G-quadruplex structures are stable and detectable in human genomic DNA

Abstract

The G-quadruplex is an alternative DNA structural motif that is considered to be functionally important in the mammalian genome for transcriptional regulation, DNA replication and genome stability, but the nature and distribution of G-quadruplexes across the genome remains elusive. Here, we address the hypothesis that G-quadruplex structures exist within double-stranded genomic DNA and can be explicitly identified using a G-quadruplex-specific probe. An engineered antibody is employed to enrich for DNA containing G-quadruplex structures, followed by deep sequencing to detect and map G-quadruplexes at high resolution in genomic DNA from human breast adenocarcinoma cells. Our high sensitivity structure-based pull-down strategy enables the isolation of genomic DNA fragments bearing single, as well as multiple G-quadruplex structures. Stable G-quadruplex structures are found in sub-telomeres, gene bodies and gene regulatory regions. For a sample of identified target genes, we show that G-quadruplex-stabilizing ligands can modulate transcription. These results confirm the existence of G-quadruplex structures and their persistence in human genomic DNA.

Figure 1

The hf2 single chain antibody specifically pulls down G-quadruplex oligonucleotides. (a) Pull-down of G-quadruplex oligonucleotides by hf2 analyzed on TBE-urea gels. Left, KIT-2 G-quadruplex oligonucleotides but not single-stranded DNA are captured by hf2. Lane 1 shows the GeneRuler™ Ultra Low Range DNA Ladder, lanes 2-4 show concentration-dependent depletion of KIT-2 quadruplex, but not single-stranded DNA (control) from the supernatant by hf2, lanes 5-7 show the specific recovery of KIT-2 quadruplex, but not single-stranded DNA with increasing hf2 concentration. Middle, KIT-2 G-quadruplex oligonucleotides but not double-stranded DNA are captured by hf2. Lane 1 shows the GeneRuler™ Ultra Low Range DNA Ladder, lanes 2-4 show concentration-dependent depletion of KIT-2 quadruplex, but not doubled-stranded DNA (control) from the supernatant by hf2, lanes 5-7 show the specific recovery of KIT-2 quadruplex, but not double-stranded DNA with increasing hf2 concentration. Right, Htelo G-quadruplex oligonucleotides are captured by hf2 in the presence of excess sonicated salmon sperm DNA. Lane 1 shows the Htelo oligonucleotide alone, lanes 2-4 show the unbound supernatant with different hf2 concentrations, lanes 5-7 show the specific recovery of KIT-2 quadruplex, but not double-stranded salmon sperm DNA. (b) Pull-down protocol used to isolate G-quadruplex DNA from genomic DNA with the hf2 antibody.

Figure 2

Peaks identified by deep sequencing after pull-down with the anti-G-quadruplex hf2 antibody. Genome browser view of four peaks (blue) present compared with input and the overlap with G-quadruplex sequences predicted by quadparser (red). RefSeq gene is shown in green. The peaks map to different chromosomal locations including the sub-telomere (top left), gene promoter (top right), exon (bottom left), and intron (bottom right).

Figure 3

Motif and circular dichroism analyses substantiate G-quadruplex identification. (a) Sequence logo of the most enriched motif as determined by MEME (e-value = 1.9e⁻¹⁹⁰, expected value from MEME expectation maximization algorithm) found in the top 200 peaks ranked by enrichment (false discovery rate < 0.05). The G-quadruplex consensus sequence is shown here for comparison. (b) Examples of circular dichroism spectra for two oligonucleotides from the identified peaks. Parallel G-quadruplexes display a characteristic peak at 263 nm and a trough at 240 nm, while anti-parallel G-quadruplexes show a peak at 295 nm. The circular dichroism spectra show characteristics of both parallel and anti-parallel G-quadruplexes indicative of hybrid-type G-quadruplexes or a mixture of parallel and anti-parallel G-quadruplexes.

Figure 4

Stabilizing small molecules modulate the expression of identified G-quadruplex-containing genes. qRT-PCR was used to examine the expression of selected genes, identified by hf2 pull-down to contain G-quadruplexes, in MCF7 cells that were treated in triplicate with the G-quadruplex-specific small molecule PDS or DMSO control. The mean and standard error (error bars) of the relative expression levels of genes in PDS-treated and DMSO control are plotted on two different scales to show the different magnitudes of changes. Two genes in particular, PVT1 and STARD8 show large changes in gene expression in PDS-treated cells compared to controls. The student’s t test was used to calculate statistical significance between PDS-treated and control cells. Asterisks indicate statistically significant changes in gene expression with P < 0.05. The P-values for ABCG1, ACTN1, DYSF, ELL, LRP1, PVT1, STARD8, TOM1, ACTB1 and B2M are 0.0251, 0.0868, 0.0840, 0.0069, 0.0051, 0.0002, 0.0015, 0.0323, 0.3720 and 0.4189.