Alcohol reverses the effects of KCNJ6 (GIRK2) noncoding variants on excitability of human glutamatergic neurons.

Experimental design and gene expression analysis.A Diagram outlining experimental design. Lymphocytes from subjects with or without AUD diagnosis and KCNJ6 haplotype variants were selected, reprogrammed into iPSC, induced into excitatory iNs, and analyzed by morphometry, immunocytochemistry, gene expression, and electrophysiology. B Sequencing alignment and depth analysis of bulk RNA sequencing confirmed expression of KCNJ6 mRNA in iN cultures, specifically the ENST00000609713 isoform, containing an 18.1 kilobase 3′UTR region. KCNJ6 exons are mapped to chromosome locations (marked in mb, megabases) and the position of the gene is indicated by the red box on the chromosome 21 pictogram, top, with transcription direction on the minus strand indicated by the broken arrow. Variant analysis of RNA sequences predicts a region of linkage disequilibrium of 22 SNPs, including the 3 SNPs used to select subjects (red, Table 1), and 19 additional SNPs (blue, Supplementary Table 1). Depth: number of sequencing reads per base aligned by position. Frequency: thickness of curved lines represents the relative frequency of splice site utilization between exons. C Single-cell RNAseq identifies a cluster of induced neurons (lower left), expressing markers consistent with neuronal function including SYP, SLC17A6, GRIN2B, SCN3A, KCNJ3, and KCNJ6; distinct from “transition neurons” that either do not express these markers or express sporadically. Additional markers are plotted in Supplementary Fig. 2E. Isolating KCNJ6 mRNA expression, aggregated by subject and treatment, AF neurons expressed a trend towards lower levels than UN neurons (p = 0.0508; Wald test), but treatment of 7d with IEE at 20 mM peak concentration increased AF expression above untreated (p = 0.0225) to levels similar to UN control (p = 0.322, not denoted on figure). D Volcano plot for untreated UN vs. AF neurons, highlighting genes significantly different (FDR > 0.05) and at least 1.5-fold changed (red dots). Genes below the fold-change cut-off are marked in blue, and those not significantly different are marked in green. Significantly different genes are listed in Supplementary Table 3. E Gene ontology (GO) enrichment of top 10 biological process (BP) terms for up- or downregulated genes. Plots indicate the number of regulated genes from the term and the color indicates the adjusted p value (q-value; key). All enriched terms are listed in Supplementary Tables 4, 5.

Validation of GIRK2 expression and function in human induced neurons.A Representative confocal images of GIRK2 immunoreactivity in mouse cortical neurons. Arrowheads indicate locations of GIRK2-staining puncta, with an example punctum enlarged in the inset, overlapping or adjacent to βIII-tubulin-positive processes. We observed two cellular expression patterns—one where the entire neuron is decorated with GIRK2 antibody (Supplementary Fig. 4), or another where GIRK2 expression is relatively faint and observed mostly on neuronal processes, shown here. B GIRK2 expression patterns in human induced neurons (iNs), showing representative confocal images from line 420. Inset shows two adjacent puncta. GIRK2 immunoreactivity matched a pattern of process-selective expression in mouse (A and Supplementary Fig. 4A), where GIRK2 was detected as relatively small (~0.5 µm diameter) puncta scattered primarily along the processes. Localization of GIRK2 immunoreactivity in human iN did not directly colocalize with synaptic vesicle marker VGLUT2 or synaptic marker Syn1 (Supplementary Fig. 5), but instead was found most frequently adjacent to synapses but overlapping the shafts of the βIII tubulin-positive processes, and less so on MAP2 positive processes (Figs. 2D.a, 3C, E). Cultured neurons express βIII tubulin throughout the cell, but not as strongly in axonal processes [69]. We previously found that processes in human iN cells stained for ankyrin G, identifying the axonal initial segment, which similarly lacked βIII-tubulin [70]. Detection of GIRK2 primarily on βIII-tubulin+/MAP- processes, therefore, suggests pre-axonal, and likely presynaptic, localization. C Following infection of iN cultures with lentivirus expressing both KCNJ6 and mCherry, large numbers of GIRK2+ puncta are seen in representative images (line 420). D Evaluation of GIRK2 function in iNs, (a) quantification of GIRK2 expression on MAP2+ vs. βIII-tubulin+ neuronal processes. GIRK2 is more abundant on βIII-tubulin processes (p = 0.0006, one-tailed Student’s t test, n = 15 cells per group, cell line 420). (b) Basal levels (upper pie plot) of the GIRK current in iNs as percent of neurons responding with hyperpolarization to the selective GIRK activator (160 nM ML297); compared with responding percentage when GIRK2 is overexpressed (lower pie chart). (c) Representative image of iN overexpressing GIRK2, as confirmed with mCherry fluorescence. (d) Representative traces of induced action potential firing before and after GIRK activation, demonstrating the contribution of GIRK function to cell excitability. (e) Representative trace of spontaneous postsynaptic current (sEPSCs) recordings during ML297 (160 nM) GIRK activator wash-in, demonstrating a shift of 7 mV holding current (amplifier-dependent compensation of GIRK-mediated membrane hyperpolarization). (f) Quantification of neuronal excitability at baseline and following GIRK activation with 160 nM ML297 (p = 0.015, paired Student’s t test, n = 9 cells before/after ML297, cell line 376).

Impact of AUD-associated KCNJ6 haplotype on neuronal properties.A Principles of morphological analysis of induced neurons: (a) Neurite area was the total TuJ1+ (βIII-tubulin+) staining area outside the cell soma. (b) Solidity is the area of the soma divided by its convex hull area. (c) Soma size was the area of the MAP2+ cell body. (d) Circularity compared the perimeter to the area. B Morphometry of iNs from KCNJ6 haplotype variant and affected (AF, cyan) or unaffected (UN, gray) individuals. Results are summed by group (left) or plotted individually by cell line (right), with subjects identified by line number (see Table 1—females identified with gray numbers). Individual cells are plotted as dots with the bar showing the mean, with error bars indicating the standard error of the mean (SEM). No significant differences were found in (a) soma size, (b) circularity, or (c) soma solidity, but (d) total neurite area was increased in the AF group (p = 0.018). C Representative images of iNs from individual lines, with arrows identifying individual GIRK2 puncta (red) localized on βIII-tubulin+ processes (gray). D GIRK2 expression was decreased in the AF as measured by puncta counts (a, p = 0.0012), circularity (c, p = 0.007), or solidity (d, p = 0.037), while there was no difference in puncta size (b). E Representative images of individual GIRK2 puncta (red) localized on βIII-tubulin+ processes (gray). F Electrophysiological analysis of passive neuronal properties, showing no difference in (a) membrane capacitance (b) membrane resistance, or (c) spontaneous EPSCs frequency. (d) Representative sEPSCs traces for each line. (e) Spontaneous EPSCs amplitude. G Electrophysiological analysis of induced neuronal properties. (a) Quantification of current required to shift resting membrane potential to −65 mV in pA: difference by group p = 1.2 × 10–5; (b) quantification of maximum number of action potentials (APs) induced with the “step” protocol, p = 0.086; (c) representative traces of APs induced with the “step” protocol; (d) quantification of number of action potentials (APs) induced with the “ramp” protocol, p = 2.0 × 10–9; (e) representative traces of APs induced with the “ramp” protocol. A generalized linear model was used to evaluate group differences for morphometry and GIRK2 expression, and generalized estimation equations was used for electrophysiology results. Numbers of cell lines and replicates for each experiment are shown in Supplementary Table 6.

Evaluation of neuronal excitability in human induced neuron populations.A Representative epifluorescence images of GcaMP6f expression in iNs from affected (233, 246) and unaffected individuals (420, 472). B, C Raster plots of the calcium spiking pattern of representative experiments from unaffected neurons (B, individual 472) and affected neurons (C, individual 233), under the stimulation protocol used. Green and purple bars show the application of 30 s pulses of Glu 10–50 µM and KCl 18 mM, respectively, on the neuron populations during the stimulation protocol. Arrows show the epochs of the calcium imaging recordings selected for calcium spike quantification: baseline, glutamate pulse 1, and KCl pulse 1. D Stimulation protocol (upper bar) and representative raw fluorescence trace (AF individual 233) from a complete recording. The zoom-ins on the first glutamate and KCl pulses depict glutamate and KCl-elicited activity, respectively. The vertical dotted lines correspond to the calcium spikes detected after analysis. E, F, G Number of spikes per ROI per minute during baseline (E, spontaneous activity), glutamate pulse 1 (F, glutamate-elicited activity) and KCl pulse 1 (G, KCl-elicited activity). The pie charts represent the proportion of neurons of UN (gray) and AF (blue) individuals that exhibited a certain number of calcium spikes during each three epochs of the recording. The bar plots show the number of spikes per ROI per minute fired by unaffected individuals (gray) and affected individuals (blue) during baseline, glutamate pulse 1 and KCl pulse 1. The sample size is depicted in the bar of each group as number of iN batches/number of experiments/number of neurons. Differences between affected and unaffected groups were evaluated by two-tailed unpaired Student’s t test (Welch’s correction for non-equal SD, **p < 0.01; ****p < 0.0001).

Ethanol treatment reduced KCNJ6 haplotype differences in iN excitability and GIRK2 expression.A Morphological analysis of IEE iNs generated from affected and unaffected individuals, showing no difference in: (a) neuronal soma size (p = 0.32); (b) soma circularity (p = 0.54); (c) soma solidity (p = 0.92); and (d) total neurite area (p = 0.98). B There was no difference in GIRK2 expression in the IEE AF group iNs compared with UN by (a) puncta counts (p = 0.46), (b) puncta size (p = 0.36), (c) puncta circularity (p = 0.052), or (d) solidity (p = 0.47). C Representative images of individual GIRK2 puncta (red) localized on βIII-Tubulin-positive processes (gray) for each cell line. D Electrophysiological analysis of passive neuronal properties in IEE iNs, showing (a) a small decrease in AF membrane capacitance (p = 9.6 × 10–9), but no difference in (b) membrane resistance (p = 0.63), (c) spontaneous EPSCs frequency (p = 0.32), or (d) spontaneous EPSCs amplitude (p = 0.19). (e) Representative sEPSCs traces for each cell line. (f) The AF group exhibited no change in resting membrane potential after IEE (p = 0.79). E Electrophysiological analysis of active neuronal properties found no difference in IEE iNs for (a) current required to shift resting membrane potential to −65 mV (p = 0.32), (b) maximum number of action potentials (APs) induced with the “step” protocol (p = 0.48), with (c) representative traces of APs induced with the “step” protocol, (d) number of action potentials (APs) induced with the “ramp” protocol (p = 0.95), with (e) representative traces of APs induced with the “ramp” protocol. F Representative images from individual lines of iNs, exposed to 7d of IEE, marked with arrows pointing to individual GIRK2 puncta (red) localized on βIII-Tubulin-positive processes (gray). G Summarized results from all lines showing differences in GIRK2 expression levels before and after 7 days of 20 mM IEE with ethanol (EtOH; p = 1.0 × 10−20). H Representative images of individual GIRK2 puncta (red) localized on βIII-Tubulin-positive processes (gray) prior and following 7 days 20 mM IEE with ethanol. Sample images are from line 246. I Representative images of FISH detection of KCNJ6 mRNA for each cell line. J Quantification of FISH. (a) The number of KCNJ6 puncta normalized to the number of cells in an image shows decreased expression in control AF compared with UN (p = 8.2 × 10−3) and increased expression following IEE (p = 8.2 × 10−3; using Tukey’s pairwise comparisons). (b) The percentage of KCNJ6-expressing MAP2+ cells substantially increase in the (c) AF group but not in the (b) UN group. Numbers of KCNJ6 puncta were analyzed by expression levels per cell, as recommended by the FISH manufacturer in (d) UN or (e) AF cells. (f) KCNJ6 puncta within the neuronal soma show increases following IEE in both UN and AF groups (p = 0.006 for genotype, Tukey’s post-hoc for UN, p = 0.01, for AF, p = 0.01). (g) A similar analysis of non-somatic puncta, presumably within neurites, showed no differences following IEE between genotypes (p = 0.23).

GIRK2 overexpression mimics ethanol response.A Current required to shift resting membrane potential to −65 mV, in untreated (control, p = 5.9 × 10−4), lentiviral KCNJ6 overexpression (over., p = 0.23), or 1 d 20 mM EtOH (p = 0.75) cultures. B Representative traces of APs induced with the “ramp” protocol. C Quantification of maximum number of action potentials (APs) induced with “ramp” protocol, control (p = 6.8 × 10−6), overexpression (p = 0.014), or 1 d 20 mM EtOH (p = 0.10). D quantification of GIRK2 puncta after overexpression (line 376, one-tailed Student’s t test p = 0.04).