KCNQ2 and KCNQ5 form heteromeric channels independent of KCNQ3

Abstract

KCNQ2 and KCNQ3 channels are associated with multiple neurodevelopmental disorders and are also therapeutic targets for neurological and neuropsychiatric diseases. For more than two decades, it has been thought that most KCNQ channels in the brain are either KCNQ2/3 or KCNQ3/5 heteromers. Here, we investigated the potential heteromeric compositions of KCNQ2-containing channels. We applied split-intein protein trans-splicing to form KCNQ2/5 tandems and coexpressed these with and without KCNQ3. Unexpectedly, we found that KCNQ2/5 tandems form functional channels independent of KCNQ3 in heterologous cells. Using mass spectrometry, we went on to demonstrate that KCNQ2 associates with KCNQ5 in native channels in the brain, even in the absence of KCNQ3. Additionally, our functional heterologous expression data are consistent with the formation of KCNQ2/3/5 heteromers. Thus, the composition of KCNQ channels is more diverse than has been previously recognized, necessitating a re-examination of the genotype/phenotype relationship of KCNQ2 pathogenic variants.

Keywords: KCNQ2; KCNQ3; KCNQ5; autism; epilepsy.

Fig. 1.

KCNQ2 and KCNQ5 protein trans-splicing leads to functional KCNQ2/5 heteromeric channels. (A) Illustration of the strategy for creating tandem KCNQ2/3 and KCNQ2/5 subunits. The KCNQ2 channel N-terminus was tagged with the DnaE-c fragment, whereas the KCNQ3 and KCNQ5 C-terminus were tagged with the DnaE-n fragment. (B) Coexpression of KCNQ2-ER-n with either KCNQ3-ER-c or KCNQ5 ER-c formed tandems in cells. HEK293T cells transfected in a 1:1 KCNQ2-ER-n: KCNQ3-ER-c or KCNQ2-ER-n: KCNQ5 ER-c ratio led to the formation of a higher molecular band (*) due to linking of the two subunits. Note that the formation is not complete as a substantial amount of monomeric KCNQ3-ER-c and KCNQ5 ER-c remains (black arrow indicates monomers). (C) Top, representative traces from cells expressing KCNQ2-ER-n, KCNQ5-ER-c, or KCNQ2-ER-n and KCNQ5-ER-c together (KCNQ2/5-tandem). Top Center and Right, representative traces from cells expressing KCNQ2-ER-n and KCNQ5-ER-c alone (KCNQ2/5-tandem) or with PIP5K and calmodulin (Calm). Bottom Left, summary graphs showing the conductance-to-voltage relationship of KCNQ2/5-tandem channels alone (V_0.5 = −11.1 ± 1.9 mV, n = 23), with PIP5K (V_0.5 = −31.5 ± 2.5 mV, n = 6; ****P = 1.4e10⁻⁷), or calmodulin (Calm; V_0.5 = −8.6 ± 2.2 mV, n = 22; **P = 0.002). Bottom Right, summary graphs of KCNQ2/5-tandem and KCNQ2/3-tandem current densities. Note that coexpressing KCNQ2/5-tandem subunits with either PIP5K or Calm increases their current density. Data are displayed as box plots showing the SD, median (solid line), and mean (square) values. Current densities were calculated from the tail currents recorded at −55 mV following a test pulse. One-way ANOVA, F(22.89), P = 2.2e10⁻⁸. Displayed P values were determined using Tukey post hoc test. (D) Left, representative recordings from cells expressing either KCNQ2/5-tandem or KCNQ2/3-tandem channels in the presence or absence of 10 mM TEA at 0 mV. Right, summary graph showing the blocking effect of 10 mM TEA in KCNQ channels with different subunit compositions. Data are displayed as mean ± SEM.

Fig. 2.

Properties of KCNQ2/3/5 heteromeric channels. (A) Left, representative recordings from cells expressing either KCNQ2/5-tandem subunits or KCNQ2/5-tandem subunits along with KCNQ3. Right, summary graphs comparing the conductance-to-voltage relationship of KCNQ2/5-tandem channels (the curve fit from Fig. 1C is illustrated for comparison), KCNQ3 channels (V_0.5 = −21.7 ± 3.2 mV, n = 11), and KCNQ2/5-tandem/KCNQ3 channels (V_0.5 = −14.4 ± 1.7 mV, n = 18). Note that coexpression of KCNQ2/5-tandem subunits with KCNQ3 led to a current-to-voltage relationship similar to one obtained for KCNQ2/5-tandem channels. Data are displayed as mean ± SEM. Inset, summary graph comparing KCNQ3 to KCNQ2/5-tandem/KCNQ3 current densities. Significance was determined using the Mann–Whitney U test (**P = 0.0063). (B) Left, representative recordings from cells expressing either KCNQ2/5-tandem subunits or KCNQ2/5-tandem subunits with KCNQ3 in the presence or absence of 10 µM ICA27243. Right, summary graph showing that ICA27243 leads to a smaller shift in the V_0.5 of KCNQ2/5-tandem/KCNQ3 (n = 5) channels compared to KCNQ2/5-tandem channels (n = 6). Statistical significance was determined using the Mann–Whitney U test (*P = 0.036). For illustration purposes, we also show the effect of 10 µM ICA27243 to unlinked KCNQ2 channels (n = 6). (C) Representative traces from HEK293T cells expressing either KCNQ3^R230C subunits or KCNQ2/5-tandem and KCNQ3^R230C subunits. Right, summary graph showing that coexpression of KCNQ2/5-tandem subunits with KCNQ3^R230C leads to a right shifted conductance-to-voltage relationship (KCNQ3^R230C V_0.5 was not determined as KCNQ3^R230C channels were constitutive open and could not be fitted with a Boltzmann function, n = 4; KCNQ2/5-tandem/KCNQ3^R230C, V_0.5 = −85 ± 5.8 mV, n = 6). Data are displayed as mean ± SEM.

Fig. 3.

Development and characterization of Kcnq2 epitope-tagged mice. (A) Top, illustration showing the location of the epitope tag on the KCNQ2 channels and Kcnq2 gene. Bottom Left, representative recordings from cells expressing either wild-type KCNQ2 channels or 3XFLAG-tagged KCNQ2 channels. Bottom Right, summary graph showing that introduction of the 3XFLAG epitope does not change the properties of the KCNQ2 channels (wild-type V_0.5= −13.7 ± 2.8 mV, n = 4; 3XFLAG-KCNQ2 V_0.5 = −11.9 ± 2.6 mV, n = 5). Data are displayed as mean ± SEM (B) Left, representative recordings from either control CA1 pyramidal neurons or neurons expressing the 3XFLAG-tagged KCNQ2 channels. The membrane potential was held at −65 mV, maintained with small DC current injections. Right, summary graph showing that introduction of the 3XFLAG epitope does not change the firing properties of CA1 pyramidal neurons (Kcnq2^+/+ n = 8, 2 mice; Kcnq2^flag/flag n = 7; 2 mice). Data are displayed as mean ± SEM (C) Coronal sections showing localization of the 3XFLAG-tagged KCNQ2 channels in axons. Slices were stained with either DAPI (blue) or an anti-FLAG M2 antibody.

Fig. 4.

KCNQ2 and KCNQ5 form a complex in the brain. (A) Representative sequence coverage of KCNQ2 channels from anti-FLAG (M2 antibody) immunoprecipitated proteins followed by mass spectrometry analysis. Yellow indicates recovered peptides. (B) Total nonnormalized spectral counts of KCNQ2, KCNQ3, and KCNQ5 channels identified in Kcnq2^+/+;Kcnq3^+/+ (cortex, n = 7; hippocampus, n = 4), Kcnq2^flag/flagKcnq3^+/+ (cortex, n = 11; hippocampus, n = 8), and Kcnq2^flag/flag;Kcnq3^−/− (cortex, n = 5; hippocampus, n = 5). Data are displayed as box and whisker plots. (C) Volcano plot comparing calculated log₂ fold change using unnormalized spectral count values of different cortical proteins identified in Kcnq2^+/+;Kcnq3^+/+ (n = 7) and Kcnq2^flag/flag;Kcnq3^+/+ (n = 7) mice. (D) Immunoblot showing immunoprecipitated proteins from HEK293T cells expressing various 3XFLAG-KCNQ2, KCNQ3, and KCNQ5 channel combinations. Immunoprecipitation was performed using an anti-KCNQ5 antibody. n indicates number of mice.