Highlights
•An efficient pipeline to generate LoF alleles in hiPSC using DNA base editing
•22 LoF alleles of neuropsychiatric risk genes were generated on 3 hiPSC lines
•The engineered LoF alleles undergo nonsense-mediated decay or protein truncation
•The SSPsyGene MiNND project provides a rich resource of LoF cell lines and data
Summary
Translating genetic findings for neurodevelopmental and psychiatric disorders (NPDs) into actionable disease biology would benefit from large-scale and unbiased functional studies of NPD genes. Leveraging the cytosine base editing (CBE) system, we developed a pipeline for clonal loss-of-function (LoF) allele mutagenesis in human induced c (hiPSCs) by introducing premature stop codons (iSTOP) that lead to mRNA nonsense-mediated decay (NMD) or protein truncation. We tested the pipeline for 23 NPD genes on 3 hiPSC lines and achieved highly reproducible, efficient iSTOP editing in 22 genes. Using RNA sequencing (RNA-seq), we confirmed their pluripotency, absence of chromosomal abnormalities, and NMD. Despite high editing efficiency, three schizophrenia risk genes (SETD1A, TRIO, and CUL1) only had heterozygous LoF alleles, suggesting their essential roles for cell growth. We found that CUL1-LoF reduced neurite branches and synaptic puncta density. This iSTOP pipeline enables a scaled and efficient LoF mutagenesis of NPD genes, yielding an invaluable shareable resource.
Introduction
In the past decade, genome-wide association studies (Schizophrenia Working Group of the Psychiatric Genomics Consortium, 2014; Grove et al., 2019; Howard et al., 2019; Meng et al., 2024; Mullins et al., 2021; Purcell et al., 2009; Ripke et al., 2011, 2013; Shi et al., 2009; Stahl et al., 2019; Stefansson et al., 2009; Trubetskoy et al., 2022; Wray et al., 2018) and whole-exome sequencing studies (Satterstrom et al., 2020; Singh et al., 2022) on neurodevelopmental and psychiatric disorders (NPDs), such as schizophrenia (SZ), autism spectrum disorder, bipolar disorder, and major depression, have identified a growing number of risk genes. However, translating these exciting genetic discoveries into translational actionable biology has been impeded by our limited knowledge of gene function and related disease mechanisms. A bottleneck in the field is that genes are often studied individually, slowing the progress and posing potential bias in functional interpretation. To overcome such limitations, the NIMH (National Institute of Mental Health)-initiated SSPsyGene (Scalable and Systematic Neurobiology of Psychiatric and Neurodevelopmental Disorder Risk Genes) Consortium (sspsygene.ucsc.edu) aims to functionally characterize the contribution of 150–250 NPD genes. The selected NPD genes mostly have disease-associated rare protein-truncating variants (PTVs) that likely cause gene loss of function (LoF) (Palmer et al., 2022; Satterstrom et al., 2020; Singh et al., 2022) and have strong effect sizes (sspsygene.ucsc.edu/resources), which will help interpret their individual biological relevance and determine any convergent or divergent biology across disorders. A large-scale, unbiased, and parallel study of these NPD genes in disease-relevant model systems will substantially deepen our understanding of the pathophysiology of NPDs.
Human induced pluripotent stem cells (hiPSCs) and their derived neural cells empowered by CRISPR-mediated gene editing provide promising cellular models for studying NPD genes (De Los Angeles et al., 2021; Duan, 2023; Michael Deans and Brennand, 2021; Muhtaseb and Duan, 2022; Wang et al., 2020; Wen et al., 2016) and for scaling up the assay. A “cell village” approach (Wells et al., 2023) enables the co-culture of tens to hundreds of hiPSC lines in a dish together, followed by assaying a specific cellular phenotype and being able to genetically infer individual cell identity. Such cell village approach may be combined with pooled screening using CRISPRi (Holtzman and Gersbach, 2018) or CRISPRoff (Nunez et al., 2021) to scale up the number of targeted genes for LoF assay. While an invaluable approach, the pooled CRISPR screening in hiPSC-derived neural models is limited by cell line-specific or LoF allele-specific unequal cellular growth, possible non-autonomous effects, and restrictive phenotypes amenable for screening.
CRISPR-Cas9 editing can be used to systematically create small DNA insertions or deletions (indels) or exon deletions in protein-coding regions through non-homologous end joining repair of double-strand breaks (DSBs) (Ran et al., 2013), resulting in protein-truncating mutations. Alternatively, LoF mutation can be generated by using CRISPR-based cytosine base editors (CBEs) to introduce premature protein stop codons (i.e., nonsense mutations; an iSTOP approach) that lead to mRNA nonsense-mediated decay (NMD) and/or protein truncates (Billon et al., 2017; Cuella-Martin et al., 2021; Hanna et al., 2021; Xu et al., 2021). Compared to the traditional CRISPR-Cas9 gene editing system, the CBE makes “C” to “T” changes in DNAs without creating cell-toxic DSBs as the Cas9 nuclease does (Ran et al., 2013) and with minimized potential off-target DNA editing (Billon et al., 2017; Cuella-Martin et al., 2021; Hanna et al., 2021; Xu et al., 2021). Furthermore, compared to the CRISPR-Cas9 editing-induced small indels that may or may not disrupt a protein sequence reading frame, a CBE can precisely introduce a premature stop codon, which makes the clonal LoF allelic confirmation more straightforward and cost-effective in a scaled LoF mutagenesis workflow. Finally, the CBE-engineered premature stop codon mutations are more reminiscent of the rare patient-specific PTVs or LoF mutations associated with NPD (Satterstrom et al., 2020; Singh et al., 2022). Recently, a DNA base editing reporter gene system has been developed to enrich the edited cells, thereby increasing the base editing efficiency of a target gene (Standage-Beier et al., 2019), including in hiPSCs (Tekel et al., 2021). However, the use of a CBE in editing hiPSC lines has been scarce (Sürün et al., 2020), and its usefulness in developing a scaled and efficient clonal LoF mutagenesis in hiPSCs has not been tested.
As part of the SSPsyGene Consortium, our Assay and Data Generation Center for the Model of iPSC-derived Neurons for NPD (MiNND) aims to employ the CBE-based iSTOP approach to generate isogenic hiPSC lines carrying LoF alleles for about 150–200 NPD genes on multiple donor genetic backgrounds. Here, leveraging an improved reporter gene editing enrichment system that can substantially increase the CBE iSTOP editing efficiency in hiPSCs, we established a semi-automated pipeline for parallel and efficient clonal LoF mutagenesis of a large number of genes. We tested the workflow on 23 NPD genes with 3 donor hiPSC lines (KOLF2.2J, CW20107, and MGS_CD14). We obtained high and reproducible iSTOP editing efficiency across all three hiPSC lines. We systematically characterized the engineered isogenic iSTOP hiPSC lines for pluripotency, karyotyping, neuron differentiation capacity, and the expected NMD and LoF.
Results
The CBEmax DNA base-editing enriching system substantially increases “C” to “T” editing in hiPSCs
A key for generating LoF alleles by using a CBE to introduce premature stop codons (C to T changes; i.e., iSTOP approach) (Billon et al., 2017; Popp and Maquat, 2016) on a large scale is to have sufficiently high gene editing efficiency. Although DNA base editors have high SNP editing efficiency (>50%) in some commonly used cell lines such as HEK293 (Rees and Liu, 2018), hiPSCs are less tested. We opted to employ a base editing reporter gene system to enrich the gene-edited cells (Standage-Beier et al., 2019), thereby increasing iSTOP editing efficiency of a target gene in selected cells. In this CBE editing enriching system (CBEmax_Enrich), a blue fluorescent protein (BFP) reporter on the reporter plasmid pEF-BFP will turn into a functional EGFP reporter when it is edited from CAC (H66) to TAC (Y66) in cells co-transfected with pEF-AncBE4max and single-guide RNAs (sgRNAs) (Figure 1A). We first individually tested the two iSTOP sgRNAs (Table S1) that target the apolipoprotein E (APOE) gene in HEK293 cells. We transiently co-transfected HEK293 cells with CBEmax_Enrich and sgRNA construct, followed by fluorescence-activated cell sorting (FACS) to enrich GFP+ cells (i.e., with the reporter gene edited) for testing editing efficiency by Sanger sequencing (Figure 1B). For each sgRNA, we found a substantial increase in the target gene editing efficiency (C to T) in FACS-sorted GFP+ cells compared to the transfected BFP+ cells (from 42% to 86% and from 29% to 81%, respectively) (Figure 1B).
Next, we similarly tested for the iSTOP editing efficiency in two hiPSC lines (Figure 1C) and whether the introduced iSTOP codons led to the expected NMD (i.e., LoF) (Figure 1D). For both iSTOP sgRNAs, we observed a robust increase, although to a less extent than in HEK293, of the target gene editing efficiency in FACS-sorted GFP+ cells compared to the transfected BFP+ cells (from 24% to 66% and from 5% to 31%, respectively) (Figure 1C). More importantly, as expected from the iSTOP-mediated NMD of mRNAs, we found 86% and 98% of APOE expression reduction in hiPSC clones homozygous for iSTOP1 and iSTOP2, respectively, and ∼50% expression reduction in hiPSC clones heterozygous for iSTOP mutations (Figure 1D).
Taken together, these results show that the CBEmax_Enrich system can significantly increase the iSTOP editing efficiency, which enables us to generate LoF alleles on a large scale by introducing premature stop codons.
A scalable workflow for efficiently deriving clonal LoF alleles in hiPSCs using CBEmax_Enrich
Our goal is to develop an efficient pipeline that involves single hiPSC cell sorting for deriving clonal LoF alleles of hundreds of NPD genes in multiple hiPSC lines. To achieve this goal, a key is to obtain a relatively high single hiPSC clonal survival rate after FACS of the enriched GFP+ cells (Figures 1B and 1C). It has been recently shown that the CEPT small molecular cocktail can increase single hiPSC cloning efficiency compared to ROCK inhibitor (Y-27632; ROCK-I) (Tristan et al., 2023). We thus tested the performance of CEPT by treating the hiPSCs with CEPT both during CBEmax_Enrich transfection (for iSTOP sgRNAs of 4 genes) and the FACS sorting of single cells into 96-well plates 48 h post-transfection. However, we observed a very low single hiPSC clonal survivability (∼5%) despite a high editing efficiency (∼70%) (Figures S1A and S1B). Combining routine ROCK-I treatment of hiPSCs at transfection with CEPT treatment during 48 h post-transfection cell sorting gave us a much higher single hiPSC clonal survivability (∼27%), and even higher survivability (∼35%) when we sorted cells 72 h post-transfection while maintaining high gene editing efficiency (Figures S1A and S1B).
After these optimizations to achieve high gene editing efficiency and single hiPSC clonal survivability, we designed a semi-automated pipeline for deriving clonal LoF alleles in hiPSCs for 23 NPD genes for each batch (Figure 2A). Briefly, the CBEmax_Enrich vector, the reporter BFP plasmid, and the sgRNA vector carrying the reporter sgRNA and a targeting sgRNA were transiently transfected into hiPSCs in a 24-well plate, each well with one of the 23 targeted LoF mutations or a non-transfection (sgRNA)-control (NTC) for 1 donor hiPSC line. We then sorted out GFP+ cells that were enriched for base editing and distributed 96 single cells per gene/LoF in a 96-well plate. A handful of single hiPSC colonies from each 96-well plate were further subjected to Sanger sequencing to verify the C to T changes (LoF allele). Then 2–3 hiPSC clones, preferably homozygous for a LoF allele, were banked. The selected hiPSC clones were also subjected to RNA sequencing (RNA-seq) to confirm the absence of chromosomal abnormality by eSNP-Karyotyping and pluripotency test. With this pipeline, we have generated LoF alleles, mostly homozygous, for 22 of the 23 selected SSPsyGene Consortium-prioritized NPD genes (no editing found for HERC1, Table S3), including 9 (ARID1B, CACNA1G, CHD8, DLL1, GABRA1, KMT2C, SCN2A, SHANK3, and SMARCC2) out of the 10 “capstone genes” (genes prioritized by the SSPsyGene Consortium to be tested for technical consistency across all consortium sites) for which we could design sgRNAs, on 3 donor lines of European ancestry (KOLF2.2J, CW20107, and MGS_CD14) (Figure 2B; Table S2). Our MiNND project within the SSPsyGene Consortium aims to produce LoF alleles for about 150–200 NPD genes on 6 different hiPSC lines. The derivation of many iSTOP LoF alleles enables us to systematically evaluate the performance of iSTOP base editing on hiPSCs and its efficiency in leading to LoF….
