THE EVOLUTION OF THE MODERN CONDITION THROUGH THE LENS OF NEURODEVELOPMENT DIVERSITY: INSIGHTS FROM A PARADIGMATIC REWIRING OF GENE REGULATION

Contemporary humans evolved unique cognitive and behavioural skills that are unprecedented in other human lineages. The recent availability of high-quality Neanderthals and Denisovans genomes has enabled the identification of human-derived genetic changes, yet their phenotypic consequences remain largely uncharted. While prior work has focused predominantly on protein-coding differences, the majority of these variants are found in noncoding regions of the genome, implicating regulatory mechanisms as the prime layer that was reconfigured during the evolution of the Sapiens brain. Understanding how regulatory changes reconfigured the neurodevelopmental trajectories that underpin hallmark features of our species’ unique cognitive abilities requires the integration of computational and experimental approaches, pairing top-down genome-scale prediction with bottom-up mechanistic dissection. As a starting point for this thesis, we leveraged a published catalogue of human-derived single-nucleotide variants found at high frequency (HFVs; ≥90%) in contemporary humans, a near-fixation threshold that accommodates the extent of Homo sapiens genetic variation, archaic admixture, and the limited availability of high-quality ancestral genomes, focusing specifically on noncoding variants. We intersected HFVs with enhancers active during early corticogenesis to define cortical regulatory islands. By computing allele-specific transcription factor binding affinities at each HFV within cortical regulatory islands, we classified each motif instance as lost, gained, affinity-decreased, affinity-increased, or unaltered. This yielded a high-resolution resource that maps, for each HFV, the affected binding sites as well as the direction and magnitude of the predicted affinity change, revealing that most variants perturb multiple sites, with TF-level burdens ranging from broad reshuffling to precise single-site changes. To anchor the differential TF affinity analysis in a biologically meaningful setting, we focused on indirect neurogenesis, a differentiation trajectory that is key for human cortical expansion. We identified KLF6 as a central regulator of a cholesterol-associated gene regulatory network in radial glia, and predicted a higher affinity of GLI3 for the modern allele in an enhancer linked to KLF6, consistent with its role in balancing proliferative and differentiative radial-glia divisions. Leveraging a prioritisation pipeline that integrates prior knowledge with deep learning-based ranking of regulatory region activity, we then moved beyond prediction to a mechanistic dissection centred on CHD2, a dosage-sensitive chromatin remodeller for which both increased and decreased levels are associated with severe neurodevelopmental disorders in humans. To this end we generated an isogenic cohort of human induced pluripotent stem cells carrying either the ancestral regulatory allele in an enhancer upstream of CHD2, resulting in higher CHD2 expression, or a loss-of-function missense mutation in the coding sequence causing haploinsufficiency. High-resolution phenotyping of human iNeurons and cortical brain organoids revealed convergent perturbations involving the autophagy pathways, synaptic programs, and cell-cycle control. While lysosome number was increased in both conditions, pH and enzymatic activity showed mirrored phenotypes in ancestral versus heterozygous genotypes. At the synaptic level, we found an enrichment of differentially expressed genes associated to Autism Spectrum Disorder and epilepsy, which displayed a temporal inversion as they were up-regulated at day 25 in haploinsufficient organoids but down-regulated at day 50 in ancestral organoids. Longitudinal imaging uncovered an early acceleration of growth trajectories in both conditions followed by sustained outgrowth by controls, and cell cycle phase scoring showed exhaustion of the proliferative progenitors’ pool in haploinsufficient organoids vis-à-vis an increased mitotic potential in ancestral organoids. Finally, we validated the prediction of an evolutionarily acquired oestradiol-dependent rewiring of CHD2 regulation, providing a mechanistic understanding of how a regulatory variant contributed to the modern pace of neuronal maturation. Through this framework we also exposed a fine-tuning of the oxygen-tension response mediated by a human-derived HFV associated to the CADPS2 gene and predicted to alter HIF1A binding. These findings implicate a modulation of the rhombic lip maturation tempo, in contrast with previous claims of heterochronous neuronal maturation being a cortex-specific feature. In sum, this thesis establishes a framework going from genome-scale regulatory predictions to mechanistic readouts for understanding how TF affinity rewiring can modulate neurodevelopmental programs, and advances testable hypotheses on the impact of noncoding variants in the evolution of the modern human brain.