A DNA methylation atlas of normal human cell types.

DNA methylation is a fundamental epigenetic mark that governs gene expression and chromatin organization, thus providing a window into cellular identity and developmental processes1. Current datasets typically include only a fraction of methylation sites and are often based either on cell lines that underwent massive changes in culture or on tissues containing unspecified mixtures of cells2-5. Here we describe a human methylome atlas, based on deep whole-genome bisulfite sequencing, allowing fragment-level analysis across thousands of unique markers for 39 cell types sorted from 205 healthy tissue samples. Replicates of the same cell type are more than 99.5% identical, demonstrating the robustness of cell identity programmes to environmental perturbation. Unsupervised clustering of the atlas recapitulates key elements of tissue ontogeny and identifies methylation patterns retained since embryonic development. Loci uniquely unmethylated in an individual cell type often reside in transcriptional enhancers and contain DNA binding sites for tissue-specific transcriptional regulators. Uniquely hypermethylated loci are rare and are enriched for CpG islands, Polycomb targets and CTCF binding sites, suggesting a new role in shaping cell-type-specific chromatin looping. The atlas provides an essential resource for study of gene regulation and disease-associated genetic variants, and a wealth of potential tissue-specific biomarkers for use in liquid biopsies.

A high-resolution enhancer atlas of the developing telencephalon.

The mammalian telencephalon plays critical roles in cognition, motor function, and emotion. Though many of the genes required for its development have been identified, the distant-acting regulatory sequences orchestrating their in vivo expression are mostly unknown. Here, we describe a digital atlas of in vivo enhancers active in subregions of the developing telencephalon. We identified more than 4,600 candidate embryonic forebrain enhancers and studied the in vivo activity of 329 of these sequences in transgenic mouse embryos. We generated serial sets of histological brain sections for 145 reproducible forebrain enhancers, resulting in a publicly accessible web-based data collection comprising more than 32,000 sections. We also used epigenomic analysis of human and mouse cortex tissue to directly compare the genome-wide enhancer architecture in these species. These data provide a primary resource for investigating gene regulatory mechanisms of telencephalon development and enable studies of the role of distant-acting enhancers in neurodevelopmental disorders.

Ep300 sequestration to functionally distinct glucocorticoid receptor binding loci underlie rapid gene activation and repression.

The rapid transcriptional response to the transcription factor, glucocorticoid receptor (GR), including gene activation or repression, is mediated by the spatial association of genes with multiple GR binding sites (GBSs) over large genomic distances. However, only a minority of the GBSs have independent GR-mediated activating capacity, and GBSs with independent repressive activity were rarely reported. To understand the positive and negative effects of GR we mapped the regulatory environment of its gene targets. We show that the chromatin interaction networks of GR-activated and repressed genes are spatially separated and vary in the features and configuration of their GBS and other non-GBS regulatory elements. The convergence of the KLF4 pathway in GR-activated domains and the STAT6 pathway in GR-repressed domains, impose opposite transcriptional effects to GR, independent of hormone application. Moreover, the ROR and Rev-erb transcription factors serve as positive and negative regulators, respectively, of GR-mediated gene activation. We found that the spatial crosstalk between GBSs and non-GBSs provides a physical platform for sequestering the Ep300 co-activator from non-GR regulatory loci in both GR-activated and -repressed gene compartments. While this allows rapid gene repression, Ep300 recruitment to GBSs is productive specifically in the activated compartments, thus providing the basis for gene induction.

Universal lung epithelium DNA methylation markers for detection of lung damage in liquid biopsies.

BACKGROUND: Circulating biomarkers for lung damage are lacking. Lung epithelium-specific DNA methylation patterns can potentially report the presence of lung-derived cell-free DNA (cfDNA) in blood, as an indication of lung cell death. METHODS: We sorted human lung alveolar and bronchial epithelial cells from surgical specimens, and obtained their methylomes using whole-genome bisulfite sequencing. We developed a PCR sequencing assay determining the methylation status of 17 loci with lung-specific methylation patterns, and used it to assess lung-derived cfDNA in the plasma of healthy volunteers and patients with lung disease. RESULTS: Loci that are uniquely unmethylated in alveolar or bronchial epithelial cells are enriched for enhancers controlling lung-specific genes. Methylation markers extracted from these methylomes revealed that normal lung cell turnover probably releases cfDNA into the air spaces, rather than to blood. People with advanced lung cancer show a massive elevation of lung cfDNA concentration in blood. Among individuals undergoing bronchoscopy, lung-derived cfDNA is observed in the plasma of those later diagnosed with lung cancer, and to a lesser extent in those diagnosed with other lung diseases. Lung cfDNA is also elevated in patients with acute exacerbation of COPD compared with patients with stable disease, and is associated with future exacerbation and mortality in these patients. CONCLUSIONS: Universal cfDNA methylation markers of normal lung epithelium allow for mutation-independent, sensitive and specific detection of lung-derived cfDNA, reporting on ongoing lung injury. Such markers can find broad utility in the study of normal and pathologic human lung dynamics.

Functional characterization of the ZEB2 regulatory landscape.

Zinc finger E-box-binding homeobox 2 (ZEB2) is a key developmental regulator of the central nervous system (CNS). Although the transcriptional regulation of ZEB2 is essential for CNS development, the elements that regulate ZEB2 expression have yet to be identified. Here, we identified a proximal regulatory region of ZEB2 and characterized transcriptional enhancers during neuronal development. Using chromatin immunoprecipitation sequencing for active (H3K27ac) and repressed (H3K27me3) chromatin regions in human neuronal progenitors, combined with an in vivo zebrafish enhancer assay, we functionally characterized 18 candidate enhancers in the ZEB2 locus. Eight enhancers drove expression patterns that were specific to distinct mid/hindbrain regions (ZEB2#e3 and 5), trigeminal-like ganglia (ZEB2#e6 and 7), notochord (ZEB2#e2, 4 and 12) and whole brain (ZEB2#e14). We further dissected the minimal sequences that drive enhancer-specific activity in the mid/hindbrain and notochord. Using a reporter assay in human cells, we showed an increased activity of the minimal notochord enhancer ZEB2#e2 in response to AP-1 and DLX1/2 expressions, while repressed activity of this enhancer was seen in response to ZEB2 and TFAP2 expressions. We showed that Dlx1 but not Zeb2 and Tfap2 occupies Zeb2#e2 enhancer sequence in the mouse notochord at embryonic day 11.5. Using CRISPR/Cas9 genome editing, we deleted the ZEB2#e2 region, leading to reduction of ZEB2 expression in human cells. We thus characterized distal transcriptional enhancers and trans-acting elements that govern regulation of ZEB2 expression during neuronal development. These findings pave the path toward future analysis of the role of ZEB2 regulatory elements in neurodevelopmental disorders, such as Mowat-Wilson syndrome.

Unraveling the transcriptional regulation of TWIST1 in limb development.

The transcription factor TWIST1 plays a vital role in mesoderm development, particularly in limb and craniofacial formation. Accordingly, haploinsufficiency of TWIST1 can cause limb and craniofacial malformations as part of Saethre-Chotzen syndrome. However, the molecular basis of TWIST1 transcriptional regulation during development has yet to be elucidated. Here, we characterized active enhancers in the TWIST1-HDAC9 locus that drive transcription in the developing limb and branchial arches. Using available p300 and H3K27ac ChIP-seq data, we identified 12 enhancer candidates, located both within and outside the coding sequences of the neighboring gene, Histone deacetyase 9 (HDAC9). Using zebrafish and mouse enhancer assays, we showed that eight of these candidates have limb/fin and branchial arch enhancer activity that resemble Twist1 expression. Using 4C-seq, we showed that the Twist1 promoter region interacts with three enhancers (eTw-5, 6, 7) in the limb bud and branchial arch of mouse embryos at day 11.5. Furthermore, we found that two transcription factors, LMX1B and TFAP2, bind these enhancers and modulate their enhancer activity. Finally, using CRISPR/Cas9 genome editing, we showed that homozygous deletion of eTw5-7 enhancers reduced Twist1 expression in the limb bud and caused pre-axial polydactyly, a phenotype observed in Twist1+/- mice. Taken together, our findings reveal that each enhancer has a discrete activity pattern, and together comprise a spatiotemporal regulatory network of Twist1 transcription in the developing limbs/fins and branchial arches. Our study suggests that mutations in TWIST1 enhancers could lead to reduced TWIST1 expression, resulting in phenotypic outcome as seen with TWIST1 coding mutations.

Reconstructing spatial organizations of chromosomes through manifold learning.

Decoding the spatial organizations of chromosomes has crucial implications for studying eukaryotic gene regulation. Recently, chromosomal conformation capture based technologies, such as Hi-C, have been widely used to uncover the interaction frequencies of genomic loci in a high-throughput and genome-wide manner and provide new insights into the folding of three-dimensional (3D) genome structure. In this paper, we develop a novel manifold learning based framework, called GEM (Genomic organization reconstructor based on conformational Energy and Manifold learning), to reconstruct the three-dimensional organizations of chromosomes by integrating Hi-C data with biophysical feasibility. Unlike previous methods, which explicitly assume specific relationships between Hi-C interaction frequencies and spatial distances, our model directly embeds the neighboring affinities from Hi-C space into 3D Euclidean space. Extensive validations demonstrated that GEM not only greatly outperformed other state-of-art modeling methods but also provided a physically and physiologically valid 3D representations of the organizations of chromosomes. Furthermore, we for the first time apply the modeled chromatin structures to recover long-range genomic interactions missing from original Hi-C data.

Contribution of epigenetic mechanisms to variation in cancer risk among tissues.

Recently, it was suggested that tissue variation in cancer risk originates from differences in the number of stem-cell divisions underlying each tissue, leading to different mutation loads. We show that this variation is also correlated with the degree of aberrant CpG island DNA methylation in normal cells. Methylation accumulates during aging in a subset of molecules, suggesting that the epigenetic landscape within a founder-cell population may contribute to tumor formation.

A hyperdynamic H3.3 nucleosome marks promoter regions in pluripotent embryonic stem cells.

Histone variants and their chaperones are key regulators of eukaryotic transcription, and are critical for normal development. The histone variant H3.3 has been shown to play important roles in pluripotency and differentiation, and although its genome-wide patterns have been investigated, little is known about the role of its dynamic turnover in transcriptional regulation. To elucidate the role of H3.3 dynamics in embryonic stem cell (ESC) biology, we generated mouse ESC lines carrying a single copy of a doxycycline (Dox)-inducible HA-tagged version of H3.3 and monitored the rate of H3.3 incorporation by ChIP-seq at varying time points following Dox induction, before and after RA-induced differentiation. Comparing H3.3 turnover profiles in ESCs and RA-treated cells, we identified a hyperdynamic H3.3-containing nucleosome at the -1 position in promoters of genes expressed in ESCs. This dynamic nucleosome is restricted and shifted downstream into the +1 position following differentiation. We suggest that histone turnover dynamics provides an additional mechanism involved in expression regulation, and that a hyperdynamic -1 nucleosome marks promoters in ESCs. Our data provide evidence for regional regulation of H3.3 turnover in ESC promoters, and calls for testing, in high resolution, the dynamic behavior of additional histone variants and other structural chromatin proteins.

Zelda is differentially required for chromatin accessibility, transcription factor binding, and gene expression in the early Drosophila embryo.

The transition from a specified germ cell to a population of pluripotent cells occurs rapidly following fertilization. During this developmental transition, the zygotic genome is largely transcriptionally quiescent and undergoes significant chromatin remodeling. In Drosophila, the DNA-binding protein Zelda (also known as Vielfaltig) is required for this transition and for transcriptional activation of the zygotic genome. Open chromatin is associated with Zelda-bound loci, as well as more generally with regions of active transcription. Nonetheless, the extent to which Zelda influences chromatin accessibility across the genome is largely unknown. Here we used formaldehyde-assisted isolation of regulatory elements to determine the role of Zelda in regulating regions of open chromatin in the early embryo. We demonstrate that Zelda is essential for hundreds of regions of open chromatin. This Zelda-mediated chromatin accessibility facilitates transcription-factor recruitment and early gene expression. Thus, Zelda possesses some key characteristics of a pioneer factor. Unexpectedly, chromatin at a large subset of Zelda-bound regions remains open even in the absence of Zelda. The GAGA factor-binding motif and embryonic GAGA factor binding are specifically enriched in these regions. We propose that both Zelda and GAGA factor function to specify sites of open chromatin and together facilitate the remodeling of the early embryonic genome.

Taf7l cooperates with Trf2 to regulate spermiogenesis.

TATA-binding protein (TBP)-associated factor 7l (Taf7l; a paralogue of Taf7) and TBP-related factor 2 (Trf2) are components of the core promoter complex required for gene/tissue-specific transcription of protein-coding genes by RNA polymerase II. Previous studies reported that Taf7l knockout (KO) mice exhibit structurally abnormal sperm, reduced sperm count, weakened motility, and compromised fertility. Here we find that continued backcrossing of Taf7l(-/Y) mice from N5 to N9 produced KO males that are essentially sterile. Genome-wide expression profiling by mRNA-sequencing analysis of wild-type (WT) and Taf7l(-/Y) (KO) testes revealed that Taf7l ablation impairs the expression of many postmeiotic spermatogenic-specific as well as metabolic genes. Importantly, histological analysis of testes revealed that Taf7l(-/Y) mice develop postmeiotic arrest at the first stage of spermiogenesis, phenotypically similar to Trf2(-/-) mice, but distinct from Taf4b(-/-) mice. Indeed, we find that Taf7l and Trf2 coregulate postmeiotic genes, but none of Taf4b-regulated germ stem cell genes in testes. Genome-wide ChIP-sequencing studies indicate that TAF7L binds to promoters of activated postmeiotic genes in testis. Moreover, biochemical studies show that TAF7L associates with TRF2 both in vitro and in testis, suggesting that TAF7L likely cooperates directly with TRF2 at promoters of a subset of postmeiotic genes to regulate spermiogenesis. Our findings thus provide a previously undescribed mechanism for cell-type-specific transcriptional control involving an interaction between a "nonprototypic" core promoter recognition factor (Trf2) and an orphan TAF subunit (Taf7l) in mammalian testis-specific gene transcription.

Extensive divergence of transcription factor binding in Drosophila embryos with highly conserved gene expression.

To better characterize how variation in regulatory sequences drives divergence in gene expression, we undertook a systematic study of transcription factor binding and gene expression in blastoderm embryos of four species, which sample much of the diversity in the 40 million-year old genus Drosophila: D. melanogaster, D. yakuba, D. pseudoobscura and D. virilis. We compared gene expression, measured by mRNA-seq, to the genome-wide binding, measured by ChIP-seq, of four transcription factors involved in early anterior-posterior patterning. We found that mRNA levels are much better conserved than individual transcription factor binding events, and that changes in a gene's expression were poorly explained by changes in adjacent transcription factor binding. However, highly bound sites, sites in regions bound by multiple factors and sites near genes are conserved more frequently than other binding, suggesting that a considerable amount of transcription factor binding is weakly or non-functional and not subject to purifying selection.

Zelda binding in the early Drosophila melanogaster embryo marks regions subsequently activated at the maternal-to-zygotic transition.

The earliest stages of development in most metazoans are driven by maternally deposited proteins and mRNAs, with widespread transcriptional activation of the zygotic genome occurring hours after fertilization, at a period known as the maternal-to-zygotic transition (MZT). In Drosophila, the MZT is preceded by the transcription of a small number of genes that initiate sex determination, patterning, and other early developmental processes; and the zinc-finger protein Zelda (ZLD) plays a key role in their transcriptional activation. To better understand the mechanisms of ZLD activation and the range of its targets, we used chromatin immunoprecipitation coupled with high-throughput sequencing (ChIP-Seq) to map regions bound by ZLD before (mitotic cycle 8), during (mitotic cycle 13), and after (late mitotic cycle 14) the MZT. Although only a handful of genes are transcribed prior to mitotic cycle 10, we identified thousands of regions bound by ZLD in cycle 8 embryos, most of which remain bound through mitotic cycle 14. As expected, early ZLD-bound regions include the promoters and enhancers of genes transcribed at this early stage. However, we also observed ZLD bound at cycle 8 to the promoters of roughly a thousand genes whose first transcription does not occur until the MZT and to virtually all of the thousands of known and presumed enhancers bound at cycle 14 by transcription factors that regulate patterned gene activation during the MZT. The association between early ZLD binding and MZT activity is so strong that ZLD binding alone can be used to identify active promoters and regulatory sequences with high specificity and selectivity. This strong early association of ZLD with regions not active until the MZT suggests that ZLD is not only required for the earliest wave of transcription but also plays a major role in activating the genome at the MZT.

Quantitative models of the mechanisms that control genome-wide patterns of transcription factor binding during early Drosophila development.

Transcription factors that drive complex patterns of gene expression during animal development bind to thousands of genomic regions, with quantitative differences in binding across bound regions mediating their activity. While we now have tools to characterize the DNA affinities of these proteins and to precisely measure their genome-wide distribution in vivo, our understanding of the forces that determine where, when, and to what extent they bind remains primitive. Here we use a thermodynamic model of transcription factor binding to evaluate the contribution of different biophysical forces to the binding of five regulators of early embryonic anterior-posterior patterning in Drosophila melanogaster. Predictions based on DNA sequence and in vitro protein-DNA affinities alone achieve a correlation of ∼0.4 with experimental measurements of in vivo binding. Incorporating cooperativity and competition among the five factors, and accounting for spatial patterning by modeling binding in every nucleus independently, had little effect on prediction accuracy. A major source of error was the prediction of binding events that do not occur in vivo, which we hypothesized reflected reduced accessibility of chromatin. To test this, we incorporated experimental measurements of genome-wide DNA accessibility into our model, effectively restricting predicted binding to regions of open chromatin. This dramatically improved our predictions to a correlation of 0.6-0.9 for various factors across known target genes. Finally, we used our model to quantify the roles of DNA sequence, accessibility, and binding competition and cooperativity. Our results show that, in regions of open chromatin, binding can be predicted almost exclusively by the sequence specificity of individual factors, with a minimal role for protein interactions. We suggest that a combination of experimentally determined chromatin accessibility data and simple computational models of transcription factor binding may be used to predict the binding landscape of any animal transcription factor with significant precision.

Particle uptake in cancer cells can predict malignancy and drug resistance using machine learning.

Tumor heterogeneity is a primary factor that contributes to treatment failure. Predictive tools, capable of classifying cancer cells based on their functions, may substantially enhance therapy and extend patient life span. The connection between cell biomechanics and cancer cell functions is used here to classify cells through mechanical measurements, via particle uptake. Machine learning (ML) was used to classify cells based on single-cell patterns of uptake of particles with diverse sizes. Three pairs of human cancer cell subpopulations, varied in their level of drug resistance or malignancy, were studied. Cells were allowed to interact with fluorescently labeled polystyrene particles ranging in size from 0.04 to 3.36 µm and analyzed for their uptake patterns using flow cytometry. ML algorithms accurately classified cancer cell subtypes with accuracy rates exceeding 95%. The uptake data were especially advantageous for morphologically similar cell subpopulations. Moreover, the uptake data were found to serve as a form of "normalization" that could reduce variation in repeated experiments.

Toxinome-the bacterial protein toxin database.

IMPORTANCE: Microbes use protein toxins as important tools to attack neighboring cells, microbial or eukaryotic, and for self-killing when attacked by viruses. These toxins work through different mechanisms to inhibit cell growth or kill cells. Microbes also use antitoxin proteins to neutralize the toxin activities. Here, we developed a comprehensive database called Toxinome of nearly two million toxins and antitoxins that are encoded in 59,475 bacterial genomes. We described the distribution of bacterial toxins and identified that they are depleted by bacteria that live in hot and cold temperatures. We found 5,161 cases in which toxins and antitoxins are densely clustered in bacterial genomes and termed these areas "Toxin Islands." The Toxinome database is a useful resource for anyone interested in toxin biology and evolution, and it can guide the discovery of new toxins.

DNA Methylation-Based Assessment of Cell Composition in Human Pancreas and Islets.

Assessment of pancreas cell type composition is crucial to the understanding of the genesis of diabetes. Current approaches use immunodetection of protein markers, for example, insulin as a marker of ß-cells. A major limitation of these methods is that protein content varies in physiological and pathological conditions, complicating the extrapolation to actual cell number. Here, we demonstrate the use of cell type-specific DNA methylation markers for determining the fraction of specific cell types in human islet and pancreas specimens. We identified genomic loci that are uniquely demethylated in specific pancreatic cell types and applied targeted PCR to assess the methylation status of these loci in tissue samples, enabling inference of cell type composition. In islet preparations, normalization of insulin secretion to ß-cell DNA revealed similar ß-cell function in pre-type 1 diabetes (T1D), T1D, and type 2 diabetes (T2D), which was significantly lower than in donors without diabetes. In histological pancreas specimens from recent-onset T1D, this assay showed ß-cell fraction within the normal range, suggesting a significant contribution of ß-cell dysfunction. In T2D pancreata, we observed increased α-cell fraction and normal ß-cell fraction. Methylation-based analysis provides an accurate molecular alternative to immune detection of cell types in the human pancreas, with utility in the interpretation of insulin secretion assays and the assessment of pancreas cell composition in health and disease.

Roles for H2A.Z and its acetylation in GAL1 transcription and gene induction, but not GAL1-transcriptional memory.

H2A.Z is a histone H2A variant conserved from yeast to humans, and is found at 63% of promoters in Saccharomyces cerevisiae. This pattern of localization suggests that H2A.Z is somehow important for gene expression or regulation. H2A.Z can be acetylated at up to four lysine residues on its amino-terminal tail, and acetylated-H2A.Z is enriched in chromatin containing promoters of active genes. We investigated whether H2A.Z's role in GAL1 gene regulation and gene expression depends on H2A.Z acetylation. Our findings suggested that H2A.Z functioned both in gene regulation and in gene expression and that only its role in gene regulation depended upon its acetylation. Our findings provided an alternate explanation for results that were previously interpreted as evidence that H2A.Z plays a role in GAL1 transcriptional memory. Additionally, our findings provided new insights into the phenotypes of htz1Delta mutants: in the absence of H2A.Z, the SWR1 complex, which deposits H2A.Z into chromatin, was deleterious to the cell, and many of the phenotypes of cells lacking H2A.Z were due to the SWR1 complex's activity rather than to the absence of H2A.Z per se. These results highlight the need to reevaluate all studies on the phenotypes of cells lacking H2A.Z.

Predicting gene knockout effects from expression data.

BACKGROUND: The study of gene essentiality, which measures the importance of a gene for cell division and survival, is used for the identification of cancer drug targets and understanding of tissue-specific manifestation of genetic conditions. In this work, we analyze essentiality and gene expression data from over 900 cancer lines from the DepMap project to create predictive models of gene essentiality. METHODS: We developed machine learning algorithms to identify those genes whose essentiality levels are explained by the expression of a small set of "modifier genes". To identify these gene sets, we developed an ensemble of statistical tests capturing linear and non-linear dependencies. We trained several regression models predicting the essentiality of each target gene, and used an automated model selection procedure to identify the optimal model and hyperparameters. Overall, we examined linear models, gradient boosted trees, Gaussian process regression models, and deep learning networks. RESULTS: We identified nearly 3000 genes for which we accurately predict essentiality using gene expression data of a small set of modifier genes. We show that both in the number of genes we successfully make predictions for, as well as in the prediction accuracy, our model outperforms current state-of-the-art works. CONCLUSIONS: Our modeling framework avoids overfitting by identifying the small set of modifier genes, which are of clinical and genetic importance, and ignores the expression of noisy and irrelevant genes. Doing so improves the accuracy of essentiality prediction in various conditions and provides interpretable models. Overall, we present an accurate computational approach, as well as interpretable modeling of essentiality in a wide range of cellular conditions, thus contributing to a better understanding of the molecular mechanisms that govern tissue-specific effects of genetic disease and cancer.