To investigate the effects of mutant KCNQ2 in human neurons, we isolated peripheral blood mononuclear cells (PBMCs) from a 13-year-old female clinically diagnosed with KCNQ2-DEE. Clinical genetic sequencing revealed a de novo heterozygous KCNQ2 variant (c.1742G>A) resulting in an arginine to glutamine missense mutation at position 581 (p.Arg581Gln, R581Q; KCNQ2 splice variant 1: NM_172107.4; Figure 1A). The arginine residue is highly conserved across species, and this mutation has been observed in at least four other individuals diagnosed with KCNQ2-DEE (ClinVar variation #21769; Olson et al., 2017; Clinical Study Group et al., 2017). Moreover, mutation of this residue to other amino acids has been reported in at least five other cases with similar pathology (R581L, R581P, and R581G; ClinVar variations #205920, #804957 and #265380, respectively), as well as in three individuals with less severe BFNS (R581Ter; ClinVar variation #21768). R581Q is located in a highly conserved coiled-coil domain of the C-terminus referred to as the C-helix. Disruption of this coiled-coil domain abolishes the ability of KCNQ2 to form functional homomeric and heteromeric (KCNQ2/KCNQ3) channels and prevents transport to the plasma membrane (Schwake et al., 2006). Thus, it is possible that R581Q precludes channel conductance by disrupting tetramerization and trafficking of the channel complex in neurons. However, this KCNQ2 C-terminus mutation has not previously been characterized functionally.

To determine the effect of KCNQ2-R581Q on channel function, CHO cells that stably expressed KCNQ3 were transiently transfected with KCNQ2 containing the KCNQ2-R581Q mutation alone (homozygous state) or in a 1:1 ratio with wild-type KCNQ2 (heterozygous state). We found that KCNQ2-R581Q alone or combined with wildtype KCNQ2 produced 81.6% (t test: p<0.0001) and 56.6% (t test: p=0.006) smaller current density, respectively, as compared to cells expressing wild-type channels (Figure 1B and Figure 1—figure supplement 1). These findings are consistent with a KCNQ2 loss-of-function.

We derived iPSCs from PBMCs using integration-free, Sendai virus-mediated reprogramming. To create a model that would allow us to attribute any phenotypic differences to the disease-associated genetic variant, we generated an isogenic mutation-corrected control iPSC line from the patient-derived line (Figure 1C). We specifically corrected the mutant allele using CRISPR/Cas9 genome editing and simultaneously introduced two silent mutations in the protospacer adjacent motif (PAM) to prevent re-cleavage (Figure 1—figure supplement 2A–C). We validated the presence of the heterozygous mutation in the patient-derived iPSCs (Q2-04^R581Q/+) and identified a mutation-corrected isogenic clonal cell line (isoQ2-04^+/+) by targeted PCR and Sanger sequencing (Figure 1D). The resulting iPSCs exhibited a normal female karyotype, typical stem cell morphology, and expressed pluripotency markers, including nuclear NANOG and the cell surface antigen SSEA4 (Figure 1E and Figure 1—figure supplement 2D–E). We found no evidence for off-target edits in any of the top 10 predicted genomic regions with homology to the targeted KCNQ2 exon using a T7 endonuclease-based assay (Figure 1—figure supplement 3A and Supplementary file 1 and 2). As an additional quality control, we performed whole genome sequencing (WGS) of samples Q2-04^R581Q/+ (29.41x coverage) and isoQ2-04^+/+ (28.12x coverage) and determined that they correspond to the same individual (pi_hat score = 0.99). We also validated the genetic correction and integrity of the targeted gene (Figure 1—figure supplement 2C), and found that there were no edits in any of all 74 predicted off-target sites indicating highly specific CRISPR activity (Figure 1—figure supplement 3B, inner arrows, Supplementary file 3).

Given the clinical presentation of KCNQ2-associated DEE and the focal source of seizures that reside in the cortex (Wilmshurst et al., 2015; Fisher et al., 2017), we chose to study cortical excitatory neurons, differentiated through a modified NGN2 overexpression protocol (Figure 1—figure supplement 4A; Zhang et al., 2013; Nehme et al., 2018). We simultaneously differentiated the Q2-04^R581Q/+ patient-derived line, the engineered isogenic control line (isoQ2-04^+/+) and three iPSC lines generated from unrelated, healthy control individuals (Figure 1—figure supplement 4B; Boulting et al., 2011). Expression of GFP marked the lentivirus-transduced cells that were co-cultured with primary mouse glia to facilitate in vitro neuronal maturation. To determine the efficiency of differentiation, we used immunocytochemistry (ICC) to quantify the percentage of MAP2, vGLUT1, and GFP-positive cells (Figure 1F,G). Over 95% of GFP-positive cells were also MAP2-positive for all iPSC lines. Furthermore, all neurons that were GFP and MAP2 positive also expressed vGLUT1 (Figure 1F,G and Figure 1—figure supplement 4C). As previously described (Zhang et al., 2013), we found that these cultures expressed high levels of vGLUT2, FOXG1, and BRN2, which are characteristic of excitatory layer 2/3 cortical neurons (Figure 1—figure supplement 4D and Supplementary file 4). Importantly, we confirmed that the differentiated neuronal cultures expressed several KCNQ2 splice variants by RT-qPCR (Figure 1H).

One of the clinical hallmarks of KCNQ2-DEE is an interictal burst-suppression pattern (i.e. paroxysmal bursts of activity interspersed with periods of electrical silence) recorded by EEG during the first few days of life (Weckhuysen et al., 2012; Milh et al., 2013; Millichap et al., 2016). The KCNQ2 iPSC-based platform that we developed presents an opportunity to dissect the functional consequences of KCNQ2 mutations during neuronal development in vitro. To assess spontaneous neuronal activity, we used multi-electrode arrays (MEAs) and performed daily recordings over a 3-week period (days 15–31, N = 3 independent differentiations; Figure 2A,B). We plated an equal number of Q2-04^R581Q/+ and isoQ2-04^+/+ neurons and carefully monitored neuronal attachment throughout the time course of experiments (p=0.541; Figure 2A,C). As neurons matured, neuronal cultures acquired significant spontaneous activity (≥10% of the electrodes active) on or after day 15 (Figure 2—figure supplement 1A). Interestingly, Q2-04^R581Q/+ neurons exhibited a significantly greater number of active electrodes (p<0.0001) and slightly higher spontaneous firing frequency (p=0.0081) early in their maturation (DIV: 10-13), relative to isoQ2-04^+/+ neurons (Figure 2—figure supplement 1A,B). This suggests that Q2-04^R581Q/+ neurons are able to fire spontaneously earlier than isoQ2-04^+/+ neurons. However, the number of active electrodes was not different between Q2-04^R581Q/+ and isoQ2-04^+/+ neurons at later times in culture (DIV: 15-31) (p=0.0896; Figure 2—figure supplement 1A). More strikingly, the distribution of spiking in Q2-04^R581Q/+ neurons was dramatically and increasingly irregular compared to isogenic control neurons, indicated by the increased inter-spike interval coefficient of variation (ISI CoV; p<0.0001; Figure 2—figure supplement 1C). This irregularity was associated with greater bursting propensity, with short intervals between spikes in bursts and longer intervals between bursts.

While firing frequency was marginally higher in Q2-04^R581Q/+ neurons (p=0.0205; Figure 2—figure supplement 1B), mere hyperexcitability may not be an accurate recapitulation of the developmental deficits in epileptic encephalopathy (Simkin and Kiskinis, 2018). Thus, we focused on the neuronal firing pattern over time in culture rather than individual time points of higher firing frequency. As illustrated by representative spike raster plots, compared to isoQ2-04^+/+ neurons, Q2-04^R581Q/+ neuronal cultures fired significantly more bursts (p<0.0001) and had a greater number of bursting electrodes (p<0.0001) that increased in number faster, as neurons matured over time (repeated measures ANOVA interaction: p<0.0001, p=0.0052, respectively; Figure 2D and Figure 2—figure supplement 1D). Furthermore, their spikes were more restricted to bursts, as demonstrated by significantly higher ISI CoV (p<0.0001), number of spikes per burst (p<0.0001), percentage of all spikes that occurred within bursts (burst %; p<0.0001) and burst frequency (p=0.0072) as compared to isogenic control neurons (Figure 2D,E and Figure 2—figure supplement 1C–E). These results indicate that Q2-04^R581Q/+ neurons were more prone to fire in bursts rather than single tonic spikes compared to isogenic control neurons (Figure 2E). Moreover, the significant interaction of genotype and day in culture suggest that this bursting phenotype became even more pronounced and phasic over time. This is further supported by the fact that Q2-04^R581Q/+ neurons developed significantly smaller inter-burst interval CoV only after day 22 in culture, and thus their bursts became more regular and phasic (p=0.0169; Figure 2—figure supplement 1F).

The altered pattern of firing activity suggested differences in the intrinsic excitability properties of Q2-04^R581Q/+ neurons. To determine the source of the progressive increase in bursting in KCNQ2-DEE neurons, we performed whole-cell current-clamp measurements and examined the AP properties and post-burst AHPs. We systematically recorded from single Q2-04^R581Q/+ and isogenic control GFP-positive cortical neurons at three time points in culture defined as week 3 (DIV: 14–16), week 4 (DIV: 22–26), and week 5 (DIV:32–35; Figure 3A,B). As neurons matured in culture, the neuronal resting membrane potentials (RMP) became more hyperpolarized and AP amplitudes increased for both genotypes. However, Q2-04^R581Q/+ neurons had a significantly more depolarized RMP (p=0.0163) and higher input resistance (p=0.0075) only during week 3 (Figure 3—figure supplement 1A and Table 1). This could be a result of the reduced M-current or a change in the way Q2-04^R581Q/+ neurons mature. The increasing AP amplitudes for both genotypes over the course of 5 weeks indicate some progress in functional neuronal maturation (Figure 3—figure supplement 1B and Table 1; Bardy et al., 2016; Nehme et al., 2018; Lindhout et al., 2020).

We observed that compared with isoQ2-04^+/+ neurons, Q2-04^R581Q/+ neurons had slower AP repolarization with longer AP half-width at week 3 (p=0.0383; Figure 3C,D; Table 1). Slower AP repolarization is consistent with a reduced M-current. However, Q2-04^R581Q/+ neurons progressively developed faster AP repolarization with shorter AP half-widths at week 5 and larger fast component of the AHP (fAHP) at week 4 (p=0.03) and at week 5 (p=0.035; Figure 3C,D and Table 1). There was no difference in AP thresholds. Faster AP repolarization would enable neurons to fire a greater number of APs more rapidly with less synaptic input, which may explain the enhanced number of spikes per burst and higher tendency of Q2-04^R581Q/+ neurons to fire within bursts (Figure 2D).

The ability of neurons to fire in bursts is facilitated by an increased AP generation capacity and by the refractory period after each period of high frequency spiking. Several K⁺ conductances activate following a burst of APs to hyperpolarize the membrane, restore ionic balance and prevent neurons from firing. Previous studies have shown the involvement of KCNQ2 channels in the medium AHP (mAHP) in cortical neurons (Guan et al., 2011; Battefeld et al., 2014). We examined the post-burst AHP using a 50 Hz train of 25 APs evoked by 2 ms/1.4 nA current pulses. Compared to controls, Q2-04^R581Q/+ neurons exhibited significantly larger mAHPs (peak of AHP) and slow post-burst AHPs (sAHP; 1 s after last stimulus; p<0.0001 and p<0.0001, respectively; Figure 3E,F and Table 1). Surprisingly, isogenic control neurons exhibited a time-dependent attenuation in the mAHP and sAHP and thus were significantly smaller than Q2-04^R581Q/+ neurons during weeks 4–5 but not earlier (Figure 3E,F and Table 1).

Although an enhanced AHP increases the refractory latency of neurons to repolarize and fire again after a burst of APs, the faster repolarization observed for Q2-04^R581Q/+ neurons may counterbalance this effect. These findings are consistent with enhanced phasic burst firing in Q2-04^R581Q/+ neurons. The changes in AP properties account for more APs fired within bursts, whereas enhanced post-burst AHPs dampen spontaneous firing between bursts (Figure 2).

Previous studies have shown that blocking M-current in rodent cortical excitatory neurons acutely with XE991 enhances neuronal excitability by lowering AP threshold, impairing AP repolarization and attenuating the post-burst AHP (Yue and Yaari, 2004; Santini and Porter, 2010; Lezmy et al., 2017). However, our analysis of Q2-04^R581Q/+ neurons demonstrated that AP repolarization was slower during week 3 (with longer AP half-width) compared to isoQ2-04^+/+, but over time AP repolarization became more pronounced with faster AP half-width and enhanced fAHP. Furthermore, the post-burst AHPs are large throughout the timeline of our experiments (Figure 3). We propose two potential explanations for these divergent mechanisms for altered excitability: either M-current inhibition in human cortical glutamatergic neurons has different effects than in rodent neurons, or chronic M-current suppression leads to enhanced burst firing and altered AP properties by indirect mechanisms that are different from those related to acute M-current inhibition.

To investigate further, we interrogated the intrinsic membrane properties of unrelated control, week 4 excitatory neurons before and after acute treatment with 20 µM XE991. Consistent with previous reports (Yue and Yaari, 2004; Santini and Porter, 2010; Lezmy et al., 2017), XE991 significantly lowered AP threshold and slowed AP repolarization (AP threshold: p=0.0104; HW: p=0.0046; fAHP: p=0.009; Figure 3—figure supplement 1C–E). Furthermore, XE991 blunted the mAHP and sAHP amplitudes in these neurons (mAHP: p=0.0014; sAHP: p=0.022; Figure 3—figure supplement 1F,G). These effects did not resemble the behavior of Q2-04^R581Q/+ neurons, suggesting that loss of M-current alone is not sufficient for phasic burst firing.

Collectively, these experiments suggest that KCNQ2-DEE neurons develop a more pronounced phasic bursting phenotype progressively as a result of chronic M-current reduction coupled with dyshomeostatic adaptation of intrinsic membrane properties at different temporal scales.

The intrinsic AP properties of Q2-04^R581Q/+ neurons are not consistent with a pure loss of M-current but rather with a gain of other fast voltage-gated and Ca²⁺-dependent K⁺ conductances that increase over time in culture on different time scales. While a number of different channels might contribute to a dyshomeostatic mechanism, strong candidates are large, intermediate, and small conductance, Ca²⁺- and voltage-gated BK (KCNMA1) and IK/SK (KCNN4/KCNN1-3) channels that participate in fast AP repolarization and slow Ca²⁺-dependent post-burst AHP, respectively (Storm, 1990; Iyer et al., 2017; Latorre et al., 2017). The kinetics of BK channels are modulated by several β subunits transforming them from non-inactivating if alone or associated with β1 (KCNMB1) or β4 (KCNMB4), to fast inactivating if associated with β2 (KCNMB2) or β3 (KCNMB3) subunits (Storm, 1989; Xia et al., 1999; Hu et al., 2003; Kaufmann et al., 2010; Li and Yan, 2016; Latorre et al., 2017). To determine if any of these channels or their accessory subunits are upregulated progressively over time in Q2-04^R581Q/+ neurons, we examined the levels of gene expression using RT-qPCR on weeks 3 and 5. We also examined the gene expression levels of several other ion channels including KCND2, KCNT1, KCNA1, KCNA2, KCNA4, KCNQ2, KCNQ3, KCNQ5, HCN1, HCN2, SCN8A, and ANKG (known to bind KCNQ2/3 in the AIS). On week 3, we found no difference in expression of KCNMB1-4, KCNN1-3, KCND2, KCNT1, KCNA1, KCNA2, KCNQ2, KCNQ3, KCNQ5, HCN1, HCN2, SCN8A, and ANKG. By contrast, KCNMA1 (BK channel), KCNN4 (IK channel), and KCNA4 (encoding K_V1.4 channel) expression was higher in Q2-04^R581Q/+ neurons (p=0.0429, p=0.016 and p=0.0216, respectively; Figure 4A and Figure 4—figure supplement 1A). By week 5, the difference in KCNMA1, KCNN4 and KCNA4 was no longer significant but other differences had emerged, including an upregulation of KCNMB3, KCNN1, KCNN2, KCNA1, and KCNA2, and downregulation of KCNQ5 expression in Q2-04^R581Q/+ neurons relative to controls (p=0.0081, p<0.0001, p=0.018, p=0.0099, p=0.0017 and p=0.0198, respectively; Figure 4B and Figure 4—figure supplement 1B). These genes, which encode a BK channel β-subunit, SK channels (SK1 and SK2), K_V1.1 and K_V1.2, respectively, have been implicated in epilepsy and developmental delay in human and animal studies (Hu et al., 2003; Lorenz et al., 2007; McKay et al., 2012; Lerche et al., 2013; Masnada et al., 2017; Paulhus et al., 2020), and their upregulation is consistent with a gain of a K⁺ conductance suggested by our electrophysiological studies (i.e. BK/SK upregulation can lead to larger post-burst AHPs and BK/K_V1.1/K_V1.2 upregulation can lead to faster AP repolarization).

To determine if SK channels were functionally altered in KCNQ2-DEE neurons, we investigated their contribution to the post-burst AHP in Q2-04^R581Q/+ and isogenic control neurons at week 4. Addition of apamin (500 nM), a SK channel antagonist, reduced the mAHP in both groups, and reduced the sAHP only in Q2-04^R581Q/+ neurons. The magnitudes of change of the mAHP (ΔmAHP: p=0.0001) and sAHP (ΔsAHP: p<0.0001) were significantly larger for Q2-04^R581Q/+ neurons compared to isoQ2-04^+/+ neurons (Figure 5A–C). These data suggest that functional enhancement of SK channels in KCNQ2-DEE neurons contributes to the increase of slowly deactivating Ca²⁺-activated currents that determine the level of post-burst AHP.

Furthermore, given the observed enhancement of AP repolarization at week 5, we applied apamin (500 nM) and paxilline (20 µM, a BK channel antagonist) during spontaneous MEA recording on day 32 in culture. Blocking SK and BK channels reduced the ISI CoV of Q2-04^R581Q/+ neurons (p=0.0015) to the level observed in untreated controls, while treatment did not affect isoQ2-04^+/+ neurons (p=0.796; Figure 5D). Addition of both drugs also had a significantly larger effect on the number of spikes/burst (ΔSpikes/Burst: p=0.0087) and burst % (ΔBurst%: p=0.0254) in Q2-04^R581Q/+ neurons than in isogenic control neurons. This pharmacological treatment specifically reduced the bursting features to the levels seen in isogenic control neurons before drug application (Figure 5D). Together, these data suggest that inhibition of SK and BK channels in Q2-04^R581Q/+ neurons reduces post-burst AHPs and normalizes phasic burst firing behavior.

To determine if the adaptive enhancement of repolarization, post-burst AHP and increased bursting by enhanced K⁺ channel functional expression is a result of early and sustained suppression of M-current, we chronically treated isogenic control neurons with a low concentration of XE991 (1 µM, starting on day 12 in culture; Figure 6A) and measured excitability during week 4. Chronically XE991-treated control neurons exhibited enhanced repolarization (half width: p=0.0002; fAHP: p=0.01) and larger post-burst AHPs (mAHP: p=0.0001; sAHP: p=0.0008) relative to untreated controls (Figure 6B–E, Figure 6—figure supplement 1A,B and Table 2). Importantly, these effects were identical to or larger than the properties observed for untreated Q2-04^R581Q/+ neurons. This experimental paradigm effectively phenocopied the electrophysiological behavior of Q2-04^R581Q/+ neurons in isoQ2-04^+/+ neurons and suggests that the enhanced AP repolarization and larger post-burst AHP that we identified in KCNQ2-DEE neurons, occur as a result of long-term reduction of M-current.

We next assessed the effects of chronic M-current inhibition on spontaneous neuronal activity using MEAs. Within 24 hr of adding XE991 to the neuronal media (DIV: 12-13), we observed an increase in the number of active electrodes approximately three times higher in XE991-treated relative to untreated isoQ2-04^+/+ neurons (p<0.0001; Figure 6—figure supplement 2A,C). This was accompanied by a sevenfold increase in burst % (p<0.0001; Figure 6—figure supplement 2B,C). However, the magnitude of change in mean firing frequency on active electrodes was not significantly affected by XE991 in isoQ2-04^+/+ neurons (p=0.4557; Figure 6—figure supplement 2A). Thus, XE991 induced a relatively fast upregulation of spontaneously firing isoQ2-04^+/+ neurons to the levels of Q2-04^R581Q/+ neurons (Figure 6—figure supplement 2). We did not find any difference in the average number of GFP-positive cells counted per well between the groups of neurons and at later time points the number of active electrodes became equal between the groups (Figure 6—figure supplement 1C,D). Furthermore, chronic M-current inhibition led to long-term enhanced burst firing parameters such as ISI CoV, number of bursting electrodes and bursts, number of spikes per burst and burst % (p<0.0001, p=0.0011, p=0.0275, p<0.0001, p<0.0001, respectively; Figure 6F and Figure 6—figure supplement 1C). The effects of chronic M-current inhibition on bursting in control neurons were more dramatic than the inherent bursting phenotype in Q2-04^R581Q/+ neurons, likely because of the less severe loss of function associated with the mutation relative to the pharmacological treatment with XE991 (Figure 1B).

Additionally, we compared the expression of K⁺ channel genes involved in AP repolarization and post-burst AHP between Q2-04^R581Q/+, isoQ2-04^+/+ and isoQ2-04^+/+ that were chronically treated with XE991 at week 5. Expression of KCNMA1, KCNMB3, KCNN1, KCNN2, KCNN4, KCNA1, and HCN1 was significantly higher in chronically XE991-treated isogenic control neurons compared to untreated neurons (p=0.0038, p=0.0148, p=0.0179, p=0.0307, p=0.0175, p=0.0443, and p=0.0002, respectively; Figure 6G and Figure 6—figure supplement 1E). Expression of other channel genes including KCND2, KCNT1, KCNA2, KCNA4, KCNQ2, KCNQ3, KCNQ5, and HCN2 was not different following chronic XE991 treatment (Figure 6—figure supplement 1E).

Lastly, acute application of paxilline and apamin on MEAs during week 5, restored the bursting behavior of Q2-04^R581Q/+ neurons and chronically XE991-treated isogenic control neurons to the levels of untreated control neurons (Figure 5 and Figure 6—figure supplement 3A,B). Collectively, these data suggest that upregulation of SK and BK channel function exacerbates bursting in Q2-04^R581Q/+ neurons, and is a result of reduced M-current.

Control 1 and 3 hiPSC line (11a and 18a; RRID:CVCL_8987 and RRID:CVCL_8993) was derived previously (Boulting et al., 2011). Control two hiPSC line (NCRM-5; NHCDR Cat# ND50031, RRID:CVCL_1E75) was obtained from the NIH Center for Regenerative Medicine (NIH CRM). KCNQ2-DEE patient-derived (Q2-04^R581Q/+) and isogenic control (isoQ2-04^+/+) iPSC lines were derived as described below. Further information on all iPSC lines can be found in Figure 1—figure supplement 4B.

Full-length cDNA encoding WT human KCNQ2 splice isoform 4 cDNA (K_V7.2; GenBank accession NM_172108) was engineered in the mammalian expression vector pIRES2_EGFP or a modified vector where EGFP was substituted by CyOFP1. Site-directed mutagenesis of KCNQ2 was performed using QuikChange II XL (Agilent technologies, Santa Clara, CA, USA; mutagenic primer sequences: 5’: TCCCAAATTAAGAGCCTGCAGTCCAGAGTGGAC, 3’: AGGCTCTTAATTTGGGACAGCATGTCCAGGTGGC) to insert the R581Q (R550Q in isoform 4) patient mutation into the wildtype construct. KCNQ3 (K_V7.3; GenBank accession NM_004519) was cloned into pcDNA5/FRT for use in generating the KCNQ3-stable CHO-K1 cells.

TetO-Ngn2-puro (Addgene plasmid #52047) and TetO-FUW-EGFP (Addgene plasmid #30130) plasmids were gifts from Marius Wernig (Vierbuchen et al., 2010; Zhang et al., 2013). FUW-M2rtTA (Addgene plasmid # 20342) was a gift from Rudolf Jaenisch (Hockemeyer et al., 2008). Lentiviruses were generated in HEK293T cells using the second-generation packaging vectors, psPAX2 and pMD2.G, as described previously (Zufferey et al., 1998) by the Northwestern University DNA/RNA Delivery Core.

Peripheral blood mononuclear cells (PBMCs) were isolated from whole blood following informed consent under protocols approved both by Ann and Robert H. Lurie Children's Hospital of Chicago and Northwestern University. Reprogramming of PBMCs into iPSCs was performed at the Northwestern Stem Cell Core Facility using Invitrogen’s CytoTune-iPS 2.0 Sendai Reprogramming system (A16517, Thermofisher), following the manufacturer’s instructions. This reprogramming system uses four transcription factors (Oct4, Sox2, Klf4, c-Myc). Briefly, PBMCs (5 × 10⁵) were seeded into one well of a 24-well plate and cultured for four days in StemSpan SFEM II PBMC complete medium (STEMCELL Technology, 09655) supplemented with 100 ng/ml SCF (PeproTech, 300–07), 100 ng/ml FLT3 (PeproTech, 300–19), 20 ng/ml IL-3 (PeproTech, 200–03), and 20 ng/ml IL-6 (PeproTech, 200–06). Immediately after plating, the cells were infected with Sendai virus for 48 hr at 37°C. The infected cells were transferred onto MEF feeders and cultured in StemSpan SFEM II. Following 21–28 days of culture, individual iPSC colonies were picked and transferred to Matrigel (BD Biosciences, BD354277) coated six-well plate for expansion and were maintained in mTeSR1 (STEMCELL Technology, 85850).

Chinese hamster ovary cells (CHO-K1, CRL 9618, American Type Culture Collection, Manassas VA, USA) stably expressing human KCNQ3 were grown in F-12 nutrient mixture medium (GIBCO/Invitrogen, San Diego, CA, USA) supplemented with 10% fetal bovine serum (ATLANTA Biologicals, Norcross, GA, USA), Zeocin (100 μg/ml) and hygromycin B (600 μg/ml), penicillin (50 units·ml⁻¹), streptomycin (50 μg·ml⁻¹) at 37°C in 5% CO₂. Unless otherwise stated, all tissue culture media was obtained from Life Technologies, Inc (Grand Island, NY, USA). Plasmids encoding WT and/or variant KCNQ2 were transiently transfected into stable KCNQ3 cells by electroporation using the Maxcyte STX system (MaxCyte Inc, Gaithersburg, MD, USA) as reported previously (Vanoye et al., 2018). We studied the KCNQ2 R581Q variant in both the homozygous and heterozygous (1:1 with wild-type KCNQ2) state in cells stably expressing KCNQ3 subunits. For the homozygous channel experiments, 15 µg KCNQ2 wild-type or variant KCNQ2 DNA were electroporated. For the heterozygous channel experiments, 20 µg of KCNQ2 wild-type or 10 µg wild-type DNA plus 10 µg variant DNA were co-electroporated.

All iPSCs were grown on Matrigel with mTeSR1 media and passaged weekly using Accutase (Sigma). All cell cultures were maintained at 37°C and 5% CO₂. All cell lines were regularly tested for presence of mycoplasma using MycoAlert PLUS Detection Kit (Lonza) and determined to be mycoplasma-free.

Primary glial cell cultures were derived from postnatal day 0–2, CD-1 mice (Charles River). Briefly, brain cortices were dissected free of meninges in dissection buffer HBSS (Thermo Fisher), then digested with trypsin (Thermo Fisher) and DNAse I (Worthington) for 10 min at 37°C. The tissue was dissociated in glia medium: DMEM (Corning, #15–013-CV) supplemented with Glutamax, D-glucose, 10% normal horse serum (Life Technologies), and penicillin-streptomycin (Thermo Fisher). After centrifugation and resuspension, cells were filtered through a 0.45 micron cell strainer and plated on poly-D-lysine coated plates with glia media at 37°C, 5% CO₂ for 2 weeks. Afterwards, glial cultures were tested for mycoplasma, dissociated for expansion, and frozen in 10% DMSO/horse serum. All animal experiments were approved and conducted in accordance with the policies and guidelines set forth by the Northwestern University Institutional Animal Care and Use Committee (IACUC).

Isogenic control iPSCs were generated using CRISPR/Cas9 from the Q2-04^R581Q/+ patient-derived iPSC line in collaboration with Applied StemCell (Milpitas, CA). Briefly, one million patient iPSCs were electroporated with a mixture of guide RNA (gRNA) and Cas9 (in the ribonucleoprotein format), and ssODN (Supplementary file 1). A small portion of the cell culture, presumably with mixed population, was subjected to Sanger sequencing analysis. Once the mixed culture showed repair with qualified HDR efficiency, the transfected cells were subjected to single cell cloning. Single-cell-derived clones were cultured for 15–20 days, followed by genotype analysis by Sanger sequencing. Clones with the desired genetic modification were identified by PCR genotyping and confirmed by Sanger sequencing (Figure 1D). All primers sequences can be found in Supplementary file 2.

Potential off-target sites were predicted with the online tool: CCTop CRISPR/Cas9 target online predictor https://cctop.cos.uni-heidelberg.de/ (Stemmer et al., 2015). We selected the top 10 genomic regions of homology and thus most likely off-target sites, and amplified each one by targeted PCR of genomic DNA from the corrected iPSC clone, for further analysis either by Sanger Sequencing or by a T7 Endonuclease assay (Figure 1—figure supplement 3A; Supplementary files 1 and 2). The same PCR conditions were used to amplify the positive control DNA template and primer mix, included in the T7 Endonuclease I Assay Kit (Genecopoeia). The amplified DNA from each potential off-target site was purified using the Wizard SV Gel and PCR Clean-Up System (Promega). The concentration of the purified DNA from the potential off-target sites and the template DNA from the positive control was assessed by using a Nanodrop 2000 Spectrophotometer (Thermo-Fisher). Amplicons (500 ng) from each potential off-target region of the isogenic and parental cell lines and 500 ng of the positive control DNA template were heated to 95°C for 5 min and subsequently allowed to cool to room temperature to denature and re-anneal the PCR products, respectively. T7 Endonuclease I was added to the re-annealed PCR products and incubated at 37°C for 60 min. The PCR products from the potential off-target sites and the positive control template were then run on a gel with 6x loading buffer, alongside a 2-log DNA ladder (New England BioLabs Inc). All primer sequences can be found in Supplementary file 2.

Whole genome sequencing was done on Q2-04^R581Q/+ and isoQ2-04^+/+ iPSC genomic DNA in collaboration with Novogene Corporation Inc The genomic DNA was randomly fragmented by sonication, then DNA fragments were end polished, A-tailed, and ligated with the full-length adapters for Illumina sequencing, and secondary PCR amplification with P5 and indexed P7 oligos. The PCR products used in the final construction of the libraries were purified with AMPure XP system. Libraries were checked for size distribution by Agilent 2100 Bioanalyzer (Agilent Technologies, CA, USA), and quantified by real-time PCR (to meet the criteria of 3 nM). Alignment to the human reference genome (build hg38) was done with the Burrows-Wheeler Aligner (BWA) v.0.7.17. Variation calling and quality controls were performed using the GATK best practices pipeline.

Whole genome sequencing reads were checked for quality using FASTQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc). Alignment to the human reference genome build hg38 was done using the Burrows-Wheeler Aligner (BWA v.0.7.17) (Li and Durbin, 2009). The variant calling process was performed following the Genome Analysis ToolKit best practices pipeline (GATK4; McKenna et al., 2010). Briefly, aligned reads in bam format followed duplicate marking, coordinate sorting and base quality score recalibration (BQSR). Next, Haplotype Caller was used to generate genomic variant call format (gVCFs) for each individual chromosome and GatherGVCF was used to concatenate them. Raw VCFs were generated using the genomicsDBimport and GenotypeGVCFs tools. Further QCs were performed using hard filtering and Variant Quality Score Recalibration (VQSR). Post-GATK QC’s were done using VCFTools v.0.1.17 (1000 Genomes Project Analysis Group et al., 2011), removing variants without PASS in the FILTER field, with read depth (DP) <20 and genotype quality (GQ) <20. Indel left-normalization and multiallelic splitting were also performed before annotation.

Variant annotation and functional deleterious prediction were done using ANNOVAR (Wang et al., 2010). Potentially deleterious variants were selected if they had (i) frequency <0.01 (rare) in gnomAD v2.1.1 exome and genome cohorts Genome Aggregation Database Consortium et al., 2020; (ii) located in exonic and splicing sites; (iii) predicted to be nonsynonymous, frameshift and nonframeshift indels, and stop-gain and stop-loss, and (iv) CADD score of >12.37 top 2% of most deleterious variants; (Rentzsch et al., 2019). Whole-genome similarity comparison was performed by IBD analysis in PLINK v.1.9 (Chang et al., 2015). We used common variants (frequency >1%) and performed linkage-disequilibrium (LD) pruning. For single variant comparison, we selected all non-missing positions with genotype differences (different zygosity) between samples Q2-04^R581Q/+ and isoQ2-04^+/+ (e.g. homozygous reference to heterozygous, homozygous reference to homozygous alternative, etc.). Variants with genotype differences were plotted as a circular ideogram using Circos plot (Krzywinski et al., 2009), generating densities in 1 MB windows, followed by potentially deleterious variants, genes harboring variants and CRISPR off-target sites.

iPSCs were differentiated into cortical glutamatergic neurons using a modified version of a protocol based on Ngn2 overexpression (Zhang et al., 2013). Stem cells were dissociated as single cells using Accutase, re-suspended in mTeSR1 with 10 µM ROCK inhibitor (Y-27632, DNSK International, #129830-38-2), then incubated with lentiviruses (FUW-M2rtTA, TetO-Ngn2-Puro, TetO-FUW-EGFP) in suspension for 5 min before plating (95,000 cells/cm²; Figure 1—figure supplement 4A). After 24 hr (day 1), lentivirus was removed and replaced with knockout serum replacement medium (KOSR) consisting of KnockOut DMEM supplemented with Knockout replacement serum KSR, nonessential amino acids (NEAA), Glutamax (Life Technologies), 55 µM β-mercaptoethanol (Gibco, Cat# 21985023), 10 µM SB431542 (DNSK International), 100 nM LDN-193189 (DNSK International), 2 µM XAV939 (DNSK International), and 3 µg/ml of doxycycline (Sigma). On the following day (Day 2), media was replaced with a 1:1 ratio of KOSR to neural induction media (NIM) composed of DMEM: F12 supplemented with NEAA, Glutamax, N2 (Gibco, Life Technologies), 0.16% D-glucose (Sigma) and 2 µg/ml heparin sulfate (Sigma). Doxycycline (2 µg/ml) and puromycin (2 µg/ml; Sigma) were added to this NIM media. On Day 3, the media was replaced with NIM containing doxycycline (3 µg/ml) and puromycin (2 µg/ml). All neurons were frozen in 10% DMSO/Hyclone FBS (VWR) on Day 4 (Figure 1—figure supplement 4A). For all experimental analysis, iPSC-derived neurons were plated on primary CD1 mouse cortical glia, derived as previously described (Di Giorgio et al., 2008). Glial cells were first plated on PDL/laminin-coated plates or coverslips in glia media composed of MEM (Life Technologies) supplemented with Glutamax (0.6%), D-glucose, and 10% horse serum (Life Technologies). After 5–7 days, neurons were thawed (Day 5 post-induction) and plated, at a density of 20,000/cm², directly onto the monolayer of mouse glia in Neurobasal medium (NBM), supplemented with NEAA, Glutamax, N2 and B27 (Life Technologies) containing BDNF (10 ng/mL, R and D systems), 2% Hyclone FBS, doxycycline (3 µg/ml), and ROCK inhibitor. Half of the media was replaced the next day and then every other day thereafter with NBM supplemented with NEAA, Glutamax, N2 and B27 containing BDNF (10 ng/mL), 2% Hyclone FBS and doxycycline (2 µg/ml).

iPSCs and neurons were plated on Matrigel or PDL/laminin-coated glass coverslips, respectively. iPSCs were fixed with 4% formaldehyde (Sigma) in 4% sucrose/PBS for 10 min at room temperature and permeabilized overnight in 0.4% PBST at room temperature. iPSCs were blocked in PBS containing 0.1% triton (PBST) and 10% normal donkey serum (NDS; Jackson Immuno Research) for 1 hr at room temperature then incubated in primary antibodies for 24 hr at 4°C. The following primary antibodies were used with iPSC: SSEA4 (DSHB, MC-813–70, 3 µg/ml), Nanog (R and D, AF1997, 1:500). iPSC were washed three times with PBS then incubated for 45 min with secondary antibodies in 0.1% PBST with 2% NDS. The secondary antibodies used were Alexa 488 donkey anti-mouse and Alexa 647 donkey anti-goat (Thermo Fisher Scientific, 1:500). After secondary antibody, coverslips were washed three times with PBS, incubated for 10 min with DAPI (Invitrogen, #33342; 1:1000) and mounted onto microscope slides with Fluoromount-g (Southern Biotech).

Neurons plated onto coverslips were fixed with 3.7% formaldehyde in 4% sucrose/PBS for 15 min at room temperature and then washed three times with cold PBS. Cells were permeabilized and blocked simultaneously in 0.1% PBST with 5% normal goat serum (NGS) for 1 hr at room temperature followed by incubation with primary antibodies overnight at 4°C. The following primary antibodies were used: GFP (Abcam ab13970, AB_300798, 1:10,000), Map2 (Millipore MAB3418, AB_94856, 1:1000) and vGLUT1 (Synaptic systems, 135 302, AB_887877, 1:200). The following day coverslips were washed three times with cold PBS then incubated with secondary antibody for 1 hr at room temperature. The following secondary antibodies were used: Alexa 488 goat anti-chicken, Alexa 568 goat anti-mouse and Alexa 647 goat anti-rabbit (Thermo Fisher Scientific, 1:1000). Primary and secondary antibodies were diluted in PBS containing 5% normal goat serum. Cells were washed three times in PBS and then briefly in distilled water and mounted onto microscope slides using ProLong Gold anti-fade reagent (Life Technologies). Neurons from the same differentiation experiment were fixed and stained at the same time with identical antibody dilutions. Images were acquired at 10x or 20x for MAP2/GFP/vGLUT1 positive neuron counting using a Leica inverted Ti microscope.

Cells were harvested by scraping from six-well plates at the indicated time points after induction of neuronal differentiation. Cells were resuspended in TRIzol Reagent (Life Technologies), and RNA was isolated following manufacturer’s protocol. First-strand cDNA was synthesized from 1.1 μg of DNase I (Invitrogen) treated RNA using SuperScript IV reverse transcriptase (Thermo) and oligo dT primers following manufacturer’s instructions. First-strand cDNA samples were diluted 1:7 and 2.5 μl was used in reactions. RT-PCR was performed using SYBR green on the CFX system (Bio-Rad). PCR conditions were 95°C for 3 min, then 40 cycles of 95°C for 10 s and 60°C for 30 s and a final melt curve step of from 65°C to 95°C in increments of 0.5°C per 5 s. All assays were performed in duplicate. The averaged cycle of threshold (Ct) value of two housekeeping genes (GPI/GAPDH) was subtracted from the Ct value of the gene of interest to obtain the ΔCt. Relative gene expression was determined as 2^−ΔCt (ΔΔCt) and expressed relative to the indicated sample in the experiment. Pooled human fetal brain total RNA (#636526, Lot #1612396A) and adult cortex total RNA (#636561, Lot #2007106) samples were purchased from TakaRa Bio. All primers sequences are listed in Supplementary file 4.

For multielectrode array (MEA) studies, 12-well MEA plates with 64 electrodes per well were coated with PDL and laminin according to Axion Biosystems protocols. Mouse glial cells were seeded at a density of 50,000 cells/well then 30–35,000 neurons/well were added 1 week later. Every other day (Mon, Wed, Fri), half of the media was removed from each well and replaced with fresh media > 5 hr before recordings were made on those days. Independent differentiations were always plated on the same day of the week thus each time point represents the same day in the feeding schedule. Data presented in Figures 2 and 6 are from the days on which media was changed. Spontaneous activity was recorded using Axion Biosystems Maestro 768 channel amplifier and Axion Integrated Studios (AxIS) v2.5 software. The amplifier recorded from all channels simultaneously using a gain of 1200x and a sampling rate of 12.5 kHz/channel. After passing the signal through a Butterworth band-pass filter (300–5000 Hz), on-line spike detection (threshold = 6 x the root-mean-square of noise on each channel) was performed with the AxIS adaptive spike detector. All recordings were conducted at 37°C in 5% CO₂/95% O₂. Spontaneous network activity was recorded for 5 min each day starting on day 10 of differentiation. Starting day 10 neurons were also electrically stimulated with 20 pulses at 0.5 and 0.25 Hz after spontaneous recordings were made. This anecdotally enhances maturation and migration of neurons to the electrode field.

All data reflects well-wide averages from active electrodes, with the number of wells per condition represented by N values. Active electrodes were defined as having ≥1 spikes/min. The mean firing frequency (Hz) was calculated as the total number of spikes divided by the number of active electrodes over the recording duration (300 s). The ISI CoV was calculated as the standard deviation divided by the mean inter-spike interval on active electrodes. Bursts were detected according to a Poisson distribution. The algorithm is adaptive to the mean firing rate on each electrode according to a ‘surprise’ threshold, adapted from Legéndy and Salcman, 1985. The burst frequency (Hz) was calculated as the total number of bursts divided by the number of bursting electrodes over the recording duration (300 s). The IBI CoV was calculated as the standard deviation divided by the mean inter-burst interval on bursting electrodes. The burst % was calculated as the percentage of all spikes which occurred in bursts. The number of bursts, number of spikes per burst and burst % were used as a measure of neuronal activity as this demonstrates maturity of neuronal functional properties. MEA recordings were done with two to three independent differentiations. Acute application of 500 nM apamin (Alomone) and 20 µM paxilline (Tocris) was done on day 32 in culture (See Figure 5D).

Automated voltage-clamp recordings were performed at room temperature using a Syncropatch 768PE (Nanion Technologies, Munich, Germany) as described previously (Vanoye et al., 2018) except that the internal solution contained 5 mM Mg-ATP. The contribution of background currents was determined by recording before and after addition of XE991 (25 μM, TOCRIS, Minneapolis, MN). Only XE991-sensitive currents were used for analysis. Whole-cell currents were elicited from a holding potential of −80 mV using 1000 ms depolarizing pulses (from −80 mV to +40 mV in +10 mV steps every 20 secs) followed by a 300 ms step to 0 mV to analyze tail currents. Cells with seal resistance ≥0.5 GΩ and series resistance ≤20 MΩ (access resistance compensation was set to 80%) were used for analysis. Peak currents were measured 999 ms after the start of the depolarizing voltage pulse and tail currents 5 ms after changing the membrane potential to 0 mV. The time-constant of activation (τ) was determined by fitting currents elicited by voltage steps between −30 mV and +40 mV (50–1000 ms after start of the voltage step) to a single exponential.

Whole-cell current-clamp recordings were made from visually identified GFP-expressing neurons using an inverted Olympus IX51 microscope equipped with a 40X objective. Recording pipettes were made of glass capillaries using a horizontal Sutter P-1000 puller yielding a 2–4 MΩ resistance pipette when filled with standard K-methyl sulfate intracellular solution containing (in mM): 120 K-MeSO₄, 10 KCl, 10 HEPES, 10 Na₂-phosphocreatine, 4 Mg-ATP, 0.4 Na₃-GTP, pH 7.35 adjusted with KOH; osmolality 285–290 mOsm/Kg. Neurons were continuously perfused with oxygenated aCSF bath solution (in mM): 125 NaCl, 26 NaHCO₃, 2.5 KCl, 1.25 NaH₂PO₄, 1 MgSO₄, 22 glucose, 2 CaCl₂, pH 7.35 at 32–35°C; osmolality 310–315 mOsm/Kg.

Current-clamp recordings were acquired using a Multiclamp 700B amplifier (Molecular Devices, USA) and digitized at 10 kHz (filtered at 3 kHz) with the neurons held at −65 mV (V_h). All reported potential values were corrected for the liquid junction potential, calculated to be −8.2 mV. Resting membrane potential (RMP) was measured immediately after establishing the whole-cell patch clamp configuration. Input resistance (R_N) was calculated as the slope of the voltage-current curve determined using 500 ms current steps from −50 pA to 30 pA in 10 pA steps. Medium (mAHP) and slow (sAHP) afterhyperpolarizations (AHPs) were measured as the difference between V_h and the negative going peak and 1 s after the offset of the last current step, respectively, induced by a 50 Hz train of 25 APs evoked by 2 ms/1.2 nA current injection pulses. Single AP properties, including fast afterhyperpolarization (fAHP), were measured using direct somatic current injection ramps (10–80 pA, 500 ms). AP amplitude was calculated as the difference between V_h to the peak of the first AP of the ramp protocol. AP threshold was calculated where the first derivative of the up phase of the trace equaled 5 mV/ms. Using a 1 ms sliding average, the fAHP measurement was taken when the mean first derivative of the trace reached 0.0 ± 0.5 after initial spike in each sweep. AP half-width measurements were taken at half the AP peak amplitude relative to V_h. Neurons meeting the following criteria were used: series resistance (R_S) <30 MΩ, membrane resistance (R_N) >200 MΩ, resting potential (V_rest) < −45 mV, and AP amplitude >80 mV from V_h. Data were analyzed using custom MATLAB protocols (Simkin et al., 2015). All MATLAB scripts are available for download at github.com/simkind/Patch-clamp-analysis.git (Simkin, 2021; copy archived at swh:1:rev:bde5c7399d9f7c789feec0ee26ab5dad4a661d90). Data collected at three time points in culture defined as week 3 (days 14–16), week 4 (days 22–26), and week 5 (days 32–35; Figure 3A). Data collected from each time-point of 3–5 days were combined for statistical analysis using Statview software.

We tested the action of chronic and acute application of 1 and 20 µM XE991 (Abcam; expected to block 50% and 100% of M-current, respectively; Wang et al., 1998). XE991 (1 µM) was chronically applied to neuronal culture media starting day 12 in differentiation (right before beginning of week three time point) and AP properties were measured on week 4. Acute application of 20 µM XE991 or 500 nM apamin (Alomone) was done during week 4 and AP/AHP properties were measured before and 10 min after continuous perfusion of aCSF with XE991 or apamin.

Drugs were prepared as stock solutions using distilled water or DMSO, and then diluted to the required concentration in aCSF or culture media immediately before use. Bath-applied drugs were perfused for at least 10 min to ensure complete equilibration within the recording chamber before recording.

Differences were evaluated using t-test, one-way or two-way ANOVA, repeated-measures ANOVA, and Fisher's protected least significant difference posthoc tests where appropriate. All data are reported as means ± SEM.

Written informed consent was received from participants prior to inclusion in the study under protocols approved both by Ann and Robert H. Lurie Children's Hospital of Chicago and Northwestern University IRB (#2015–738).