Finally, we assayed the activity of thousands of CREs isolated from their genomic context across multiple timepoints in vivo using a cell-type-specific massively parallel reporter assay. We showed that cell type-specificity is encoded in the sequence, in agreement with our quantitative model. 12/14

We also show explicitly that NSC methylation has a repressive effect on IPC accessibility, and that cooperativity between proximal CREs is correlated with accessibility both in terms of sequence content and total proximal activation. 11/14

We developed a machine-learning model that integrates sequence (putative TF motifs, dinucleotide content) and epigenomic features (ATAC, methylation, proximal activity) to predict NSC<->IPC specificity of CREs, that yields R^2 = 0.62 on held-out data. 10/14

Moreover, temporally activating CREs in NSCs get compacted significantly closer over time to astrocyte- or NSC-specific TSSs, an effect not observed with IPC TSSs. 9/14

Our Hi-C analysis shows that while overall chromatin topology is conserved across time in NSCs, there are specific hotspots whose differential insulation across time is correlated with the presence or depletion of proximal CREs. 8/14

Temporally activating CREs are significant: they are enriched with astrocyte-specific CREs, and astro CREs undergo extensive demethylation within NSCs across time, in contrast to other cell type-specific CREs. 7/14

We found that ATAC-activating CREs are tightly concomitant with methylation decrease. But the opposite was not true! Deactivating CREs in NSCs stay unmethylated across our time series, and this is not due to absence of DNMTs. 6/14

Furthermore, we were able to identify CREs with dynamic accessibility within cell types, across time. We focused on CREs that either activate/increase accessibility, or deactivate over time in NSCs. 5/14

Our metacell model of ATAC states provides a high-resolution timeline of cis-regulatory element (CRE) accessibility in corticogenesis. We use it to show that CREs specific to corticofugal neurons are activated late in maturation, in stark contrast to callosal projection neuron-specific CREs. 4/14

We show that astrocyte genes in NSCs are upregulated gradually over time, not suddenly, along with erosion of NSC proliferative capacity. This erosion is also correlated with lengthening of the cell cycle, suggesting that NSC fate bias is coupled to the cell cycle. 3/14

We have generated one of the most comprehensive multiome datasets in the context of brain development, integrating time-series measurements of scRNA-seq, scATAC-seq, 5mC methylation and Hi-C of mouse somatosensory cortex across 6 developmental timepoints. 2/14

Species-specific genes were often associated with multiple differentially accessible regions, suggesting that synergistic enhancer activation is a key mechanism driving epigenome evolution.
#EnhancerSynergy #GeneRegulation

High-resolution Hi-C revealed unexpected global shifts in 3D genome architecture in the chimpanzee and gorilla NSC, while topologically associating domains remained remarkably conserved.
#3DGenome #HiC

We identified novel human-biased genes and epigenetic signatures and further validated them using cortical organoids and MPRA data to generate a set of highly confident putative enhancers - paving the way for future functional studies of human brain development.
#ComparativeGenomics

We have generated one of the most comprehensive datasets of primate epigenome evolution to date — profiling and integrating genome-wide 3D genome organization, chromatin accessibility and gene expression in human, chimpanzee, gorilla and macaque IPSC and neural stem cells (NSC).