The one gene – one protein dogma has been very successful in identifying the mutant genes and dysfunctional proteins associated with a range of inherited conditions, from muscular dystrophy to Alzheimer’s disease. The approach has been less useful with many diseases affecting the nervous system and behaviour. Over the past few years it has become increasingly apparent that gene transcription and translation into proteins is only a small part of a complex story. Rather like the queen bee in a beehive, each gene is embedded in a network of molecules, now known as the epigenome, that links the environment to the mechanisms regulating gene expression. At this IPSEN meeting, researchers from leading laboratories will explore how knowledge of epigenetic regulation is contributing to understanding the links between nervous system function and behaviour in health and disease. The meeting has been organised by Paolo Sassone-Corsi (University of California at Irvine, Irvine, USA) and Yves Christen (Fondation IPSEN, Paris, France).
The standard picture of the naked DNA helix is misleading: in cells, the DNA ribbon is wound round small spherical proteins termed histones to form the complex known as chromatin. The relationship between DNA and histones is one part of the complex of mechanisms, known as epigenetic regulation, that organise the transcription of genes: of the roughly 20,000 genes in every adult human cell only a small percentage are transcribed and the functions of each differentiated cell are determined by which set of genes this is. Although the mechanisms of epigenetic regulation are still being unravelled, it is already clear that they enable very subtle changes in gene transcription that fine tune the cell to current needs and that they often coordinate several or many genes involved in a particular function, such as metabolism and synaptic transmission (Sassone-Corsi; Abel). Long-term alterations in epigenetic regulation may even be inherited without any changes to the DNA (Isabelle Mansuy, University of Zurich, Zurich, Switzerland; Michael Meaney, McGill University, Montréal, Canada).
One important mechanism of epigenetic regulation is chromatin remodelling through acetylation and methylation of histones: enzymes alter the shape and function of particular histones and their relationship to the surrounding DNA by attaching or removing one or more acetyl or methyl groups on specific amino acids. Much of the subtlety and diversity of epigenetic regulation derives from the position and number of these attached groups. Methyl groups also attach to DNA itself and lower the probability of a gene being transcribed.
The transcription of many genes is stimulated by transcription factors, molecules produced in the cytoplasm in response to external or internal signals that link changes in the environment or in demand to the production of new proteins. These factors trigger transcription by binding to a special region of the gene, the promoter, an interaction that requires various helper molecules. One class of helper, involved in most transcription processes, is the nucleosome remodelling factors, which are enzymes that give flexibility to the chromatin. According to their precise structure, nucleosome remodelling factors can target specific genes and either activate or repress them (Peter Becker, Ludwig-Maximilians-Universität München, Munich, Germany). In yeast cells, nucleosome remodelling factors are one of several epigenetic pathways regulating the aging of cells (Shelley Berger, University of Pennsylvania School of Medicine, Philadelphia, USA). Another pathway uses Sirtuins, a type of histone deacetylase that removes acetyl groups from histones in the end region of the chromosome, the telomere, long known to have a role in aging. These multiple controls are typical of the complexity of epigenetic regulation. In neurons, a specific histone methyl transferase has been identified that has a key role in regulating the expression of genes that code for neuron-specific proteins in adult brain; this enzyme seems to participate in at least two neuron-specific pathways (Anne Schaefer, The Rockefeller University, New York, USA).
The circadian clock, the internal mechanism that couples physiological processes to the 24-hour cycle of day and night, is a good example of epigenetic coordination: the many genes that control metabolism are coherently regulated by the influence of clock proteins through histone acetylation and methylation (Sassone-Corsi). A homologue of the yeast Sirtuins provides such a link and a specific methyl transferase has been found that is rhythmically recruited to the promoters of clock genes.
Synaptic transmission and plasticity are also functions requiring the coordination of many intracellular processes and so it is not surprising that epigenetic mechanisms play a large part in their regulation (Jean-Pierre Changeux; Institut Pasteur, Paris, France; Ted Abel, University of Pennsylvania, Philadelphia, USA; David Sweatt, University of Alabama at Birmingham, Birmingham, USA; Li-Huei Tsai, Massachusetts Institute of Technology, Boston, USA; Eric Nestler, Mount Sinai School of Medicine, New York, USA). Chromatin remodelling through histone acetylation and methylation is crucial for various types of memory formation and storage in animal models (Abel; Sweatt; Tsai) and in the modified neuronal plasticity that leads to abnormal behaviour in drug addiction (Nestler). Identifying sites where chromatin remodelling has occurred as a result of exposure to drugs is aiding in the identification of affected genes. Addictive drugs may also have a direct effect on chromatin remodelling enzymes.
Similarities between the mechanisms being discovered in the regulation of synaptic plasticity and memory and those found in embryonic development and cell differentiation indicate that these processes have a common origin (Sweatt). On a more theoretical level, the interaction between environment and the state of the synaptic network that leads, through epigenetic regulation, to the stability of synapses may have contributed significantly to the increase in the complexity of connections in the brain in primate evolution, while in humans the long post-natal period allows for social and cultural inputs to brain development (Changeux).
Mutations affecting the functioning of proteins that participate in epigenetic regulation may of course result in incorrect or absent activation of genes essential for normal neuronal function and mental health (Schaefer; Tsai; Thomas Bourgeron, Institut Pasteur, Paris, France; Adrian Bird, University of Edinburgh, Edinburgh, UK; Lisa Monteggia, University of Texas Southwestern Medical Center at Dallas, Dallas, USA). Defects in synaptic transmission are being identified in some forms of autistic spectrum disorder, which may in some cases be associated with abnormal sleep homeostasis and changes in the melatonin/serotonin pathway that is integral to sleep regulation (Bourgeron). Rett syndrome, a profound autistic condition, is characterised by reductions in neuron size, the complexity of connections, neurotransmission and synaptic plasticity (Bird; Monteggia). Central to the condition is the loss of a protein known as methyl CpG binding protein 2 (MeCP2), which is very abundant in normal neurons. MeCP2 binds to particular methylated sites in the promoters of many genes, alters chromatin structure and is involved in histone acetylation; its loss affects the transcription of various genes required for synaptic plasticity and learning (Monteggia; Sweatt). In a mouse model, Rett-like symptoms have been reversed by activating a silent Mecp2 gene (Bird).
A challenging aspect of epigenetic regulation is that changes in the epigenome may be transmitted to future generations if they are present in the germ cells. Stress in early life is known to be a risk factor for later psychiatric illness, which in some families seems to heritable (Mansuy). In mice, chronic and unpredictable maternal separation results in depressive and impulsive behaviour that has been transmitted to offspring. The early trauma is associated with persistent changes in DNA methylation in several promoters in germ cells and in the brains of the offspring, with altered gene expression in the brain when these mice become adult. The quality of maternal care also affects the transcription of genes implicated in the stress response, particularly in the hypothalamus-pituitary-adrenal axis (Meaney). In rats, changes in DNA methylation have been found across the genome in response to variations in maternal care, including in genes that regulate neural development.
The contributions to this meeting provide a sample of the huge palette of epigenetic effects. As with any cellular regulatory mechanism, the potential for the development of diagnostic and therapeutic tools is huge but the complexity of the epigenome and the wide-ranging consequences of even small changes should be a warning that pitfalls may lie ahead on the road to the clinic.
The one gene – one protein dogma has been very successful in identifying the mutant genes and dysfunctional proteins associated with a range of inherited conditions, from muscular dystrophy to Alzheimer’s disease. But this view of the genome is inadequate to explain the variability between organisms and the different ways in which they respond to their environments. Rather, it has become increasingly apparent that genes are embedded, rather like the queen bee in a beehive, in a network of molecules, now known as the epigenome, that regulates gene expression to meet environmental demands. At the 19th annual IPSEN colloquium in the Neuroscience series, held in Paris on April 18, 2011, researchers from leading laboratories explored the epigenetic links between nervous system function and behaviour in health and disease and the new directions that are opening up for therapy. The meeting has been organised by Paolo Sassone-Corsi (University of California at Irvine, Irvine, USA) and Yves Christen (Fondation IPSEN, Paris, France).
Although the cellular exploration of epigenetic mechanisms has only recently begun, the concept dates back to C.H. Waddington, who in 1942 defined epigenetics as the interaction of genes with their surroundings to produce a phenotype, that is, the precise form of the adult body (Jean-Pierre Changeux; Institut Pasteur, Paris, France). Waddington envisaged an ‘epigenetic landscape’, a slope formed of bifurcating valleys, down which a ball rolls. The exact shape of the ridges and valleys, representing a combination of genetic and environmental influences that is different for each cell type or organism, determines where the ball ends up.
Today the term epigenetics is used in developmental biology to denote the regulation of cellular differentiation (Peter Becker, Ludwig-Maximilians-Universität München, Munich, Germany; Adrian Bird, University of Edinburgh, Edinburgh, UK); the organism’s accommodation to the environment in adult life (Shelley Berger, University of Pennsylvania School of Medicine, Philadelphia, USA; Sassone-Corsi; Ted Abel, University of Pennsylvania, Philadelphia, USA; David Sweatt, University of Alabama at Birmingham, Birmingham, USA), and the processes modelling the human brain to fit its social and cultural environment (Changeux). And evidence is accumulating that at least some environmentally driven epigenetic changes may be passed on to later generations (Michael Meaney, McGill University, Montréal, Canada; Isabelle Mansuy, University of Zurich, Zurich, Switzerland).
At the heart of epigenetics is the mechanisms that regulate the transcription of the genes coded in the DNA into messenger RNAs in response to the needs of the cell and the whole organism. The identification of some of the molecules involved in these communication and regulation processes is key to understanding how the epigenome works.
As cells differentiate into various types with specialised functions, the subset of genes being transcribed narrows to just those required for the specific needs of each cell, be it a skin epithelial cell or a neuron. Which genes are actively transcribed and which remain silent depends on a combination of molecular mechanisms: regulation of the way the DNA ribbon is packed into the cell nucleus, a process known as chromatin remodelling; and DNA methylation, the addition of a methyl group at particularly sites on the DNA ribbon, which in most cases prevents genes being transcribed. Both mechanisms depend on specific enzymes and regulators that are recruited to the DNA in response to signals received by the cell that activate complex signalling pathways in the cytoplasm and nucleus. In mature cells, similar mechanisms increase or decrease the activity of genes in response to metabolic demands or experience, or switch on silent genes required by changing conditions.
Packing it in and shutting it up
The packing of the DNA into the nucleus determines which stretches of the DNA are exposed to the transcription process. Each human cell contains over 2m of DNA, representing about 20,000 genes, fitted into a nucleus about 10m in diameter. This is achieved in an orderly fashion by winding the DNA ribbon round a series of structures known as nucleosomes, rather like a thread around cotton reels, to form the dense material called chromatin. In the lengths of DNA where genes are not being transcribed, the packing is very tight; where genes are active, the DNA ribbon is spooled out and becomes free of the nucleosomes (Becker). Molecular machines known as chromatin remodellers orchestrate this relaxation and condensation of the chromatin structure. They bind to a pioneer adaptor factor inserted into the DNA ribbon between two closely arrayed nucleosomes; once attached to the DNA, this complex slides the nucleosomes apart.
During differentiation, remodellers help to establish and maintain the integrity of chromatin allowing the genome flexibility to respond to developmental, metabolic and environmental signals. One developmental remodeller, chromatin accessibility complex or CHRAC, is involved in the spacing of nucleosomes and their higher-order packing: by regulating nucleosome spacing, CHRACs may lay the foundation for repressing genes that the differentiated cell does not need (Becker).
Chromatin plasticity and the availability of DNA for transcription is also promoted by changes in the shapes of the nucleosomes around which the DNA winds. Each has a core of four globular histone molecules (H2A, H2B, H3 and H4), with one or two tails sticking out of the chromatin. The conformation of the histone molecule and its packing in the nucleosome is modified by attaching acetyl, methyl or other groups to the amino acids that make up the tails and to a few amino acids in the globular part of the histone. The nature and position of the group determines whether the neighbouring chromatin is relaxed or condensed (Berger; Schaeffer). Attaching the side chains requires a range of enzymes; for example, histone acetyl transferases add acetyl groups, while histone deacetylases remove them.
One example of histone acetylation is in aging cells, where the chromatin structure is more relaxed than in young cells. In yeast cells, a histone deacetylase belonging to the sirtuin family, slows down aging by removing the acetyl group from a lysine in the tail of histone H4 (Berger). Sir2 is antagonized by an acetylase, Sas2. These enzymes are particularly significant because they are concentrated at the ends of the chromosomes, the telomeres, which unravel and wear away as the cell ages. Evidence for similar mechanisms in human aging is being examined, as enzymes homologous to Sir2 and Sas2 have been identified.
Genes in differentiated cells are also silenced by double methylation of a lysine on the tail of histone H3H meH
by the histone lysine methyltransferase GLP/G9a (Schaeffer). When GLP/G9a is experimentally inactivated in adult neurons, many non-neuronal genes are activated and transcription of genes involved in serotonin and dopamine synthesis and function increases. In mice, switching off GLP/G9a in postnatal neurons produced impaired learning, memory and environmental adaptation, all symptoms of a rare and severe mental retardation syndrome in humans who have a mutation in GLP.
Another way to silence genes is the direct attachment of methyl groups to DNA. The methyl groups are located on specific sequences of bases on one strand of the DNA, in the promoter region, the section of the gene sequence that regulates translation. Promoters contain runs of cytosine and guanine bases connected by a phosphate group that are termed CpG islands. Methyl groups attached to these CpGs bind repressor complexes that prevent transcription – about 70% of CpGs in the human genome are in this state (Bird). In neurons, methyl-CpG-binding protein 2 (MeCP2) is an important repressor, binding directly to CpG islands and acting as an anchor for other corepressors and a histone deacetylase. The actions of MeCP2 ran like a refrain through the meeting.
In most cells, a fifth histone, H1, acts like a catch, securing the turns of the DNA ribbon round the core, but in neurons MeCP2 seems to substitute for about half the H1s. Loss of MeCP2 disrupts epigenetic control in mature neurons, resulting in twice as much histone acetylation and altered transcription. In humans, mutations in MeCP2 causes Rett syndrome, a severe developmental disorder on the autistic spectrum seen in about 1:15,000 girls (Bird; Lisa Monteggia, University of Texas Southwestern Medical Center at Dallas, Dallas, USA).
Fundamental to biological function is the entraining of metabolic pathways to the cycle of day and night, known as the circadian clock. As this requires the rhythmic activation and shutting down of the production of about 15% of the cell’s proteins, it is not surprising to find that epigenetic mechanisms are involved: in mammals, chromatin in ‘clock central’, the suprachiasmatic nucleus in the hypothalamus is remodelled as light intensity changes and modifications of histone tails act as metabolic sensors (Sassone-Corsi). The CLOCK protein, a transcription factor driven by light/dark changes, stimulates histone acetylation, which is balanced by the histone deacetyase SIRT1, a sirtuin related to yeast Sir2, that responds to metabolic signals driven by the circadian clock. The CLOCK-stimulated acetylation is driven indirectly by another transcription factor, MLL1, a methyl transferase originally discovered through its role in leukaemia, which adds three methyl groups to histone H3 in a circadian fashion. MLL1 also associates with CLOCK but only at specific times in the daily cycle, activating the acetylation of other sites on histone H3 and promoting gene expression. What drives MLL1 awaits further investigation.
Synapses and memory
Chief among the characteristics of the brain is the complexity of its connections and its plasticity. The connectivity pattern is laid down during post-natal development as dendritic trees are remodelled in response to the animal’s experience. This stable network is further modified throughout life as learning reinforces transmission at synapses that are regularly used and decreases the potency of those that are rarely active. The epigenetic programmes that maintain the dynamic state of the synaptic network are slowly becoming visible.
The repressor MeCP2 seems to be essential for synapse function. Although
wide-spread across the genome, its action is limited and precise, at least within experimental parameters. Neurons from mice lacking the mecp2 gene do not show large changes in gene expression or altered neuron morphology but careful electrophysiological analysis is revealing loss of synaptic transmission specifically at excitatory synapses (Lisa Monteggia, University of Texas Southwestern Medical Center at Dallas, Dallas, USA). Synaptic stimulation seems to decrease MeCP2 repression of genes coding for proteins involved in excitatory transmission, though these target genes are yet to be identified. Extending this to Rett syndrome, where MeCP2 is lacking, depressed excitatory transmission will alter the establishment of synaptic connectivity during the development of the nervous system.
It may well be that the epigenetic mechanisms controlling cellular differentiation are also at work in learning and memory, determining specific changes in synaptic function and connectivity (Sweatt). The transition from short-term, labile memory to long-term stable storage requires new protein synthesis, which results from activation of genes in response to transmission at specific synapses. Histone acetylation is one mechanism that promotes this transition (Abel). It is mediated by a pathway that stimulates a protein, CBP, to bind to a complex on gene promoters that stimulates histone acetylation, relaxes the chromatin and allows gene transcription. Mice lacking CBP have several memory deficits: they fail to associate electric shocks with a particular place and to recognise novel objects. These deficits can be reversed by inhibiting the histone deacetylase. The complex is which CBP binds is specific to certain target genes, including members of the nuclear hormone receptor family; one function of these is enhancing the production of the synaptic growth factor BDNF.
DNA methylation is required for synaptic plasticity, memory formation and spatial memory in the hippocampus (Sweatt). Several of the genes targeted participate in synapse stabilization, including BDNF and calcineurin, an enzyme that activates another transcription pathway. The hippocampus is only a staging post in laying down memories and the changes in DNA methylation disappear within a day. Longer-term storage requires stabilized patterns of synaptic connectivity in the cortex. In the anterior cingulate cortex, a shift in the methylation of the calcineurin gene promoter is essential for stablizing synapses and the maintenance of memories over a long period. Such a change may be an epigenetic ‘mark’ that identifies a neuron made receptive by experience.
Social and cultural experience
Looking up from the minutiae of synaptic mechanisms to the connectivity of the brain as a whole gives a wider view of epigenetic mechanisms at work in creating and maintaining brain plasticity (Changeux). It is startling to realise that while mice and humans have approximately the same number of genes, mice have only 40×106 neurons, compared to 50-100×109 in humans – a good illustration that genes provide just a blueprint for the developing nervous system. Epigenetic mechanisms become particularly important in shaping this basic plan, especially during the long post-natal period in humans. Differing social and cultural influences create individual differences and provide for extra-genetic cultural heritage. The learning of new skills can take over old circuits, even in adults.
The environment in which you are raised also affects your subsequent health and well-being: children who suffer neglect or abuse tend to develop both physical and mental health problems as adults (Meaney; Mansuy). Similarly, rat pups that do not get adequate maternal care have higher stress responses than those with more attentive mothers, although this can be reversed if pups are placed with mothers of the opposite tendency (Meaney). The numbers of receptors for the glucocorticoid hormone in the hippocampus are reduced in poorly mothered offspring and epigenetic regulation seems to be central to this effect. Attention stimulates serotonin release, which triggers an intracellular pathway, again involving CBP, which here binds to the promoter of the glucocorticoid receptor gene. The result is lower DNA methylation and increased histone acetylation, allowing increased binding of a specific transcription factor. Glucocorticoid receptors are also reduced in human victims of childhood abuse who have committed suicide.
So can nurture become nature? Evidence is accumulating for the inheritance of some stable epigenetic changes in gene transcription that result from individual experience. Poorly nurtured rat pups themselves pay less attention to their offspring and epigenetic markers in the genes coding for the estrogen receptor, the ‘nurturing’ hormone oxytocin and the neuromodulator dopamine seem to be transmitted to the offspring (Meaney). In a study where baby mice were exposed to unpredictable maternal deprivation and stressed mothers, they and their offspring in the following two generations show signs of depression, impulsive behaviour and impaired social skills (Mansuy). The possibility that these negative effects of early stress were transmitted soley through nurture is ruled out by mating each generation of stressed males with normally reared females. One of the signals involved in coping with and adaptation to stress is corticotrophin releasing hormone, mediated by type 2 receptors in the hypothalamus. Second and third generation offspring of the stressed male mice have fewer of these receptors, which seems to result from altered methylation of CpG islands in the promoter of the receptor gene in the father’s sperm and the offsprings’ brains. Genes such as mecp2 are also affected, hinting at more widespread modifications in gene transcription.
Epigenetics and pathology
Clearly, epigenetic mechanisms can play it both ways: ensuring the stable, long-term changes in gene transcription required by differentiated cells and at the same time allowing dynamic changes such as the circadian fluctuations of protein production. Because some epigenetic processes promote plasticity, there is hope that they can be harnessed in the treatment of pathological changes. One demonstration of this is the reversal of cognitive impairments and neurodegeneration in a genetically engineered mouse that shows all the pathological features of Alzheimer’s disease (Li-Huei Tsai, Massachusetts Institute of Technology, Boston, USA). Chemically inhibiting histone deacetylation restored learning and longer-term memory storage; synaptic density also improved. The search is now on for a more effective small-molecule histone deacetylase inhibitor.
Use of addictive drugs also leads to stable changes in synaptic transmission in some brain circuits, which are essentially a form of memory (Eric Nestler, Mount Sinai School of Medicine, New York, USA). In the nucleus accumbens, a brain area affected by addictive drugs, the genes whose transcription is altered by chronic cocaine use are being identified by a three-pronged mapping process that identifies first, genes that are being transcribed when exposed to cocaine; second, which of these show changes in acetylation and methylation; and third, among these, which are regulated by two transcription factors known to mediate aspects of addictive behaviour. One surprising discovery is that cocaine causes the relaxation of large parts of the chromatin in neurons in the nucleus accumbens by repressing methylation of histone 3, leading to a permissive state of gene regulation. The histone deacetylase SIRT1, also important in circadian cycles, is heavily involved, regulating the genes for several transcription factors involved in dendritic growth and induction and the stabilisation of synapses, including two already discussed: MECP2 and G9a. The circuitry is clearly complex but the recurrence of similar mechanisms in different contexts should facilitate unravelling it and help to open up novel ways of treating drug addiction.
A proof-in-principle that some epigenetic changes can be reversed has been demonstrated in a mouse model of Rett syndrome (Bird). The human mecp2 gene is on the X-chromosome, so Rett syndrome affects only girls who have one mutated gene but a normal one on their second X-chromosome – boys with a mecp2 mutation on their single X-chromosome do not survive. When mice with the mutant gene that had developed severe Rett-like behaviour were given another normal gene, their movement and breathing, two typical Rett symptoms, became almost normal. Although such gene substitution is unlikely to be feasible in humans, the experiment demonstrates that loss of MeCP2 function does not result in irrevocable brain damage, and so manipulations to increase the effectiveness of the normal gene on the other X-chromosome may be fruitful.
Unlike Rett syndrome, other disorders on the autistic spectrum do not have clear-cut genetics origins (Thomas Bourgeron, Institut Pasteur, Paris, France). The mutations that have been identified so far mostly affect synaptic stability and sleep–wake cycles but it seems likely that the behavioural variability seen in autistic disorders is the result of more than one mutation in each affected individual. A gene coding for the synaptic scaffold protein SHANK2 has been traced through several generations in some families but other mutations seem to be spontaneous. Disturbed sleep patterns may mean that autistic children do not have adequate opportunity to consolidate the information they receive when awake and so suffer from synaptic overload. Although epigenetic regulation has not yet been investigated in these conditions, the information on synaptic plasticity and circadian rhythms should provide leads to mechanisms worth investigating.
Given the infinite variation among humans – even identical twins are never 100% alike – it is hardly surprising that epigenetic regulation is turning out to be so complex but its study so fruitful. Pictures of the concerted changes in networked pathways are essential for a proper understanding but this will take time (Nestler). Ultimately, such pictures may provide on the one hand a more complete explanation of our humanity, on the other a map that will help with precision targeting of new therapies
To add back into AlzF version
Abel: CREB, a transcription factor that acts a molecular switch by binding to Cre, a response element in gene promoters (Abel). But this is not enough: once CREB binds to Cre, it attracts the CREB-
The contextual fear response