Specific Transcription Factor Binding Research Articles

Transcription Factor Binding Affinities and DNA Shape Readout.

SummaryAn essential event in gene regulation is the binding of a transcription factor (TF) to its target DNA. Models considering the interactions between the TF and the DNA geometry proved to be successful approaches to describe this binding event, while conserving data interpretability. However, a direct characterization of the DNA shape contribution to binding is still missing due to the lack of accurate and large-scale binding affinity data. Here, we use a binding assay we recently established to measure with high sensitivity the binding specificities of 13 Drosophila TFs, including dinucleotide dependencies to capture non-independent amino acid-base interactions. Correlating the binding affinities with all DNA shape features, we find that shape readout is widely used by these factors. A shape readout/TF-DNA complex structure analysis validates our approach while providing biological insights such as positively charged or highly polar amino acids often contact nucleotides that exhibit strong shape readout.

Learning Useful Representations of DNA Sequences From ChIP-Seq Datasets for Exploring Transcription Factor Binding Specificities.

Deep learning has been successfully applied to surprisingly different domains. Researchers and practitioners are employing trained deep learning models to enrich our knowledge. Transcription factors (TFs)are essential for regulating gene expression in all organisms by binding to specific DNA sequences. Here, we designed a deep learning model named SemanticCS (Semantic ChIP-seq)to predict TF binding specificities. We trained our learning model on an ensemble of ChIP-seq datasets (Multi-TF-cell)to learn useful intermediate features across multiple TFs and cells. To interpret these feature vectors, visualization analysis was used. Our results indicate that these learned representations can be used to train shallow machines for other tasks. Using diverse experimental data and evaluation metrics, we show that SemanticCS outperforms other popular methods. In addition, from experimental data, SemanticCS can help to identify the substitutions that cause regulatory abnormalities and to evaluate the effect of substitutions on the binding affinity for the RXR transcription factor. The online server for SemanticCS is freely available at http://qianglab.scst.suda.edu.cn/semanticCS/.

A Comparative Study of Supervised Machine Learning Algorithms for the Prediction of Long-Range Chromatin Interactions

The role of three-dimensional genome organization as a critical regulator of gene expression has become increasingly clear over the last decade. Most of our understanding of this association comes from the study of long range chromatin interaction maps provided by Chromatin Conformation Capture-based techniques, which have greatly improved in recent years. Since these procedures are experimentally laborious and expensive, in silico prediction has emerged as an alternative strategy to generate virtual maps in cell types and conditions for which experimental data of chromatin interactions is not available. Several methods have been based on predictive models trained on one-dimensional (1D) sequencing features, yielding promising results. However, different approaches vary both in the way they model chromatin interactions and in the machine learning-based strategy they rely on, making it challenging to carry out performance comparison of existing methods. In this study, we use publicly available 1D sequencing signals to model cohesin-mediated chromatin interactions in two human cell lines and evaluate the prediction performance of six popular machine learning algorithms: decision trees, random forests, gradient boosting, support vector machines, multi-layer perceptron and deep learning. Our approach accurately predicts long-range interactions and reveals that gradient boosting significantly outperforms the other five methods, yielding accuracies of about 95%. We show that chromatin features in close genomic proximity to the anchors cover most of the predictive information, as has been previously reported. Moreover, we demonstrate that gradient boosting models trained with different subsets of chromatin features, unlike the other methods tested, are able to produce accurate predictions. In this regard, and besides architectural proteins, transcription factors are shown to be highly informative. Our study provides a framework for the systematic prediction of long-range chromatin interactions, identifies gradient boosting as the best suited algorithm for this task and highlights cell-type specific binding of transcription factors at the anchors as important determinants of chromatin wiring mediated by cohesin.

Dissection of c-AMP Response Element Architecture by Using Genomic and Episomal Massively Parallel Reporter Assays.

In eukaryotes, transcription factors (TFs) orchestrate gene expression by binding to TF-binding sites (TFBSs) and localizing transcriptional co-regulators and RNA polymerase II to cis-regulatory elements. However, we lack a basic understanding of the relationship between TFBS composition and their quantitative transcriptional responses. Here, we measured expression driven by 17,406 synthetic cis-regulatory elements with varied compositions of a model TFBS, the c-AMP response element (CRE) by using massively parallel reporter assays (MPRAs). We find CRE number, affinity, and promoter proximity largely determines expression. In addition, we observe expression modulation based on the spacing between CREs and CRE distance to the promoter, where expression follows a helical periodicity. Finally, we compare library expression between an episomal MPRA and a genomically integrated MPRA, where a single cis-regulatory element is assayed per cell at a defined locus. These assays largely recapitulate each other, although weaker, non-canonical CREs exhibit greater activity in a genomic context.

Engaging the nucleosome

Structural Biology Cell identity is defined by gene expression patterns that are established through the binding of specific transcription factors. However, nucleosomal units limit access of transcription factors to specific DNA motifs within the mammalian genome. To study how transcription factors bind such chromatinized, nucleosome-embedded motifs, Michael et al. focused on the pluripotency factors OCT4 and SOX2. They systematically quantified the relative affinities of these factors at different motif positions throughout the nucleosome, enabling structure determination of OCT4-SOX2–bound nucleosomes by cryo–electron microscopy. OCT4 and SOX2 bound cooperatively to strengthen DNA-binding affinity and resulted in DNA distortions that destabilized the nucleosome. This analysis reveals position-dependent binding modes that were validated in vivo, providing insights on how transcription factors read out chromatinized motifs. Science , this issue p. [1460][1] [1]: /lookup/doi/10.1126/science.abb0074

A Generative Neural Network for Maximizing Fitness and Diversity of Synthetic DNA and Protein Sequences.

SUMMARYEngineering gene and protein sequences with defined functional properties is a major goal of synthetic biology. Deep neural network models, together with gradient ascent-style optimization, show promise for sequence design. The generated sequences can however get stuck in local minima and often have low diversity. Here, we develop deep exploration networks (DENs), a class of activation-maximizing generative models, which minimize the cost of a neural network fitness predictor by gradient descent. By penalizing any two generated patterns on the basis of a similarity metric, DENs explicitly maximize sequence diversity. To avoid drifting into low-confidence regions of the predictor, we incorporate variational autoencoders to maintain the likelihood ratio of generated sequences. Using DENs, we engineered polyadenylation signals with more than 10-fold higher selection odds than the best gradient ascent-generated patterns, identified splice regulatory sequences predicted to result in highly differential splicing between cell lines, and improved on state-of-the-art results for protein design tasks.

Mechanisms of OCT4-SOX2 motif readout on nucleosomes.

Transcription factors (TFs) regulate gene expression through chromatin where nucleosomes restrict DNA access. To study how TFs bind nucleosome-occupied motifs, we focused on the reprogramming factors OCT4 and SOX2 in mouse embryonic stem cells. We determined TF engagement throughout a nucleosome at base-pair resolution in vitro, enabling structure determination by cryo-electron microscopy at two preferred positions. Depending on motif location, OCT4 and SOX2 differentially distort nucleosomal DNA. At one position, OCT4-SOX2 removes DNA from histone H2A and histone H3; however, at an inverted motif, the TFs only induce local DNA distortions. OCT4 uses one of its two DNA-binding domains to engage DNA in both structures, reading out a partial motif. These findings explain site-specific nucleosome engagement by the pluripotency factors OCT4 and SOX2, and they reveal how TFs distort nucleosomes to access chromatinized motifs.

Kethoxal-assisted single-stranded DNA sequencing captures global transcription dynamics and enhancer activity in situ.

Transcription is a highly dynamic process that generates single-stranded DNA (ssDNA) in the genome as ‘transcription bubbles’. Here we describe a kethoxal-assisted single-stranded DNA sequencing (KAS-seq) approach, based on the fast and specific reaction between N3-kethoxal and guanines in ssDNA in live cells and mouse tissues. KAS-seq enables rapid (within 5 min), sensitive, and genome-wide capture and mapping of ssDNA produced by transcriptionally active RNA polymerases or other processes in situ by using as few as 1,000 cells. KAS-seq defines a group of enhancers that are single-stranded, which enrich unique sequence motifs and are associated with specific transcription factor binding and more enhancer-promotor interactions. Under protein condensation inhibition conditions, KAS-seq uncovers a rapid release of RNA polymerase II (Pol II) from a group of promotors. KAS-seq thus facilitates fast, comprehensive, and accurate analysis of transcription dynamics and enhancer activities simultaneously in a low input and high-throughput manner.

Predicting transcription factor binding using ensemble random forest models.

Background: Understanding the location and cell-type specific binding of Transcription Factors (TFs) is important in the study of gene regulation. Computational prediction of TF binding sites is challenging, because TFs often bind only to short DNA motifs and cell-type specific co-factors may work together with the same TF to determine binding. Here, we consider the problem of learning a general model for the prediction of TF binding using DNase1-seq data and TF motif description in form of position specific energy matrices (PSEMs). Methods: We use TF ChIP-seq data as a gold-standard for model training and evaluation. Our contribution is a novel ensemble learning approach using random forest classifiers. In the context of the ENCODE-DREAM in vivo TF binding site prediction challenge we consider different learning setups. Results: Our results indicate that the ensemble learning approach is able to better generalize across tissues and cell-types compared to individual tissue-specific classifiers or a classifier built based upon data aggregated across tissues. Furthermore, we show that incorporating DNase1-seq peaks is essential to reduce the false positive rate of TF binding predictions compared to considering the raw DNase1 signal. Conclusions: Analysis of important features reveals that the models preferentially select motifs of other TFs that are close interaction partners in existing protein protein-interaction networks. Code generated in the scope of this project is available on GitHub: https://github.com/SchulzLab/TFAnalysis (DOI: 10.5281/zenodo.1409697).

Predicting transcription factor binding using ensemble random forest models

Background: Understanding the location and cell-type specific binding of Transcription Factors (TFs) is important in the study of gene regulation. Computational prediction of TF binding sites is challenging, because TFs often bind only to short DNA motifs and cell-type specific co-factors may work together with the same TF to determine binding. Here, we consider the problem of learning a general model for the prediction of TF binding using DNase1-seq data and TF motif description in form of position specific energy matrices (PSEMs). Methods: We use TF ChIP-seq data as a gold-standard for model training and evaluation. Our contribution is a novel ensemble learning approach using random forest classifiers. In the context of the ENCODE-DREAM in vivo TF binding site prediction challenge we consider different learning setups. Results: Our results indicate that the ensemble learning approach is able to better generalize across tissues and cell-types compared to individual tissue-specific classifiers or a classifier built based upon data aggregated across tissues. Furthermore, we show that incorporating DNase1-seq peaks is essential to reduce the false positive rate of TF binding predictions compared to considering the raw DNase1 signal. Conclusions: Analysis of important features reveals that the models preferentially select motifs of other TFs that are close interaction partners in existing protein protein-interaction networks. Code generated in the scope of this project is available on GitHub: https://github.com/SchulzLab/TFAnalysis (DOI: 10.5281/zenodo.1409697).

Molecular mechanisms of the protein-protein interaction-regulated binding specificity of basic-region leucine zipper transcription factors.

It is well known that the DNA-binding specificity of transcription factors (TFs) is influenced by protein-protein interactions (PPIs). However, the underlying molecular mechanisms remain largely unknown. In this work, we adopted the cAMP-response element-binding protein (CREB) of the basic leucine zipper (bZIP) TF family as a model system, and a workflow of combined bioinformatics and molecular modeling analysis of protein-DNA interaction was tested. First, the multiple sequence alignment and SDPsite method were used to find potential bZIP family binding specificity determining positions (SDPs) within the protein-protein interaction region. Second, the mutation system was analyzed using molecular dynamics simulation. Molecular mechanics Poisson-Boltzmann surface area (MM/PBSA) free energy calculations confirmed the enhancement of the binding affinity of the mutation, which was in agreement with experimental results. The root mean square fluctuation (RMSF) and hydrogen bonding changes suggested an open and close protein dimerization process after the system was mutated, which resulted in the change of the hydrogen bonding of the protein-DNA interface and a slight conformational change. We believe that this work will contribute to understanding the protein-protein interaction-regulated binding specificity of bZIP transcription factors.

Low-Affinity Binding Sites and the Transcription Factor Specificity Paradox in Eukaryotes.

Eukaryotic transcription factors (TFs) from the same structural family tend to bind similar DNA sequences, despite the ability of these TFs to execute distinct functions in vivo. The cell partly resolves this specificity paradox through combinatorial strategies and the use of low-affinity binding sites, which are better able to distinguish between similar TFs. However, because these sites have low affinity, it is challenging to understand how TFs recognize them in vivo. Here, we summarize recent findings and technological advancements that allow for the quantification and mechanistic interpretation of TF recognition across a wide range of affinities. We propose a model that integrates insights from the fields of genetics and cell biology to provide further conceptual understanding of TF binding specificity. We argue that in eukaryotes, target specificity is driven by an inhomogeneous 3D nuclear distribution of TFs and by variation in DNA binding affinity such that locally elevated TF concentration allows low-affinity binding sites to be functional.

Similarity regression predicts evolution of transcription factor sequence specificity.

Transcription factor (TF) binding specificities (motifs) are essential for the analysis of gene regulation. Accurate prediction of TF motifs is critical, because it is infeasible to assay all TFs in all sequenced eukaryotic genomes. There is ongoing controversy regarding the degree of motif diversification among related species that is, in part, because of uncertainty in motif prediction methods. Here we describe similarity regression, a significantly improved method for predicting motifs, which we use to update and expand the Cis-BP database. Similarity regression inherently quantifies TF motif evolution, and shows that previous claims of near-complete conservation of motifs between human and Drosophila are inflated, with nearly half of the motifs in each species absent from the other, largely due to extensive divergence in C2H2 zinc finger proteins. We conclude that diversification in DNA-binding motifs is pervasive, and present a new tool and updated resource to study TF diversity and gene regulation across eukaryotes.

QBiC-Pred: quantitative predictions of transcription factor binding changes due to sequence variants.

Non-coding genetic variants/mutations can play functional roles in the cell by disrupting regulatory interactions between transcription factors (TFs) and their genomic target sites. For most human TFs, a myriad of DNA-binding models are available and could be used to predict the effects of DNA mutations on TF binding. However, information on the quality of these models is scarce, making it hard to evaluate the statistical significance of predicted binding changes. Here, we present QBiC-Pred, a web server for predicting quantitative TF binding changes due to nucleotide variants. QBiC-Pred uses regression models of TF binding specificity trained on high-throughput in vitro data. The training is done using ordinary least squares (OLS), and we leverage distributional results associated with OLS estimation to compute, for each predicted change in TF binding, a P-value reflecting our confidence in the predicted effect. We show that OLS models are accurate in predicting the effects of mutations on TF binding in vitro and in vivo, outperforming widely-used PWM models as well as recently developed deep learning models of specificity. QBiC-Pred takes as input mutation datasets in several formats, and it allows post-processing of the results through a user-friendly web interface. QBiC-Pred is freely available at http://qbic.genome.duke.edu.

Impact of cancer mutational signatures on transcription factor motifs in the human genome

BackgroundSomatic mutations in cancer genomes occur through a variety of molecular mechanisms, which contribute to different mutational patterns. To summarize these, mutational signatures have been defined using a large number of cancer genomes, and related to distinct mutagenic processes. Each cancer genome can be compared to this reference dataset and its exposure to one or the other signature be determined. Given the very different mutational patterns of these signatures, we anticipate that they will have distinct impact on genomic elements, in particular motifs for transcription factor binding sites (TFBS).MethodsWe used the 30 mutational signatures from the COSMIC database, and derived a theoretical framework to infer the impact of these signatures on the alteration of transcription factor (TF) binding motifs from the JASPAR database. Hence, we translated the trinucleotide mutation frequencies of the signatures into alteration frequencies of specific TF binding motifs, leading either to creation or disruption of these motifs.ResultsMotif families show different susceptibility to alterations induced by the mutational signatures. For certain motifs, a high correlation is observed between the TFBS motif creation and disruption events related to the information content of the motif. Moreover, we observe striking patterns regarding for example the Ets-motif family, for which a high impact of UV induced signatures is observed. Our model also confirms the susceptibility of specific transcription factor motifs to deamination processes.ConclusionOur results show that the mutational signatures have different impact on the binding motifs of transcription factors and that for certain high complexity motifs there is a strong correlation between creation and disruption, related to the information content of the motif. This study represents a background estimation of the alterations due purely to mutational signatures in the absence of additional contributions, e.g. from evolutionary processes.

Mice Exposed to Early Life Stress Display Sex‐Specific Upregulation of Leptin Gene Expression in Adipose Tissue

Epidemiological studies have established early life stress (ELS) as an independent risk factor for increased BMI and cardio‐metabolic disease later in life. We have previously shown that a mouse model of ELS, Maternal Separation and Early Weaning (MSEW), increases adiposity in weanlings and exacerbates high fat diet (HFD)‐induced fat expansion in a sex‐specific manner. Therefore, the aim of this study was to investigate the sex‐specific effects of MSEW on adipocyte‐derived production of important hormones for body weight homeostasis, such as leptin and adiponectin. MSEW was performed by separating the pups from the dam for 4 to 8 hours during postnatal days (PD) 2 to 16, and weaned on PD 17. Control mice remained undisturbed and were weaned on PD 21. Following, eight‐week‐old mice were then fed on either a low fat diet (LFD, n=6) or HFD (10 or 60 % Kcal from fat, n=6) for 12 weeks. Each litter was represented by one mice of each sex randomly assigned to each diet. Body composition trajectory was assessed by EchoMRI. At the end of the experiment, mice were sacrificed for adipose tissue collection. Leptin and adiponectin gene expression was performed in gonadal white adipose tissue (gWAT) and subcutaneous WAT (scWAT). After 12 weeks of HFD, male MSEW mice showed increased BW compared with controls (47.9 ± 1.1 vs. 40.2 ± 2.6 g, p<0.05) as well body adiposity (36.7 + 1.1 vs. 22.2 ± 4.1 % body fat, respectively). In female MSEW mice fed a HFD, but not in males, the gene expression of leptin in gWAT was significantly upregulated compared with controls (5.0 ± 0.5 vs. 3.6 ± 0.5, p<0.05), but not in scWAT (2.1 ± 0.6 vs. 1.5 ± 0.5, p<0.05). Despite sex‐specific changes in adiponectin gene expression in response to HFD, there was not a significant effect due to MSEW. To begin to address the mechanism by which MSEW increases leptin expression in a sex‐specific manner, we analyzed the site‐specific methylation of the leptin promoter in gWAT from mice fed a HFD (n=4). The methylation landscape in the leptin promoter was different between male and female mice. However, we found that CpG sites 3 and 15 were significantly hypomethylated only in fat from female MSEW mice comparted to controls (site 3: 21±1 vs. 52±1 %5mC, p<0.05; site 15: 84±1 vs. 100 %5mC, p<0.05). Both sites in the leptin promoter are strong areas for transcription factor binding. In addition, DNMT3A was upregulated in MSEW mice (p<0.05). Taken together, our current working hypothesis is focused on elucidating whether exposure to MSEW may promote the binding of specific transcription factors that upregulate leptin gene expression in a depot and sex‐specific manner.Support or Funding InformationR01HF135158 to ASL, R01HL135158‐01A1S1 to JRLThis abstract is from the Experimental Biology 2019 Meeting. There is no full text article associated with this abstract published in The FASEB Journal.

Uncovering DNA binding properties of the GATA4 and TBX5 transcription factor complex

Transcription factors (TFs) are essential master regulators of gene expression. These DNA binding proteins are fundamental for cellular differentiation during development. Eukaryotic TFs are notorious for not working alone and they often bind DNA as multimeric protein complexes to regulate gene expression. GATA4 and TBX5 are transcription factors that are central components of the gene regulatory network of heart development and function. Recent studies have determined the binding specificity of multiple transcription factors as monomers. However, the DNA‐binding properties of the cooperative complex between GATA4 and TBX5 remain undetermined. Based on this, we want to know the intrinsic DNA binding preferences and the emergent properties that result from the cooperative complex formed by GATA4 and TBX5. We used a cell‐free protein expression system to produce GATA4 and TBX5, and we successfully validated their DNA‐binding activity through an Electrophoretic Mobility Shift Assay (EMSA). Subsequently, we will determine the in vitro DNA‐binding specificity of the GATA4:TBX5 complex using Cognate Site Identification (CSI) by High‐Throughput Systematic Evolution of Ligands by Exponential Enrichment (HT‐SELEX). The findings of this study will help understand the DNA‐recognition rules of the GATA4:TBX5 complex and its potential roles in normal heart development.Support or Funding InformationGrant Acknowledgement: This research is funded by the National Institutes of Health ‐ NIH SC1GM127231 and the Puerto Rico Louis Stokes Alliance for Minority Participation (PR‐LSAMP)‐Undergraduate Fellowship.This abstract is from the Experimental Biology 2019 Meeting. There is no full text article associated with this abstract published in The FASEB Journal.

FactorNet: A deep learning framework for predicting cell type specific transcription factor binding from nucleotide-resolution sequential data.

Due to the large numbers of transcription factors (TFs) and cell types, querying binding profiles of all valid TF/cell type pairs is not experimentally feasible. To address this issue, we developed a convolutional-recurrent neural network model, called FactorNet, to computationally impute the missing binding data. FactorNet trains on binding data from reference cell types to make predictions on testing cell types by leveraging a variety of features, including genomic sequences, genome annotations, gene expression, and signal data, such as DNase I cleavage. FactorNet implements several convenient strategies to reduce runtime and memory consumption. By visualizing the neural network models, we can interpret how the model predicts binding. We also investigate the variables that affect cross-cell type accuracy, and offer suggestions to improve upon this field. Our method ranked among the top teams in the ENCODE-DREAM in vivo Transcription Factor Binding Site Prediction Challenge, achieving first place on six of the 13 final round evaluation TF/cell type pairs, the most of any competing team. The FactorNet source code is publicly available, allowing users to reproduce our methodology from the ENCODE-DREAM Challenge.

Epigenetic regulation of REX1 expression and chromatin binding specificity by HMGNs.

HMGN proteins localize to chromatin regulatory sites and modulate the cell-type specific transcription profile; however, the molecular mechanism whereby these ubiquitous nucleosome binding proteins affect gene expression is not fully understood. Here, we show that HMGNs regulate the expression of Rex1, one of the most highly transcribed genes in mouse embryonic stem cells (ESCs), by recruiting the transcription factors NANOG, OCT4 and SOX2 to an ESC-specific super enhancer located in the 5' region of Rex1. HMGNs facilitate the establishment of an epigenetic landscape characteristic of active chromatin and enhancer promoter interactions, as seen by chromatin conformation capture. Loss of HMGNs alters the local epigenetic profile, increases histone H1 occupancy, decreases transcription factors binding and reduces enhancer promoter interactions, thereby downregulating, but not abolishing Rex1 expression. ChIP-seq analyses show high colocalization of HMGNs and of REX1, a zinc finger protein, at promoters and enhancers. Loss of HMGNs preferentially reduces the specific binding of REX1 to these chromatin regulatory sites. Thus, HMGNs affects both the expression and the chromatin binding specificity of REX1. We suggest that HMGNs affect cell-type specific gene expression by modulating the binding specificity of transcription factors to chromatin.

The chromatin remodeling protein Lsh alters nucleosome occupancy at putative enhancers and modulates binding of lineage specific transcription factors

ABSTRACTDynamic regulation of chromatin accessibility is a key feature of cellular differentiation during embryogenesis, but the precise factors that control access to chromatin remain largely unknown. Lsh/HELLS is critical for normal development and mutations of Lsh in human cause the ICF (Immune deficiency, Centromeric instability, Facial anomalies) syndrome, a severe immune disorder with multiple organ deficiencies. We report here that Lsh, previously known to regulate DNA methylation level, has a genome wide chromatin remodeling function. Using micrococcal nuclease (MNase)-seq analysis, we demonstrate that Lsh protects MNase accessibility at transcriptional regulatory regions characterized by DNase I hypersensitivity and certain histone 3 (H3) tail modifications associated with enhancers. Using an auxin-inducible degron system, allowing proteolytical degradation of Lsh, we show that Lsh mediated changes in nucleosome occupancy are independent of DNA methylation level and are characterized by reduced H3 occupancy. While Lsh mediated nucleosome occupancy prevents binding sites for transcription factors in wild type cells, depletion of Lsh leads to an increase in binding of ectopically expressed tissue specific transcription factors to their respective binding sites. Our data suggests that Lsh mediated chromatin remodeling can modulate nucleosome positioning at a subset of putative enhancers contributing to the preservation of cellular identity through regulation of accessibility.