Skip to main content
European Commission logo print header

Epigenetics towards systems biology

Final Report Summary - EPIGENESYS (Epigenetics towards systems biology)

Executive Summary:
The “EpiGeneSys” Network of Excellence has strived to build a bridge between two areas of European excellence, epigenetics and systems biology and has established a framework to catalyse interdisciplinary exchanges and training, as well as to foster the sharing of tools, resources and knowledge. The common objective has been to address fundamental epigenetic mechanisms in quantitative terms both spatially and temporally, with the ultimate of expressing the underlying dynamic events in mathematical terms in order to model and predict how the balance between maintenance and erasure of epigenetic information varies in specific developmental contexts under normal or pathological conditions. The four scientific WPs have had the common goal of bringing fundamental questions in epigenetics towards systems level analysis.

1. The Dynamics of Epigenetic Regulators
EpiGeneSys has contributed to gaining a quantitative understanding of the molecular driving forces that govern robustness and sensitivity of the binding of epigenetic regulators to chromatin and which thus are the key points for regulation during normal cellular function and for deregulation upon disease.
2. Linking Genotype to Epigenotype
EpiGeneSys has helped to elucidate the crosstalk between genome and epigenome by applying rigorous high-throughput systems biology approaches for several epigenetic variables in selected model organisms and populations of known sequence diversity.
3. Signalling to the Epigenome
EpiGeneSys has contributed to determining how the environment, stress, metabolic status and growth factors signal to the epigenome to induce new programmes of gene expression that may either be transient or lead to long-term heritable phenotypes.
4. An Integrated Computational Epigenetics Framework
EpiGeneSys has helped to facilitate better integration between systems biology tools and techniques and epigenetic research through a dedicated work package.

As a cross-cutting effort, the understanding of epigenetic inheritance was advanced through the study of stability/plasticity of the various epistates at all levels. Through its more than 350 publications, many in leading high impact factor journals, and the many meetings and workshops organised by the network, EpiGeneSys research has reached a wide scientific audience. The RISE1 scheme, which integrated 20 junior group leaders into the network, aided the integration of epigenetics and systems biology through its multi-targeted approach. The EpiGeneSys training programme has benefited around 850 participants and laid the ground for research combining epigenetics and systems biology to keep on prospering in the future. The network has extensively collaborated with other European and international initiatives, thereby ensuring the dissemination of a new culture joining epigenetics and systems biology across Europe. The network has promoted new scientific actions by providing stimuli to members to apply to project calls and has already planned an EpiGeneSys meeting for 2017 to ensure the long term impact of its activities.
EpiGeneSys has engaged in a range of communication activities aimed at the wider public, involving links with music, design, visual or performing arts. These initiatives have focused on explaining basic epigenetic mechanism and have shown how the investigation of epigenetic mechanisms supported by interdisciplinary approaches provides novel insights into health and disease and offers hope for new treatments.
These cohesive efforts have significantly contributed to structuring the European Research Area (ERA) in both epigenetics and systems biology and to advancing our understanding of epigenetic systems. The success of EpiGeneSys demonstrates the effectiveness such large, integrative projects and calls for a continuation of such vital contribution to the sustainability of the ERA through new instruments and continued European funding.

Project Context and Objectives:
1. Science and Technology Objectives

1.1 Context

The “EpiGeneSys” Network of Excellence has strived to build a bridge between two areas of European excellence, namely epigenetics and systems biology. To organise and structure “EpiGeneSys” research rationally, the network has established a framework to catalyse interdisciplinary exchanges and training, and to enable sharing of tools, resources and knowledge. The aim has been to create an essential basis for enabling European epigenetics research to enter the arena of systems biology.
Epigenetics research has evolved exponentially in terms of its importance and impact over recent years. This is partly because epigenetics has emerged as a main driver to our understanding of how genes respond to environmental factors across lifespan. The influence of the environment or nutrition on epigenetic processes can affect human health, for many human diseases, ranging from cancer, diabetes, to neuropsychiatric and infectious diseases, epigenetics offers the hope for new interventions. The reversible nature of epigenetic traits makes them an attractive target for future actions in prediction of disease and precision medicine. Epigenetics will therefore be a key pillar for the improvement of treatments, health outcomes and the promotion of healthy aging. Thus, it is becoming critical to assess how exactly Epigenetics can so impact on many complex biological processes relevant to human health.
Epigenetics research includes the study of how patterns of gene expression defined during development can be maintained and passed from one cell to its descendants; how gene expression changes during cellular differentiation; and how environmental factors can alter, sometimes even over multiple generations, the way that genes are expressed. European epigenetics research has made enormous progress in the last years, revealing several key parameters that provide new insights into the regulation of genome function. These parameters include chromatin modifications such as DNA methylation, histone modifications, nucleosome positioning, the binding of other chromatin proteins as well as non-coding RNAs. New technologies are providing a degree of resolution at several levels (genomic, spatial and temporal) never previously achieved. Tools for dissecting the spatio-temporal dynamics of gene expression and molecular genetics tools for interfering with different parameters, provide powerful new insights into the kinetics and epigenetic mechanisms underlying development, differentiation and heritability. The wealth of data that is being produced at these different levels has to be analysed and synthesized. It is clear that integration of all these parameters to explain, model and predict how these epigenomes are formed during development, how they dynamically react to the environment, and defining which parameters are critical for heritability or reversibility of gene expression states, is a major challenge as well as a potential bottleneck.
The need to translate epigenetics questions into a systems biology approach becomes even more evident when one considers their common fundamental properties. A fundamental and fascinating characteristic of all epigenetic systems is the balance between stability and flexibility: epigenetic mechanisms ensure robust maintenance of gene expression states but also allow changes following developmental, metabolic, or environmental cues. To understand this balance between stability and flexibility, and how it can create variability, it is essential to model the dynamic nature of the system.
Reciprocally, systems biology by being able to access high quality quantitative data in epigenetics also benefits from new opportunities to develop creative research potentially opening a whole range of applications into the many aspects of biology onto which epigenetics impacts.

1.2 The main objective - Integration of epigenetics and systems biology

The EpiGeneSys Network of Excellence has aimed to bring together the two thriving fields of epigenetics and systems biology within the European Research Area (ERA). The network has set out to gather multidisciplinary expertise to define an original research portfolio and provide platforms for exchange of resources and knowledge. It has placed importance on providing interdisciplinary training and a competitive career development platform for a future generation of researchers in this novel area of European research. A thematically related objective has been to ensure dissemination of results and long term integration of epigenetic research in the area of systems biology by establishing links between European scientific communities of epigenetics and systems biology, and by connecting with other European and worldwide systems biology and epigenetics efforts, that will endure way beyond the running time of the project.
To achieve its goals, the EpiGeneSys has sought to combine the efforts of the consortium, whose members have expertise in multiple disciplines, (development, molecular biology, genomics, proteomics, biochemistry, structural biology, computational biology and mathematics) and attested excellence in epigenetics research in a variety of complementary model systems (yeast, Drosophila, mouse, Xenopus, plants, and human cells) including human diseases (cancer, genomic imprinting disorders).
The four scientific WPs have had the common goal of bringing a fundamental question in epigenetics towards systems level analysis. As a further integrative component, EpiGeneSys has aimed to advance our understanding of epigenetic inheritance, i.e. by assessing stability versus plasticity during the cell-cycle, throughout multiple divisions and even several generations.

1.3 Scientific Objectives

WP2 Dynamics of Epigenetic Regulators
This WP has been founded on the principle that the quantitative study of epigenetic systems at the single molecule and single cell level, combined with computational modelling of these systems in terms of dynamic chromatin binding behaviour of several interacting components, would bring invaluable insights into the inherent stability and plasticity of these systems.
The goal of this work package was to understand how the dynamic binding of epigenetic regulators to chromatin enables both stability and flexibility of chromatin states. As model epigenetic systems, the WP considered heterochromatin states involved in gene silencing, dosage compensation and Polycomb/Trithorax regulation as well as compare the protein composition of pericentric heterochromatin in diverse cell types and in diverse mutant backgrounds.
How do these epigenetic systems ensure both stability and plasticity? How can they show at the same time a robust epigenetic memory and be flexible so as to adapt and respond to various changes: during the cell cycle, during development and other environmental stress?

WP 3 Linking Genotype to Epigenotype
Epigenetic modifications in chromatin provide a regulatory layer that modulates genome function in a reversible fashion. The main effector mechanisms are nucleosome positioning, histone modifications and DNA methylation. Recent genome-wide approaches have shown these mechanisms to be highly specific with respect to chromosomal regions and developmental time-points and to be dependent upon inter and intra species sequence variation.
The WP has aimed to decipher the crosstalk between DNA sequence and epigenomic landscapes, from the chemical modification of DNA, to the positions of nucleosomes and the recruitment of epigenetic machineries. It has strived to elucidate this crosstalk by applying rigorous high-throughput systems biology approaches to several epigenetic variables in selected model organisms and populations of known sequence diversity. The generated datasets and tools have been utilized to model genome-wide epigenetic states and regulatory networks, and to predict regulatory interactions between genotype and epigenotype. The objective has been to validate and characterise these interactions through experimental testing using synthetic sequences in selected model systems.

WP4: Signalling to the Epigenome
The goal of this work package has been to understand how the environment, nutrients and metabolism, growth factors, cytokines and developmental forces shape the epigenome through signalling pathways from the cell surface to DNA organised into chromatin. Cells respond continually to their environment and to their nutritional state by inducing new programmes of gene expression, which in turn modulates cell behaviour. Tight and precise, but dynamic regulation of these processes is essential both for correct embryonic development and for the health and viability of the adult organism. Positive and negative regulation of transcription requires the remodelling of chromatin via histone and DNA modifications, the alteration of nucleosomal positioning and the modulation of higher order chromatin structure. The systems biology goal has been to quantify components of signalling pathways from cell surface molecules to chromatin components, and to build mathematical models representing the signalling events, to understand their interplay with epigenetic gene regulation in normal and perturbed situations. The approach has aimed to integrate whole genome approaches with quantitative measurements at the single cell and population levels to ultimately generate both kinetic and network models describing how environmental stimuli and metabolic status influence the epigenome, how cell lineages are specified in early vertebrate development, and how short term signals induce long term heritable changes in gene expression.

WP5: An Integrated Computational Epigenetics Framework
The purpose of this work package has been to transform, integrate and amplify the efforts of systems biologists and epigeneticists both within the network and outside it. The objective has been to promote efficient data integration and foster productive collaborations among partners, with the mission of actively engaging the systems and computational biology communities by articulating epigenetic challenges and promoting their research.

To understand epigenetic inheritance, throughout all WPs, a focus on robust epigenetic systems such as DNA methylation, X chromosome inactivation, and heterochromatin has pursued the goal of investigating at all levels the issue of stability/plasticity of the various epistates. How they are assembled, how they are maintained through replication, mitosis and meiosis, and how and to what extent they are erased in germ line cells. The systems biology goal has been first to characterise and quantify the components of each step in the assembly, replication and disassembly of these systems, and secondly to examine the relative timing of these events during development and during the cell cycle or across generations. The long term objective is to express these dynamic events in mathematical terms to understand the balance between maintenance and erasure of epigenetic information in specific developmental contexts under normal or pathological conditions, including epigenetic diseases related to methylation, imprinting and various chromatin/ nuclear organisation disorders.

1.4 Non-scientific objectives

To reach the objective of integrating epigenetics and systems biology EpiGeneSys designed the RISE1 programme Research Integrating Systems Biology & Epigenetics), the competitive recruitment and integration of young investigators working at the interface of biological, computational, and/or engineering sciences with a focus on applying tools of systems biology. This action has had the goal of securing a dynamic expansion of the NoE in terms of fostering the convergence of the two fields, integrating new concepts and technologies as well as reaching out to other related scientific communities in Europe. An associated non-beneficiary membership has sought to support this main objective by widening the scope of the network to the community beyond the consortium.
The objective of generating a durable combined approach shared between both the epigenetic and systems biology communities was to be critically strengthened through the integration, exchange and further training of postdoctoral researchers and PhD students as well as by organising open international conferences and workshops on linking epigenetic research and systems biology to stimulate and promote European research of a world-class quality. The aim has been to provide a forum that allows EpiGeneSys issues and progress to be discussed and reviewed with all members of the network. The network has aspired to establish, curate and maintain a website for knowledge transfer and information exchange within the scientific community, which serves as an information resource about systems biology and epigenetics and links researchers in epigenetics and systems biology for exchange about experimental approaches, strategies or problems.
EpiGeneSys also has sought to communicate epigenetics and systems biology in an accessible and interesting fashion to the wider public through various media, such as animations, videos, print media, with a special focus on making use of arts and humanities to communicate science to demonstrate the importance of science to people’s lives.
A long-term objective of EpiGeneSys has been to pave the way for a continued initiative that represents, channels and sustains visibly a long-term consolidation of the scientists and scientific knowledge related to the new discipline integrating epigenetics and systems biology. In particular, EpiGeneSys has strived to set a fertile ground for novel research directions as well as facilitate the interchange of information and ideas among researchers. An important goal has been to ensure the continuation of the communication and dissemination web platforms that have linked up with the wider scientific and general community and thereby have durably become the main virtual contact point and web interface in the field of epigenetics and systems biology. EpiGeneSys has aimed to secure a lasting integration platform for junior researchers (postdoctoral researchers, PhD and Master students) that exposes them to different approaches and applications combining epigenetics and systems biology. It has also been an objective to facilitate the development of long-term collaborations on project proposals and publications as well as of dissemination activities and to co-ordinate the organisation of Europe-wide workshops, conferences and summer schools dedicated to current or new scientific challenges in the two fields.

Project Results:
2. Science and Technology

2.1 WP2 Dynamics of Epigenetic Regulators.

Epigenetic systems comprise multiprotein complexes, many of which interact with chromatin in a dynamic manner, in rapid exchange between bound and free pools. Epigenetic regulators can be anchored to chromatin via many interaction partners, including other proteins, modified histone tails, metabolites, DNA and non-coding RNA. The strength and nature of these interactions changes dynamically during replication, mitosis, and upon developmental transitions. Many individual interactions have been genetically and biochemically determined in exquisite detail. Although some of these observations have been quantified, a large gap in our understanding of the dynamic nature of these complex systems has been the lack of comprehensive quantitative in vivo measurements of chromatin binding in real time. Such knowledge is essential to enable mathematical models describing the interplay of these interactions, and to inform experiments in which specific interactions are perturbed. The aim of this work package was to gain a quantitative understanding of the molecular driving forces that govern robustness and sensitivity of the binding of epigenetic regulators to chromatin and which thus are the key points for regulation during normal cellular function and for deregulation upon disease.

We focused on three main questions:
1) Single interactions. What is the molecular structure and affinity of the interaction between a given regulator and its binding partner in vitro? What is the affinity and residence time of a given regulator binding to chromatin in vivo?
2) Interaction maps. Which interactions determine the binding of a given epigenetic regulator to chromatin in interphase cells? To what extent, in quantitative terms, are these interactions cooperative or competitive?
3) Dynamic changes. How does the chromatin binding behaviour of a given regulator change during replication and mitosis, in terms of quantity bound, residence time and quantitative extent of interaction with other components? How do these parameters change upon developmental cell fate transitions? How do they change upon challenges to the system such as DNA damage?
Below we summarize the major achievements separately for individual labs within EpiGeneSys.

Robin Allshire, University of Edinburgh, UK
The Allshire group successfully developed protocols allowing chromatin regions of interest to be affinity selected from cells and their composition, both proteins and histone post-translation modifications, to be quantified in an unbiased manner using next generation mass spectrometry. In addition we identified the Eic1 protein as being required to connect with the main kinetochore complex (CCAN) and allow the temporal recruitment of the key CENP-A loading factors Mis18 and HJURPScm3. Related to this, we solved the structure of the Mis18 YIPPEE-like domain. The resulting structure suggests testable models with respect to Mis18 function in CENP-A loading.
Furthermore, we demonstrated that sequence-encoded features of fission yeast centromeric DNA create an environment of pervasive low quality RNAPII transcription that is an important determinant of CENP-A assembly. Our observations and resulting model emphasizes roles for both genetic and epigenetic processes in centromere specification. adapted the Recombination Induced Tag Exchange (RITE) system to follow histone turnover in fission yeast. Our analysis contributed to resulting models concerning nucleosome turnover on transcription units. Finally, we developed a system to generate regions of conditional synthetic heterochromatin at specific loci in fission yeast. Using this synthetic system, we demonstrated the important principle that the lysine 9 methylation histone mark can carry information through cell division, and even through meiosis into subsequent generations, independently of DNA sequence. Our findings support a model where histone post-translational modifications on nucleosomes can act as heritable epigenetic marks.

Geneviève Almouzni, CNRS, Institut Curie, Paris, France
The general objective of the Almouzni lab was to characterize histone chaperone and variant dynamics through the cell cycle and at defined chromatin landmarks. The lab assessed these dynamic processes from nucleosome assembly to the establishment of chromatin domains bridging different scales and contexts.
Taken together, the EpiGeneSys network enabled the Almouzni lab to define how distinct histone chaperones escort variants through the cell cycle and deposit them in a defined spatial and temporal manner. Importantly, the lab identified the importance of dosage in the chromatin regulator network and the biological consequences of an imbalance. This leads to a new way of thinking about chromatin dynamics, whereby the network of chaperones and variants is dynamic and participates in extensive crosstalk. This enables cells to adapt to various needs during the cell cycle and its lifetime. How these dynamic changes and/or rewiring are controlled will be the next challenge.

Elizabeth Bayne, University of Edinburgh, UK
The Bayne lab has dissected the molecular structure of the interactions of Stc1, a key factor coupling RNAi to chromatin modification in fission yeast. Having determined the structure of Stc1 by NMR, we went on to show that Stc1 binds the RNAi effector Ago1 via its conserved N-terminal tandem zinc finger domain and binds the chromatin modifying complex CLRC via its disordered C-terminal domain, thereby elucidating the molecular mechanism by which RNAi and chromatin modification are connected. In addition, through genetic approaches we identified several novel factors required to facilitate heterochromatin formation in fission yeast (including the COP9 signalosome and novel splicing-related factors), as well as factors involved in antagonising the formation of heterochromatin (including the PAF complex component Leo1), allowing us to further refine the regulatory network governing heterochromatin assembly in S. pombe.

Sigurd Braun, Ludwig Maximilians-Universität, München, Germany
Through systematic genetic screens and functional genomics, the Braun lab has identified a large number of mutants with defects in silencing for all major heterochromatin domains in the fission yeast S. pombe. This collection comprises many novel candidates and reveals distinct requirements for different factors and pathways at different heterochromatin domains (Barrales et al., manuscript in preparation). To understand the function of these novel candidates, we focused on the inner nuclear membrane protein Lem2, which is a homolog of lamin-associated proteins belonging to the conserved LEM domain family. We demonstrated that Lem2 controls heterochromatin silencing and its recruitment to the nuclear periphery. However, quite unexpectedly, we found that chromatin binding and heterochromatin silencing are mediated independently through the two conserved domains of Lem2, the LEM and MSC domains, respectively. Through advancing the synthetic gene array (SGA) technology to analyse heterochromatin-specific genetic interactions, we further showed that Lem2 acts redundantly in silencing and cooperates with multiple pathways (including the RNAi machinery). All these pathways have in common that they originate from the nuclear envelope, reinforcing the concept of the nuclear periphery as a specialized compartment for gene repression. We employed this functional genomics approach to study the functions of other factors in heterochromatin formation: In collaboration with the Bayne lab, we uncovered functional redundancy and epistatic interactions of the PAF complex in preventing the spreading of heterochromatin. We further showed that the PWWP protein Pdp3, a subunit of the histone acetyltransferase Mst2 complex, prevents Mst2 from interfering with heterochromatin, possibly through anchoring the HAT to euchromatin. Finally, to study the dynamics of heterochromatin, we developed with the Al-Sady lab (UCSF) a fluorescence-based silencing reporter assay in S. pombe, which allows the simultaneous detection of multiple reporter genes at the single-cell level by flow cytometry, providing a cutting-edge level of precision in monitoring silencing in vivo.

Stephen Cusack, EMBL, Grenoble, France
Our main contribution towards objectives of WP was to biophysically and structurally analyse interactions of several epigenetic regulators. Our work focused on three main regulators: the MSL complex, the NSL complex (both in collaboration with the Akhtar group) and the PRDM9 histone methyltransferase (in collaboration with the deMassey group).
The MSL complex is a key player in dosage compensation in Drosophila. We determined crystal structures of the complex between the predicted coiled-coil region of human MSL1 and the N-terminal portion of MSL2. Unexpectedly, this structure revealed that these two proteins form a hetero-tetramer rather than a dimer as previously thought. Two MSL1 subunits form a dimeric coiled-coil, which then serves as a binding platform for two molecules of MSL2. This data thus suggested that the entire MSL complex functions as dimeric, likely containing all subunits in two copies. Using structure-based MSL1 mutants we were able to show that dimerization of the entire MSL complex via MSL1 is required for dosage compensation in flies. Together with our previous work we have now determined structures of all defined sub-complexes of the MSL complex.
The NSL complex is a general transcription regulator. We identified interactions of WDR5 with KANSL1 and KANSL2 subunits, mapped interacting domains and determined crystal structure of the minimal complex which revealed how WDR5 is recruited into the NSL complex via conserved linear motifs of KANSL1 and KANSL2. Using structure-based KANSL1 mutants in transgenic flies we could show that the KANSL1-WDR5 interaction is required for proper assembly, efficient recruitment of the NSL complex to target promoters and for fly viability. Our data also showed that the interactions of WDR5 within the MOF- containing NSL complex and MLL/COMPASS histone methyltransferase complexes are mutually exclusive and that WDR5 cannot be a shared subunit linking these two complexes, as previously believed.
PRDM9 is a histone methyltransferase and a key determinant of the localisation of meiotic recombination hotspots in humans and mice. We determined the structure of its catalytic PR-SET domain, which is the first structure of a PR-SET domain of any member of the PRDM family bound to a histone substrate and S-adenosyl-L-homocysteine co-factor. The structure clarified the catalytic mechanism of these proteins. Together with the apo-structure our work also revealed a unique mechanism of regulation of SET protein activity by an unprecedented rearrangement of the substrate and cofactor binding sites by a concerted action of the pre-SET and post-SET domains

Jerôme Déjardin, CNRS, IGH Montpellier, France
The Déjardin lab has successfully developed a quantitative means to analyse changes in protein composition at a given locus that is called quantitative Proteomics of Isolated Chromatin segments (qPICh). This technological development was aimed at evaluating the function of major epigenetic marks at pericentromeric regions, a major locus for heterochromatin formation in mammals. The biological functions of three critical epigenetic pathways were evaluated: DNA methylation, H3K9me3 and H4K20me3. In a very fruitful collaboration with the Imhof and Torres-Padilla labs, we also addressed quantitative changes in histone modifications at these loci, to uncover potential crosstalk between distinct epigenetic pathways. We find that DNA methylation, not H3K9me3, controls the epigenetic status of pericentromeric regions and identify compensatory mechanisms leading to Polycomb recruitment at this locus. In particular, we find new proteins able to sense DNA methylation levels and able to recruit PcG factors to chromatin. In particular, we also uncovered that a protein named BEND3 is important to target PRC2 to heterochromatin in the absence of DNA methylation, a condition which occurs during development and also in cancer cells. We believe we have identified an important pathway for PcG recruitment during development and in cancer.
The Déjardin lab also developed a novel locus specific purification approach named ePIC (end targeted PICh) in which chromatin fragments harbouring complex sequences can be purified. The lab used ePICh to purify the rDNA promoter chromatin from a mouse cancer cell line and identified a novel transcription factor, ZFP106, that is important to recruit the RNA polymerase I machinery to the rDNA genes.

Axel Imhof, Ludwig Maximilians-Universität, München, Germany
Epigenetic information is transmitted by a complex interaction of chromatin associated proteins, histone modifications and the underlying DNA. The ultimate goal of the EpiGeneSys network was to unravel the intrinsic mechanisms that mediate these processes allowing us to potentially modulate them. During the five years of EpiGeneSys, we could uncover a set of heterochromatin associated proteins that play fundamental roles in the separation of biological species. Our findings allowed a better understanding of the evolutionary forces that result in hybrid incompatibility at a molecular level and will hopefully result in a much better understanding of why and how species form from a common ancestor.
In addition, we managed to perform highly quantitative studies of histone modifications thereby discovering a major role of the cell cycle timing in maintaining epigenetic memories. The systematic investigation of quantitative changes of histone modifications upon removing individual histone acetyltransferases not only facilitated the generation of a mathematical model to describe the in vivo network of chromatin modulating enzymes but also uncovered an unexpected connection between histone acetyltransferases and the life span of a multicellular organism.
Finally, in a collaborative effort with other members of the EpiGeneSys network we established a technology that allowed us to identify, quantify and describe the molecular works of the epigenome in its entirety at a particular region of the genome. This technology enabled us to make a substantial step forward in our understanding of the way we transmit information imprinted on our cells by a constantly changing environment.

Andreas Ladurner, Ludwig Maximilians-Universität, München, Germany
The general objective of the Ladurner laboratory was to characterize at the molecular and mechanistic level how histone chaperone proteins interact with their substrates. The lab successfully determined for the first time the high resolution crystal structure of several domains within the essential histone chaperone FACT, including a domain that contributes to the binding of histones H2A-H2B, and a domain that interacts with the FACT subunit Pob3. This research was complemented by extensive biochemical validation and mutagenesis assays. In addition, the laboratory identified the allosteric mechanism through which the ATP- and poly-ADP-ribose-(PAR) dependent chromatin remodeller Alc1/ChdL1, a human oncogene, is activated. Biochemical, real-time imaging, and biophysical assays including H/D-MS exchange analysis reveal that the two globular domains of Alc1, the Snf2 ATPase domain and the PAR-binding macrodomain, interact with each other or with high affinity in the absence of PAR ligand. Upon binding of at least a dimer or trimer of ADP-ribose (which binds with low nanomolar affinity), a conformational change in the macrodomain leads to the dissociation from the Snf2 domain. These experiments, together with additional integrative structural approaches such as NMR spectroscopy, reveal the first experimental evidence for the allosteric regulation of an ATP-dependent chromatin remodeller by a DNA damage- and stress-activated signalling molecule such as PAR.

Manolis Papamichos-Chronakis, INSERM, Paris, France
Genomic stability is essential for organismal survival and prevention of devastating diseases. Chromosomal DNA is tightly packaged and organized into the chromatin nucleoprotein structure. Chromatin and chromatin-regulatory factors, such as histone post-translational modifications, histone variants and ATP-dependent chromatin-remodelling enzymes, control the spatiotemporal access of DNA regulators to the genome and protect the integrity of the genome. Although biochemical roles have been assigned to many of these factors, both the context in which these activities promote maintenance of genome stability and their biological functions are far from being understood. Our aim is to understand how chromatin is integrated into the genome stability network and to decipher the interconnections between chromatin regulators and genome stability factors. Using budding yeast as a model system we approach our question with a combination of yeast genetics, biochemistry and high-throughput technologies. Our primary research investigates the functional role of the following two evolutionary conserved chromatin factors, the histone variant H2A.Z and the ATP-dependent chromatin remodelling enzyme INO80 in maintenance of genome stability. First, using genetic analysis we revealed an unanticipated functional link of INO80 with the DNA damage and replication-induced proteolysis of RNA Polymerase II by the ubiquitin proteasome system. Second, functional genomics analysis revealed that regulation of H2A.Z is at the interface of transcription with DNA replication and repair. In conclusion, our results suggest a novel regulatory role for chromatin in maintenance of genomic integrity.

Leonie Ringrose, IMBA, Vienna, Austria
We developed an “in- vivo biochemistry” approach, allowing the absolute quantification of molecule numbers, mobilities and residence times for members of the Polycomb/Trithorax group of epigenetic regulators in living Drosophila during developmental and cell cycle transitions. This enables quantitative modelling using absolute concentrations and measured kinetic parameters, and delivered insights into dynamic changes in the kinetic properties of the system upon developmental and cell cycle transitions. Experimental perturbation revealed molecular mechanisms by which chromatin attachment and dissociation are regulated, and thus that affect the stability of epigenetic memory. Later work via collaborations with theoretical groups within the network (Martin Howard) and beyond (Kim Sneppen, Ian Dodd) focused on stochastic modelling of events at single loci, informed by the above analyses. This analysis revealed a rich repertoire of system properties beyond simple memory, and has broad implications for our understanding of the genome-wide function of the PcG and TrxG proteins on their several hundred chromatin targets. Experimental testing of model predictions is currently underway.

Maria Elena Torres-Padilla, IGBMC, Strasbourg, France
The Torres-Padilla lab successfully generated a mathematical epigenetic landscape of mammalian pre-implantation development using the mouse embryo as a model system. Using robust single cell expression data, they showed that chromatin modifiers define early developmental transitions better than at later stages, when lineages can be resolved based on the expression of specific transcription and signalling factors. This work introduced for the first time mathematical and computational methods based on robust quantitative data on gene expression from single cells in vivo. Moreover, the data generated paved the ground for generating new models for cell fate transitions during early development.

Juanma Vaquerizas, MPI for Molecular Biomedicine, Münster, Germany
Plasmodium falciparum is a eukaryotic pathogen causing the most malignant for of malaria in humans. Its ability to express different virulence genes in order to avoid detection by the host makes it especially dangerous. Recent analyses of the genomic organisation of P. falciparum revealed a highly dynamic nuclear architecture, and a strong association between DNA-DNA contacts and gene expression. Most notably, virulence genes appear to form spatial clusters with distinct, domain-like structures. Despite these general architectural findings, it is currently unknown how the establishment and loss of specific DNA-DNA contacts between genomic elements during the P. falciparum infection cycle contributes to different virulence phenomena.
The Vaquerizas group has focused on studying the genomic organization of P. falciparum from a complex network perspective. Specifically, they constructed differential networks of contact changes between infection stages, permitting the time-resolved analysis of individual genomic contacts. This framework allowed them to investigate how network properties and contact changes between specific genomic elements are linked to important gene regulatory mechanisms, including virulence gene selection, activation, and silencing. A first major observation is the dramatic change in the amount of specific DNA-DNA contacts among virulence genes during the course of the infection cycle, prompting a more detailed analysis of the association between virulence and the 3D organization of the P. falciparum genome.
In parallel, in collaboration with the Torres-Padilla laboratory (Helmholtz Munich, Germany), the Vaquerizas group has studied the gene expression pattern of a specific family of repetitive elements in the mouse genome, namely MERVLs. In a paper that published last year, they showed that removal of the replication-dependent histone chaperone complex CAF-1, results in a remarkable increase of the expression of 2-cell-specific markers, and the gain of chromatin marks and characteristics associated with totipotent-like states. In particular, their analysis of RNA-seq data for protein-coding genes and repetitive elements showed that cells with a depletion of CAF-1 levels significantly up-regulate gene expression programmes and repetitive elements exclusive for 2-cell embryos. The Vaquerizas laboratory was also able to demonstrate that there is a specific association between gene expression and the up-regulation of close-by MERVL elements, with a marked enrichment within a 40 kilobase window surrounding MERVL repetitive elements.

Michiel Vermeulen, Radboud University Nijmegen, Netherlands
Our lab contributed to the EpiGeneSys project with an inventory of ‘readers’ for methylated, hydroxymethyled and non-methylated DNA in several cell types and species. We also developed several workflows facilitating the identification of protein-protein and protein-DNA interactions using quantitative mass spectrometry-based proteomics.

2.2 WP3 Linking Genotype to Epigenotype

Epigenetic modifications in chromatin provide a regulatory layer that modulates genome function in a reversible fashion. The main effector mechanisms are nucleosome positioning, histone modifications and DNA methylation but their effect on gene regulation is poorly understood. Recent genome-wide approaches have shown these mechanisms to be highly specific with respect to chromosomal regions and developmental time-points and to be dependent upon inter and intra species sequence variation. The goal of WP3 was to elucidate the crosstalk between genome and epigenome by applying rigorous high-throughput systems biology approaches for several epigenetic variables in selected model organisms and populations of known sequence diversity. The generated datasets and tools were utilized to model genome-wide epigenetic states and regulatory networks, and to predict regulatory interactions between genotype and epigenotype. Predictions were validated and characterized through experimental testing using synthetic sequences in selected model systems.
Our work program had three major pillars and in all of these the member labs of WP3 achieved milestones and deliverables. Indeed, work within WP3 and by its member labs over the course of EpiGeneSys have contributed substantially to a systems level understanding of gene regulation and the epigenome. Member labs published so far 364 publications of which 113 specifically acknowledge network funding. These papers were published in all major journals including Cell, Nature and Science and the results were presented and discussed at international scientific meetings. Several of these publications were accompanied by press releases of the respective institute or University. Thus the generated results have been recognized and disseminated within the scientific community and the general public.
Below we summarize the major achievements separately for individual labs within EpiGeneSys.

Stephan Beck, UCL London, UK
The Beck group contributed to Linking Genotype to Epigenotype in particular by developing novel methods, the identification of disease-specific epigenetic variants, novel epigenetic mechanisms and training/outreach, resulting in 28 publications in total.
Highlights of the methodological work include experimental and computational improvements of array- and sequencing-based technologies for methylome analysis. Implemented under open source R and publicly available via Bioconductor, ChAMP has become one of the most popular analysis pipelines for 450k BeadChips and nanoMeDIP-seq enrichment-based methylome analysis. ChAMP offers a range of normalization and correction methods as well as SNP filtering/flagging, calling of differentially methylated position (DMPs) and regions (DMRs) and detection of copy number variation (CNV). The ability to call CNVs from the same sample used for methylome analysis is particularly important in cancer because of the high levels of genetic heterogeneity observed in many tumours. To meet this requirement, the Beck group developed a method for calling CNVs directly from 450K DNA methylation data. In addition, they also developed a method for detecting 5-hydroxymethylation using 450K arrays. Using saturation analysis, they identified a previously unreported problem (50% DMP loss) in whole-genome bisulfite sequencing (WGBS) at the recommended coverage of 30X and developed an innovative solution, exploiting the comethylation (COMET) structure of methylomes. Through best-fit analysis of COMETs and haplotypes, they identified a strong correlation indicating that this may be a suitable approach to generate the epigenetic equivalent of the HapMap once WGBS data become available on the population level. In summary, these methods demonstrate new ways of linking genotype to epigenotype.
The majority of the disease-specific work was conducted in the context of epigenome-wide association studies (EWAS). To this end, EWAS were conducted for three phenotypes: [a] Wilms tumour, a mutationally ‘quite’ cancer and thus more likely to be driven by epimutations; [b] pain sensitivity, a complex phenotype which reflects the contribution of multiple biological, psychological and environmental risk factors; [c] inflammatory disease, with focus on the Leukotriene B4 receptor (LTB4R) which was identified as the most epigenetically divergent human gene in peripheral blood in comparison with other primates. Using a newly developed analysis pipeline described above, the group furthermore identified a number to of novel candidate loci for targeted reprogramming and or meQTL analysis.
By linking genotype to epigenotype, the Beck group was able to elucidate two novel epigenetic mechanisms. Using comparative methylome analysis across primates, they identified human-specific CpG sites (CpG Beacons) and showed them to be involved in an evolutionary mechanism which transitions genes from facultative to constitutive expression and vice versa. Using meta-analysis of IDH mutant cancers, they identified a hypermethylation phenotype that was enriched for a tissue-specific transcription factor binding site. Biochemical and ChIP analyses showed the transcription factor (EBF1) to interact with TET2 which is lacking the CXXC CpG-binding domain, thus mediating active demethylation in a sequence-dependent manner.
Finally, the Beck group organized two training workshops, which were attended by over 100 researchers from over 20 countries on both occasions and contributed to the development of a participant-centred proposal on epigenome data release.

Giacomo Cavalli, CNRS, IGH Montpellier, France
The Cavalli group has contributed to understanding the DNA logic of Polycomb recruitment and their contribution to chromosomal organization. More specifically thy mapped Polycomb binding sites in various species of Drosophila and, in collaboration with the Tanay group, has found that Polycomb binding sites are astoundingly well conserved in the genomes of fly species as evolutionarily distant as 30 million years. They have identified cases where specific sequence motifs are added or deleted for one species that result in specific gain or losses of Polycomb protein binding at the cognate site. They have produced transgenes carrying these motifs, wt or mutated, to identify critical sequence elements responsible for the evolution of Polycomb distribution at their target genes. They collected data showing that transgenes carrying sequences from other species that are more weakly bound by Polycomb factors behave as weaker PREs in Drosophila melanogaster. Conversely, transgenes carrying regions bound in other species maintain the binding in melanogaster, while the corresponding homologous melanogaster region is not bound because of a lack of sequence elements that have been acquired in other species. The data suggest that small sequence changes can induce or prevent Polycomb recruitment. Moreover, the data also show that 3D organization of recruiter binding sites contributes to Polycomb recruitment. The Cavalli lab has performed Hi-C in the same Drosophila species to correlate 3D contact maps to PcG recruitment. The Cavalli group furthermore used the Hi-C method to map chromatin contacts genome-wide. They have published such maps for Drosophila embryos and showed that the preparation can robustly confirm previously mapped 3C and 4C contacts. They have thus analysed the genome-distribution of contacts and, in collaboration with the Tanay lab, they tested them against mathematical models of genome folding using the simplest possible principles to predict chromosome-wide contact maps. These principles, including the power law regime controlling contact probabilities given linear genomic distance, the adjustment of such linear distances by local scaling factors, the organization of chromosomes into contact domains, and the clustering of such domains into several epigenetically defined classes, allowed the Tanay and Cavalli group to model Hi-C contact maps accurately and to infer from them new architectural and potentially regulatory structures. The Cavalli lab has now extended this work to map chromosome contacts in S2 cultured cells. In addition, they have increased the sequencing depth of the maps in embryos by one order of magnitude. Extending ChIP-Seq analysis of Polycomb recruitment in four different Drosophila species, they have also produced Hi-C maps in the same samples. In order to test the function of Polycomb proteins in setting up the architecture of fly chromosomes, the Cavalli lab produced Hi-C maps in two mutants of the PcG, such as Ph, a member of the PRC1 complex, and E(z), a critical component of the PRC2 complex. The Cavalli lab also collaborated with the Schwartz lab and has produced Hi-C maps of Cp190- as well as CTCF-deficient cells. Their data show that these two proteins have limited effects on the global architecture of the genome. However, their absence does affect specific chromosomal regions that are bound by them. In order to test the role of DNA sequences that can recruit PcG proteins in vivo, the Cavalli lab has produced transgenes that contain wild type and mutated sequences and constructed fly strains carrying these transgenes. The results show that some of the binding sites are necessary and sufficient for recruitment, whereas others are not sufficient on their own. Therefore, these analyses showing support the idea that chromatin binding proteins shape chromosome architecture but chromosome structure feeds back onto function by influencing how proteins can recognize their cognate DNA target sites.

Nicole Sorenzo, Sanger Institute, Cambridge, UK
The Soranzo group has explored at the population level the effects of human genetic variation to the epigenome and gene expression of blood cells and the coordinated effects between these three dimensions. They used blood as a model system of cell differentiation and focused on megakaryocytes, monocytes, neutrophils and T cells, which are key cells in human responses to immunity and inflammation. Overall, this revealed a positive correlation between gene expression and either enhancer/promoter histone modifications or pioneer transcription factors. The correlation between gene expression and DNA methylation is context specific.
The group determined that there is gain in using epigenomic annotations to rank putative causal genetic variants and gain in using multidimensional phenotypes in genetic association analysis. QTL mapping in a large population cohort has delivered novel and less common eQTLs. Furthermore, large population and multiple molecular assays in multiple cell types have afforded to quantify that gene expression variability is largely due to cis genetic variation than the local epigenome, and to identify cell specific regulatory regions that may explain an individual risk to certain diseases.

Marcel Méchali, CNRS, IGH Montpellier, France
The Méchali group have performed the task to identify sequence determinants of DNA replication. This was achieved by developing a powerful method for genome-wide origin mapping by HiSeq and establish a robust bioinformatics procedure for the analysis of genome-wide data. The first important conclusion is that replication origins belong to three different classes, according to their genetic and epigenetic signatures. Moreover, replication origins are flexible, as their usage may vary according to the growth conditions or differentiation programmes. The second important conclusion is the unexpected finding that origin G-rich rich elements, particularly G quadruplex elements are regulatory elements of replication origins. The group has been the first laboratory to describe the Origin G-Rich Elements (OGRE) that can form G4 structures at replication origins. They also discovered that Polycomb marks, which are important elements regulating embryonic development and the organization of the chromosome domains, are up to 90% predictive of replication origins. Another task was to analyse the reprograming of origins during neural differentiation of mouse ES cells. This has been experimentally entirely achieved and the results are under thorough bioinformatic analysis. A third task was to functionally validate the replication origins that were identified. This has also been achieved by introducing endogenous mutations in the origins and creating new origins at ectopic sites. Finally, the Méchali group could design an experimental procedure to be able to map replication origins in an animal, and this was done using C. Elegans embryos. This analysis of metazoan origins represents the most robust and deepest study achieved in metazoan cells so far. It may pave the way to the design of new replicating vectors, as well as to investigate how the deregulation of replication origins may lead to genetic diseases associated to dwarfism, ageing and cancer.

Amos Tanay, Weizmann Institute, Rehovot Israel
The Tanay group work for EpiGeneSys involved a combination of development of computational methodologies and collaboration with members of the network. The group was active in three major domains, A) studies of chromosomal topologies (together with the Cavalli and Hadjur groups, as well as with associate member group Fraser), led to discovery of topological domains and several lines of investigations toward characterizing their mechanistic origins and possible functional roles. B) Models and experiments to understand the dynamics of DNA methylation, promoting a stochastic approach to the role of DNA methylation as a carrier of cellular memory C) single cell genomics and in particular single cell RNA-seq, for defining diverse epigenetic states at the single cell level. Through the network, the Tanay group had the benefit of linking systems biology and epigenetics through many different interactions with members of the epigenetic community. Such interactions involved both providing tools and computational advice to non-computational groups, as well as receiving valuable support with the adaptation and development of experimental strategies to explore epigenetics in a quantitative and systematic way in the lab.

Reini Luco, CNRS, IGH Montpellier, France
The Luco lab has taken advantage of publically available genome-wide data to identify chromatin modifications that differentially mark alternatively spliced events depending on their pattern of splicing. They have shown that there are specific subsets of alternatively spliced cassette exons differentially enriched in specific combinations of histone and DNA modifications that are responsible for the inclusion or exclusion of the marked exon. Interestingly, those differentially marked splicing events are not defined by weaker splice sites or higher GC content, as one would have expected, but by enrichment of specific RNA binding motifs along the regulated exon, suggesting that these chromatin signatures might be involved in the differential recruitment of specific splicing regulators to the pre-mRNA (Agirre et al., manuscript in preparation). Moreover, using an inducible and dynamic cell reprogramming system, based on the epithelial-to-mesenchymal transition (EMT), they have shown that in the model gene FGFR2, there are cell-specific chromatin signatures responsible for the differential recruitment of the splicing factor PTB to FGFR2 pre-mRNA, which has a direct impact on the final splicing outcome. Importantly, as soon as the first changes in splicing are observed during the EMT, there are also changes in those chromatin enrichment levels, but not in the expression levels of the splicing regulators involved, such as PTB, suggesting that highly dynamic changes in splicing might be more dependent on changes in chromatin than in changes in expression levels of the regulatory proteins. Finally, towards the understanding of what establishes those splicing-specific chromatin signatures, the Luco lab has identified a new role for antisense lncRNAs in alternative splicing regulation. They have discovered a new lncRNA, expressed within FGFR2 gene, only in epithelial cells and in the antisense direction, responsible for the direct recruitment of the chromatin complexes, amongst Polycomb, responsible for the chromatin signature necessary for inclusion of the epithelial isoform. In the absence of this lncRNA, there is enrichment of the chromatin marks specific of mesenchymal cells, which inhibits the inclusion of the epithelial exon. This was an important finding for the field that was published in Nature Structural Molecular Biology and opened new perspectives in the integrative role of lncRNAs and chromatin in cell-specific alternative splicing regulation (Gonzalez et al., 2015).

Dirk Schübeler, Friedrich Miescher Institute, Basel, Switzerland
The Schübeler group used mouse stem cells to ask how DNA sequence and chromatin determine local DNA methylation states. In particular, they studied how CG content and binding of transcription factors (TF) create hypomethylated states. Through repetitive genomic targeting of the same chromosomal site they could show how binding motifs for TFs increase the likelihood of reduced DNA methylation (Krebs et al., eLife, 2014). In parallel they showed that TF based hypomethylation includes active demethylation (Feldmann et al., PLoS Genetics 2013). The relation between TF motif and methylation however is highly factor specific. Some factors can cause local demethylation (Stadler et al., Nature 2011) while others are highly sensitive. This can create a hierarchy where methylation insensitive factors create local region of hypomethylation that creates binding opportunities for sensitive factors (Domcke, Bardet et al., Nature 2015). This let them to postulate that DNA methylation can function in form of indirect cooperativity between transcription factors. In parallel efforts the group has furthermore mapped the genomic binding of MBD domain proteins (Baubec et al., Cell 2013) and of DNA methyltransferases (Baubec et al., Nature 2015). This revealed how histone modifications that are present at active genes can recruit de novo methylation activity highlighting the crosstalk between chromatin and DNA methylation.

Vincent Colot, CNRS, ENS Paris, France
The Colot group used the population of Arabidopsis epigenetic recombinant inbred lines (epiRILs) it has generated to establish in plants the principles that govern the inheritance patterns of DNA methylation over TE and other repeat sequences across generations as well as their phenotypic consequences (Colomé-Tatché et al, PNAS, 2012; Cortijo et al, Science 2014). The central role of RNA-directed DNA methylation (RdDM) in preventing irreversible DNA methylation loss was confirmed, but why some repeat sequences are efficiently targeted by the RdDM machinery when others are not remains largely unknown. Nonetheless, the group could show that for RdDM targets, CG sites within nucleosomal DNA acquire DNA methylation progressively over successive generations, which CG sites within nucleosomal DNA become methylated immediately. These findings have important implications, as they indicate intricate dynamics of DNA methylation establishment over repeat sequences.
In order to generate predictive models the Colot group collaborated with Vincent Hakim (Statistical Physics, ENS Paris) to model progressive DNA remethylation at the single cytosine level. It was determined that progressivity likely reflects the existence of a very narrow window of time (a few cell divisions at most following fertilization) during which DNA methylation can be apposed, in a probabilistic manner, on CG sites that are located within nucleosomal DNA.
The Colot group further developed tools to target DNA methylation or demethylation to specific regions using the CRISPR-Cas9 system.
Thanks to EpiGeneSys, the Colot group produced >20 publications and two of the personnel involved in the project (C. Chica and M. Kassam) went on to obtain permanent positions at the Pasteur Institute and Nestlé-Lausanne, respectively.

Peter Rugg-Gunn, Babraham Institute, Cambridge, UK
The Rugg-Gunn group has generated genome-wide DNA interaction maps in naïve and primed mouse pluripotent cells and identified key differences in promoter-enhancer interactions between the cell types that are likely to influence cell state. Bioinformatic approaches to integrate the DNA interaction data with chromatin marks and pluripotency factor binding profiles revealed that pluripotency factors are highly enriched at enhancer-associated interacting regions. Using mutant cell lines, they have mapped changes in promoter-enhancer contacts that are associated with the loss of selected pluripotency factors. These results provide new insight into how sequence-specific DNA binding transcription factors can establish key gene regulatory networks that enable the establishment and maintenance of the pluripotent state. In addition, the Rugg-Gunn group has also identified new roles for pluripotency factors in regulating heterochromatin organisation in mouse embryonic stem cells (Novo et al., 2016). These results establish a direct connection between the pluripotency network and chromatin organisation and emphasise that maintaining a specific heterochromatin architecture is a highly regulated process in embryonic stem cells.

Eran Segal, Weizmann Institute, Rehovot, Israel
In this project, the Segal group worked on combined regulatory models for predicting expression patterns from genotype and epigenotype, and determining epigenome variation for transcription factors and nucleosomes in yeast in models of known genetic variation. A wide range of analyses and experimental data were generated throughout, including analyses of libraries of the expression of thousands of different regulatory sequences that were fully designed and accurately measured which tested the effect on the epigenome (nucleosome positioning and expression) of transcription factor binding sites and nucleosome disfavouring sequences. The libraries were based on technology that the Segal developed which allowed for accurate measurements of the expression of thousands of different regulatory sequences of our choice. Several questions pertaining to DNA-encoded nucleosome positioning were examined. These included effects of DNA sequence on the positioning of nucleosomes, effects of nucleosome disfavouring sequences, and a multitude of different manipulations to the location, strength, multiplicity and identity of transcription factor binding sites. New methods were also developed for revealing the mechanisms by which the sequence drives expression. The Segal lab devised a method for simultaneously measuring the binding events of thousands sequences at single cell resolution, thus providing information on the joint binding configurations of all functional elements within each tested sequence as well as the fraction of cells in the population found in each of these binding configurations. To achieve this, the system was adapted to high-throughput, and specifically tailored to large-scale libraries of regulatory sequences, a method that can provide single cell binding data and which was applied to a few yeast and mammalian promoters. Combined with the above expression assays that was developed, this method provided a unique opportunity to study, at unprecedented scale, the mechanism by which sequence affects expression, by separately studying how sequence dictates DNA binding events and how these binding events, in turn, dictate expression.

Petra Hajkova, Imperial College, London, UK
The Hajkova laboratory has investigated the interplay of chromatin and DNA modifications (5mC and 5hmC) during epigenetic reprogramming in vivo. The team has developed an ultra-sensitive LC/MS approach as well as novel base resolution genome-wide mapping approaches (in collaboration with NEB). The work has contributed to main objectives in WP3 by a) elucidating relationship between chromatin structure and DNA methylation (Nashun et al, Mol Cell 2015) and b) by identification of sequence features of the genomic regions displaying 5hmC enrichment during the periods of developmental reprogramming.

Guillaume Filion, CRG, Barcelona, Spain
The Filion lab developed and used the TRiP technology (Thousands of Reporters in Parallel) to understand how the genomic and chromosomal context influences the activity of enhancers. Their results show that the activity of the same enhancer is site-dependent. This means that the global organization of the genome, in terms of chromatin domains or three-dimensional organization, has an influence on the capacity of transcription factors to form active regulatory elements. These results illustrate that the sequence of a regulatory element is not sufficient to predict its activity: the knowledge of the surrounding genomic context is also necessary. This underscores the importance of understanding how the large-scale context affects the activity of regulatory elements to fully understand how transcription is regulated.

Yuri Schwartz, Umea University, Sweden
The Schwartz group focused on understanding chromosomal interactions and testing their design principles by genetic manipulation and introduction of synthetic sequences. To this effect they contributed to identification of novel interacting partners of the key chromatin insulator protein Cp190. They integrated the information on genomic binding of these proteins to define 29 combinatorial classes of insulator protein binding sites. The Schwartz group tested representative elements from some of the new classes of insulator protein binding sites for their enhancer-blocking ability (Savitsky et al., 2016). They also used the new classification to test predictions of mechanistic interrelations between different insulator proteins. To test the design principles of chromosomal interactions the Schwartz group used the Drosophila homeotic gene cluster as a model to look for functional relations between key insulator proteins CTCF, Mod(mdg4) and Cp190. They for the first time showed that Mod(mdg4) acts in concert with CTCF, that Mod(mdg4) and Cp190 are not redundant and that Cp190 is critical for exceptionally strong Fub insulator to partition the bithorax complex into two topological domains (Savitsky et al., 2016). In addition, the Schwartz group derived and characterized new cultured Drosophila cell lines that lack individual key insulator proteins Cp190 and CTCF. In collaboration with the Cavalli group they have produced Hi-C maps of genomic contacts in these mutant cells and matching wild-type control cells. Joint analysis of Hi-C results by Schwartz and Cavalli groups showed that that effects of Cp190 and CTCF mutations on contact patterns are very specific and that the changes of contact frequencies across insulator protein binding sites always increase in mutant cells. The classes of insulator protein binding sites that show changes of contact frequencies always co-bind multiple proteins. Finally, they found that in some instances the increase in the contact frequencies across insulator protein binding sites correlates with merging of two neighbouring Topologically Associated Domains (TADs) and increase in the long-range contacts.

Suzana Hadjur, UCL London, UK
Within EpiGeneSys the Hadjur group has explored the mechanisms that underlie the evolution of chromosomal domain structures. Towards this goal they compared four mammalian species in regards to nuclear organization using Hi-C and by determining genomic binding sites for CTCF and cohesin proteins by Chromatin-IP. Their study reveals a direct link between insulator site divergence and the evolution of chromatin domain structure, pointing to a direct role for CTCF/cohesin in driving structural change in the genome. The analysis also revealed that oriented CTCF motifs determine the directionality of CTCF-mediated interactions (Vietri Rudan 2015 Cell Reports).

Nick Luscombe, UCL London, UK
The Luscombe lab's work focused on developing computational methods and applying them to analyse primary experimental data in collaboration with EpiGeneSys network members. The work focused on three main areas. (a) Understanding the epigenomic mechanisms for regulating dosage compensation in flies (together with Asifa Akhtar), led to the discovery of new mechanisms of molecular action by components of the dosage compensation complex. (b) Developing the Capture HiC chromatin capture technology (together with Peter Fraser's lab), which enabled to investigate promoter-enhancer interactions in the human and mouse genomes to unprecedented resolution, and also introduced a method for directly linking intergenic SNPs to sites of action. (c) Developing statistical models for understanding the relationships between regulatory proteins and target gene expression, with proof of concept studies in early fly development. Through the network, the Luscombe group benefited greatly from interacting with experimental laboratories. This ensured that computational methods we developed were relevant to biological investigations; allowed to help train experimentalists in computational analysis methods (for instance, through workshops and by hosting them in the laboratory); and finally it allowed to contribute to cutting edge research through wet/dry collaborations. Our laboratory will continue these new-founded collaborations and we shall continue to develop ways to investigate the impact of epigenetics on gene regulation.

Asifa Akhtar, MPI of Immunobiology and Epigenetics, Freiburg, Germany
To study the mechanism underlying X chromosomal dosage compensation in Drosophila, the Akhtar lab, in collaboration with the Luscombe lab, generated genome wide profiles of the RNA polymerase II from male and female salivary glands isolated from the third instar larvae. This work revealed that there is an enhanced recruitment of RNA Pol II on promoters and gene bodies in male versus female X chromosome. By performing analogous analysis in MSL2 mutants, they could show the increased Pol II recruitment is MSL complex dependent. Thus, the MSL complex acts to facilitate transcription already at early steps of transcription. Consistent with these results, their recent observations revealed that the MSL1 regulates bulk Ser5 phosphorylation of RNA Pol II. They identified that MSL1 interacts with CDK7, a kinase that phosphorylates Pol II Ser5. Interestingly, they also observed that MSL1 is phosphorylated in vivo and that MSL1 phosphomutants led to loss of dosage compensation and male lethality revealing a novel function of MSL1 post-translation modification in X chromosome regulation. Thus, there is a functional interplay between MSL1 and CDK7 to regulate transcription of target genes.
In addition, the Akhtar lab studied the contribution of chromosome conformation in X chromosome regulation. They identified that TAD organization is sex independent and that high affinity sites cluster along the X chromosome thus providing an interaction platform for the MSL complex to efficiently spread along the X chromosome. By using transgenic flies where high affinity sites were ectopically inserted on the autosomes, they could show that the targeting of the MSL complex can occur “in trans” on X chromosomal sequences, but the spreading of the complex requires X chromosome configuration. They proposed that the X chromosome uses “conformation based affinity” to spread on the X chromosome.

2.3 WP4 Signalling to Chromatin

The aim of this work package was to determine how the environment, stress, metabolic status and growth factors signal to the epigenome to induce new programmes of gene expression that may either be transient or lead to long-term heritable phenotypes. WP4 was divided into three scientific tasks and a general joint task. Task 1 investigated how the metabolic status of a cell and environmental stimuli influence the epigenome to dictate transcriptional programmes that can have far reaching and long term consequences for the organism. Task 2 was dedicated to determining how developmental signalling pathways influence transcriptional programmes through the epigenome to specify cell lineages. Task 3 determined how transient signalling leads to long term, heritable changes in gene expression. In a common effort we have sought to integrate our data with the aim to establish common platforms for data generation and modelling.
The work has gone extremely well over the last five years, and we have made substantial discoveries as outlined below.

Jane Mellor, University of Oxford, UK
The Mellor lab have discovered more about the link between metabolism and ageing in the yeast Saccharomyces cerevisiae. Acetylation at lysine 18 on histone H3, deposited by the SAGA complex and subsequently bound by the SAGA component, Spt7 (Howe et al. Mol Cell 55: 733-44, 2014), plays a key role in controlling translation of ribosomal protein genes during the growth phase when acetyl CoA levels are high (unpublished). Analysis of the effect of constitutive acetylation at K18 helps define how translation is related to longevity and how this links to the metabolic state of the cell. In addition, they have developed a state-switching model which explains how transcriptional interference and insulation, by altering chromatin structure, and involving antisense transcription (Murray et al. Nucleic Acids Res 43: 7823-37, 2015; Murray et al Nucleic Acids Res 40: 2432-44, 2012) controls the repression and activation of transcription leading to cycling levels of transcripts Nguyen et al. ELife 3: 2014).

Anne Ferguson-Smith, University of Cambridge, UK
The Ferguson-Smith lab determined how a compromised in utero environment, which leads to growth and developmental defects, influences postnatal phenotype. During the networking period, they have successfully studied several mammalian models addressing relationships between in utero environment, epigenetic states, prenatal development and postnatal phenotype. They took hypothesis-driven approaches and conducted both independent research and collaborative studies with colleagues within and outside Europe. Using a mouse model of in utero under-nutrition leading to intergenerational transmission of metabolic disease in offspring and grand-offspring, they have shown that the resistance of imprinted genes to early developmental epigenetic reprogramming does not render them more (or less) susceptible to environmental perturbations or more (or less) contributory to metabolic disease, at least in this model (Radford et al., PLoS Genetics 2012). However, their results identified regions of the sperm methylome that were perturbed in males exposed to in utero undernutrition. These regions, which have regulatory potential, were often adjacent to metabolism-associated genes whose expression was perturbed in embryos sired by these males prior to the onset of disease symptoms. Nonetheless, abnormal methylation levels did not persist in these offspring. Their findings suggest that abnormal methylation levels in sperm can act as a biomarker of compromised in utero exposure, but that other mechanisms exist to transmit non-genetic information across the generations (Radford et al., Science 2014). An epigenetic biomarker of compromised differentiation capacity in stem cells was also identified in a project conducted with members of WP3 (Butcher et al., Nat Comm 2015) resulting in a predictive model for ‘good’ versus ‘bad’ iPSCs. In a genetic model of defective folate metabolism during pregnancy, the Ferguson-Smith lab contributed to the characterisation of transgenerational phenotypes over multiple generations underpinned by epigenetic abnormalities (Padmanhaban et al., Cell 2013). The role of abnormal expression levels of developmentally regulated imprinted genes, their regulation and the signalling pathways that they interact with to control growth, metabolism and disease was further elucidated (Charalambous et al., PNAS 2014; Boucher et al PNAS 2014; Charalambous et al., Acta Physiol 2013; Martinez et al., 2014; Sferruzzi-Perri et al., FASEB J 2013) contributing to the understanding of the mechanisms and pathways contributing to the developmental origins of adult health and disease. Finally, a collaboration between the Ferguson-Smith lab and that of Andrew Pospisilik (EpiGeneSys – RISE programme; this WP), led to the exciting finding that a stochastic bi-stable obese phenotype regulated by a well-characterised epigenetic regulator (Trim28) interacting with an imprinted gene network. Most importantly, this bi-modal obesity phenotype and associated mechanism identified in mice appears to be found in humans too (Daalgard et al., Cell 2016).

Michael Weber, CNRS, University of Strasbourg, France
The laboratory of Michael Weber led a project to study how environmental pollutants modify the epigenome in germ cells. It has been postulated that in utero exposure to environmental pollutants such as endocrine disruptors during early life could induce persistent alterations in DNA methylation in germ cells that could be transmitted across generations and could lead to increased risks of disease in the descendants. The Weber lab analysed, in collaboration with Jesus del Mazo, a mouse model of prenatal exposure to the endocrine disruptor vinclozolin, a common fungicide with antiandrogenic effects. Female mice were exposed to vinclozolin during pregnancy and germ cells, somatic cells and phenotypic data were collected in F1-F3 generations. The results showed that vinclozolin induces a transgenerational reduction of the fertility rate and increased apoptosis in adult testis of male mice along three generations. The molecular data indicate that exposure to vinclozolin, even at low dosage, alters the Lin28-let-7-Blimp1 pathway over three generations. They then performed large scale DNA methylation profiling in male gametes of exposed versus control individuals to reveal potential epigenetic mechanisms of action. This revealed very little perturbation in gametic DNA methylation reprogramming as a consequence of vinclozolin exposure, suggesting that other epigenetic pathways mediate the transmission of phenotypes.
The Weber lab also studied how epigenetic marks such as DNA methylation contribute to establish cell identities in the embryo. During development, the cells of the embryo progressively acquire cell-type specific gene expression profiles that create the different tissues of the organism. It is believed that epigenetic marking helps to reinforce and stabilize these cellular decisions. The Weber lab studied this question using the hematopoietic lineage as a model. They used systematic epigenomic and transcriptome analysis in hematopoietic progenitors in several time points of development, which revealed epigenetic changes accompanying lineage specification in vivo. To test the role of these epigenetic changes, they created and analysed a series of genetic mutants to interfere with DNA methylation reprograming in hematopoietic progenitors. These mutations induce lethality after birth, indicating that DNA methylation is crucial to establish the proper hematopoietic identity. The integration of methylome, transcriptome and phenotypic data is now in progress and will reveal how DNA methylation contributes to establish specific gene expression profiles that create the hematopoietic cell lineage identity in mammals.

Jon Houseley, Babraham Institute, Cambridge, UK
The Houseley team worked on modelling the influence of current epigenetic information on future actions. His lab set out to discover how epigenetic modifications impact the ability of organisms to adapt to new environments. They found that certain epigenetic marks are highly dynamic across the ageing process, playing a critical role in regulating gene expression across lifespan. Importantly, they also found that the ageing process is a substantial contributor to organismal adaptability, a property that unexpectedly increases with age. They concluded that a major mechanism by which epigenetics can impact adaptability is through the modulation of the rate of ageing.

Wolf Reik, Babraham Institute, Cambridge, UK
During the tenure of EpiGeneSys the Reik lab investigated how cell lineages are determined in embryo and made a number of breakthrough discoveries in this topic. This includes the realization that FGF signalling (and it¹s inhibition) underlying primed and naïve pluripotency regulates global DNA methylation states and hence epigenetic memory in mouse and human ESCs (Ficz et al 2013 Cell Stem Cell, Takashima et al 2014 Cell, Guo et al 2016 Stem Cell Reports, von Meyenn et al 2016 Mol Cell). Also, key signalling pathways important for adhesion and migration of trophoblast cells and for trophoblast function in vivo were found to be regulated by maternal (oocyte derived) methylation marks (Branco et al 2016 Dev Cell). In collaboration, the lab also invented new methods for single cell sequencing of the methylome (Smallwood et al 2014 Nature Methods) and for sequencing of the transcriptome and methylome from the same single cell (Angermueller et al 2016 Nature Methods). Dr Tim Hore who was the main postdoc funded by EpiGeneSys obtained a lectureship position at Otago University New Zealand and is being considered for a prestigious Rutherford Research Fellowship.

Caroline Hill, Cancer Research UK/The Crick Institute, London, UK
The aim of the work of the Hill lab to discover how growth factor signalling between cells controls patterning of the germ layers of the early embryo, in particular they strived to understand how TGF-β superfamily ligands, which include the TGF-s, Activin, Nodal, BMPs and GDFs, signal to chromatin to regulate transcriptional programmes and define different cell fates and behaviours. The aim was also to build computational models of the TGF-β signalling pathway that could explain how cells continuously monitor the dose and duration of signalling to elicit appropriate responses at the level of transcription. The work has been very successful and they have achieved much of what they set out to do. They succeeded in first building a computational model of TGF- signalling that explained how the pathway is modulated over time (Vizan et al 2013, Science Signalling). This work highlighted the fact that acute TGF- signalling leads to cells becoming refractory to further stimulation. In contrast, this is not a feature of other TGF- family ligands, in particular Activin, Nodal and BMPs. Instead they could show for these ligands that receptors constantly monitor the levels of extracellular ligand resulting in highly sustained signalling over long periods of time. For the Activin/Nodal ligands they undertook a genome-wide analysis of how signalling through Smad2–Smad4 complexes regulates transcription, and defined the sequence of events that occur from Smad binding to chromatin to chromatin remodelling to transcriptional regulation at the level of RNA polymerase II (Gaarenstroom et al, eLife, in review). This work has also revealed how the transcriptional programme is modulated during sustained signalling as a result of a dynamic transcriptional network.
Lastly, to put these mechanistic studies into an in vivo context the Hill lab have defined how domains of active BMP and Nodal signalling are established in early embryos using zebrafish as a model system (Ramel and Hill, 2013 Dev Biol; Reichert et al., 2013 Development; van Boxtel et al., 2015 Dev Cell). Most importantly, the mechanistic findings can now be tied in with the in vivo spatial/temporal regulation for the Nodal pathway. The Hill lab have shown that the key determinant of cell fate is duration of Nodal signalling, rather than ligand dose as has previously been assumed.

Vijay Tiwari. IMB, Mainz, Germany
The Tiwari group has studies how signalling pathways communicate with chromatin to mediate transcription reprogramming underlying cell-fate changes during Epithelial to Mesenchymal Transition (EMT). They investigated gene regulatory mechanisms driving epithelial to mesenchymal transition (EMT), a process by which cells lose their cell contacts and gain migratory and invasive properties. EMT is essential for developmental processes such as neural crest migration, and it occurs aberrantly in cancer metastasis. In this context, they are currently pursuing two objectives: (1) discovering novel regulators of the gene expression program underlying mesenchymal fate specification, and (2) the interplay of transcription factors with epigenetic mechanisms in driving EMT. Their highly innovative and multidisciplinary study holds promise to lay milestones and open new horizons for a wider community interested in fundamental concepts of gene regulation and translational medicine.

Renato Paro, ETH, Basel, Switzerland
The Paro lab determined how a short-term stress signal induces long-term changes in gene expression patterns that can be transmitted through generations. Environmental epigenetics attempts to understand how environmental cues interact with cellular epigenetic networks affecting gene expression, and possibly leading to the inheritance of altered gene expression states. The project of the Paro group assessed one such interactions, between heat shock protein 90 (Hsp90) and the cellular memory system (PcG/TrxG). While the former mediates stress response, the latter are involved in maintenance of gene expression crucial for development and homeostasis. By mapping chromatin-binding sites of Hsp90 at high resolution across the Drosophila genome, they uncovered an unexpected mechanism by which this chaperone orchestrates cellular physiology. Hsp90 localizes near promoters of many coding and non-coding genes including miRNAs. Using computational and biochemical analyses they found that Hsp90 regulates many PcG/TrxG target genes by affecting RNA polymerase II pausing, via stabilizing the Negative Elongation Factor complex. Hsp90 is required for optimal activation of paused genes in Drosophila cells in response to environmental stimuli.
By performing similar analyses in mammalian cells the Paro group could demonstrate a conservation of Hsp90´s chromatin function. They showed that Hsp90 inhibition leads to an immediate and transient up-regulation of a number of genes implicated in stress responses as well as intrinsic apoptotic signalling. They observed that Hsp90 physically associates with nearly 200 gene loci. Interestingly, Hsp90-targeted genes were themselves enriched for gene expression regulators including chromatin and transcription factors, allowing the identification of connected gene expression networks. This now permits to resolve questions pertaining to possible effects of environmental stress on gene regulation, linking cellular exterior to epigenetic gene regulation via Hsp90 and PcG/TrxG.
A microbial infection is another environmental stress stimulus, which could impose heritable chromatin changes. Lipopolysaccharides (LPS) are the principal cell wall components of gram-negative bacteria having the potential to lead to detrimental effects. In mammals for example, LPS induces septic shock through a Toll-like receptor-dependent signalling pathway. In Drosophila S2 cells, resembling embryonic hemocytes, exposure to LPS leads to the differential expression of a wide range of genes including defence and immunity genes as well as signalling factors and cytoskeletal and cell adhesion regulators. The Paro group generated reporter cell lines that induce green fluorescent protein (GFP) expression upon LPS treatment. The isolated fraction of responders is subjected to repeated LPS stimuli and the chromatin changes are studied. Perturbation experiments uncovered the role of the PcG/TrxG system in memorizing stress response. To assess whether pathogen exposures are epigenetically memorized in a mitotic fashion but also in a transgenerational manner, appropriate reporter fly lines were created. Oral infection of Drosophila larvae by the phytopathogenic bacteria Erwinia cc15 (Ecc15) activates an immune response in the gut including the induction of stress signalling by JNK (puc). Using fluorescent Drosophila reporter lines and quantitative RT-PCR, they investigated gene expression changes in the larval gut after repeated exposure to bacteria. Experiments show that expression of puc after oral Ecc15 infection of Drosophila larvae is affected by previous exposure to Ecc15, pointing to a potential memory effect.

Edith Heard, Institut Curie, Paris, France
Heard lab worked on determining how the X-chromosome is inactivated in female embryos and how this is state maintained as stable and heritable. Their ultimate aim was to model the X-inactivation process in the context of other changes in early development. For the initiation of XCI this was achieved with the development of a mathematical model (Schulz et al in preparation) as well as several key discoveries, including the definition of the extent of Xist’s regulatory landscape by chromosome conformation capture, which also led to the discovery of TADs (Nora et al, Nature 2012); as well as the block to differentiation that the presence of two active X chromosome represents that can be relieved by induction of XCI (Schulz et al, Cell Stem Cell 2014). This demonstrates the tight integration between the regulation of differentiation and the initiation of XCI and has opened up several new exciting avenues of research. Both these studies represented collaborations within EpiGeneSys as well as with international groups. The Heard lab have also generated an X-chromosome wide integrated analysis of chromosome structure, accessibility and transcriptional status using in vitro differentiated ESC into neural progenitor cells (NPCs) (Giorgetti et al, under revision). This has revealed that the inactive X is organized into two megadomains, separated by the DXZ4 macrosatellite region, and that it is globally devoid of TADs, except at regions that escape from XCI. These findings on the global changes in organization of the X chromosome will be integrated with epigenomic data being generated and will be used to establish the dynamic changes using mathematical models in a continuation of the collaborative efforts within EpiGeneSys.

Kerstin Kaufmann, Universität Potsdam, Germany
The Kaufmann team determined the trans-generational regulation of recombination initiation by epigenetic factors in Arabidopsis thaliana. They aimed to understand the functional role of epigenetic factors in the transgenerational response of the Arabidopsis thaliana plant to stress stimuli, by impacting homologous recombination initiation during meiosis. Over the course of the project period, genome-wide DNA binding maps of SPO11, which is a key factor for meiotic recombination in Arabidopsis, were created and analysed. For this purpose, the DamID-seq method was optimized and used to study the SPO11 genomic binding landscape during meiosis under normal and stress conditions. Computational tools for the integrative and comparative analysis of ChIP-seq data in plants were established and optimized. The SPO11 DNA-binding data were integrated with available genome-wide data on several types of histone modifications and DNA methylation, in order to determine potential mechanisms of SPO11 recruitment. The Kaufmann lab also compared SPO11 binding as a marker of recombination initiation with available re-sequencing data of successful meiotic recombination events. This research therefore provides therefore a deeper insight into molecular mechanisms of meiotic recombination initiation under different environmental conditions.

J. Andrew Pospisilik, MPI of Immunobiology and Epigenetics, Freiburg, Germany
The Pospisislik lab studied the acute daily plasticity of the epigenome and its dependence on PcG/TrxG control. Over the course of the funding scheme his group established and screened multiple PcG/TrxG conditional deletion mutants. Adipose and liver-specific (Ap2-cre / Alb-cre) Mll, Ezh2, and Eed mutants. To date, the liver mutants have all been screened and now the adipose lines are undergoing phenotyping. Reported during the last period liver-specific Eed mutants grow and develop normally, and interestingly, despite liver-specific deletion, a strong adipose tissue related metabolic phenotype is observed. Specifically, Alb-Eed mutants exhibit reprogramming of their subcutaneous white adipose tissues (scWAT) towards a beige fate and function. Oral glucose tolerance, body weight, and insulin tolerance are all improved in these animals, a phenotype that persists upon high-fat feeding. Consistent with an apparent browning of their adipose, these mice exhibit an increased body temperature. More recent examinations also of adipose-tissue targeted Eed deletion revealed a similarly disease resistant phenotype. Intersecting H3K27me3 ChIP-seq data sets with transcriptional profiles to better understand the likely transcriptional rewiring that explains the enhanced metabolic phenotypes the Pospisilik lab identified multiple candidate effector proteins that might comprise the liver-adipose axis. In addition, human samples comprising HDL/LDL/VLDL measures from normal and metabolically diseased individuals have confirmed biologically active ligand in theses regards. These data provide evidence that a parallel axis is active in humans. Also achieved over the course of the funding scheme, the Pospisilik lab generated complete epigenomes for adipocyte circadian function (day and night time points). These comprise DNA-Me, coding and non-coding RNA-seq, ChIP-seq for the core 6 histone modifications. These data are currently being intersected with their recently published model of mammalian polyphenism, a Trim28 haploinsufficient mutant that emerges from development to become either lean or obese, but not in between. Intriguingly they found that the transcriptional changes, particularly in obese/lean-defining imprinted genes, exhibit clear evidence of chromatin state associations. Thus, novel biological insights are being made into the Polycomb and circadian plasticity of adipocyte chromatin.

Claire Rougeulle, CNRS, Université Paris-Diderot, France
The Rougeulle lab has centered its research on assess the epigenetic instability of human ES cells (hESC) and how it impacts on the biology of these cells. They investigated the epigenetic instability of hESC by focusing on the activity of the X chromosome, which is epigenetically regulated in females through X-chromosome inactivation (XCI). Most female hESCs have already undergone XCI, but XCI is indeed highly unstable in female hESCs and there is a spontaneous loss of several markers of XCI occurring in culture, in a process called erosion of XCI. The Rougeulle lab first analysed the epigenomic landscape of the inactive X (Xi) in hESC, and showed that it is partitioned in distinct, non-overlapping epigenomic domains characterized by either H3K9me3 or H3K27me3. They further demonstrated that such organization is not conserved in differentiated cells, indicating that the chromatin organization of the inactive X varies according to the cellular context.
The bi-partite organization of the Xi in hESCs also appears to condition the instability of XCI, as erosion of XCI was shown to be restricted to H3K27me3-rich domains, in which chromatin and transcriptional alterations occur.
The Rougeulle lab has also identified a human-specific X-linked ncRNA, XACT that has the unique property of coating active X chromosomes in hESCs. They have shown that re-expression of XACT from the inactive X is, so far, the first sign of XCI erosion, occurring prior to the loss of other XCI markers, and in particular before the loss of XIST expression and before the partial reactivation of the chromosome. In addition, the expression of XACT is restricted to pluripotent contexts in which XCI is unstable. Altogether, these data suggest that XACT could play a causative role in the instability of XCI in human pluripotent stem cells. Studies aiming at understanding how XACT could have such an effect suggested that XACT could impact on the ability of XIST to coat and silence the chromosome.
The team of Rougeulle have also studied the contribution of epigenetic misregulation in the etiology of autism spectrum disorders (ASD). They have investigated transcriptomic and miRNA profiles as well as the distribution of chromatin marks (H3K4me3 and H3K27me3) and of DNA methylation in olfactory mucosa stem cells (OMSC) derived from autistic patients and sex- and age matched controls. Together with their collaborators (teams of Drs. Colleaux and Humeau), they have identified a signature of 4 miRNAs that are commonly deregulated in ASD. In addition, several transcripts were found differentially expressed in patients’ cells compared to controls. These genes could represent novel candidates participating to the etiology of ASD.
The ambitious aim of this last task was to try to establish common platforms for data generation and modelling. In terms of the modelling, this has not turned out to be possible as the systems used in the work package have been too disparate. However, with regard to data generation, the Mellor lab have developed techniques for internal controls for ChIP-seq, RNA-seq, ribosome profiling and NET-seq in yeast, using a related yeast, which should be applicable to other organisms. Moreover, the Reik lab have invented new methods for single cell analysis. One is for sequencing of the methylome (Smallwood et al 2014 Nature Methods) and the other for sequencing of the transcriptome and methylome from the same single cell (Angermueller et al 2016 Nature Methods). These methods will undoubtedly be used by other members of EpiGeneSys, as well as widely in the epigenetics field in general. During the course of this project the WP4 members also got together several times at annual meetings to discuss issues around performing, controlling and validating next generation datasets and best practice. They were able to share ideas, published lists of antibodies and protocols.

2.4 WP5 – An Integrated Computational Epigenetics Framework

Rapid progress was achieved in computational epigenetics during 2010-2015. The overarching goal of WP5 was to facilitate better integration between systems biology tools and techniques and epigenetic research. We originally identified several domains that were challenging (at 2010) for most experimental groups, and wished to open these by making tools more accessible and usable (Task 1-3). Over the last five years, through work by a large number of groups (Task 4), and most importantly, by deeper and deeper integration of bioinformatics and computational skills into research group, the challenges identified originally are for the most part resolved. Nevertheless, the deeper theoretical question underlying the new techniques requires additional research and perhaps new interdisciplinary ideas.
- Analysis of “linear” epigenomic marks, including ChIP-seq. Including normalization, peak finding and comparative analysis. ChIP-seq analysis is now considered routine, implemented in standard tools (e.g. bioconductor), and being used by essentially all groups within the network. While the high-level interpretation of ChIP-seq profiles remains difficult in some cases, the technical aspects can be approached by any student with minimal access to computer work.
- Analysis of nucleosome positioning. Nucleosome positioning was being advanced by improvements in read depth, and by linking footprinting techniques (DNase-seq, ATAC-seq) and even ChIP-seq, with inference of nucleosome distributions. While the computational problems were technically resolved, the interactions and causal relationship between nucleosomes and epigenomic control are still unresolved completely.
- Analysis of DNA methylation – normalization and filtering of common techniques, identification of differentially methylated regions and its interpretation. Extensive work by EpiGeneSys members and others, and large scale data production by projects such as BLUEPRINT and IHEC, have made the technical challenge of processing bisulfite sequencing data quite straightforward. For DNA methylation, mapping and normalization is easier than in other techniques, and methylation profiles are being generated rapidly and efficiently by techniques varying from arrays to WGBS. Interpretation and integrative modelling of methylation and other data is also technically feasible through widely available tools. Again, quantitative and systematic modelling of epigenetic control and the effect of DNA methylation on it is lagging behind these technical advances.
- Hi-C and 4C analysis and integration. Chromosome conformation capture was booming during 2010-2015, stimulating rapid development of computational tools for analysis of modern experiment. Since the techniques are newer, the computational methodologies are less mature, and EpiGeneSys contributed greatly to the acceleration of their development and distribution. 4C is now reasonable standardized, with several pipelines available, which Hi-C analysis is being standardized, in particular through a major NIH project (4D Nucleomics).
These four domains are exemplifying the evolution of computational epigenomics over the project period. Progress was so rapid and parallel, that some of the formal Tasks (e.g. toolbox) were supressed by broader goals that involved considerable interaction between EpiGeneSys and other projects (BLUEPRINT, ENCODE). Nevertheless, the less anticipated effect of massive recruitment of bioinformaticians into experimental research groups resulted in remarkably efficient realisation of the major WP5 goals, around year 3-4 of the project, while additional work by computational group aimed at expanding the horizon of epigenetic research with new frontiers such as single cell genomics.
Challenges ahead – can we truly use mathematics and quantitative models to understand epigenetics?
While the progress in integration of “systems” techniques into epigenetics over the last five years seems remarkable, its long term and more fundamental effect of epigenetics (and biology in general) is yet to be determined. Modern epigenetic research is data-rich, and consequently the toolbox used by epigenetic groups now include algorithms for high throughput data manipulation and advanced statistics. Nevertheless, it may be argued that at the current state, quantitative and systems methods are used in an ad-hoc fashion, to analyse specific experiment and derive from them qualitative models. We are still far away, in almost all fields of epigenetics, from the ability to write down quantitative equations that can represent our experimental data and predict new ones, and the wealth of data we currently derived experimentally is serving us in many cases to screen through thousands or even millions of possibilities, and identify one or few qualitative hypotheses that we follow up on. Substantial additional work, and further strengthening of the interaction between computational scientists and epigeneticists, are required to continue and move forward, beyond the use of computational methods for phenomenological analysis.
Challenges ahead – translating systems biology and epigenetics.
Another major challenge ahead lies in the application of the new experimental and computational techniques to study patients' data. There are already several emerging applications, mostly using DNA methylation analysis to study cancer. Nevertheless, it is likely that massive DNA sequencing of various patients' cohorts, and its combination with various epigenomics techniques, will pose new unexpected challenges to our interdisciplinary community. The strong ties we already established, and the new generation of young researchers who are mastering both experimental and computational aspects of modern epigenetic research, promise to serve us well while approaching these challenges.

Potential Impact:
3. Impact

3.1 Impact on science and technology

EpiGeneSys has created resources and approaches to deliver a long-term impact on structuring the European Research Area (ERA) in both epigenetics and systems biology, as a result increasing the competitiveness of the ERA in other areas and supporting EU policy. Through its integrated research effort gathering multidisciplinary expertise (e.g. computational epigenomics, modelling, biochemistry, construction of databases, novel single cell approaches, live imaging) with substantial cross fertilisation, the network has advanced our understanding of epigenetic systems, particularly in four broad areas.
First, EpiGeneSys has contributed to gaining a quantitative understanding of the molecular driving forces that govern robustness and sensitivity of the binding of epigenetic regulators to chromatin and which thus are the key points for regulation during normal cellular function and for deregulation upon disease. The network has provided deep insights into the mechanistic and quantitative details of how epigenetic modifications are established, maintained, and erased. Second, EpiGeneSys has helped to elucidate the crosstalk between genome and epigenome by applying rigorous high-throughput systems biology approaches for several epigenetic variables in selected model organisms and populations of known sequence diversity. The generated datasets and tools were utilized to model genome-wide epigenetic states and regulatory networks, and to predict regulatory interactions between genotype and epigenotype. Predictions were validated and characterized through experimental testing using synthetic sequences in selected model systems. Third, EpigeneSys has contributed to determining how the environment, stress, metabolic status and growth factors signal to the epigenome to induce new programmes of gene expression that may either be transient or lead to long-term heritable phenotypes. Fourth, in addition to these achievements, EpiGeneSys has helped to facilitate better integration between systems biology tools and techniques and epigenetic research through a dedicated work package. The analysis of “linear” epigenomic marks, including ChIP-seq has become a standard tool used by essentially all groups within the network. Nucleosome positioning has been advanced by improvements in read depth, and by linking footprinting techniques (DNase-seq, ATAC-seq) and even ChIP-seq, with inference of nucleosome distributions. In terms of analysis of DNA methylation, extensive work by EpiGeneSys members and others, and large scale data production by projects such as BLUEPRINT and IHEC, have made the technical challenge of processing bisulfite sequencing data quite straightforward. EpiGeneSys has contributed greatly to the acceleration the development of chromosome conformation capture and the stimulation of rapid development of computational tools for analysis of modern experiment. Progress has been rapid and parallel, stimulating broader goals that involved considerable interaction between EpiGeneSys and other projects (BLUEPRINT, ENCODE). These synergies have also produced the first single cell epigenome sequencing technologies and multimodal techniques together with new computational algorithms.
The impact on scientific knowledge and technology development has occurred on a global scale on account of more than 350 scientific publications stemming from the research efforts of the network, many in leading high impact factor journals such as Nature, Cell, or Science with a very wide audience, publishing in open access mode wherever possible. The network has organised 10 conferences and thematic workshops (not even including the workshops of the training programme) and co-organised 13 more, thereby reaching more than 2000 scientists. Moreover, members of EpiGeneSys have also presented their exciting research findings at the leading international meetings such as EMBO, Gordon Research, Keystone, and FASEB conferences.

3.2 Impact on integration of Epigenetics and Systems Biology communities in Europe

The scientific achievements of EpiGeneSys alone have greatly contributed to a sustainable integration of epigenetics and systems biology. The goal of bringing systems biology and epigenetic methodologies together has been achieved by both collaborations between computational and experimental groups as well as by developing computational expertise within experimental epigenetics groups. A large number of published and upcoming papers represent the significant incorporation of computational methods into mainstream epigenetics and epigenomics. EpiGeneSys members have contributed to software compendia in large community efforts such as the EC collaborative project BLUEPRINT, IHEC (International Human Epigenome Consortium) and projects such as the US NIH Epigenomics and 4D-Nucleomics Roadmaps. Software implementing systems biology approaches to epigenetic problems is being deployed throughout the large EpiGeneSys community, and is being applied daily to generate a large number of publications. Partly as a result of these integration efforts many participating Institutes and Universities in Europe are making group leader appointments of mathematicians, physicists, and modellers into biology based departments. The reverse is also beginning to happen in some places.
EpiGeneSys has extensively collaborated with other European initiatives (BLUEPRINT, Nucleosome4D, 4DCellFate, EPIGEN, Modhep, ENCODE), thereby ensuring the dissemination of a new culture joining epigenetics and systems biology across Europe. In order to further support this goal, joint meetings and workshops organised with other initiatives, research institutes or funding bodies always featured sessions/talks focusing on research at the interface of the two disciplines.
Beyond the European stage the network has had key impacts by linking up with international research actions such as the International Human Epigenome Consortium (IHEC). Similarly, EpiGeneSys has been a unique partner for interactions through the NIH Roadmap initiatives Epigenomics and 4DNucleome, as well as further collaboration worldwide (Japan, China, Canada).
The impact of the network has been further facilitated and augmented by the EpiGeneSys project website, which has become a hub for the epigenetics and systems biology communities in Europe, but, as the website analytics have shown, is also an important source of information for the scientific community world-wide. The network has promoted new actions by providing stimuli to members to apply to new project calls (COST applications, application for a French Groupe de Recherche International, the 4DNucleome Initiative in Europe and collaboration with the NIH Roadmap programmes, etc.) to ensure a long term impact of our action. A new meeting of the epigenetics/systems biology community has already been planned for 2017. Although it is possible to fund minor actions like a conference, dedicated funding is necessary to ensure that the momentum of the progress towards true inter- and transdisciplinarity achieved by EpiGeneSys is not lost. The continuation of this vital contribution to the sustainability of the ERA calls for new instruments and continued European funding.
The EpiGeneSys training programme, which addressed mostly PhD students and postdoctoral researchers, has laid the ground for research combining epigenetics and systems biology to keep on flourishing in the future.

3.3 Impact on science and health policies

Further impacts will as a result occur in science policy making, and science funding policy developments nationally and including greater European integration leading to substantial European contributions to international flagship projects. By structuring research in Epigenetics to move into systems biology, EpiGeneSys has increased the competitiveness of the European Research Area (ERA) in many different aspects ranging from scientific knowledge and technology, to medical applications and human health, to education and training, and public education as well as supporting EU policy. Round table discussions on long-term strategy (Palermo 2013 and Paris 2016) united policy makers, scientists and representatives of funding bodies in order to identify potential approaches to continue the successful integration of epigenetics and systems biology so far facilitated by EpiGeneSys. A feasibility study undertaken by the network showed that greater engagement with policy makers and other scientific bodies could ensure the right support for scientific research, both in the field of epigenetics and more generally, including promoting the right environment for investment in research, ensuring better collaboration across disciplines and the EU, and ensuring the right policies are in place in terms of education, long-term support and funding. As an open and inclusive network EpiGeneSys has had the cohesive strength to catalyse interdisciplinarity and pave the way for future translational research. Systems biology approaches are expected to be useful at the clinical level and in generating new hypothesis to understand mechanism behind disease origin.
Epigenetics has the potential to be a key element in a paradigm change of our understanding of health and disease and fundamentally change public health policies. Epigenetic modifications are normally used during the development and maintenance of different cell types, but faulty epigenetic regulation can cause lasting damage, leading to cancer and other diseases ranging from metabolic disorders such as diabetes, to heart disease and mental health conditions. In the last 30 years, epigenetically induced changes in gene expression have been linked to many different cancers, among them bowel, breast, lung, prostate, liver, ovarian and pancreatic cancer. DNA methylation levels seem also to affect factors influencing cancer risk, such as obesity, smoking and ageing. Epigenetics also play a role in the autoimmune diseases such as rheumatoid arthritis with aberrant levels of DNA methylation and histone modifications having been connected to increased levels of inflammatory proteins in affected bone joints. Epigenetics has also been linked to a range of neurodegenerative diseases such as Alzheimer’s, Parkinson’s and Huntington’s. Epigenetic changes have been observed in mental retardation disorders with severe intellectual disabilities, such as Fragile X, Rett, Prader-Willi and Angelman syndromes. The last few years have seen increasing interest in investigating the epigenetic dimension of neuropsychiatric diseases such as schizophrenia, bipolar disorder and depression. The reversible nature of epigenetic traits makes them an attractive target for therapeutic interventions, especially in the context of precision medicine, and for the prediction of disease. Precision medicine involves the collaboration of many disciplines, from molecular biology and chemistry to physics and computational science; it requires a seamless transition from basic research to translation medicine and on to the clinic. The interdisciplinary approach of EpiGeneSys has greatly contributed to the continuing development of technologies and analysis methodologies which are necessary to advance the understanding of disease, particularly for sequencing techniques such as ChIP-seq, footprinting techniques (DNase-seq, ATAC-seq) with inference of nucleosome distributions, bisulfite sequencing in the area of DNA methylation as well as chromosome conformation capture. Single cell methods have added the power to understand heterogeneity of the genome, transcriptome and epigenome which may underlie disease progression and ageing. Research conducted in the context of EpiGeneSys has added to our knowledge about how epigenetic patterns can serve as biomarkers for the diagnosis of disease, the early identification of patients at risk of a disease and to trace or predict the efficiency of particular treatments in patients. This helps the stratification of patients into groups according to their type of response to a treatment. To give a short example: Chromatin structure influences the efficiency of many cancer therapeutics and resistance mechanisms may be susceptible to alterations in chromatin organisation. Work carried out in the framework of EpiGeneSys and supported by an EpiGeneSys Small Collaborative Project Award through the EpiGeneSys Training Programme points toward chromatin regulators as biomarkers for drug response as well as therapeutic targets to sensitise patients toward docetaxel and combat drug resistance (Gurard-Levin, Zachary A., et al. "Chromatin regulators as a guide for cancer treatment choice." Molecular cancer therapeutics (2016): molcanther-1008.)
EpiGeneSys views basic research and applied research as being complementary to each other and therefore both important and essential, a dichotomy basic – applied research should be avoided and instead the continuum and non-linear nature of research be recognised. The increasing number of cancer therapies would not have been possible without publicly funded research supporting investigation of cellular mechanisms. Improved communication with policy makers is needed to draw attention to the contribution that basic research makes to applied science.
The network has spent considerable effort on showcasing the importance of investigating epigenetic mechanisms not only in contributing to scientific advancement but also as a crucial factor for the development of new therapeutic interventions, one example for this is the exhibition #Hashtag Visions of Epigenetics, which was addressed to the general public. Investigating the molecular basis of life has led and will again lead to many game-changing breakthroughs.
As an important consideration for science policy makers it should also be mentioned that EpiGeneSys through its interdisciplinary approach and its contribution to all levels of research, from elucidating basic molecular mechanisms to developing new technologies and methodologies as well as providing novel types of diagnostic and therapeutic approaches plays an important role in contributing to the economy through a process of research leading to deployment and subsequently driving economic growth. A case in point is the founding of the company Cambridge Epigenetix in 2013 based on sequencing technology developed by EpiGeneSys supported research. The company has recently attracted substantial funding and currently employs about 30 research and development scientists.

3.4 Impact on training and career development in Europe

EpiGeneSys has had major impact on this important area at many different levels, which will help to ensure the future competitiveness of the European science education and training system, research skill base, and development of new job markets. The combinations of skills in quantitative methods, modelling, epigenomics and epigenetics, as well as data management will be particularly sought after in academic research, industry, higher education, teaching and public engagement and education job sectors. To prepare young scientists for these challenges of the future, the network has provided a large training scheme in epigenetics and systems approaches for undergraduates, PhD students and postdoctoral researchers, which has advanced the mobility of wet to dry as well as dry to wet research, and their reciprocal integration. Around 850 participants benefited from the training events organized by EpiGeneSys, among them ten workshops, eight science courses, two summer schools, 4 rounds of open calls for collaborative projects joining epigenetics and systems biology and 600 days of lab exchanges. The majority of these training events focused on generating a sound knowledge base for furthering our understanding of epigenetic phenomena using the approach, tools and technologies of systems biology. There were also opportunities for training in data base management, quality control, and technology development. This comprehensive programme has ensured a lasting effect on how a future generation of scientists will address new research questions. The RISE1 scheme, which integrated 20 junior group leaders in the network, has developed a strong effect through its varied approach: it supported the young PIs during the crucial first years of an independent research career through providing networking opportunities, peer-coaching, the possibility to form new collaborations with colleagues (which would otherwise not have happened), enhanced visibility and exposure and the chance to validate scientific ideas with leaders in the field. Several of the RISE scientists have since been appointed to senior group leader positions in leading European Institutes. The RISE1 programme provided training opportunities, not just through the training activities of the network, but also a tailored workshop on how to meet the challenges of being the leader of a team. It aided the integration of epigenetics and systems biology through the fact the successful candidate had to propose a project that combined both disciplines, which firmly established the young group leaders at the core of the network and placed their projects at the forefront of the scientific goals of the project. Very importantly, the EpiGeneSys community has grown to be an embodiment of the best virtues of European integration, collaboration, and free movement thus majorly contributing to a European led philosophy of how science is best done.

3.5 Impact on society and public debate

Today, with modern European society moving continually towards a “knowledge society”, knowledge acts as the key contributor and major creative force underlying social development, but at the same time, uncertainty about risks and unpredictability of modern technologies is growing. In recent years, the media have increasingly contributed to the popularity of the term "epigenetics"; many articles and documentaries have fuelled the imagination of the audience. Although the characteristic epigenetic status of the genome in most cells of the human body is believed to be remarkably robust, there are indications that lifestyle, such as diet, or exposure to pollutants, may have an impact on the activity of our genes through disruption of epigenetic marks, and lead to undesirable effects, not just for us, but perhaps even for our children and grandchildren. However, this still remains to be shown. Healthcare is experiencing a paradigm shift towards precision medicine, which aims to provide the right treatment, for the right person, at the right time based on genetic, epigenetic, environmental, and lifestyle data of the patient. It is necessary for all citizens and stakeholders to fully understand the concept to be able to participate and exploit the full benefits. Truly individualised care has to be science-based and build a bridge from basic science to treatment. These considerations and the availability of epigenetic biomarkers and, in the future, of whole epigenomes of individuals and single cell information (together with their genetic information) will result in new ethical debates, new rules and policies in the public sector and perhaps new ways of viewing inheritance that transcend the current ones based solely on natural selection. EpiGeneSys has created dialogue platforms for all stakeholders, to discuss the potential of biomedical research and show a realistic outlook into its future, while giving mindful attention to the public’s fears and objections raised through fast scientific developments that are hard to grasp for a lay audience. In order to make scientific knowledge durable and of sustainable impact for humankind and to contribute to informal, lifelong learning, public education in science and its technological and medical applications is of critical importance. As a primary objective EpiGeneSys has designed an informative and engaging web site including articles, videos and podcasts. Website analytics have shown that the section of the website addressed to the general public has attracted considerably more visitors than the scientific one, which serves as indication that the general public is interested in the subject and there is a need for producing material to inform the lay public about epigenetics and its interdisciplinary approach. Supported by the website, EpiGeneSys has engaged with the public in manifold ways, seeking to reach citizens, who up to date have remained removed from science. In a truly transdisciplinary effort, EpiGeneSys has established links with music, design, visual or performing arts.

EpiGeneSys has engaged with the public:
- via interactive art workshops/exhibitions, such as the events in Barcelona (November 2013), Paris (#Hashtag – Visions of Epigenetics, May 2015) and Edinburgh (June 2015).
- via events and lectures aimed at the general public (e.g. Epigenetics: Myths, mysteries and molecules in Cambridge in December 2013)
- via print media, stand-alone articles or special editions on epigenetics in magazines like Biofutur (monthly magazine in the field of Life Sciences distributed in France, Belgium, Switzerland and Canada) or the book “revisions” about the EpiGeneSys Art & Science programme
- via radio: interviews with scientists
- via film: two video clips explaining scientific principals via animations, a video about epigenetics & systems biology, a video about training, videos of scientific lectures, interviews with scientists about EpiGeneSys and a 20-minute video about the achievements of the network (available on the EpiGeneSys website), movie clips about art & science. A number of EpiGeneSys scientists have participated in the making of the ARTE documentary Les nouveaux secrets de notre hérédité/ Epigenetik - Sind wir Gene oder Umwelt.
Through its initiatives, EpiGeneSys has made significant steps towards greater transparency and improved communication of scientific research and its implications for the health and future of all European citizens.

List of Websites:
www.epigenesys.eu

almouzni@curie.fr
dorthe.nickel@curie.fr