SUMMARY EndoC-βH1 is emerging as a critical human β cell model to study the genetic and environmental etiologies of β cell (dys)function and diabetes. Comprehensive knowledge of its molecular landscape is lacking, yet required, for effective use of this model. Here, we report chromosomal (spectral karyotyping), genetic (genotyping), epigenomic (ChIP-seq and ATAC-seq), chromatin interaction (Hi-C and Pol2 ChIA-PET), and transcriptomic (RNA-seq and miRNA-seq) maps of EndoC-βH1. Analyses of these maps define known (e.g., PDX1 and ISL1 ) and putative (e.g., PCSK1 and mir-375 ) β cell-specific transcriptional cis -regulatory networks and identify allelic effects on cis -regulatory element use. Importantly, comparison with maps generated in primary human islets and/or β cells indicates preservation of chromatin looping but also highlights chromosomal aberrations and fetal genomic signatures in EndoC-βH1. Together, these maps, and a web application we created for their exploration, provide important tools for the design of experiments to probe and manipulate the genetic programs governing β cell identity and (dys)function in diabetes.

Megabase-scale intervals of active, gene-rich and inactive, gene-poor chromatin are known to segregate, forming the A and B compartments. Fine mapping of the contents of these A and B compartments has been hitherto impossible, owing to the extraordinary sequencing depths required to distinguish between the long-range contact patterns of individual loci, and to the computational complexity of the associated calculations. Here, we generate the largest published in situ Hi-C map to date, spanning 33 billion contacts. We also develop a computational method, dubbed PCA of Sparse, Super Massive Matrices (POSSUMM), that is capable of efficiently calculating eigenvectors for sparse matrices with millions of rows and columns. Applying POSSUMM to our Hi-C dataset makes it possible to assign loci to the A and B compartment at 500 bp resolution. We find that loci frequently alternate between compartments as one moves along the contour of the genome, such that the median compartment interval is only 12.5 kb long. Contrary to the findings in coarse-resolution compartment profiles, we find that individual genes are not uniformly positioned in either the A compartment or the B compartment. Instead, essentially all (95%) active gene promoters localize in the A compartment, but the likelihood of localizing in the A compartment declines along the body of active genes, such that the transcriptional termini of long genes (>60 kb) tend to localize in the B compartment. Similarly, essentially all active enhancers elements (95%) localize in the A compartment, even when the flanking sequences are comprised entirely of inactive chromatin and localize in the B compartment. These results are consistent with a model in which DNA-bound regulatory complexes give rise to phase separation at the scale of individual DNA elements.

Balanced chromosomal rearrangements (BCRs), including inversions, translocations, and insertions, reorganize large sections of the genome and contribute substantial risk for developmental disorders (DDs). However, the rarity and lack of systematic screening for BCRs in the population has precluded unbiased analyses of the genomic features and mechanisms associated with risk for DDs versus normal developmental outcomes. Here, we sequenced and analyzed 1,420 BCR breakpoints across 710 individuals, including 406 DD cases and the first large-scale collection of 304 control BCR carriers. We found that BCRs were not more likely to disrupt genes in DD cases than controls, but were seven-fold more likely to disrupt genes associated with dominant DDs (21.3% of cases vs. 3.4% of controls; P = 1.60x10-12). Moreover, BCRs that did not disrupt a known DD gene were significantly enriched for breakpoints that altered topologically associated domains (TADs) containing dominant DD genes in cases compared to controls (odds ratio [OR] = 1.43, P = 0.036). We discovered six TADs enriched for noncoding BCRs (false discovery rate < 0.1) that contained known DD genes (MEF2C, FOXG1, SOX9, BCL11A, BCL11B, and SATB2) and represent candidate pathogenic long-range positional effect (LRPE) loci. These six TADs were collectively disrupted in 7.4% of the DD cohort. Phased Hi-C analyses of five cases with noncoding BCR breakpoints localized to one of these putative LRPEs, the 5q14.3 TAD encompassing MEF2C, confirmed extensive disruption to local 3D chromatin structures and reduced frequency of contact between the MEF2C promoter and annotated enhancers. We further identified six genomic features enriched in TADs preferentially disrupted by noncoding BCRs in DD cases versus controls and used these features to build a model to predict TADs at risk for LRPEs across the genome. These results emphasize the potential impact of noncoding structural variants to cause LRPEs in unsolved DD cases, as well as the complex interaction of features associated with predicting intolerance to alteration of three-dimensional chromatin topology.