Chromatin gene-gene loops support the cross-regulation of genes with related function.

Chromatin loops between gene pairs have been observed in diverse contexts in both flies and vertebrates. Combining high-resolution Capture-C, DNA fluorescence in situ hybridization, and genetic perturbations, we dissect the functional role of three loops between genes with related function during Drosophila embryogenesis. By mutating the loop anchor (but not the gene) or the gene (but not loop anchor), we disentangle loop formation and gene expression and show that the 3D proximity of paralogous gene loci supports their co-regulation. Breaking the loop leads to either an attenuation or enhancement of expression and perturbs their relative levels of expression and cross-regulation. Although many loops appear constitutive across embryogenesis, their function can change in different developmental contexts. Taken together, our results indicate that chromatin gene-gene loops act as architectural scaffolds that can be used in different ways in different contexts to fine-tune the coordinated expression of genes with related functions and sustain their cross-regulation.

A transcription factor collective defines cardiac cell fate and reflects lineage history.

Cell fate decisions are driven through the integration of inductive signals and tissue-specific transcription factors (TFs), although the details on how this information converges in cis remain unclear. Here, we demonstrate that the five genetic components essential for cardiac specification in Drosophila, including the effectors of Wg and Dpp signaling, act as a collective unit to cooperatively regulate heart enhancer activity, both in vivo and in vitro. Their combinatorial binding does not require any specific motif orientation or spacing, suggesting an alternative mode of enhancer function whereby cooperative activity occurs with extensive motif flexibility. A fraction of enhancers co-occupied by cardiogenic TFs had unexpected activity in the neighboring visceral mesoderm but could be rendered active in heart through single-site mutations. Given that cardiac and visceral cells are both derived from the dorsal mesoderm, this "dormant" TF binding signature may represent a molecular footprint of these cells' developmental lineage.

Dual functionality of cis-regulatory elements as developmental enhancers and Polycomb response elements.

Developmental gene expression is tightly regulated through enhancer elements, which initiate dynamic spatio-temporal expression, and Polycomb response elements (PREs), which maintain stable gene silencing. These two cis-regulatory functions are thought to operate through distinct dedicated elements. By examining the occupancy of the Drosophila pleiohomeotic repressive complex (PhoRC) during embryogenesis, we revealed extensive co-occupancy at developmental enhancers. Using an established in vivo assay for PRE activity, we demonstrated that a subset of characterized developmental enhancers can function as PREs, silencing transcription in a Polycomb-dependent manner. Conversely, some classic Drosophila PREs can function as developmental enhancers in vivo, activating spatio-temporal expression. This study therefore uncovers elements with dual function: activating transcription in some cells (enhancers) while stably maintaining transcriptional silencing in others (PREs). Given that enhancers initiate spatio-temporal gene expression, reuse of the same elements by the Polycomb group (PcG) system may help fine-tune gene expression and ensure the timely maintenance of cell identities.

The hourglass model of evolutionary conservation during embryogenesis extends to developmental enhancers with signatures of positive selection.

Inter-species comparisons of both morphology and gene expression within a phylum have revealed a period in the middle of embryogenesis with more similarity between species compared with earlier and later time points. This "developmental hourglass" pattern has been observed in many phyla, yet the evolutionary constraints on gene expression, as well as the underlying mechanisms of how this is regulated, remain elusive. Moreover, the role of positive selection on gene regulation in the more diverged earlier and later stages of embryogenesis remains unknown. Here, using DNase-seq to identify regulatory regions in two distant Drosophila species (D. melanogaster and D. virilis), we assessed the evolutionary conservation and adaptive evolution of enhancers throughout multiple stages of embryogenesis. This revealed a higher proportion of conserved enhancers at the phylotypic period, providing a regulatory basis for the hourglass expression pattern. Using an in silico mutagenesis approach, we detect signatures of positive selection on developmental enhancers at early and late stages of embryogenesis, with a depletion at the phylotypic period, suggesting positive selection as one evolutionary mechanism underlying the hourglass pattern of animal evolution.

Expanding the Circuitry of Pluripotency by Selective Isolation of Chromatin-Associated Proteins.

Maintenance of pluripotency is regulated by a network of transcription factors coordinated by Oct4, Sox2, and Nanog (OSN), yet a systematic investigation of the composition and dynamics of the OSN protein network specifically on chromatin is still missing. Here we have developed a method combining ChIP with selective isolation of chromatin-associated proteins (SICAP) followed by mass spectrometry to identify chromatin-bound partners of a protein of interest. ChIP-SICAP in mouse embryonic stem cells (ESCs) identified over 400 proteins associating with OSN, including several whose interaction depends on the pluripotent state. Trim24, a previously unrecognized protein in the network, converges with OSN on multiple enhancers and suppresses the expression of developmental genes while activating cell cycle genes. Consistently, Trim24 significantly improved efficiency of cellular reprogramming, demonstrating its direct functionality in establishing pluripotency. Collectively, ChIP-SICAP provides a powerful tool to decode chromatin protein composition, further enhanced by its integrative capacity to perform ChIP-seq.

A conserved role for Snail as a potentiator of active transcription.

The transcription factors of the Snail family are key regulators of epithelial-mesenchymal transitions, cell morphogenesis, and tumor metastasis. Since its discovery in Drosophila ∼25 years ago, Snail has been extensively studied for its role as a transcriptional repressor. Here we demonstrate that Drosophila Snail can positively modulate transcriptional activation. By combining information on in vivo occupancy with expression profiling of hand-selected, staged snail mutant embryos, we identified 106 genes that are potentially directly regulated by Snail during mesoderm development. In addition to the expected Snail-repressed genes, almost 50% of Snail targets showed an unanticipated activation. The majority of "Snail-activated" genes have enhancer elements cobound by Twist and are expressed in the mesoderm at the stages of Snail occupancy. Snail can potentiate Twist-mediated enhancer activation in vitro and is essential for enhancer activity in vivo. Using a machine learning approach, we show that differentially enriched motifs are sufficient to predict Snail's regulatory response. In silico mutagenesis revealed a likely causative motif, which we demonstrate is essential for enhancer activation. Taken together, these data indicate that Snail can potentiate enhancer activation by collaborating with different activators, providing a new mechanism by which Snail regulates development.

Identification and in silico modeling of enhancers reveals new features of the cardiac differentiation network.

Developmental patterning and tissue formation are regulated through complex gene regulatory networks (GRNs) driven through the action of transcription factors (TFs) converging on enhancer elements. Here, as a point of entry to dissect the poorly defined GRN underlying cardiomyocyte differentiation, we apply an integrated approach to identify active enhancers and TFs involved in Drosophila heart development. The Drosophila heart consists of 104 cardiomyocytes, representing less than 0.5% of all cells in the embryo. By modifying BiTS-ChIP for rare cells, we examined H3K4me3 and H3K27ac chromatin landscapes to identify active promoters and enhancers specifically in cardiomyocytes. These in vivo data were complemented by a machine learning approach and extensive in vivo validation in transgenic embryos, which identified many new heart enhancers and their associated TF motifs. Our results implicate many new TFs in late stages of heart development, including Bagpipe, an Nkx3.2 ortholog, which we show is essential for differentiated heart function.

Opbp is a new architectural/insulator protein required for ribosomal gene expression.

A special class of poorly characterized architectural proteins is required for chromatin topology and enhancer-promoter interactions. Here, we identify Opbp as a new Drosophila architectural protein, interacting with CP190 both in vivo and in vitro. Opbp binds to a very restrictive set of genomic regions, through a rare sequence specific motif. These sites are co-bound by CP190 in vivo, and generally located at bidirectional promoters of ribosomal protein genes. We show that Opbp is essential for viability, and loss of opbp function, or destruction of its motif, leads to reduced ribosomal protein gene expression, indicating a functional role in promoter activation. As characteristic of architectural/insulator proteins, the Opbp motif is sufficient for distance-dependent reporter gene activation and enhancer-blocking activity, suggesting an Opbp-mediated enhancer-promoter interaction. Rather than having a constitutive role, Opbp represents a new type of architectural protein with a very restricted, yet essential, function in regulation of housekeeping gene expression.

Dynamix: dynamic visualization by automatic selection of informative tracks from hundreds of genomic datasets.

MOTIVATION: Visualization of genomic data is fundamental for gaining insights into genome function. Yet, co-visualization of a large number of datasets remains a challenge in all popular genome browsers and the development of new visualization methods is needed to improve the usability and user experience of genome browsers. RESULTS: We present Dynamix, a JBrowse plugin that enables the parallel inspection of hundreds of genomic datasets. Dynamix takes advantage of a priori knowledge to automatically display data tracks with signal within a genomic region of interest. As the user navigates through the genome, Dynamix automatically updates data tracks and limits all manual operations otherwise needed to adjust the data visible on screen. Dynamix also introduces a new carousel view that optimizes screen utilization by enabling users to independently scroll through groups of tracks. AVAILABILITY AND IMPLEMENTATION: Dynamix is hosted at http://furlonglab.embl.de/Dynamix . CONTACT: charles.girardot@embl.de. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Coordinated repression and activation of two transcriptional programs stabilizes cell fate during myogenesis.

Molecular models of cell fate specification typically focus on the activation of specific lineage programs. However, the concurrent repression of unwanted transcriptional networks is also essential to stabilize certain cellular identities, as shown in a number of diverse systems and phyla. Here, we demonstrate that this dual requirement also holds true in the context of Drosophila myogenesis. By integrating genetics and genomics, we identified a new role for the pleiotropic transcriptional repressor Tramtrack69 in myoblast specification. Drosophila muscles are formed through the fusion of two discrete cell types: founder cells (FCs) and fusion-competent myoblasts (FCMs). When tramtrack69 is removed, FCMs appear to adopt an alternative muscle FC-like fate. Conversely, ectopic expression of this repressor phenocopies muscle defects seen in loss-of-function lame duck mutants, a transcription factor specific to FCMs. This occurs through Tramtrack69-mediated repression in FCMs, whereas Lame duck activates a largely distinct transcriptional program in the same cells. Lineage-specific factors are therefore not sufficient to maintain FCM identity. Instead, their identity appears more plastic, requiring the combination of instructive repressive and activating programs to stabilize cell fate.

Qualitative Dynamical Modelling Can Formally Explain Mesoderm Specification and Predict Novel Developmental Phenotypes.

Given the complexity of developmental networks, it is often difficult to predict the effect of genetic perturbations, even within coding genes. Regulatory factors generally have pleiotropic effects, exhibit partially redundant roles, and regulate highly interconnected pathways with ample cross-talk. Here, we delineate a logical model encompassing 48 components and 82 regulatory interactions involved in mesoderm specification during Drosophila development, thereby providing a formal integration of all available genetic information from the literature. The four main tissues derived from mesoderm correspond to alternative stable states. We demonstrate that the model can predict known mutant phenotypes and use it to systematically predict the effects of over 300 new, often non-intuitive, loss- and gain-of-function mutations, and combinations thereof. We further validated several novel predictions experimentally, thereby demonstrating the robustness of model. Logical modelling can thus contribute to formally explain and predict regulatory outcomes underlying cell fate decisions.

Subtle changes in motif positioning cause tissue-specific effects on robustness of an enhancer's activity.

Deciphering the specific contribution of individual motifs within cis-regulatory modules (CRMs) is crucial to understanding how gene expression is regulated and how this process is affected by sequence variation. But despite vast improvements in the ability to identify where transcription factors (TFs) bind throughout the genome, we are limited in our ability to relate information on motif occupancy to function from sequence alone. Here, we engineered 63 synthetic CRMs to systematically assess the relationship between variation in the content and spacing of motifs within CRMs to CRM activity during development using Drosophila transgenic embryos. In over half the cases, very simple elements containing only one or two types of TF binding motifs were capable of driving specific spatio-temporal patterns during development. Different motif organizations provide different degrees of robustness to enhancer activity, ranging from binary on-off responses to more subtle effects including embryo-to-embryo and within-embryo variation. By quantifying the effects of subtle changes in motif organization, we were able to model biophysical rules that explain CRM behavior and may contribute to the spatial positioning of CRM activity in vivo. For the same enhancer, the effects of small differences in motif positions varied in developmentally related tissues, suggesting that gene expression may be more susceptible to sequence variation in one tissue compared to another. This result has important implications for human eQTL studies in which many associated mutations are found in cis-regulatory regions, though the mechanism for how they affect tissue-specific gene expression is often not understood.

Je, a versatile suite to handle multiplexed NGS libraries with unique molecular identifiers.

BACKGROUND: The yield obtained from next generation sequencers has increased almost exponentially in recent years, making sample multiplexing common practice. While barcodes (known sequences of fixed length) primarily encode the sample identity of sequenced DNA fragments, barcodes made of random sequences (Unique Molecular Identifier or UMIs) are often used to distinguish between PCR duplicates and transcript abundance in, for example, single-cell RNA sequencing (scRNA-seq). In paired-end sequencing, different barcodes can be inserted at each fragment end to either increase the number of multiplexed samples in the library or to use one of the barcodes as UMI. Alternatively, UMIs can be combined with the sample barcodes into composite barcodes, or with standard Illumina® indexing. Subsequent analysis must take read duplicates and sample identity into account, by identifying UMIs. RESULTS: Existing tools do not support these complex barcoding configurations and custom code development is frequently required. Here, we present Je, a suite of tools that accommodates complex barcoding strategies, extracts UMIs and filters read duplicates taking UMIs into account. Using Je on publicly available scRNA-seq and iCLIP data containing UMIs, the number of unique reads increased by up to 36 %, compared to when UMIs are ignored. CONCLUSIONS: Je is implemented in JAVA and uses the Picard API. Code, executables and documentation are freely available at http://gbcs.embl.de/Je . Je can also be easily installed in Galaxy through the Galaxy toolshed.

Combinatorial binding predicts spatio-temporal cis-regulatory activity.

Development requires the establishment of precise patterns of gene expression, which are primarily controlled by transcription factors binding to cis-regulatory modules. Although transcription factor occupancy can now be identified at genome-wide scales, decoding this regulatory landscape remains a daunting challenge. Here we used a novel approach to predict spatio-temporal cis-regulatory activity based only on in vivo transcription factor binding and enhancer activity data. We generated a high-resolution atlas of cis-regulatory modules describing their temporal and combinatorial occupancy during Drosophila mesoderm development. The binding profiles of cis-regulatory modules with characterized expression were used to train support vector machines to predict five spatio-temporal expression patterns. In vivo transgenic reporter assays demonstrate the high accuracy of these predictions and reveal an unanticipated plasticity in transcription factor binding leading to similar expression. This data-driven approach does not require previous knowledge of transcription factor sequence affinity, function or expression, making it widely applicable.

Model-based method for transcription factor target identification with limited data.

We present a computational method for identifying potential targets of a transcription factor (TF) using wild-type gene expression time series data. For each putative target gene we fit a simple differential equation model of transcriptional regulation, and the model likelihood serves as a score to rank targets. The expression profile of the TF is modeled as a sample from a Gaussian process prior distribution that is integrated out using a nonparametric Bayesian procedure. This results in a parsimonious model with relatively few parameters that can be applied to short time series datasets without noticeable overfitting. We assess our method using genome-wide chromatin immunoprecipitation (ChIP-chip) and loss-of-function mutant expression data for two TFs, Twist, and Mef2, controlling mesoderm development in Drosophila. Lists of top-ranked genes identified by our method are significantly enriched for genes close to bound regions identified in the ChIP-chip data and for genes that are differentially expressed in loss-of-function mutants. Targets of Twist display diverse expression profiles, and in this case a model-based approach performs significantly better than scoring based on correlation with TF expression. Our approach is found to be comparable or superior to ranking based on mutant differential expression scores. Also, we show how integrating complementary wild-type spatial expression data can further improve target ranking performance.

CTCF, BEAF-32, and CP190 are not required for the establishment of TADs in early Drosophila embryos but have locus-specific roles.

The boundaries of topologically associating domains (TADs) are delimited by insulators and/or active promoters; however, how they are initially established during embryogenesis remains unclear. Here, we examined this during the first hours of Drosophila embryogenesis. DNA-FISH confirms that intra-TAD pairwise proximity is established during zygotic genome activation (ZGA) but with extensive cell-to-cell heterogeneity. Most newly formed boundaries are occupied by combinations of CTCF, BEAF-32, and/or CP190. Depleting each insulator individually from chromatin revealed that TADs can still establish, although with lower insulation, with a subset of boundaries (~10%) being more dependent on specific insulators. Some weakened boundaries have aberrant gene expression due to unconstrained enhancer activity. However, the majority of misexpressed genes have no obvious direct relationship to changes in domain-boundary insulation. Deletion of an active promoter (thereby blocking transcription) at one boundary had a greater impact than deleting the insulator-bound region itself. This suggests that cross-talk between insulators and active promoters and/or transcription might reinforce domain boundary insulation during embryogenesis.

A stem cell zoo uncovers intracellular scaling of developmental tempo across mammals.

Differential speeds in biochemical reactions have been proposed to be responsible for the differences in developmental tempo between mice and humans. However, the underlying mechanism controlling the species-specific kinetics remains to be determined. Using in vitro differentiation of pluripotent stem cells, we recapitulated the segmentation clocks of diverse mammalian species varying in body weight and taxa: marmoset, rabbit, cattle, and rhinoceros. Together with mouse and human, the segmentation clock periods of the six species did not scale with the animal body weight, but with the embryogenesis length. The biochemical kinetics of the core clock gene HES7 displayed clear scaling with the species-specific segmentation clock period. However, the cellular metabolic rates did not show an evident correlation. Instead, genes involving biochemical reactions showed an expression pattern that scales with the segmentation clock period. Altogether, our stem cell zoo uncovered general scaling laws governing species-specific developmental tempo.

4DXpress: a database for cross-species expression pattern comparisons.

In the major animal model species like mouse, fish or fly, detailed spatial information on gene expression over time can be acquired through whole mount in situ hybridization experiments. In these species, expression patterns of many genes have been studied and data has been integrated into dedicated model organism databases like ZFIN for zebrafish, MEPD for medaka, BDGP for Drosophila or GXD for mouse. However, a central repository that allows users to query and compare gene expression patterns across different species has not yet been established. Therefore, we have integrated expression patterns for zebrafish, Drosophila, medaka and mouse into a central public repository called 4DXpress (expression database in four dimensions). Users can query anatomy ontology-based expression annotations across species and quickly jump from one gene to the orthologues in other species. Genes are linked to public microarray data in ArrayExpress. We have mapped developmental stages between the species to be able to compare developmental time phases. We store the largest collection of gene expression patterns available to date in an individual resource, reflecting 16 505 annotated genes. 4DXpress will be an invaluable tool for developmental as well as for computational biologists interested in gene regulation and evolution. 4DXpress is available at http://ani.embl.de/4DXpress.

Lineage-Resolved Enhancer and Promoter Usage during a Time Course of Embryogenesis.

Enhancers are essential drivers of cell states, yet the relationship between accessibility, regulatory activity, and in vivo lineage commitment during embryogenesis remains poorly understood. Here, we measure chromatin accessibility in isolated neural and mesodermal lineages across a time course of Drosophila embryogenesis. Promoters, including tissue-specific genes, are often constitutively open, even in contexts where the gene is not expressed. In contrast, the majority of distal elements have dynamic, tissue-specific accessibility. Enhancer priming appears rarely within a lineage, perhaps reflecting the speed of Drosophila embryogenesis. However, many tissue-specific enhancers are accessible in other lineages early on and become progressively closed as embryogenesis proceeds. We demonstrate the usefulness of this tissue- and time-resolved resource to definitively identify single-cell clusters, to uncover predictive motifs, and to identify many regulators of tissue development. For one such predicted neural regulator, l(3)neo38, we generate a loss-of-function mutant and uncover an essential role for neuromuscular junction and brain development.

The Role of Chromatin Accessibility in cis-Regulatory Evolution.

Transcription factor (TF) binding is determined by sequence as well as chromatin accessibility. Although the role of accessibility in shaping TF-binding landscapes is well recorded, its role in evolutionary divergence of TF binding, which in turn can alter cis-regulatory activities, is not well understood. In this work, we studied the evolution of genome-wide binding landscapes of five major TFs in the core network of mesoderm specification, between Drosophila melanogaster and Drosophila virilis, and examined its relationship to accessibility and sequence-level changes. We generated chromatin accessibility data from three important stages of embryogenesis in both Drosophila melanogaster and Drosophila virilis and recorded conservation and divergence patterns. We then used multivariable models to correlate accessibility and sequence changes to TF-binding divergence. We found that accessibility changes can in some cases, for example, for the master regulator Twist and for earlier developmental stages, more accurately predict binding change than is possible using TF-binding motif changes between orthologous enhancers. Accessibility changes also explain a significant portion of the codivergence of TF pairs. We noted that accessibility and motif changes offer complementary views of the evolution of TF binding and developed a combined model that captures the evolutionary data much more accurately than either view alone. Finally, we trained machine learning models to predict enhancer activity from TF binding and used these functional models to argue that motif and accessibility-based predictors of TF-binding change can substitute for experimentally measured binding change, for the purpose of predicting evolutionary changes in enhancer activity.