G-quadruplexes in promoters throughout the human genome

Abstract

Certain G-rich DNA sequences readily form four-stranded structures called G-quadruplexes. These sequence motifs are located in telomeres as a repeated unit, and elsewhere in the genome, where their function is currently unknown. It has been proposed that G-quadruplexes may be directly involved in gene regulation at the level of transcription. In support of this hypothesis, we show that the promoter regions (1 kb upstream of the transcription start site TSS) of genes are significantly enriched in quadruplex motifs relative to the rest of the genome, with >40% of human gene promoters containing one or more quadruplex motif. Furthermore, these promoter quadruplexes strongly associate with nuclease hypersensitive sites identified throughout the genome via biochemical measurement. Regions of the human genome that are both nuclease hypersensitive and within promoters show a remarkable (230-fold) enrichment of quadruplex elements, compared to the rest of the genome. These quadruplex motifs identified in promoter regions also show an interesting structural bias towards more stable forms. These observations support the proposal that promoter G-quadruplexes are directly involved in the regulation of gene expression.

Figure 1

(a) Structure of a G-tetrad, showing hydrogen bonds and monovalent cation. (b) Schematic of an intramolecular G-quadruplex. G-tetrads are shown as blue squares, and monovalent cations as grey spheres. The structure shown is folded in an antiparallel conformation, with the strands of the G-tetrads alternately running up and down. (c) Model for transcription modulation via formation of a quadruplex in a promoter region.

Figure 2

(a) Density of PQS with distance upstream from the TSS. The genome as a whole has a density of 0.13 PQS/kb, shown by the dashed line. (b) The percentage of promoter regions containing at least one PQS increases as the size of the promoter increases. This increase is extremely fast over the first 1000 bases. The dashed line shows what percentage would be predicted if the density of PQS were equal to that across the genome as a whole.

Figure 3

(a) Percent of PQS identified in given ranges 5′ of a TSS (‘promoter regions’), which have at least one single-base loop. The genome wide average figure is 64% (dashed line). (b) Proportion of PQS upstream of a TSS and within 500 bp of an NHS cluster which have at least one single-base loop decreases rapidly with increasing promoter size. (c) Proportion of PQS with at least one single-base loop in bulk genomic DNA, within 1000 bp upstream of a TSS (promoter), within 500 bp of an NHS cluster, and in promoter/NHS cluster overlap regions. The whole-genome figure is shown as a dashed line.

Figure 4

CD spectra of the sequences individually characterized in this study. Peaks at 260 nm indicate a parallel folded quadruplex, whereas peaks at 295 nm indicate an anti-parallel fold.