G-quadruplexes Are Transcription Factor Binding Hubs In Human ...

Transcription factors (TFs) control gene expression and chromatin structure through precise protein-DNA interactions at specific genome locations [1]. Preferred binding sites for hundreds of TFs exhibit short, defined DNA recognition motifs, commonly called “consensus sequences,” based on in vitro binding studies [2,3,4] and also in chromatin using ChIP-seq [5]. Two modes of protein-DNA recognition are described to contribute to TF binding specificity [6]. The first, based on the nucleotide readout, involves hydrogen bonding and hydrophobic interactions between amino acid side chains of the TF with base pairs primarily in the major groove of the DNA helix [7]. The second mode uses shape readout and is mediated by local structural features of the DNA double helix, such as minor groove width, base roll, and helix twist [8,9,10]. TF binding specificity can also be influenced by co-binding proteins [4] as well as epigenetic features such as CpG-methylation [11] and nucleosome positioning [12]. Despite this progress, experimentally observed binding sites for many TFs have not been explained [13]. As it is an open question as to what possible genomic features determine such binding events, we set out to explore how alternative DNA secondary structures, called G-quadruplexes, contribute to TF binding.

DNA G-Quadruplexes (G4s) are secondary structures made up of stacked G-tetrads, with each tetrad formed from the co-planar arrangement of four Hoogsteen-bonded guanine bases (Additional file 1: Fig. S1a) [14]. G4 structures have been visualized in human cells [15] and mapped in chromatin to regulatory regions particularly in promoters of highly expressed cancer genes [16, 17]. Analysis of patient-derived breast cancer tumor xenograft models has recently revealed a relationship of G4s with somatic copy-number aberrations and underlying transcriptional programs [18]. This together with small molecule perturbation experiments [19] is suggestive of important roles for G4s in transcriptional regulation. Biophysical and biochemical affinity experiments have identified proteins, such as helicases and DNA binding proteins, that show selective recognition for G4s over double-stranded DNA in vitro [20, 21]. The detailed molecular and functional relationship between endogenous G4s and components of the transcription machinery therefore warrants thorough investigation.

Herein, we report that numerous TFs are recruited to sites of endogenous G4s in human chromatin. Supporting this, the binding of several TFs to G4 structures is shown to have affinities comparable to that of canonical DNA double-strand interactions. Promoter G4s also appear to be bound by a surprisingly large number of TFs, particularly for highly expressed genes. Moreover, within a chromatin context, we provide robust evidence to demonstrate that TF binding to G4s can be competed out with a G4-selective small molecule. We posit that G4s are a previously overlooked key element of gene regulation that serves as high-affinity hubs enabling the recruitment of many different TFs to the same site to promote active transcription.