GC skew at the 5′ and 3′ ends of human genes links R-loop formation to epigenetic regulation and transcription termination

Abstract

Strand asymmetry in the distribution of guanines and cytosines, measured by GC skew, predisposes DNA sequences toward R-loop formation upon transcription. Previous work revealed that GC skew and R-loop formation associate with a core set of unmethylated CpG island (CGI) promoters in the human genome. Here, we show that GC skew can distinguish four classes of promoters, including three types of CGI promoters, each associated with unique epigenetic and gene ontology signatures. In particular, we identify a strong and a weak class of CGI promoters and show that these loci are enriched in distinct chromosomal territories reflecting the intrinsic strength of their protection against DNA methylation. Interestingly, we show that strong CGI promoters are depleted from the X chromosome while weak CGIs are enriched, a property consistent with the acquisition of DNA methylation during dosage compensation. Furthermore, we identify a third class of CGI promoters based on its unique GC skew profile and show that this gene set is enriched for Polycomb group targets. Lastly, we show that nearly 2000 genes harbor GC skew at their 3′ ends and that these genes are preferentially located in gene-dense regions and tend to be closely arranged. Genomic profiling of R-loops accordingly showed that a large proportion of genes with terminal GC skew form R-loops at their 3′ ends, consistent with a role for these structures in permitting efficient transcription termination. Altogether, we show that GC skew and R-loop formation offer significant insights into the epigenetic regulation, genomic organization, and function of human genes.

GC skew defines distinct RNA polymerase pause sites in CpG island promoters

“…3B). The compiled analysis of GC skew at all CGI-associated genes oriented to the TSS and scaled to the CGI showed that whereas there is generally positive skew downstream from the TSS as previously described (Ginno et al 2013), the proximally paused genes have a sharp peak in GC skew located just downstream from the TSS and the distally paused genes have a sharp peak of GC skew at the CGI edge (Fig. 3C).…”

“…Previous work by the Chedin group has shown a strong correlation between global GC skew, R-loop formation, and the lack of DNA methylation across CGI domains (Ginno et al 2012(Ginno et al , 2013. By use of a sequence-based algorithm, SkewR, that takes into account the degree, length, and direction of GC skew, as well as sequence composition (C + G content, CpG density), the group has classified CGIs into three classes (strong, weak, and reverse) based on a predicted propensity for R-loop formation.…”

Factors affecting the persistence of drug-induced reprogramming of the cancer methylome

“…CGIs represent unique chromatin environment in the genome, and are characterized by constrained divergent (bidirectional) transcription, marking by H3K4me3, and high levels of GC-skew, a sequence-based feature associated with the formation of unusual secondary structures. 35 Transcription through such regions leads to the formation of R-loops formed by the pairing of a G-rich nascent RNA to its C-rich template behind the progressing polymerase. R-loops have been suggested to play a key role in preventing DNA methylation at CGI.…”

“…35 We limited our analysis to the subset of our significantly demethylated CpGs common between the platform used here (Illumina 27K array) and that used for the TCGA samples (Illumina 450K array). This resulted in a total of 3,989 CpG sites (Resistant D 1,213, Slow D 1,527, Moderate D 906, Rapid D 342).…”

Fragile X-associated tremor/ataxia syndrome

“…In this regard, transcription through the highly GC-rich FMR1 5′ UTR region promotes (co-transcriptional) R-loop formation, 76 whereby the G-rich RNA transcript reinvades the DNA duplex and forms a stable RNA:DNA hybrid with the C-rich template strand, thereby displacing the non-template DNA strand. Although R-loop formation normally occurs at multiple loci throughout the genome, 81 defects in mRNA processing can result in an Rloop-dependent activation of the DNA damage response and lead to the accumulation of the phosphorylated H2A variant, γH2AX, which is associated with the DNA damage repair process. 82,83 We have recently reported R-loop formation both at the endogenous FMR1 locus and in an inducible episomal system in which the expanded CGG-repeat (~95 CGG) is located upstream of a GFP reporter.…”