisted in all variants compared to the WT indicating a dominant negative effect of the variants on the WT. WT, n = 12; WT + R359C, n = 10; WT + L692V, n = 10; WT + Q735R, n = 10; CTRL, n = 10. (F) Comparison of maximum peak current density at +60 mV. All variants show a significant reduction compared to the WT. (G) Voltage-dependent activation curves. Lines represent Boltzmann functions fit to the normalized tail current. Currents of the WT + R359C were still too small to establish such a relationship. (H) Peak K+ currents normalized by cell capacitances and plotted versus voltage of cells either transfected with WT (1 µg) or WT and one mutated channel subunit (1 µg + 1 µg) in a CHO line stably transfected with KCNQ3-WT. The dominant negative effect of the variants on the WT and the significant reduction persisted in all variants compared to the WT. WT, n = 13; R359C, n = 10; L692V, n = 10; Q735R, n = 10; CTRL, n = 10. (I) Comparison of maximum peak current density at +60 mV. All variants show a significant reduction compared to the WT. (J) Voltage-dependent activation curves. Lines represent Boltzmann functions fit to the normalized tail current. Shown are means ± SEM (B, D, E, G, H, J). Scatter-and-whisker plots (C, F, I) show median (horizontal line) and the interquartile ranges. Dots indicate maximum values of single cells. *p ≤ 0·05; ** p ≤ 0·01; *** p ≤ 0·001; **** p ≤ 0·0001; Table 2 provides exact values and statistical analyses.

Figure 3

Functional effects of K_v7.5 WT and mutant channels in Chinese hamster ovarian cells (+ antibiotics cohort). (A) Representative K⁺ current traces from KCNQ5 WT (black) and c.901dupG (turquoise) during voltage steps from -80 mV to +60 mV in 10 mV increments. (B) Peak K+ currents of cells either transfected with WT or c.901dupG channel subunit were normalized by cell capacitances and plotted versus voltage. WT, n = 70; 901dupG, n = 16. (C) Comparison of maximum peak current density at +60 mV. c.901dupG results in a significant decrease in current density compared to the WT. (D) Peak K+ currents normalized by cell capacitances and plotted versus voltage of cells either transfected with WT (2·5 µg) or WT and 901dupG (2·5 µg + 2·5 µg). No significant differences were observed. WT, n = 40; WT+901dupG, n = 27. (E) Comparison of maximum peak current density at +60 mV. No significant differences were observed. (F) Voltage-dependent activation curves. Lines represent Boltzmann functions fit to the normalized tail current. Shown are means ± SEM (B, D, F). Scatter-and-whisker plots (C and E) show median (horizontal line) and the interquartile ranges. Dots indicate maximum values of single cells. **** p ≤ 0·001; ns non-significant; Table 2 provides exact values and statistical analyses.

Figure 4

Western blot analysis of KCNQ5 expression and phospholipid binding abilities in CHO cells. (A) Western blot of CHO cell lysates of transiently transfected cells (20µg per lane; n = 3). Controls consisted of untransfected CHO cells. (B) Comparison of K_V7.5-WT expression to variants showed no significant changes (n = 3). (C) Western blot of K_V7.3 and K_V7.5 expression in the K_V7.3 stable cell line (n = 4). (D) Biotinylation assays of transfected CHO cells for cell surface expression analysis (n = 3). (E) No significant changes in cell surface expression between K_V7.5-WT and K_V7.5 variants were observed (n = 3). Controls consisted of untransfected CHO cells. (F) Representative PIP strips of the phospholipid interaction of WT versus R359C (n = 3). 1 = Lysophosphatic acid, 2 = Lysophosphatidylcholine, 3 = Phosphatidylinositol, 4 = Phosphatidylinositol 3-phosphate, 5 = Phosphatidylinositol 4-phosphate, 6 = Phosphatidylinositol 5-phosphate, 7 = Phosphatidylethanolamine, 8 = Phosphatidylcholine, 9 = Sphingosine 1-phosphat, 10 = Phosphatidylinositol 3,4-bisphosphate, 11 = Phosphatidylinositol 3,5-bisphosphate, 12 = Phosphatidylinositol 4,5-bisphosphate, 13 = Phosphatidylinositol (3,4,5)-trisphosphate, 14 = Phosphatidic acid, 15 = Phosphatidylserine, 16 = blank. (G) Quantitative analysis of interactions on PIP strips (n = 3) using Student's unpaired t-test. ns non-significant; *p ≤ 0·05; ** p ≤ 0·01; *** p ≤ 0·001; **** p ≤ 0·0001.

Figure 5

Model predictions for K_V7.5-WT and K_V7.5-R359C and their interaction with PI(4,5)P₂. (A) Two K_V7.5 consensus homology models were generated and K_V7.5-R359C was introduced in each model. PIP₂ molecules were positioned in the K_V7.5-WT and K_V7.5-R359C models in the virtually identical position as described for the K_V7.4-WT structure 7VNP.pdb.In silico, two PIP₂ molecules bind to two distinct sites per K_V7.5 subunit (PIP₂-A is shown in magenta and PIP₂-B is colored in cyan). The position of residues R359 are encircled by a red square (upper right), whereas each two R359 are positioned opposite in a tetrameric assembly and allow for calculation of two Cα cross distances in a membrane parallel plane (cartoon right). (B) Cα-cross distances in both 180° tilted directions were calculated resulting is two values per simulation. The cross distance of residue 359 tends to be slightly, however not significantly (Student's paired t-test), larger in K_V7.5-R359C in absence of PIP₂ in the simulations compared to the K_V7.5-WT protein. (C) The more tight a PIP₂ molecule binds the more restricted is its flexibility is which can be calculated as Root Mean Square Fluctuation (RMSF). The RMSF of PIP₂ molecules -A and -B in the K_V7.5-WT(filled symbols) and K_V7.5 R359C (open symbols) channel complexes were calculated per channel subunit for both simulation approaches (simulation 1 in red and simulation 2 in black). In all simulations, both PIP₂ binding sites for K_V7.5-R359C mutant channels showed lower RMSF values compared to K_V7.5-WT channels in silico (Paired Student's t-test, p= 0.0006). (D) The cross distance of K_V7.5-R359 is significantly (Student's paired t-test) larger in presence of PIP₂ (p=0.041) whereas this effect is not detected in the K_V7.5-R359C.

Figure 6

Functional effects of PIP₂ overexpression in K_V7.5 WT vs. R359C channels in CHO cells. (A) Representative K⁺ current traces from K_V7.5-WT + PIP5K (black), K_V7.5-R359C + PIP5K (green), and PIP5K control cells (black) during voltage steps from -80 mV to +60 mV in 10 mV increments. (B) Peak K⁺ currents of cells either transfected with WT alone (grey dots), WT + PIP5K (black), R359C (grey triangles) or R359C + PIP5K (green triangle) channel subunit were normalized by cell capacitances and plotted versus voltage. The currents generated by K_V7.5-R359C + PIP5K remain largely reduced as compared to K_V7.5-WT + PIP5K. K_V7.5-WT, n = 12; K_V7.5-WT + PIP5K, n = 10; K_V7.5-R359C, n = 10; K_V7.5-R359C + PIP5K, n = 10; PIP5K, n = 10; CTRL n = 10. (C) Comparison of maximum peak current density at +60 mV. PIP5K co-expression significantly increased the K_V7.5-R359C peak current density, yet it is still significantly reduced as compared to K_V7.5-WT + PIP5K. (D) Voltage-dependent activation curves. Lines represent Boltzmann functions fit to the normalized tail current. The activation curve for K_V7.5-R359C + PIP5K is significantly shifted towards more positive voltages. Shown are means ± SEM (B, D). Scatter-and-whisker plots (C) show median (horizontal line) and the interquartile ranges. Dots indicate maximum values of single cells. **** p ≤ 0·001; **p ≤ 0·01; * p ≤ 0·05; ns non-significant; Table 3 provides exact values and statistical analyses.

Figure 7

Functional effects of PIP₂ depletion in K_V7.5-WT vs. K_V7.5-R359C channels in CHO cells. (A) Representative K⁺ current traces from WT (control), K_V7.5-WT + VSP, K_V7.5-WT + VSP + PIP5K (all black), K_V7.5-WT + R359C + VSP, and K_V7.5-WT + R359C + VSP + PIP5K (both green) cells responding to the displayed voltage protocol. The dotted line indicates 0 pA. (B) Time dependence of current decrease under VSP activation in the absence (WT black dots, R359C green triangles; n = 5 for both) or presence of PIP5K (WT grey dots, R359C grey triangles; n = 5). These values were calculated by normalizing the values immediately after the +100 mV step to those immediately prior to it and a Boltzmann function was fit to the data points. (C) Time dependence of current recovery after VSP activation in the absence (WT black dots, R359C green triangles; n = 5 for both) or presence of PIP5K (WT grey dots, R359C grey triangles; n = 5). These values were calculated by normalizing the values every second after the +100 mV step to that immediately prior to it. Shown are means ± SEM (B, C).