Fig. 1: Overview of the study design and summary results of RNA-sequencing analysis from iPSC-derived neurons.

a Study design and analysis workflow. Figure 1a shows a summary of our study workflow, including major findings. The study materials and methods are detailed in Supplementary Methods. Our comprehensive findings are described in the ‘Results’ section. Step I: iPSC-derived neurons of three sample groups (6 controls [CT], 6 BD Li responders [BD.LR], and 5 BD Li non-responders [BD.NR]; all samples were obtained from male subjects) were tested with in vitro lithium. Part of step I is modified from Welham, et al. (2015) [81]. Step II: RNA-seq pipeline was used for analyzing transcriptomic profiles of all three sample groups under Li-treated (Li+) and untreated (Li-) conditions. The RNA-seq analyses were classified into ‘within-group’ and ‘across-group’. Forty-one significantly DE genes were identified in LR vs NR regardless treatment conditions. Step III: The PGBD [12]/VA GWAS for lithium response in BD (n = 256) was used in a GWA-boosting (GWAB) analysis [31], which identified the top 5% of GWAB-prioritized genes (1119 genes). Step IV: Network propagation analysis [20] of the top 500-proximal gene network derived from the 41 DE genes revealed a highly significant overlap with the 1119 GWAB-prioritized genes, containing 103 genes (73 GWAB and 30 DE/seed) genes that were significantly proximal to the 41 seed genes with a hypergeometric P of 1.28E–09. Step V: Functional enrichment analysis of gene clusters and KEGG pathways in the top 500-proximal gene network identified 189 terms significantly enriched in three clusters and 37 significantly enriched KEGG pathways for lithium response in BD. b Distribution of RNA-seq genes identified in iPSC-derived neurons. Bar plots of serial filtering RNA-seq genes for ‘within-group’ (left panel) and ‘across-group’ (right panel) analyses. Each analysis was categorized into subgroup comparisons (total n = 9; see comparison details in Supplementary Methods). A total of 13 691 RNA-seq genes for each subgroup comparison were removed by serial filters. The RNA-seq genes that passed the serial filtering threshold of CPM > 1, P-value < 0.05, log₂FC ≥ |1|, and B-H q-value ≤ 0.20 were considered as ‘differentially expressed’ (DE) genes. Each subplot shows a plot of the serial filters (y-axis) against the number of genes (x-axis). RNA-seq genes identified in subgroup comparisons with detailed expression data are presented in Supplementary Tables 4–7. c, d Distribution and expression profiles of 45 DE genes identified in across-group analysis of iPSC-derived neurons. A total of 45 DE genes (P < 0.05, log₂FC of ≥ |1|, and B-H q-value ≤ 0.20) were identified in ‘across-group’ analysis: 39 genes in untreated (Li-) condition and 31 genes in Li-treated (Li+) condition. Note that out of 45 DE genes, almost all (n = 43) were found in LR vs NR comparisons: 37 genes in untreated condition and 28 genes in Li-treated condition, including 22 genes that overlapped between them. Lists of DE genes with detailed expression data under two treatment conditions are presented in Supplementary Table 8. c Venn diagrams of 45 DE genes comparisons among CT, BD LR, and BD NR under two treatment conditions. d Expression of 45 DE genes classified into subgroup comparisons under untreated (Li-) and Li-treated (Li+) conditions. Each subgroup comparison displays each graph comprising one heatmap and one bar plot. Heatmaps (left panel) display hierarchical clustering of gene expression levels for sets of DE genes. Gene symbols (rows) are listed and indicate direction of regulation (down-regulated, light red; up-regulated, light blue). The color scale (top left) represents the degree of differential expression (low, blue; high, red). The color boxes above the heatmaps represent sample groups (CT, green; BD.LR, blue; BD.NR, red). Texts (columns) below the heatmap represent samples and are colored by treatment conditions (Li-, dark blue; Li+, orange). Bar plots (right panel) present FC expression values (log₂ transformed) and significance of gene expression (P < 0.05). The color scale (right) represents the degree of significance in expression for each gene (low, dark purple; high, yellow), displayed as nominal P-values (-log transformed). e RNA-seq validation of selected DE genes using RT-qPCR. Bar plot and scatter plot show a FC expression comparison of RNA-seq and RT-qPCR results for the four selected DE genes (out of total 37; Fig. 1d; Supplementary Table 8a) in Li-.LR vs Li-.NR. Bar plot (left panel) represents relative FC expression values (y-axis) of the selected four DE genes (x-axis), measured by RNA-seq (grey bars) and RT-qPCR (yellow bars; Supplementary Fig. 4): HEY1, KLF10, PTP4A3 were down-regulated; and POU3F1 was up-regulated. FC values were presented as in log₂ unit. RT-qPCR were calculated using the 2^−ΔΔCt method. Error bars represent the mean ± SEM of triplicate RT-qPCR data. Scatter plot (right panel) shows a high correlation (Pearson R² = 0.971, P = 0.015) of log₂FC expression between RNA-seq (x-axis) and RT-qPCR (y-axis) methods. B-H Benjamini and Hochberg, CPM counts-per-million, CT controls, DE differentially expressed, FC fold-change, GWAS genome-wide association study, iPSC induced pluripotent stem cell, KEGG Kyoto Encyclopedia of Genes and Genomes pathways (https://www.genome.jp/kegg/pathway.html), LR BD Li responders, NR BD Li non-responders, RNA-seq RNA-sequencing, RT-qPCR reverse transcription quantitative real-time PCR, SEM the standard error of the mean.