In the loop: promoter-enhancer interactions and bioinformatics

Abstract

Enhancer-promoter regulation is a fundamental mechanism underlying differential transcriptional regulation. Spatial chromatin organization brings remote enhancers in contact with target promoters in cis to regulate gene expression. There is considerable evidence for promoter-enhancer interactions (PEIs). In the recent years, genome-wide analyses have identified signatures and mapped novel enhancers; however, being able to precisely identify their target gene(s) requires massive biological and bioinformatics efforts. In this review, we give a short overview of the chromatin landscape and transcriptional regulation. We discuss some key concepts and problems related to chromatin interaction detection technologies, and emerging knowledge from genome-wide chromatin interaction data sets. Then, we critically review different types of bioinformatics analysis methods and tools related to representation and visualization of PEI data, raw data processing and PEI prediction. Lastly, we provide specific examples of how PEIs have been used to elucidate a functional role of non-coding single-nucleotide polymorphisms. The topic is at the forefront of epigenetic research, and by highlighting some future bioinformatics challenges in the field, this review provides a comprehensive background for future PEI studies.

Keywords: CTCF; Chromatin loops; SNPs; chromosome conformation capture; enhancer prediction; histone modifications; promoter–enhancer interactions; transcription factories.

Figure 1

Models of chromatin organization. A diagram of different models of chromatin organization in the nuclear space. Interphase chromatin that interacts with the nuclear lamina (grey), nucleolus (green), nuclear pores (red), transcription factories (orange) and splicing speckles (black) are depicted here. Generally, lamin- and nucleolar-associated domains are transcriptionally repressed and have a more condensed chromatin, whereas chromatin that loops to the nuclear pore, transcription factories and splicing speckles are transcriptionally active and therefore have a more open chromatin structure (here, depicted as 10 nm chromatin fibre). Enhancers can activate gene expression over a distance and contain binding sites for TFs that recruit co-factors (activators or repressors). A promoter–enhancer looping mechanism mediated by cohesin (brown), CTCF (purple) and the mediator complex (red) that brings the enhancer into close proximity to its target promoter are presented in the enlarged box. The enhancer and promoter are marked with white boxes, and the transcription start site of the transcribed target gene is annotated with an arrow. TFs (green) and co-factors (yellow) bind the enhancer and are brought close to the basal transcription machinery at the promoter. RNAPolII (orange) transcribes pre-mRNA from the target gene and eRNA from the enhancer. Some of these models may co-exist for different PEIs; however, there are also other models that we could not show. A colour version of this figure is available online at BIB online: https://academic.oup.com/bib.

Figure 2

An example of a Hi-C contact map. Hi-C contact map of a segment of mouse chromosome 11, generated using Sushi [90] from Dixon et al. [85] data. A TAD and a long-range interaction between two loci are annotated. A colour version of this figure is available online at BIB online: https://academic.oup.com/bib.

Figure 3

Long-range interactions functionally connect disease-associated SNPs with disease candidate genes. (A) Physical proximity between DEXI gene locus and autoimmune disease-associated SNPs in the intron of CLEC16A. Hi-C data from human foetal lung (IMR-90) cells (from the Ren lab, [85]) show interactions between CLEC16A intron 19 and the DEXI locus. The enhancer marks in IMR-90 cells for H3K4me1 and H3K27ac are shown in green and blue, respectively, and the filter threshold for the Hi-C data was set to 5. SNPs in the region are in black, and the eQTL SNP rs12708716 is marked in red. The arc (pink) for interacting regions (grey) is highlighted with an arrow. (B) Long-range interactions links obesity-associated variants in FTO with the IRX3 locus. Hi-C data in human foetal lung (IMR-90) cells show interactions between the first intron of FTO with IRX3. The tracks for H3K4me1 and H3K27ac are shown in green and blue from IMR-90 cells and different human brain tissues from the NIH Roadmap Epigenomics Mapping Consortium. The filter threshold for the Hi-C data was set to 10. SNPs in the region are in black, and the BMI-associated SNP rs9930506 is marked in red. Arcs (pink) for interacting regions (grey) are highlighted with arrows. These public data sets are available and visualized with the WashU EpiGenome Browser (http://epigenomegateway.wustl.edu/browser/). dbSNP release 137 is shown in dark green, and the The National Human Genome Research Institute (NHGRI) Catalogues of GWAS are visualized in UCSC browser (http://genome-euro.ucsc.edu) [166]. A colour version of this figure is available online at BIB online: https://academic.oup.com/bib.