Atlas and developmental dynamics of mouse DNase I hypersensitive sites

Abstract

<jats:title>Abstract</jats:title><jats:p>Early mammalian development is orchestrated by genome-encoded regulatory elements populated by a changing complement of regulatory factors, creating a dynamic chromatin landscape. To define the spatiotemporal organization of regulatory DNA landscapes during mouse development and maturation, we generated nucleotide-resolution DNA accessibility maps from 15 tissues sampled at 9 intervals spanning post-conception day 9.5 through early adult, and integrated these with 41 adult-stage DNase-seq profiles to create a global atlas of mouse regulatory DNA. Collectively, we delineated &gt;1.8 million DNase I hypersensitive sites (DHSs), with the vast majority displaying temporal and tissue-selective patterning. Here we show that tissue regulatory DNA compartments show sharp embryonic-to-fetal transitions characterized by wholesale turnover of DHSs and progressive domination by a diminishing number of transcription factors. We show further that aligning mouse and human fetal development on a regulatory axis exposes disease-associated variation enriched in early intervals lacking human samples. Our results provide an expansive new resource for decoding mammalian developmental regulatory programs.</jats:p>

Expanded encyclopaedias of DNA elements in the human and mouse genomes

“…Using an improved DNase treatment followed by sequencing (DNase-seq) assay that requires only small numbers of input cells, we expanded ENCODE human DHS maps by more than 200 different cell types and states, chiefly primary cells and tissues 11 ( Supplementary Table 4a). By incorporating both a multi-tissue developmental series and a larger range of adult tissues ( Supplementary Table 4b), we also greatly expanded mouse DHS maps 17 of coordinated tissue-selective DHS activation patterns, which were then used to annotate DHSs, genes, and genetic variation 11 . The number of human ENCODE biosamples with deep DNase-seq data (more than 200 million uniquely mapped reads) was tripled to more than 300, enabling delineation of 4.4 million consensus human DNase I footprints within DHSs, enhanced annotation of tissue selectivity, and identification of functional variants that directly affect regulatory factor occupancy 11,12 .…”

“…• Characterize the landscape of 3D chromatin interactions across 24 different cell types 13 . • Expand annotation of mouse chromatin modification, DNA accessibility, DNA methylation, and RNA transcription landscapes in early developmental stages not readily accessible in human [14][15][16][17] . To enhance the utility and accessibility of ENCODE data for studies of gene regulation, in this report, we have now:…”

Chromatin Velocity reveals epigenetic dynamics by single-cell profiling of heterochromatin and euchromatin

“…76 ), and H3K9me3-enriched regions were identified using SICER v2 (ref. 77 ) using default parameters.…”

“…We segmented the genome according to DHSs, as previously described 78 . Briefly, we downloaded the index of DHSs for human 79 and mouse genomes 77 ; intervals closer than 500 bp were merged using bedtools 80 to create the interval set for accessible chromatin (named 'DHS'). We then took the complement of the set to create the interval set for compacted chromatin (named 'complement').…”

Dimensionality reduction and statistical modeling of scGET-seq data

“…At time of writing, the data fusion method we adopted is no more developed or maintained, which prompts to the pursuit of an alternative method. In addition, we used to count tn5 and tnH reads over two sets of genomic regions, representative of accessible and compact chromatin, based on the DnASE I Hypersentitive Sites census available for human [3] or mouse [4]. Since the number of regions in both sets are different, we needed to obtain a low-dimensional representation (20-dimensional UMAP) of each count matrix before data fusion, at the risk of introducing distortion artifacts.…”