CRISPR-Enabled functional genomics in hPSCs-derived neural models for autism spectrum disorder

Abstract

Autism Spectrum Disorder (ASD) is a genetically heterogeneous neurodevelopmental condition in which hundreds of individually rare risk variants converge on a small number of shared biological pathways, including synaptic scaffolding, chromatin remodeling, excitation-inhibition balance, and cellular energy metabolism. Translating this genetic heterogeneity into mechanistic insight requires experimental systems capable of interrogating individual gene functions in human-relevant neural contexts at scale. CRISPR-enabled functional genomics in human pluripotent stem cell (hPSC)-derived neural models, spanning neural progenitors, cortical and inhibitory neurons, astrocytes, microglia, and brain organoids, provides precisely this capability. By integrating pooled perturbation screens with multimodal readouts including single-cell and spatial transcriptomics, chromatin accessibility profiling, proximity labeling proteomics, multi-electrode array electrophysiology, and metabolic flux analysis, these platforms enable systematic, causal mapping of ASD gene function at system resolution. Early applications have already revealed convergent mechanisms: BAF complex disruption expands the ventral progenitor pool and biases its fate toward oligodendrocyte and interneuron lineages; ADNP loss impairs microglial synaptic pruning through altered endocytic trafficking; and mTOR pathway dysregulation in PTEN- and TSC2-perturbed models links genetic risk directly to metabolic and mitochondrial dysfunction. Computational frameworks including MIMOSCA and SCEPTRE enable causal network reconstruction and pseudotime inference from these datasets, moving the field from gene lists toward pathway-level models of ASD pathobiology. Translational applications leverage isogenic iPSC panels and variant-level base and prime editing to stratify ASD variants by functional impact, informing gene therapy design for haploinsufficient targets such as CHD8 and SCN2A via AAV or antisense oligonucleotide delivery. Remaining challenges, including model developmental immaturity, batch variability, and the difficulty of modeling polygenic risk, are addressed by a roadmap integrating spatial perturbomics, AI-driven causal inference, and population-scale standardized biobanks. This review synthesizes the current state of CRISPR-based functional genomics in human stem cell neural models as a coherent experimental framework for converting ASD genetic associations into mechanistic understanding and therapeutic opportunity. Patient-derived induced pluripotent stem cells (iPSCs) are reprogrammed and differentiated into neuronal, glial, and three-dimensional organoid models, representing disease-relevant cell types. Genome-scale pooled CRISPR perturbations including knockout (CRISPR-Cas9), repression (CRISPRi), activation (CRISPRa), and precise editing (base and prime editors) are applied to dissect ASD-associated genes and variants. High-throughput readouts, including single-cell multi-omics (scRNA-seq, scATAC-seq), spatial transcriptomics, electrophysiology, and high-content imaging, capture molecular and cellular phenotypes. Computational pipelines integrate multi-modal data, applying AI/ML for causal inference, network modeling, and prioritization of high-confidence targets. Translational workflows validate targets through isogenic rescue assays, small-molecule and biologic screens, and gene-therapy design (AAV, antisense oligonucleotides, base/prime editing). Ethical, legal, and social implications (ELSI) are embedded across the pipeline, including data privacy for patient-derived iPSCs, responsible governance of organoid models, and stakeholder engagement to ensure socially responsible translation. Collectively, these platforms bridge genetic associations to mechanistic understanding and therapeutic innovation in ASD, enabling precision medicine while upholding ethical and community standards.