Integrating genomic data and patient specific iPSC models to investigate disease mechanisms in autism spectrum disorder

Autism spectrum disorder (ASD) and related neurodevelopmental disorders (NDDs) are clinically and genetically heterogeneous conditions arising from the interplay between rare high penetrance variants, a broader polygenic background and environmental and epigenetic factors. Genetic risk architectures range from Mendelian like cases with a single major rare variant, to individuals carrying multiple rare high impact variants in neurodevelopmental genes, to scenarios in which a rare variant acts on an unfavourable polygenic background, and finally to highly polygenic cases without identifiable rare high impact variants. Within this framework, the thesis combines exome and genome sequencing, copy number variant (CNV) analysis, X chromosome inactivation (XCI) studies and functional models based on human induced pluripotent stem cells (iPSCs) to dissect complex genetic architectures in ASD and NDDs. In a highly selected cohort of 113 individuals with ASD and intellectual disability, after exclusion of Fragile X syndrome and pathogenic CNVs at chromosomal microarray, whole exome sequencing (ES) yielded a pathogenic or likely pathogenic variant in 11 patients (~10%), in line with similarly ASD/ID series. Beyond these solved cases, some patients carried multiple rare variants converging on shared pathways or variants of uncertain significance in biologically plausible genes, highlighting the importance of going beyond a purely diagnostic view of ES and using it as a starting point for mechanistic hypotheses on gene–gene and pathway level interactions. To explore the contribution of CNV “multi hit” architectures, the work focuses on seven patients harbouring recurrent NDD associated CNVs (NRXN1 deletion, 15q11.2 microdeletion or 16p11.2 duplication) together with additional rare CNVs of independent parental origin. In these individuals, secondary CNVs disrupt genes involved in synaptic adhesion, glutamatergic transmission, axon guidance, calcium signalling, cytoskeletal dynamics, epigenetic regulation and neuronal metabolism, with functional enrichment analyses revealing convergence on pathways such as nervous system development, axon guidance, synaptic transmission and ion transport. Clinically, the resulting phenotypes only partially overlap with canonical syndromes associated with the primary CNV, supporting an oligogenic “multi hit” model in which recurrent CNVs act as major vulnerability factors modulated by secondary genomic hits rather than as isolated causes. The thesis also investigates X chromosome inactivation as a physiological modifier of X linked variant expressivity. In a cohort of 98 females (probands and mothers), skewed or severely skewed XCI patterns were observed in approximately one quarter of individuals, in line with population based datasets. Integrating XCI assays with the X pose computational pipeline, which combines DNA and RNA seq data to infer allele specific expression on the X chromosome, enabled a refined interpretation of X linked variants, exemplified by the re evaluation of a de novo truncating SLC6A8 variant in a female proband in the context of non random XCI. These findings underscore the value of XCI profiling as a complementary tool that can unmask or down weight the clinical impact of X linked variants in diagnostic settings. A central part of the work is a detailed case study of a male patient (OS5467) with severe neurodevelopmental regression and drug resistant epilepsy, carrying rare variants in three DNA damage repair genes: FAN1, ERCC2 and SFR1. Functional studies in lymphoblastoid cell lines demonstrate reduced FAN1 protein levels in the proband and his carrier mother and marked an hypersensitivity to the interstrand crosslinking agent mitomycin C. Short and long read genome sequencing in the proband further revealed an excess of somatic short tandem repeat (STR) expansions, particularly at AT rich motifs and within or near neurodevelopmental genes such as SHANK2 and FHIT, with enrichment of affected loci in neuronal and synaptic pathways. Together, these data support the presence of a FAN1 centred DNA repair defect that promotes somatic instability of repetitive DNA and may contribute, in combination with other DDR variants, to the patient’s regressive phenotype. To functionally interrogate polygenic and oligogenic architectures, the thesis employs iPSC based models in selected cases. In BF1498, who carries multiple inherited variants in nine genes mapping to the KEGG “glutamatergic synapse” pathway (including SHANK1, SHANK2, SHANK3, GRIN2A, GRIA1, SYNGAP1, APBA2, KCNJ10 and LRP1), patient specific iPSCs were generated and differentiated into both mixed cortical cultures via neural progenitor cells and NGN2 induced glutamatergic neurons, co cultured with rat astrocytes on multielectrode arrays. This platform was established to test whether a burden of modest effect variants converging on excitatory synapses translates into measurable alterations in neuronal maturation and network activity compared with controls, even in the absence of a single high impact mutation. Finally, as part of a collaborative project in Professor Gaia Novarino’s laboratory at the Institute of Science and Technology Austria (ISTA), this thesis contributed to an organoid-based platform aimed at investigating cellular and molecular mechanisms underlying developmental and epileptic encephalopathies, including SLC13A5-related disease. Cortical organoids were generated from control, patient and CRISPR-corrected iPSC lines according to established protocols, and my work focused on the histological and morphological characterisation of both in vitro organoids and human organoids transplanted into neonatal rat cortex, using stage-specific immunohistochemistry for progenitor, neuronal and glial markers at multiple time points. In addition, I helped to set up sparse viral labelling and three-dimensional imaging strategies, including tissue clearing, to visualise single-neuron morphology in vitro, and I performed three-dimensional reconstruction and Sholl analysis on labelled neurons in engrafted organoids, revealing complex dendritic arbors consistent with morphologically mature human neurons in a three-dimensional context. Together with the iPSC-derived two-dimensional neuronal cultures used elsewhere in the thesis, these experiments illustrate how complementary human cellular models, ranging from mixed neuronal networks to three-dimensional organoids and in vivo grafts, can be combined to capture distinct facets of neurodevelopmental pathology and provide a flexible methodological framework for dissecting disease mechanisms in genetically complex NDD.