L’épigénétique : « Ensemble des processus qui déterminent comment le génotype détermine le phénotype » dans les années 40, puis un peu plus tard « ensemble des phénomènes d’héritabilité qui ne sont pas strictement dépendants de la séquence primaire de l’ADN », la définition de l’épigénétique est toujours en devenir.
Liste des participants
Geneviève Almouzni (France), Frédéric Bantignies (France), Denise Barlow (Austria), Giacomo Cavalli, organisateur (France), Vicky Chandler (USA), Vincent Colot (France), Claude Desplan (USA), Rebecca Doerge (USA), Denis Duboule (Switzerland), Hugues Roest Crollius (France), Anne Ferguson-Smith (UK), Michel Georges (Belgium), Edith Heard (France), organisateur, Amar Klar (USA), Rob Martienssen (USA), Eric Meyer (France), Vincenzo Pirrotta (USA), Minoo Rassoulzadegan (France), Wolf Reik (UK), Shahragim Tajbakhsh (France), Geneviève Thon (Denmark)
by Frédéric Bantignies
21 – 26 avril 2008
A meeting aimed at deciphering some of the secrets of Epigenetics took place at the Treilles, from April 21st to 26th. This intimate meeting was organized by Edith Heard, from the Curie Institute, Paris and Giacomo Cavalli, from the Institute of Human Genetics, Montpellier. The organizers put together an interesting list of scientists (epigeneticists, developmental biologists, geneticists…see the list of participants above), who fitted together well, in a very relaxed and open atmosphere. The meeting was organized in 8 independent sessions (listed at the end), including a round table for open discussions. Most aspects of Epigenetics were covered, from molecular to developmental, evolutionary, inheritance aspects, in many model organisms, including yeast, paramecia, drosophila, mice, sheep, as well as plants.
As an introduction, Giacomo Cavalli gave a historical perspective on the definition of Epigenetics. The term was first coined by Conrad Waddington in 1942. He used the term Epigenetics for the stochastic decisions taken by cells as they move toward their ultimate fate. His model was illustrated by a ball rolling down a hill via different paths (valleys), ending with a particular fate. Approximately, 50 years later, in 1994, Robin Holliday defined Epigenetics as heritable changes that do not involve alterations in DNA sequence. However, these days the uses of the term Epigenetics are so diverse that its modern definition has become rather vague. The Treilles meeting started on the establishment of this fact!
In mammals, two favorite “Epigenetic Models” that have emerged are genomic imprinting and X-chromosome inactivation. Denise Barlow opened up the first scientific session by presenting her recent work on the Igf2r imprinted locus, which depends on the activity of the Air non-coding RNA. She showed that imprinting does not necessarily mean the complete silencing of one parental allele, but rather up-regulation of one allele, the maternal allele in the case of Igf2r, compared to the other one. Specific heterochromatin marks and DNA methylation appear to be involved in this process and Igf2r locus is still continuing to provide important insights into the way in which a large locus can be regulated monoallelically. The next speaker was Anne Ferguson-Smith who described her work on another imprinted locus, Dlk1 and the fascinating evolutionary considerations underlying the imprinting of this region. She also described the identification of a new factor, Zfp57, which appears to be important for protecting specific imprinting regions from genome-wide DNA demethylation which occurs after fertilization in mammals. Related to imprinting, Edith Heard proposed that a maternal imprint, marking the non-coding Xist RNA locus as silent, as well as a predisposition of the paternal allele of Xist to be active, could constitute the imprint for paternal X-chromosome inactivation (Xi) in mice, which occurs early on during embryogenesis and is maintained in the placenta. However, in the embryo, where random X inactivation takes place, and for which ES cells are a model system, imprinting does not seem to be involved. Instead, homologous pairing events (chromosome “kissing”) between specific parts of the X inactivation centre may explain the monoallelic regulation of the Xist gene. Edith speculated that this type of “pairing to assess ploidy” might represent a more genome-wide process, especially considering that up to 15% of autosomal genes may have a random, monoallelic expression pattern in humans. Another important epigenetic model is the Polycomb memory system in Drosophila. Giacomo Cavalli and Vince Pirrotta presented their recent data on this cell memory system, mostly based on their genome-wide binding profiles of Polycomb and Trithorax proteins. Giacomo presented interesting data indicating that recruiters cannot explained the recruitment of PcG proteins on chromatin, but rather the ratio between two of these recruiters, PHO and PHOL, could largely influence this recruitment. Vince insisted on the fact that the PcG repression is a dynamic process that is neither stable nor permanent, probably involving chromatin “bivalent states”, although this still needs to be proven.
The next session focused on Epigenesis and Developmental patterning. Claude Desplan described data supporting the idea that “bi-stable loops” stabilize the cell fate of photoreceptors during eye development in Drosophila. This provided a stimulating perspective on epigenetics, beyond chromatin marks. Wolf Reik presented interesting data on DNA methylation profiling in different mouse cell lineages, and showed that the status of a few specific genes could be responsible for a positive feedback loop to reinforce the expansion of specific stem cell lineages, such as the trophoblast lineage, during early development. During this session, Denis Duboule also described the importance of “genomic landscapes” in Hox gene clusters. These highly organized landscapes, which harbour the Hox clusters, are characterized by complex long-range interactions between enhancers and promoters, in order to precisely control the levels of expression of Hox genes in space and time during mouse development. Confronting the genomic and epigenomic landscapes of these regions, as well as their organization in the nucleus, is the next challenge in order to gain insight into the regulatory events underlying such precise gene expression patterns.
In the session on Epigenetics and Genome evolution, Anne Ferguson-Smith raised the issue that imprinting may be a relatively recent event, allowing adaptation to crucial developmental cues, such as energy metabolism efficiency in the case of the Dlk1 locus. Therefore, instead of being a sign of “parental conflict”, one of the favored hypotheses for why imprinting evolved, imprinting may be a tool to optimize the dosage of critical genes. Wolf Reik made the important point that invasive placentation may increase the selective pressure for imprinting in mammals, leading to the emergence of larger and more complex imprinting clusters in eutherians. On this evolutionary theme, Hugues Roest Crollius described work examining genome evolution using different approaches. These included the study of genome compaction in Tetraodon nigroviridis, the smallest vertebrate genome. He also described his analysis of vertebrate ancestral genomes to reconstitute chromosomal organization and of recent genomes to identify regions that resist polymorphism. Finally, he described an exciting new venture attempting to reconstruct the ancestral vertebrate genome. Our understanding of epigenetics could clearly benefit hugely from the availability of ancestral genomes, much as paleontology profits from fossils. Denis Duboule also made the rather provocative point that the highly organized Hox clusters that are found only in higher vertebrates, may have evolved from an ancestral disorganized cluster. Under evolutionary pressure, these clusters may have become highly organized because they acquired “global gene regulation”, via compaction and duplications, which allowed for even more functional diversity. Finally, Claude Desplan described an elegant series of experiments demonstrating how different insects can achieve segmentation in different ways. He proposed that “short germ-band” insects were an ancestral stage from which “long germ-band” insects, such as Drosophila, evolved, and the evolutionary forces that may have driven this were discussed.
A session was also devoted to non-mammalian “Epigenetic Models”. One fascinating model organism presented by Eric Meyer is the ciliate, Paramecium, which has both a germ line micronucleus and a somatic macronucleus. The germ line genome of ciliates is extensively rearranged during development of the somatic macronucleus. Numerous sequences are eliminated, while others are amplified to a high ploidy level. Eric described the remarkable strategy used in this organism to target DNA elimination from the macronucleus. This process is based on scanning of the macronucleus genome (or transcriptome) by RNA molecules that are produced by the germline micronucleus. This RNA-mediated genome scanning process appears to involve components of the RNAi machinery. Another classic epigenetic model is the process of paramutation in plants. Paramutation involves an interaction between two alleles of a single locus, resulting in a heritable change of one allele that is induced by the other allele. Vicky Chandler described an example of paramutation (the B locus) in maize, where this process was first discovered, and showed recent data demonstrating that components of the RNAi machinery and small RNA molecules are involved in the maintenance of this process, though probably not in its establishment. The manner in which paramutation is actually set up during early development remains a mystery. A well studied model for the role of the RNAi machinery in epigenetic phenomena concerns the silencing of heterochromatin in S. Pombe. The paradox that “silent” heterochromatin must actually be transcribed was discussed, and Rob Martienssen provided an explanation for this: he showed that transcription of heterochromatin actually occurs at early S-phase, when repressive marks and factors are under-represented at the loci he examined. Thus a dynamic cycle of transcription, small RNA production and targeting of silencing complexes and histone marks to the locus must during the cell cycle.
A session was dedicated on possible mechanisms leading to the perpetuation of chromatin states during replication and cell division. Geneviève Almouzni presented her recent data concerning the role of Asf-1, a chaperon protein for the H3-H4 histones. This factor could serve as a bridge between H3-H4 dimers and MCM replication proteins for the reloading of histones into the daughter molecules after the passage of the replication fork. This function could be important for the maintenance of chromatin marks, i.e. the memory of chromatin states during cell cycle. Geneviève Thon mentioned that at the mating-type locus in S. Pombe, the silencing that is dependent on Clr4, a histone methyltransferase, can be released upon temporary removal of this enzyme, but that silencing is not immediately re-established, indicating some form of memory of the induced, de-repressed state. She also described a theoretical model to understand the dynamic interactions between nucleosomes containing acetylation and methylation epigenetic marks. Sharagim Tajbakhsh described studies on asymmetric cell division in muscle stem cells. Interestingly, in a few cases, clear asymmetrical segregation of the DNA template could be observed, which could confirm the “immortal DNA strand” hypothesis proposed by Cairns (Nature, 1975). This phenomenon was proposed as a possible means to protect the stem cell genome from DNA replication errors. In addition, it could help to preserve epigenetic signatures present in stem cells, something that Sharagim is currently investigating. Amar Klar also discussed the phenomenom of asymmetric DNA strand segregation, for which ther is evidence both in S. Pombe, at the mating-type locus, and in the mouse. He discussed data from published studies that he suggests provide evidence for this process in the mouse, but clearly this process is likely to be complex and chromosome specific.
During a session on transgenerational inheritance, several models were discussed. First, Vicki Chandler presented the fact that the level of heritability of the paramutated phenotype of the B allele in maize is dependent on the numbers of repeated elements that induce this phenotype. Under a certain number of these elements (<5), the heritability of paramutation is incomplete, and this might reflect quantitative difference at the level of chromatin and DNA methylation. Vincent Colot also described some interesting aspects of epigenetic inheritance in Arabidopsis. He found some genomic regions that can be remethylated after the removal of ddm1 and its reintroduction, and showed that this remethylation seems to be dependent on the RNAi machinery. He proposed that this could reflect a mechanism for correcting changes (or errors) in epigenetic states at certain sequences in plants, where heritable epimutations are common and may play an important role in controlling transposable elements and associated sequences. In mammals, it is thought that epigenetic states are generally not heritable across generations and only a few examples of meiotic heritability have been reported. Minoo Rassoulzadegan previously reported one such locus (Kit) that shows such epigenetic heritability, and she showed that transmission somehow relies on an RNA intermediate. In this workshop, she presented new data involving the injection of micro RNAs (miRs) into fertilized eggs. The injected miRs induced the up-regulation of their targets, and produced specific phenotypes in the embryos and mice, such as cardiac hypertrophy in the case of miR-1, for which Cdk9, a key regulator of cardiac growth, is a target. Strikingly, they also found that the production of miRs is sustained in these mice in the sperm nucleus, and therefore, could be responsible for the transgenerational inheritance of these phenotypes that they observed over at least one generation. Although these data support the hypothesis of RNA-mediated inheritance, it is unclear whether the RNA, or the epigenetic changes it induces, are propagated during embryogenesis and into the germ line. Frédéric Bantignies went on to describe evidence in Drosophila that the specific nuclear organization imposed by the Polycomb-group proteins could influence the maintenance of repression states at Hox genes during development. This raised the important issue of the role that nuclear organization might play in the heritability (mitotic in this case) or epigenetic states. Finally, Amar Klar is trying to understand the molecular basis of “Left-Right” brain asymmetry. He gave an interesting and provocative talk proposing that genes involved in asymmetric strand segregation could be involved in this process and is currently trying to test this hypothesis.
The last day of the meeting included a session dedicated to Epigenetics/Epigenomics and the integration of epigenetic information. Michel Georges described some of his recent work concerning the sheep Callipyge phenotype, which is related to the Dlk1 imprinted locus. Vincent Colot and Denise Barlow presented work on epigenetic landscapes in Arabidopsis and Mouse cells, respectively. These landscapes revealed very interesting features. In Arabidopsis, surprisingly, the H3K9me3 chromatin mark is on active genes like H3K4me3, and it extends into the gene. In contrast, the H3K9me2 mark is at heterochromatin. In mice, 17% of chromosome 17 was found to carry the H3K27me3 chromatin mark, which forms huge blocks of up to 1Mb. The blocks are gene-poor, LINE-rich regions and drop abruptly at active genes. Denise suggested that this could be related to metaphase chromosome organization and banding patterns, where LINEs are found in condensed chromatin bands, while interbands would be the SINE-rich, more active genic regions. The meeting ended with the talk of Rebecca Doerge, who explained different statistical methods that could be used to integrate Epigenomic results. Using Statistical Meta-Analysis, a discipline in itself, they can combine results from different experimental designs, such as expression data, DNA methylation and histone methylation profiles. This clearly is becoming very important, given the exponential amount of epigenomic data available.
The concluding remarks of this meeting were organized as a round table animated by Edith Heard and Giacomo Cavalli. This took place outside, on the terrace on a sunny afternoon. During this round table discussion, various definitions of epigenetics and heterochromatin were discussed. No clear consensus for new definitions became apparent – probably reflecting the rapid evolution of the term “epigenetics”! The notion of heritability during replication was also raised. It was proposed that a compartmentalized nucleus would be a homeostatic machine that may self-propagate functional states during replication. Also, the semi-conservative mode of replication is not proven but remains possible, based on the fact that the pre-deposition form of H3/H4 is a dimer. This mode of replication might apply to some specific situations. Another topic that was discussed was why plants seem different from animals in term of the heritability and plasticity of epigenetic phenomena. Actually, several examples reveal that they might not be so different, as this workshop highlighted. Clearly however, the relatively “immobile” life of a plant compared to animals has implications for the processes plants can use to survive and evolve! Next, the issues of what can drive the emergence of imprinting, what pressures maintain it, and how epialleles might drive adaptability and speciation, were discussed. One theory is that imprinting may have evolved from “parental conflict” (where the interests of the mother – to distribute her resources equally between multiple offspring from different fathers, and those of the father – to ensure his offspring obtain maximal maternal resources – are in competition). However, several other theories need to be considered also, including the idea that imprinting may be a tool to optimize expression of highly dosage-dependent genes. Indeed, one idea proposed by Anne Ferguson Smith is that imprinting usually arises as a maternal process – a fluctuation in gene expression that results in monoallelic expression at a low frequency but that can be selected on. Clearly, different classes of imprinted genes (eg those expressed in the placenta only, versus those expressed in the embryo and placenta or those only expressed in adults) may have evolved imprinting for different reason and also in different ways. Another point that was discussed was that imprinted loci may arise by “hitchhiking” existing molecular mechanisms. Epialleles could generate variation allowing an isolate to survive, and then perhaps enable a mutation to arise to consolidate the phenotype. Alternatively, epialleles may simply allow a genome to survive during hard times.
Finally, one of the recurrent themes during this round table was that non-coding RNAs would certainly be major players in all kinds of epigenetic phenomena!