Quantitative modeling of transcription factor binding specificities using DNA shape

Statistical analysis of structural determinants for protein-DNA-binding specificity

DNA-binding proteins play critical roles in biological processes including gene expression, DNA packaging and DNA repair. They bind to DNA target sequences with different degrees of binding specificity, ranging from highly specific (HS) to nonspecific (NS). Alterations of DNA-binding specificity, due to either genetic variation or somatic mutations, can lead to various diseases. In this study, a comparative analysis of protein-DNA complex structures was carried out to investigate the structural features that contribute to binding specificity. Protein-DNA complexes were grouped into three general classes based on degrees of binding specificity: HS, multispecific (MS), and NS. Our results show a clear trend of structural features among the three classes, including amino acid binding propensities, simple and complex hydrogen bonds, major/minor groove and base contacts, and DNA shape. We found that aspartate is enriched in HS DNA binding proteins and predominately binds to a cytosine through a single hydrogen bond or two consecutive cytosines through bidentate hydrogen bonds. Aromatic residues, histidine and tyrosine, are highly enriched in the HS and MS groups and may contribute to specific binding through different mechanisms. To further investigate the role of protein flexibility in specific protein-DNA recognition, we analyzed the conformational changes between the bound and unbound states of DNA-binding proteins and structural variations. The results indicate that HS and MS DNA-binding domains have larger conformational changes upon DNA-binding and larger degree of flexibility in both bound and unbound states. Proteins 2016; 84:1147-1161. © 2016 Wiley Periodicals, Inc.

Bayesian Markov models consistently outperform PWMs at predicting motifs in nucleotide sequences

Position weight matrices (PWMs) are the standard model for DNA and RNA regulatory motifs. In PWMs nucleotide probabilities are independent of nucleotides at other positions. Models that account for dependencies need many parameters and are prone to overfitting. We have developed a Bayesian approach for motif discovery using Markov models in which conditional probabilities of order k - 1 act as priors for those of order k This Bayesian Markov model (BaMM) training automatically adapts model complexity to the amount of available data. We also derive an EM algorithm for de-novo discovery of enriched motifs. For transcription factor binding, BaMMs achieve significantly (P    =  1/16) higher cross-validated partial AUC than PWMs in 97% of 446 ChIP-seq ENCODE datasets and improve performance by 36% on average. BaMMs also learn complex multipartite motifs, improving predictions of transcription start sites, polyadenylation sites, bacterial pause sites, and RNA binding sites by 26-101%. BaMMs never performed worse than PWMs. These robust improvements argue in favour of generally replacing PWMs by BaMMs.

How motif environment influences transcription factor search dynamics: Finding a needle in a haystack

Transcription factors (TFs) have to find their binding sites, which are distributed throughout the genome. Facilitated diffusion is currently the most widely accepted model for this search process. Based on this model the TF alternates between one-dimensional sliding along the DNA, and three-dimensional bulk diffusion. In this view, the non-specific associations between the proteins and the DNA play a major role in the search dynamics. However, little is known about how the DNA properties around the motif contribute to the search. Accumulating evidence showing that TF binding sites are embedded within a unique environment, specific to each TF, leads to the hypothesis that the search process is facilitated by favorable DNA features that help to improve the search efficiency. Here, we review the field and present the hypothesis that TF-DNA recognition is dictated not only by the motif, but is also influenced by the environment in which the motif resides.

SiteOut: An Online Tool to Design Binding Site-Free DNA Sequences

DNA-binding proteins control many fundamental biological processes such as transcription, recombination and replication. A major goal is to decipher the role that DNA sequence plays in orchestrating the binding and activity of such regulatory proteins. To address this goal, it is useful to rationally design DNA sequences with desired numbers, affinities and arrangements of protein binding sites. However, removing binding sites from DNA is computationally non-trivial since one risks creating new sites in the process of deleting or moving others. Here we present an online binding site removal tool, SiteOut, that enables users to design arbitrary DNA sequences that entirely lack binding sites for factors of interest. SiteOut can also be used to delete sites from a specific sequence, or to introduce site-free spacers between functional sequences without creating new sites at the junctions. In combination with commercial DNA synthesis services, SiteOut provides a powerful and flexible platform for synthetic projects that interrogate regulatory DNA. Here we describe the algorithm and illustrate the ways in which SiteOut can be used; it is publicly available at https://depace.med.harvard.edu/siteout/.

Binding of transcription factor GabR to DNA requires recognition of DNA shape at a location distinct from its cognate binding site

Mechanisms for transcription factor recognition of specific DNA base sequences are well characterized and recent studies demonstrate that the shape of these cognate binding sites is also important. Here, we uncover a new mechanism where the transcription factor GabR simultaneously recognizes two cognate binding sites and the shape of a 29 bp DNA sequence that bridges these sites. Small-angle X-ray scattering and multi-angle laser light scattering are consistent with a model where the DNA undergoes a conformational change to bend around GabR during binding. In silico predictions suggest that the bridging DNA sequence is likely to be bendable in one direction and kinetic analysis of mutant DNA sequences with biolayer interferometry, allowed the independent quantification of the relative contribution of DNA base and shape recognition in the GabR-DNA interaction. These indicate that the two cognate binding sites as well as the bendability of the DNA sequence in between these sites are required to form a stable complex. The mechanism of GabR-DNA interaction provides an example where the correct shape of DNA, at a clearly distinct location from the cognate binding site, is required for transcription factor binding and has implications for bioinformatics searches for novel binding sites.

The Effects of Sequence Variation on Genome-wide NRF2 Binding--New Target Genes and Regulatory SNPs

Transcription factor binding specificity is crucial for proper target gene regulation. Motif discovery algorithms identify the main features of the binding patterns, but the accuracy on the lower affinity sites is often poor. Nuclear factor E2-related factor 2 (NRF2) is a ubiquitous redox-activated transcription factor having a key protective role against endogenous and exogenous oxidant and electrophile stress. Herein, we decipher the effects of sequence variation on the DNA binding sequence of NRF2, in order to identify both genome-wide binding sites for NRF2 and disease-associated regulatory SNPs (rSNPs) with drastic effects on NRF2 binding. Interactions between NRF2 and DNA were studied using molecular modelling, and NRF2 chromatin immunoprecipitation-sequence datasets together with protein binding microarray measurements were utilized to study binding sequence variation in detail. The binding model thus generated was used to identify genome-wide binding sites for NRF2, and genomic binding sites with rSNPs that have strong effects on NRF2 binding and reside on active regulatory elements in human cells. As a proof of concept, miR-126–3p and -5p were identified as NRF2 target microRNAs, and a rSNP (rs113067944) residing on NRF2 target gene (Ferritin, light polypeptide, FTL) promoter was experimentally verified to decrease NRF2 binding and result in decreased transcriptional activity.

Decoding the non-coding genome: elucidating genetic risk outside the coding genome

Current evidence emerging from genome-wide association studies indicates that the genetic underpinnings of complex traits are likely attributable to genetic variation that changes gene expression, rather than (or in combination with) variation that changes protein-coding sequences. This is particularly compelling with respect to psychiatric disorders, as genetic changes in regulatory regions may result in differential transcriptional responses to developmental cues and environmental/psychosocial stressors. Until recently, however, the link between transcriptional regulation and psychiatric genetic risk has been understudied. Multiple obstacles have contributed to the paucity of research in this area, including challenges in identifying the positions of remote (distal from the promoter) regulatory elements (e.g. enhancers) and their target genes and the underrepresentation of neural cell types and brain tissues in epigenome projects - the availability of high-quality brain tissues for epigenetic and transcriptome profiling, particularly for the adolescent and developing brain, has been limited. Further challenges have arisen in the prediction and testing of the functional impact of DNA variation with respect to multiple aspects of transcriptional control, including regulatory-element interaction (e.g. between enhancers and promoters), transcription factor binding and DNA methylation. Further, the brain has uncommon DNA-methylation marks with unique genomic distributions not found in other tissues - current evidence suggests the involvement of non-CG methylation and 5-hydroxymethylation in neurodevelopmental processes but much remains unknown. We review here knowledge gaps as well as both technological and resource obstacles that will need to be overcome in order to elucidate the involvement of brain-relevant gene-regulatory variants in genetic risk for psychiatric disorders.

Building accurate sequence-to-affinity models from high-throughput in vitro protein-DNA binding data using FeatureREDUCE

Transcription factors are crucial regulators of gene expression. Accurate quantitative definition of their intrinsic DNA binding preferences is critical to understanding their biological function. High-throughput in vitro technology has recently been used to deeply probe the DNA binding specificity of hundreds of eukaryotic transcription factors, yet algorithms for analyzing such data have not yet fully matured. Here, we present a general framework (FeatureREDUCE) for building sequence-to-affinity models based on a biophysically interpretable and extensible model of protein-DNA interaction that can account for dependencies between nucleotides within the binding interface or multiple modes of binding. When training on protein binding microarray (PBM) data, we use robust regression and modeling of technology-specific biases to infer specificity models of unprecedented accuracy and precision. We provide quantitative validation of our results by comparing to gold-standard data when available.

A human transcription factor in search mode

Transcription factors (TF) can change shape to bind and recognize DNA, shifting the energy landscape from a weak binding, rapid search mode to a higher affinity recognition mode. However, the mechanism(s) driving this conformational change remains unresolved and in most cases high-resolution structures of the non-specific complexes are unavailable. Here, we investigate the conformational switch of the human mitochondrial transcription termination factor MTERF1, which has a modular, superhelical topology complementary to DNA. Our goal was to characterize the details of the non-specific search mode to complement the crystal structure of the specific binding complex, providing a basis for understanding the recognition mechanism. In the specific complex, MTERF1 binds a significantly distorted and unwound DNA structure, exhibiting a protein conformation incompatible with binding to B-form DNA. In contrast, our simulations of apo MTERF1 revealed significant flexibility, sampling structures with superhelical pitch and radius complementary to the major groove of B-DNA. Docking these structures to B-DNA followed by unrestrained MD simulations led to a stable complex in which MTERF1 was observed to undergo spontaneous diffusion on the DNA. Overall, the data support an MTERF1-DNA binding and recognition mechanism driven by intrinsic dynamics of the MTERF1 superhelical topology.

DNAshapeR: an R/Bioconductor package for DNA shape prediction and feature encoding

Summary: DNAshapeR predicts DNA shape features in an ultra-fast, high-throughput manner from genomic sequencing data. The package takes either nucleotide sequence or genomic coordinates as input and generates various graphical representations for visualization and further analysis. DNAshapeR further encodes DNA sequence and shape features as user-defined combinations of k-mer and DNA shape features. The resulting feature matrices can be readily used as input of various machine learning software packages for further modeling studies.

Availability and implementation: The DNAshapeR software package was implemented in the statistical programming language R and is freely available through the Bioconductor project at https://www.bioconductor.org/packages/devel/bioc/html/DNAshapeR.html and at the GitHub developer site, http://tsupeichiu.github.io/DNAshapeR/.

Contact:rohs@usc.edu

Supplementary information:Supplementary data are available at Bioinformatics online.

Favorable genomic environments for cis-regulatory evolution: A novel theoretical framework

Cis-regulatory changes are arguably the primary evolutionary source of animal morphological diversity. With the recent explosion of genome-wide comparisons of the cis-regulatory content in different animal species is now possible to infer general principles underlying enhancer evolution. However, these studies have also revealed numerous discrepancies and paradoxes, suggesting that the mechanistic causes and modes of cis-regulatory evolution are still not well understood and are probably much more complex than generally appreciated. Here, we argue that the mutational mechanisms and genomic regions generating new regulatory activities must comply with the constraints imposed by the molecular properties of cis-regulatory elements (CREs) and the organizational features of long-range chromatin interactions. Accordingly, we propose a new integrative evolutionary framework for cis-regulatory evolution based on two major premises for the origin of novel enhancer activity: (i) an accessible chromatin environment and (ii) compatibility with the 3D structure and interactions of pre-existing CREs. Mechanisms and DNA sequences not fulfilling these premises, will be less likely to have a measurable impact on gene expression and as such, will have a minor contribution to the evolution of gene regulation. Finally, we discuss current comparative cis-regulatory data under the light of this new evolutionary model, and propose that the two most prominent mechanisms for the evolution of cis-regulatory changes are the overprinting of ancestral CREs and the exaptation of transposable elements.

Integrated microfluidic approach for quantitative high-throughput measurements of transcription factor binding affinities

Protein binding to DNA is a fundamental process in gene regulation. Methodologies such as ChIP-Seq and mapping of DNase I hypersensitive sites provide global information on this regulation in vivo In vitro methodologies provide valuable complementary information on protein-DNA specificities. However, current methods still do not measure absolute binding affinities. There is a real need for large-scale quantitative protein-DNA affinity measurements. We developed QPID, a microfluidic application for measuring protein-DNA affinities. A single run is equivalent to 4096 gel-shift experiments. Using QPID, we characterized the different affinities of ATF1, c-Jun, c-Fos and AP-1 to the CRE consensus motif and CRE half-site in two different genomic sequences on a single device. We discovered that binding of ATF1, but not of AP-1, to the CRE half-site is highly affected by its genomic context. This effect was highly correlated with ATF1 ChIP-seq and PBM experiments. Next, we characterized the affinities of ATF1 and ATF3 to 128 genomic CRE and CRE half-site sequences. Our affinity measurements explained that in vivo binding differences between ATF1 and ATF3 to CRE and CRE half-sites are partially mediated by differences in the minor groove width. We believe that QPID would become a central tool for quantitative characterization of biophysical aspects affecting protein-DNA binding.

Exploring comprehensive within-motif dependence of transcription factor binding in Escherichia coli

Modeling the binding of transcription factors helps to decipher the control logic behind transcriptional regulatory networks. Position weight matrix is commonly used to describe a binding motif but assumes statistical independence between positions. Although current approaches take within-motif dependence into account for better predictive performance, these models usually rely on prior knowledge and incorporate simple positional dependence to describe binding motifs. The inability to take complex within-motif dependence into account may result in an incomplete representation of binding motifs. In this work, we applied association rule mining techniques and constructed models to explore within-motif dependence for transcription factors in Escherichia coli. Our models can reflect transcription factor-DNA recognition where the explored dependence correlates with the binding specificity. We also propose a graphical representation of the explored within-motif dependence to illustrate the final binding configurations. Understanding the binding configurations also enables us to fine-tune or design transcription factor binding sites, and we attempt to present the configurations through exploring within-motif dependence.

Modeling co-occupancy of transcription factors using chromatin features

Regulation of gene expression requires both transcription factor (TFs) and epigenetic modifications, and interplays between the two types of factors have been discovered. However study of relationships between chromatin features and TF–TF co-occupancy remains limited. Here, we revealed the relationship by first illustrating distinct profile patterns of chromatin features related to different binding events, including single TF binding and TF–TF co-occupancy of 71 TFs from five human cell lines. We further implemented statistical analyses to demonstrate the relationship by accurately predicting co-occupancy genome-widely using chromatin features including DNase I hypersensitivity, 11 histone modifications (HMs) and GC content. Remarkably, our results showed that the combination of chromatin features enables accurate predictions across the five cells. For individual chromatin features, DNase I enables high and consistent predictions. H3K27ac, H3K4me 2, H3K4me3 and H3K9ac are more reliable predictors than other HMs. Although the combination of 11 HMs achieves accurate predictions, their predictive ability varies considerably when a model obtained from one cell is applied to others, indicating relationship between HMs and TF–TF co-occupancy is cell type dependent. GC content is not a reliable predictor, but the addition of GC content to any other features enhances their predictive ability. Together, our results elucidate a strong relationship between TF–TF co-occupancy and chromatin features.

RNA:DNA hybrids in the human genome have distinctive nucleotide characteristics, chromatin composition, and transcriptional relationships

Background

RNA:DNA hybrids represent a non-canonical nucleic acid structure that has been associated with a range of human diseases and potential transcriptional regulatory functions. Mapping of RNA:DNA hybrids in human cells reveals them to have a number of characteristics that give insights into their functions.

Results

We find RNA:DNA hybrids to occupy millions of base pairs in the human genome. A directional sequencing approach shows the RNA component of the RNA:DNA hybrid to be purine-rich, indicating a thermodynamic contribution to their in vivo stability. The RNA:DNA hybrids are enriched at loci with decreased DNA methylation and increased DNase hypersensitivity, and within larger domains with characteristics of heterochromatin formation, indicating potential transcriptional regulatory properties. Mass spectrometry studies of chromatin at RNA:DNA hybrids shows the presence of the ILF2 and ILF3 transcription factors, supporting a model of certain transcription factors binding preferentially to the RNA:DNA conformation.

Conclusions

Overall, there is little to indicate a dependence for RNA:DNA hybrids forming co-transcriptionally, with results from the ribosomal DNA repeat unit instead supporting the intriguing model of RNA generating these structures intrans. The results of the study indicate heterogeneous functions of these genomic elements and new insights into their formation and stability in vivo.

Electronic supplementary material

The online version of this article (doi:10.1186/s13072-015-0040-6) contains supplementary material, which is available to authorized users.

Searching target sites on DNA by proteins: Role of DNA dynamics under confinement

DNA-binding proteins (DBPs) rapidly search and specifically bind to their target sites on genomic DNA in order to trigger many cellular regulatory processes. It has been suggested that the facilitation of search dynamics is achieved by combining 3D diffusion with one-dimensional sliding and hopping dynamics of interacting proteins. Although, recent studies have advanced the knowledge of molecular determinants that affect one-dimensional search efficiency, the role of DNA molecule is poorly understood. In this study, by using coarse-grained simulations, we propose that dynamics of DNA molecule and its degree of confinement due to cellular crowding concertedly regulate its groove geometry and modulate the inter-communication with DBPs. Under weak confinement, DNA dynamics promotes many short, rotation-decoupled sliding events interspersed by hopping dynamics. While this results in faster 1D diffusion, associated probability of missing targets by jumping over them increases. In contrast, strong confinement favours rotation-coupled sliding to locate targets but lacks structural flexibility to achieve desired specificity. By testing under physiological crowding, our study provides a plausible mechanism on how DNA molecule may help in maintaining an optimal balance between fast hopping and rotation-coupled sliding dynamics, to locate target sites rapidly and form specific complexes precisely.

Plasticity in Repressor-DNA Interactions Neutralizes Loss of Symmetry in Bipartite Operators

Transcription factor-DNA interactions are central to gene regulation. Many transcription factors regulate multiple target genes and can bind sequences that do not conform strictly to the consensus. To understand the structural mechanism utilized by the transcription regulators to bind diverse target sequences, we have employed the repressor AraR from Bacillus subtilis as a model system. AraR is known to bind to eight different operator sites in the bacterial genome. Although there are differences in the sequences of four of these operators, ORE1, ORX1, ORA1, and ORR3, the AraR-DNA binding domain (AraR-DBD) as well as full-length AraR unexpectedly binds to each of these sequences with similar affinities as measured by fluorescence anisotropy experiments. We have determined crystal structures of AraR-DBD in complex with two different natural operators ORE1 and ORX1 up to 2.07 and 1.97 Å resolution, respectively. These structures were compared with the previously reported structures of AraR-DBD bound to two other natural operators (ORA1 and ORR3). Interactions of two molecules of AraR-DBD with the symmetric operator, ORE1, are identical, but their interaction with the non-symmetric operator ORX1 results in breakdown of the symmetry in protein-DNA interactions. The novel interactions observed are accompanied by local conformational change in the DNA. ChIP-sequencing (ChIP-Seq) data on other transcription factors has shown that they can bind to diverse targets, and hence the plasticity exhibited by AraR may be a general phenomenon. The ability of transcription factors to form alternate interactions may be important for employment in new functions and evolution of novel regulatory circuits.

ChEC-seq kinetics discriminates transcription factor binding sites by DNA sequence and shape in vivo

Chromatin endogenous cleavage (ChEC) uses fusion of a protein of interest to micrococcal nuclease (MNase) to target calcium-dependent cleavage to specific genomic loci in vivo. Here we report the combination of ChEC with high-throughput sequencing (ChEC-seq) to map budding yeast transcription factor (TF) binding. Temporal analysis of ChEC-seq data reveals two classes of sites for TFs, one displaying rapid cleavage at sites with robust consensus motifs and the second showing slow cleavage at largely unique sites with low-scoring motifs. Sites with high-scoring motifs also display asymmetric cleavage, indicating that ChEC-seq provides information on the directionality of TF-DNA interactions. Strikingly, similar DNA shape patterns are observed regardless of motif strength, indicating that the kinetics of ChEC-seq discriminates DNA recognition through sequence and/or shape. We propose that time-resolved ChEC-seq detects both high-affinity interactions of TFs with consensus motifs and sites preferentially sampled by TFs during diffusion and sliding.

In chromatin endogenous cleavage (ChEC), micrococcal nuclease (MNase) is fused to a protein of interest and its cleavage is thus targeted to specific genomic loci in vivo. Here, the authors show that time-resolved ChEC-seq (high-throughput sequencing after ChEC) can detect DNA shape patterns regardless of motif strength.

Varying levels of complexity in transcription factor binding motifs

Binding of transcription factors to DNA is one of the keystones of gene regulation. The existence of statistical dependencies between binding site positions is widely accepted, while their relevance for computational predictions has been debated. Building probabilistic models of binding sites that may capture dependencies is still challenging, since the most successful motif discovery approaches require numerical optimization techniques, which are not suited for selecting dependency structures. To overcome this issue, we propose sparse local inhomogeneous mixture (Slim) models that combine putative dependency structures in a weighted manner allowing for numerical optimization of dependency structure and model parameters simultaneously. We find that Slim models yield a substantially better prediction performance than previous models on genomic context protein binding microarray data sets and on ChIP-seq data sets. To elucidate the reasons for the improved performance, we develop dependency logos, which allow for visual inspection of dependency structures within binding sites. We find that the dependency structures discovered by Slim models are highly diverse and highly transcription factor-specific, which emphasizes the need for flexible dependency models. The observed dependency structures range from broad heterogeneities to sparse dependencies between neighboring and non-neighboring binding site positions.

Reconstruction of novel transcription factor regulons through inference of their binding sites

Background

In most sequenced organisms the number of known regulatory genes (e.g., transcription factors (TFs)) vastly exceeds the number of experimentally-verified regulons that could be associated with them. At present, identification of TF regulons is mostly done through comparative genomics approaches. Such methods could miss organism-specific regulatory interactions and often require expensive and time-consuming experimental techniques to generate the underlying data.

Results

In this work, we present an efficient algorithm that aims to identify a given transcription factor’s regulon through inference of its unknown binding sites, based on the discovery of its binding motif. The proposed approach relies on computational methods that utilize gene expression data sets and knockout fitness data sets which are available or may be straightforwardly obtained for many organisms. We computationally constructed the profiles of putative regulons for the TFs LexA, PurR and Fur in E. coli K12 and identified their binding motifs. Comparisons with an experimentally-verified database showed high recovery rates of the known regulon members, and indicated good predictions for the newly found genes with high biological significance. The proposed approach is also applicable to novel organisms for predicting unknown regulons of the transcriptional regulators. Results for the hypothetical protein Dde0289 in D. alaskensis include the discovery of a Fis-type TF binding motif.

Conclusions

The proposed motif-based regulon inference approach can discover the organism-specific regulatory interactions on a single genome, which may be missed by current comparative genomics techniques due to their limitations.

Electronic supplementary material

The online version of this article (doi:10.1186/s12859-015-0685-y) contains supplementary material, which is available to authorized users.

Motif signatures in stretch enhancers are enriched for disease-associated genetic variants

Background

Stretch enhancers (SEs) are large chromatin-defined regulatory elements that are at least 3,000 base pairs (bps) long, in contrast to the median enhancer length of 800 bps. SEs tend to be cell-type specific, regulate cell-type specific gene expression, and are enriched in disease-associated genetic variants in disease-relevant cell types. Transcription factors (TFs) can bind to enhancers to modulate enhancer activity, and their sequence specificity can be represented by motifs. We hypothesize motifs can provide a biological context for how genetic variants contribute to disease.

Results

We integrated chromatin state, gene expression, and chromatin accessibility [measured as DNase I Hypersensitive Sites (DHSs)] maps across nine different cell types. Motif enrichment analyses of chromatin-defined enhancer sequences identify several known cell-type specific “master” factors. Furthermore, de novo motif discovery not only recovers many of these motifs, but also identifies novel non-canonical motifs, providing additional insight into TF binding preferences. Across the length of SEs, motifs are most enriched in DHSs, though relative enrichment is also observed outside of DHSs. Interestingly, we show that single nucleotide polymorphisms associated with diseases or quantitative traits significantly overlap motif occurrences located in SEs, but outside of DHSs.

Conclusions

These results reinforce the role of SEs in influencing risk for diseases and suggest an expanded regulatory functional role for motifs that occur outside highly accessible chromatin. Furthermore, the motif signatures generated here expand our understanding of the binding preference of well-characterized TFs.

Electronic supplementary material

The online version of this article (doi:10.1186/s13072-015-0015-7) contains supplementary material, which is available to authorized users.

A widespread role of the motif environment in transcription factor binding across diverse protein families

Transcriptional regulation requires the binding of transcription factors (TFs) to short sequence-specific DNA motifs, usually located at the gene regulatory regions. Interestingly, based on a vast amount of data accumulated from genomic assays, it has been shown that only a small fraction of all potential binding sites containing the consensus motif of a given TF actually bind the protein. Recent in vitro binding assays, which exclude the effects of the cellular environment, also demonstrate selective TF binding. An intriguing conjecture is that the surroundings of cognate binding sites have unique characteristics that distinguish them from other sequences containing a similar motif that are not bound by the TF. To test this hypothesis, we conducted a comprehensive analysis of the sequence and DNA shape features surrounding the core-binding sites of 239 and 56 TFs extracted from in vitro HT-SELEX binding assays and in vivo ChIP-seq data, respectively. Comparing the nucleotide content of the regions around the TF-bound sites to the counterpart unbound regions containing the same consensus motifs revealed significant differences that extend far beyond the core-binding site. Specifically, the environment of the bound motifs demonstrated unique sequence compositions, DNA shape features, and overall high similarity to the core-binding motif. Notably, the regions around the binding sites of TFs that belong to the same TF families exhibited similar features, with high agreement between the in vitro and in vivo data sets. We propose that these unique features assist in guiding TFs to their cognate binding sites.

Deconvolving the recognition of DNA shape from sequence

Protein-DNA binding is mediated by the recognition of the chemical signatures of the DNA bases and the 3D shape of the DNA molecule. Because DNA shape is a consequence of sequence, it is difficult to dissociate these modes of recognition. Here, we tease them apart in the context of Hox-DNA binding by mutating residues that, in a co-crystal structure, only recognize DNA shape. Complexes made with these mutants lose the preference to bind sequences with specific DNA shape features. Introducing shape-recognizing residues from one Hox protein to another swapped binding specificities in vitro and gene regulation in vivo. Statistical machine learning revealed that the accuracy of binding specificity predictions improves by adding shape features to a model that only depends on sequence, and feature selection identified shape features important for recognition. Thus, shape readout is a direct and independent component of binding site selection by Hox proteins.