Genome-wide transcription factor activities are explained by intrinsic conformational dynamics of binding-sites and distal flanking-regions

Abstract

Background:Transcription factors (TFs) recognize small DNA sequence motifs directly or through their sequence--dependent structure. While sequence composition and degeneracy are verified to be the defining factors of TF binding specificity, the role of conformational dynamics of the DNA remains poorly understood. With growing evidence from next generation sequencing (NGS) data suggesting the inadequacy of sequence--only models, alternative models for describing the TF binding preferences are required, wherein the conformational dynamics presents an attractive option. Results:Here, we report a novel method (DynaSeq) which accurately predicts DNA-conformational ensembles for genomic targets of TFs. Using DynaSeq we demonstrate how the dynamics of binding sites and their distal flanking regions can be used to elucidate TF--binding patterns for two model systems: cell type-specific binding of STAT3 and chromatin structural specificity of 3 functional TF classes viz. pioneers, settlers and migrants. We find that TF preferences in both these systems can be accurately explained by the conformational dynamics of their binding sites and their distal flanking DNA regions. Conclusions:Conformational dynamics not only distinguishes binding sites from genomic backgrounds in STAT3; it also points to a modular organization of their surrounding regions. Further, the differential binding modes of STAT3--DNA reveal a potential mechanism of cellular specificity. Our model identifies clear signatures to accurately classify pioneer, migrant and settler TF targets from the dynamics of distal flanking regions. This suggests that the chromatin preferences of TFs are significantly influenced by the intrinsic conformational dynamics of the DNA surrounding the TF binding sites. Introduction:The physical basis of protein--DNA interactions has thus far been investigated from the perspective of direct recognition of nucleic acid bases by complementary TF residues or through an indirect recognition of sequence-dependent DNA structure [1--4]. While the former ignores the differential accessibilities of DNA bases in the double helix the latter assumes the existence of a unique and exclusive structure of the DNA. We have previously, developed techniques to thread DNA sequences onto the structure of a known protein--DNA complex and determine the energy of native and designed sets of sequences [5--11]. Using these predicted intrinsic DNA energies, sequence specificity of the DNA structure in a given protein--DNA complex was successfully explained. However, focused on the assumption of an energetically most favorable structure, none of these methods addresses the dynamics of the molecules. Since the system has been assumed to be rigid the role of conformational dynamics in shape component or indirect readout mechanism of protein--DNA recognition has been overlooked. In addition to the assumption of a static structure of the binding site the role of its flanking regions and their structure at a genomic scale also remains unexplored. Since most high r...

Most brain disease-associated and eQTL haplotypes are not located within transcription factor DNase-seq footprints in brain

“…However, variation within classical TF motifs was found to poorly predict changes in binding (11) and flanking sequences located far from the actual TF binding site also appear to be important (12). Even when TF binding is altered, the level of expression from adjacent genes is often not substantially affected (13).…”

“…One or more variants in strong linkage disequilibrium (LD) with sentinel GWAS or eQTL variants are expected to be causal, in part by altering transcription factor (TF) binding and/or chromatin architecture ( 10 ). However, variation within classical TF motifs was found to poorly predict changes in binding ( 11 ) and flanking sequences located far from the actual TF binding site also appear to be important ( 12 ). Even when TF binding is altered, the level of expression from adjacent genes is often not substantially affected ( 13 ).…”