H1 hESCs and all derivative lines were cultured in feeder-free condition. All hESCs were maintained on Matrigel (BD Biosciences) coated tissue cultures with mTeSR1 media (STEMCELL Technologies), and were passaged every 7 days via Dispase (0.83 U/mL) treatment. All experiments involving hESCs were approved by the Yale Embryonic Stem Cell Research Oversight Committee (approval number: E-13-077) (Embryonic Stem Cell Research Oversight).

The sgRNAs (Set1 oligo 1: 5’-CACCGGGCTCACTGT TGGCTATTTC-3’, oligo 2: 5’-AAACGAAATAGCCAACAGTGAGC-3’, Set2 oligo 1: 5’-CACCGTAATTGACAAGGCTCACTGT-3’, oligo 2: 5’-AAACACAGTGAGCCTTGTCAATTAC-3’) were designed by using CRISPR Design Tool (https://chopchop.cbu.uib.no/) and then cloned into pSpCas9(BB)-2A-Puro (PX459) plasmid (Fig. 1). The desired point mutation was introduced by designing a single stranded oligonucleotide donor (ssODN) containing a ≥45 bp homologous arm at each side. H1 hESCs were singularized with Accutase (100 mL, Cat# AT104; STEMCELL Tech-nologies), and 2 million single cells were electroporated with 2 μg of PX459-sgRAN and 2 μg of ssODN using the Human Stem Cell Nucleofector Solution 1 (Lonza VPH-5012). Cells were then seeded onto Matrigel-coated 10-cm dishes in mTeSR1 media supplemented with 10 μM Y27632 (STEMCELL Technologies). Puromycin selection was performed after 24 hours post-nucleofection and maintained for ∼7 days until stable colonies appeared. Single colonies were picked and cultured in Matrigel-coated 96-well plate. After 7 days, gDNA isolation, polymerase chain reaction (PCR) amplification of the target region and sequencing were performed. Successfully targeted clones were then expanded for downstream use.

The human induced neurons (iNs) were generated by using the protocol described earlier (15). Briefly, hESCs were singularized with Accutase, and 1×10⁴ cells were seeded on the Matrigel-coated 24-well plate supplemented with 10 μM Y27632 on Day −2 (or two days before induction) (Fig. 2A). On Day −1 (or one day before induction), hESCs were infected with lentivirus containing the FUW-TetO-NGN2 and rTTA in mTESR1 media. In the following day (Day 0), media was replaced with N2/DMEM-F12 containing 1% NEAA (Invitrogen), 10 mg/mL human BDNF (PeproTech), 10 mg/mL human NT-3 (PeproTech) and 0.2 mg/L mouse laminin (Invitrogen), and NGN2 induction is started with addition of 2 μM doxycycline (Dox). A 24 hours puromy-cin selection (1 mg/mL) was performed on Day 1. In the next day, media was replaced with the Neurobasal medium supplemented with B27 containing 1% (v/v) Glutamax (35050; Life Technologies), BDNF, NT3, and 2 g/L Ara-C (Sigma-Aldrich). After Day 2, half of the media in each well was replenished every 2 days. The efficiency of conversion of hESCs into iN cells was measured by expression assay of MAP2 and TUBB3.

The immunofluorescence analysis was performed as described previously (17). The iNs were fixed in 4% paraformaldehyde for 20 minutes at room temperature (RT) following 3 times washing with phosphate buffered saline (PBS), and incubation with 0.2% Triton in PBS at RT for 15 minutes and blocked with 3% bovine serum albumin (BSA) in PBS at RT for 1 hour. Then, cells were incubated with primary antibody diluted in 3% BSA in PBS at RT for 1 hour, incubated with secondary antibody diluted in 3% BSA in PBS at RT for 1 hour and then stained for nuclei via DAPI. The following antibodies were used for this study: human anti-MAP2 (1：500; Millipore), human anti-β-Tubulin III (1：1,000; Sigma-Aldrich), and Alexa fluor 555-conjugated donkey secondary antibodies (1：1,000; Invitrogen).

The human iNs were used for total RNA isolation via the RNeasy Mini Kit (QIAGEN). One microgram of RNA was converted to cDNA using the iScript Select cDNA Synthesis Kit. For the quantification of gene expression, quantitative PCR (qPCR) was carried out on the CFX96-Real-Time PCR system (Bio-Rad) using the SsoFast EvaGreen Supermix (Bio-Rad). The PCR conditions were: 95℃ for 15 minutes, followed by 40 two-step cycles at 94℃ for 10 seconds and 60℃ for 45 seconds. Primers used were as follows: NANOG forward: 5’-TGCTGAGATGCCTCACACGGA-3’, reverse: 5’-TTTTTGCGACACTCTTCTCTGC AGA-3’, OCT4 forward: 5’-CCTCACTTCACTGCACTGTA-3’, reverse: 5’-CAGGTTTTCTTTCCCTAGCT-3’, PAX6 forward: 5’-TGTTCCAACTGATATCGTGCCT-3’, reverse: 5’-ATGGCTGTTAGAGCCGCTTC-3’, TUBB3 forward: 5’-GCCGCTACCTGACGGTGGC-3’, reverse: 5’-GGGCGGGATGTCACACACGG-3’, FOXG1 forward: 5’-GAGGTGCAATGTGGGGAGAA-3’, reverse: 5’-TTCTCAAGGTCTGCGTCCAC-3’, NGN2 forward: 5’-ATGTTCGTCAAATCTGAGA-3’, reverse: 5’-CTAGATACAGTCCCTGGCG-3’, FYN forward: 5’-AAGATGCTGAGCGACAGCTA-3’, reverse: 5’-GAATAGGCACCTTTGGTGGT-3’, FOXP2 forward: 5’-CTGTGCAAGCACCATAGAAACA-3’, reverse: 5’-TGGCTGCTTCTGTCCTTGAG-3’, NCOA3 forward: 5’-GGACAAATGAGACCCAAAGAC-3’, reverse: 5’-AATCTTCCCCTTCCTCCATC-3’, CDH2 forward: 5’-CATCATTGCCATCCTGCTC-3’, reverse: 5’-TCTTCTTCTCCTCCACCTTC-3’, MSL3 forward: 5’-ATGCCAGACCAACATCATAAC-3’, reverse: 5’-TCCATCCACCATCTCCTTAC-3’, vGAT forward: 5’-AACGCCATCCAGGGCATGTT-3’, reverse: 5’-CCTCGCCG TCTTCATTCTCC-3’, vGLUT1 forward: 5’-AGCTGGG ATCCAGAGACTGT-3’, reverse: 5’-CCGAAAACTCTGTTGGCTGC-3’, vGLUT2 forward: 5’-TCAGATTCCGGGAGGCTACA-3’, reverse: 5’-TGGGTAGGTCACACCCTCAA-3’, GATA4 forward: 5’-CGAGGAGATGCGTCCCATCAAGAC-3’, reverse: 5’-AGTCCTGCTTGGAGCTGGTCTGTG-3’, SOX17 forward: 5’-TCGGGGACATGAAGGTGAAG-3’, reverse: 5’-TAGCCCACACCATGAAAGC-3’, FYN-P1: 5’-GGGCCCAGTTTGAAACACTTC-3’, FYN-P2: 5’-CAACGAACGACGTGCAACTT-3’, FYN-P3: 5’-GGTAAGCCTTGGCATCCCTT-3’.

For lentivirus production, each plasmid (FUW-TetO-Ngn2) expressing full-length mouse Ngn2, and FUW-rTTA plasmid was transfected along with GAG-POL and VSV-G vectors into HEK293T cells with 70% confluence by using X-tremeGENE9. The medium was collected after 48 and 72 hours post-transfection. After filtration and concentration, the lentivirus was titrated in HEK293T cells and then puromycin selected to measure the titer.

SNP T137346C is defined in an old human genome manner. We found its exact location before performing fur-ther experiments. To find its location, in silico PCR was performed using UCSC Genome Browser via primers reported previously (8). T137346C appears most likely to be rs706895, which is located in the intron rather than in 5’UTR. We studied the polymorphism of FYN in human primary cells and used genomic DNA as a template for PCR to clone the 250 bp upstream and 250 bp downstream of rs706895 into pGL3-basic luciferase vector. We successfully cloned two reporters with the T allele (FYN-rs706895*T) or the C allele (FYN- rs706895*C) and performed Dual-Luciferase Reporter Assay (Promega). Briefly, luciferase reporters were transfected into 293T cells with renila control plasmid. At 24 hours of transfection, cells were 50 mM ethanol treated for 5 minutes up to 4 days, and lysed to measure the firefly luciferase. Renila activity was used for reporter gene expression normalization.

Human genomic DNA sequences, CpGI and RefSeq gene coordinates (version hg19) were downloaded from UCSC Genome Browser. RNA-Seq reads were mapped to the human genome by Tophat2 (v2.1.0) with default parameters (18). Gene expression level (FPKM value) was calculated by Cufflinks (v1.2.1) using RefSeq genes as reference annotation (19). For RNA-Seq read validation, alignment was performed using the STAR package (v2.7.11b) to generate BAM files (20). To calculate TPM, the BAM files were sorted using Samtools (v1.20) (21), and transcript abundance was quantified using Salmon (v1.10.3) (22). Differe-ntially-expressed genes were identified with a criteria of more than 2 fold changes of gene expression between CC-iN, CT-iN, or TT-iN with and without ethanol treatment. Gene Ontology (GO) analysis were performed by GOstats bioconductor package in R. False discovery rate (FDR) was then estimated by Benjamini-Hochberg method with p adjust function in R. GO terms with FDR<0.05 were used as statistical significance. Enrichment of the differentially-expressed genes in iNs was analyzed using the GSEA software (v2.2.2 from Broad Institute) by 1,000 permutations of gene set and weighted statistic without collapsing dataset. We counted the number of RNA-Seq reads in each exon-exon junction of FYN and divided them by total number of RNA-Seq reads. To draw a heatmap, we calculate z-score of the normalized read count.

To study the effect of FYN*rs706895 on FYN expre-ssion, we use the CRISPR-Cas9 tool to edit rs706895 in H1 hESCs. H1 hESCs are homozygous for FYN SNP rs706895*C (CC-H1). We designed a guide RNA (sgRNA) to target the rs706895 locus as well as a single-stranded DNA template to replace the rs706895 C allele of H1 with the rs706895 T allele (Fig. 1A). After introducing CAS9, sgRNA, and the template into H1 cells and selected with puromycin, we isolated two clones with T/C (TC-H1) and two clones with T/T (TT-H1) from CC-H1 clones (Fig. 1A). The pluripotency of each hESC clone with different rs706895 alleles was confirmed by staining with pluripotency markers (Fig. 1B). Karyotype analysis showed that all edited clones carried chromosome 12 trisomy (Supplementary Fig. S1), which is a predominant cytogenetic abnormality in human pluripotent stem cells (23). Despite trisomy 12, same karyotypes among the clones enable studying the impact of FYN genotypes. Thus, RT-qPCR showed that CC-H1, CT-H1, and TT-H1 hESC lines expressed pluripotency genes (OCT4, NANOG, KLF4, and TFCP2L) as well as FYN (Fig. 1C) at similar levels. Collectively, we obtained hESC lines containing different FYN rs706895 genotypes (CC/, C/T, and T/T) with similar pluripotency.

To determine whether rs706895 regulates the expression of FYN in human neurons, we induced neural differentiation of the edited and wild-type H1 hESC lines. To expedite the neurogenesis and to obtain pure excitatory neurons, we utilized the Ngn2 overexpression system by which induction of Ngn2 by rTTA over 7∼10 days converted hESCs into cortical excitatory neurons (Fig. 2A) (15). We found that all hESC lines were readily differentiated into neurons (Fig. 2B, left). We stained the iNs with neuronal markers, TUBB3 and MAP2, to quantify the neurogenesis of hESC lines. TUBB3 and MAP2 staining showed that CC-H1, CT-H1 and TT-H1 iNs showed equally efficient neurogenesis (Fig. 2B, Right). All iNs derived from H1 lines with the FYN variant showed a complete absence of pluripotency markers, such as NANOG and OCT4, but expressed well-known neuronal markers, including PAX6,TUBB3, and FOXG1 (Fig. 2C). Moreover, we examined the expression of FYN and found that its transcription had not significant difference in TT-iNs compared to CC- and CT-iNs (Supplementary Fig. S2A).

Numerous studies have demonstrated that SNPs disrupt the TF binding by either destroying a binding the site for TFs or creating a binding site for TFs, in turn, associated with global transcriptome regulation (24). Thus, we investigated the role of FYN SNP rs706895 in altering transcriptome profiles of iNs by RNA-Seq (Fig. 3). Hierarchical clustering was performed to characterize the transcrip-tome patterns in CC-, CT-, and TT-iNs (Fig. 3A). The clustering dendrogram distinguished TT-iNs from CC- and CT-iNs (Fig. 3A). We further compared genome-wide expression patterns of CC-, CT-, and TT-iNs. There were 456 up- and 46 down-regulated genes in iNs with CC or CT genotypes compared to iNs with the TT genotype (Fig. 3B). qPCR analysis further confirmed the down-regulation of NCOA3, CDH2, FOXP2, and MSL3 and up-regulation of vGAT in TT-iNs compared to CC- and CT-iNS (Sup-plementary Fig. S2B). GO analysis revealed that the upregulated genes in iNs with CC or CT genotypes are involved in cerebral cortex, pallium, histone, and forebrain development as well as covalent chromatin modification (Fig. 3C). The accumulation of transcriptome datasets indicates that FYN has multiple FYN transcripts with additional exons as well as with the first exon of FYN extended and the original 5’ UTR SNP rs706895 being placed in intron 4 of the longest FYN transcript. To determine whe-ther SNP rs706895 affects the splicing variants of FYN, we examined splice junctions within FYN by following the methods described previously (25). In total, 15 spliced junctions in FYN were identified, as potentially regulated through SNP rs706895 (Fig. 3D). We simplified names of the FYN transcripts with different lengths as FYN-l (FYN-long, ENST00000354650), FYN-m (FYN-medium, ENST00000229471), and FYN-s (FYN-short, ENST00000538466) (Supplementary Fig. S2C). Overall, the expression of FYN did not show difference among CC-, CT-, and TT-iNs (Fig. 3E). Finally, we performed direct comparison of CC-, CT-, and TT-iNs to determine whether FYN SNP rs706895 influences the expression of genes related with alcoholism (26). The expression of these alcoholic genes was reduced in TT-iNs compared to CC- or CT-iNs (Fig. 3F, Supplementary Table S1). Collectively, FYN SNP rs706895 moderates the expression of genes involved in neuronal development, particularly alcoholism, as well as those involved in splicing junctions of FYN.

To examine ethanol-induced changes in FYN expre-ssion, we treated hESC-derived iNs carrying different genotypes of SNP rs706895 with 50 mM ethanol for 24 hours. iNs from CC-H1 showed a dramatic induction of FYN by ethanol treatment (Supplementary Fig. S2D). In order to examine global effect of ethanol treatment on iNs with different genotypes of FYN SNP rs706895, we performed RNA-Seq. We assessed whether there was any correlation between differentially expressed genes with ethanol treatment across all lines. We found the least correlation of ethanol-induced gene expressions between CC- and TT-iN cells (correlation coefficient=0.059; Fig. 4A), among the comparisons of other cell lines. These results imply that CC-iNs and TT-iNs respond to ethanol treatment in a most distinct manner. The GO analysis of differentially expressed genes with ethanol treatment demonstrated that genes related to synaptic signaling, neurotransmission, and amino acid transports were highly enriched in TT-iNs compared to those in CC- and CT-iNs (Fig. 4B). In contrast, development-related genes were largely depleted in TT-iNs upon ethanol treatment (Fig. 4B). Indeed, qPCR results further confirmed the induction of vesicular glutamate transporters, vGLUT1 and vGLUT2, and reduction of developmental related genes, SOX17 and GATA4 in TT-iNs compared to CC- and CT-iNs with ethanol treatment (Supplementary Fig. S2E). We also investigated whether ethanol treatment affected the gene regulation of specific neuronal subtypes, which were previously described by single-cell transcriptome of human brains (27). Upon ethanol treatment, TT-iNs demonstrated a significant up-regulation of genes specific for inhibitory neurons classified as inhibitory-3 (Inh) subgroup (SST^＋, RELN^＋, and CALB2^＋) (Fig. 4C). Lastly, expression of FYN isoforms were investigated in all three types of iNs (with C/C, C/T, or T/T genotypes). In TT-iN cells, the FYN-s isoform was drastically elevated after ethanol treatment (Fig. 4D, Sup-plementary Fig. S2F). Collectively, SNP rs706895 in FYN not only regulated clusters of genes involved in neuronal connection and subtypes but also FYN isoforms in res-ponse to ethanol treatment.

To determine if SNP rs706895 C and T alleles contribute differently to FYN expression, we performed a Dual-Luciferase Reporter Assay using luciferase Renilla as an internal control. We found that the reporter gene with the C allele (FYN-C) of rs706895 was associated with a higher level of the luciferase activity, further supporting the influence of SNP rs706895 on FYN expression (Fig. 4E). Fyn is known to regulate the phosphorylation of NR2B that enhances the channel activity, counteracting the inhibitory action of ethanol (28). To investigate whether there is any relationship between rs706895 at FYN and the susceptibility to alcoholism, we tested the response of the rs706895 T or C FYN reporter to ethanol treatment. We found that FYN-C reporter showed a higher level of the luciferase activity responding to ethanol treatment (Fig. 4F). Overall, these results suggest that the AD risk C allele of FYN SNP rs706895 drives a higher expression of FYN and a higher response of cortical neurons to ethanol treatment.