Polymer physics predicts the effects of structural variants on chromatin architecture

Abstract

Structural variants (SVs) can result in changes in gene expression due to abnormal chromatin folding and cause disease. However, the prediction of such effects remains a challenge. Here we present a polymer-physics-based approach (PRISMR) to model 3D chromatin folding and to predict enhancer-promoter contacts. PRISMR predicts higher-order chromatin structure from genome-wide chromosome conformation capture (Hi-C) data. Using the EPHA4 locus as a model, the effects of pathogenic SVs are predicted in silico and compared to Hi-C data generated from mouse limb buds and patient-derived fibroblasts. PRISMR deconvolves the folding complexity of the EPHA4 locus and identifies SV-induced ectopic contacts and alterations of 3D genome organization in homozygous or heterozygous states. We show that SVs can reconfigure topologically associating domains, thereby producing extensive rewiring of regulatory interactions and causing disease by gene misexpression. PRISMR can be used to predict interactions in silico, thereby providing a tool for analyzing the disease-causing potential of SVs.

Models of polymer physics for the architecture of the cell nucleus

“…Those genomic regions play a key role during cell differentiation and tissue development in mammals and structural variants in the loci are linked to human congenital diseases (Franke et al, ; Lupiáñez et al, and references therein). The SBS model of the loci was shown to reproduce Hi‐C data with good accuracy (Figure a, Chiariello et al, , Bianco et al, ). In fact, the contact frequency matrices derived from the simulated ensembles of 3D conformations of the different loci have a comparatively high Pearson, r , and distance corrected Pearson correlation coefficient, r ′ ( r ′ accounts for trivial genomic proximity effects, Bianco et al, ) with corresponding experimental Hi‐C matrices, ranging around r = 0.95 and r ′ = 0.60, respectively, below and above the scale of TADs.…”

“…Further progresses from polymer physics can shed light on such molecular mechanisms and their implications in gene regulation. By those advancements, polymer models have been used to predict the effects of structural variants on the 3D organization of chromatin loci and the link with their associated phenotype (Bianco et al, ). Hopefully, the progress in our understanding of chromosome architecture can help developing novel approaches to tackle diseases linked to chromatin misfolding, such as congenital disorders and cancer (Lupiáñez et al, ; Valton & Dekker, ; Weischenfeldt et al, ).…”

“…In fact, the contact frequency matrices derived from the simulated ensembles of 3D conformations of the different loci have a comparatively high Pearson, r , and distance corrected Pearson correlation coefficient, r ′ ( r ′ accounts for trivial genomic proximity effects, Bianco et al, ) with corresponding experimental Hi‐C matrices, ranging around r = 0.95 and r ′ = 0.60, respectively, below and above the scale of TADs. The model has been validated against independent capture Hi‐C (cHi‐C) data by its predictions on how chromatin refolds across mutants bearing structural variants, such as large deletions or duplications (Bianco et al, ). The ensemble of thermodynamics conformations of the loci derived from the model helps visualizing in 3D the patterns seen in Hi‐C data, such TADs and metaTADs (Figure a).…”

“…Importantly, the SBS model has been shown to explain Hi‐C, GAM, and fluorescence in situ hybridization (FISH) data, and the 3D structure of genomic loci across chromosomal scales and cell types (Barbieri et al, , 2017, Fraser et al, , Brackley et al, , Chiariello et al, , Bianco et al, ).…”

“…In particular, the SBS was used to describe the folding of loci such as the Sox9 locus in mouse embryonic stem cells (mESC) (Chiariello et al, ) and the Epha4 locus in mouse and human cell types (Bianco et al, ). Those genomic regions play a key role during cell differentiation and tissue development in mammals and structural variants in the loci are linked to human congenital diseases (Franke et al, ; Lupiáñez et al, and references therein).…”