epigenetics, brain evolution and behaviour · review epigenetics, brain evolution and behaviour...

Available online at www.sciencedirect.com

ARTICLE IN PRESS

www.elsevier.com/locate/yfrne

Frontiers inNeuroendocrinology

Frontiers in Neuroendocrinology xxx (2008) xxx–xxx

Review

Epigenetics, brain evolution and behaviour

Eric B. Keverne a,*, James P. Curley b

a Sub-Department of Animal Behaviour, University of Cambridge, Madingley, Cambridge, CB23 8AA, UKb Department of Psychology, Columbia University, New York, 10025, USA

Abstract

Molecular modifications to the structure of histone proteins and DNA (chromatin) play a significant role in regulating the transcrip-tion of genes without altering their nucleotide sequence. Certain epigenetic modifications to DNA are heritable in the form of genomicimprinting, whereby subsets of genes are silenced according to parent-of-origin. This form of gene regulation is primarily under matri-lineal control and has evolved partly to co-ordinate in-utero development with maternal resource availability. Changes to epigeneticmechanisms in post-mitotic neurons may also be activated during development in response to environmental stimuli such as maternalcare and social interactions. This results in long-lasting stable, or short-term dynamic, changes to the neuronal phenotype producinglong-term behavioural consequences. Use of evolutionary conserved mechanisms have thus been adapted to modify the control of geneexpression and embryonic growth of the brain as well as allowing for plastic changes in the post-natal brain in response to external envi-ronmental and social cues.� 2008 Elsevier Inc. All rights reserved.

Keywords: Epigenetics; Genomic imprinting; Histone modification; DNA methylation; Brain; Placenta; Development

1. Introduction

There has been an exponential growth in research onepigenetics at the molecular level which has focussed onthe study of covalent and non-covalent modification ofchromatin, notably DNA methylation and the acetylation,methylation and phosphorylation of histone proteins thatinfluence overall chromatin structure and hence geneexpression [6,11]. Such epigenetic mechanisms maintaintight control over gene transcription in pluripotent cellsof all multicellular organisms. Chromatin structure is, how-ever, dynamic and can be influenced by the environment,raising the possibility that acquired changes through thephysical and social environments can produce lifetimechanges in phenotypes that may even be passed on to thenext generation.

In this paper, we discuss how both heritable and non-heritable epigenetic processes have been critical in shapingmammalian brain evolution and behaviour. We propose

0091-3022/$ - see front matter � 2008 Elsevier Inc. All rights reserved.

doi:10.1016/j.yfrne.2008.03.001

* Corresponding author. Fax: +44 01223 741802.E-mail address: [email protected] (E.B. Keverne).

Please cite this article in press as: E.B. Keverne, J.P. Curley, Epige(2008), doi:10.1016/j.yfrne.2008.03.001

that genomic imprinting, the epigenetically regulated par-ent-of-origin specific expression of genes, coadaptively reg-ulates the mammalian embryonic development andreproductive behaviour. We also consider how mammalsare able to demonstrate extensive plasticity in behaviourthroughout life as a consequence of environmental experi-ences inducing non-heritable changes in those epigeneticmechanisms regulating gene transcription in post-mitoticneurons. The ‘‘agouti” locus is one such example where,in mice, genetically identical parents with agouti genes indifferent epigenetic states produce offspring with differentand variable coat colour [111].

Epigenetics pervades all aspects of development includ-ing the brain. The mammalian brain is very special in thiscontext and differs from that of other vertebrates in that itdevelops intimately within a maternal environment bothpre-natally in-utero and post-natally during suckling. Thehuman brain is further exceptional in that developmentcontinues long beyond weaning and parts of the cortexundergo radical reorganisation during the post-pubertalperiod [95]. Human studies with monozygotic twins havelinked the environment to long-lasting epigenetic effects

netics, brain evolution and behaviour, Front. Neuroendocrinol.

mailto:[email protected]

2 E.B. Keverne, J.P. Curley / Frontiers in Neuroendocrinology xxx (2008) xxx–xxx

ARTICLE IN PRESS

on the phenotype in that disease susceptibility (e.g. schizo-phrenia, manic depression) is not concordant for geneti-cally identical twins [12]. Moreover, it has recently beenshown that young identical twins have similar amountsof epigenetically methylated DNA, but with ageing theydiffer significantly both in the amounts and the patterningof DNA methylation, and hence gene expression [32]. Epi-genetics is not unique to mammals since the whole develop-mental process of multicellular organisms, invertebrate orvertebrate, requires that developmentally important genesare sustained in an epigenetically silenced state in pluripo-tent cells, such as stem cells, in order to be activated later.

2. Epigenetic mechanisms

There exist a number of mechanisms regulating the com-plex spatial and temporal pattern of deoxyribonucleic acid(DNA) transcription, which are critical for cellular and tis-sue differentiation. Some of these mechanisms can beclassed as ‘epigenetic’, defined as those modifications toDNA or the nucleosome that do not alter the sequenceof nucleotides, but do modify the transcription of genes.Significantly, these modifications can be extremely stable,being transmitted in non-neural tissues from mother todaughter cells, and in post-mitotic neurons permanentlymodifying the phenotype. It is also becoming clear thatsome of these epigenetic modifications in the brain areresponsive to environmental factors, creating long-lastingstable effects on neural phenotypes and behaviour [45].

Within the cell nucleus, DNA is not freely moving but isfound within nucleosomes [52,35,66]. These compriseapproximately 147 nucleotide base pairs enveloped aroundan octamer of histone proteins (two each of H2A, H2B, H3and H4 proteins) (Fig. 1). A fifth histone protein, H1, linksindividual nucleosomes together with approximately 80nucleotide base pairs intervening. Nucleosomes, togetherwith other non-histone scaffold proteins and enzymes, suchas proteases and nucleases involved in replication andrepair, make up chromatin (Fig. 1). Chromatin exists sta-bly in broadly two main states, inactive (heterochromatin)and active (euchromatin), though in reality individualregions of chromatin are at any one time somewhere on acontinuum between these states. Heterochromatin is tightlypacked and restricts the access of RNA polymerase II toDNA thereby repressing gene transcription. Conversely,the DNA within euchromatin is relatively loose aroundthe histone proteins facilitating the access of RNA poly-merase II and other proteins of the transcriptional machin-ery and increasing gene expression. The transition betweenthese two states of chromatin can be induced through epi-genetic modifications to the individual histone proteins andDNA itself. At least eight different types of covalent mod-ifications of histone proteins can occur at their globulardomain, but the majority of these occur at several of theover 60 amino acid residues along the N-terminal tails thatprotrude outwards from the octamer [52]. Variability in thetypes of covalent modification that occur, the histones and


amino acid residues at which they occur, as well as theirinteractions with other transcription factors, coactivatorsand repressor molecules all combine to create variation inthe levels of transcription of specific genes.

Histone acetylation involves the transfer of an acetylgroup (cleaved from acetyl-coenzyme A) to the e-NH+

group of lysine residues [31]. The opposite reaction is his-tone deacetylation which involves the transfer of acetylgroups from histones to coenzyme A and the increasedwrapping of histone tails to DNA and decreased transcrip-tion factor to activate gene expression. This neutralises thepositive charge of the histone tail and reduces its attractionto the negatively charged phosphate backbone of DNA,thereby loosening the nucleosome and allowing access ofadditional transcriptional proteins and enzymes and thusincreasing gene transcription. Histone acetylation anddeacetylation are catalysed by histone acetyltransferases(HATs) and histone deacetylases (HDACs), respectively.Over 20 HATs have been identified with those regulatinggene transcription belonging to class A HATs (class BHATs regulate chromatin synthesis and nucleosome assem-bly in the cytoplasm) [43]. Moreover, several transcriptionfactors and coactivator proteins such as CREB bindingprotein (CBP) have intrinsic HAT activity and can there-fore increase gene transcription through their interactionswith DNA and their ability to recruit other HATs [69].There are three classes of HDACs (I, II and III) with thefirst two being most significant in mammals. Class IHDACs (Numbers 1, 2, 3, 8) have widespread tissueexpression, but class II HDACs (Numbers 4, 5, 6, 7, 9,10) have particularly high expression within muscle andthe brain [98]. Histone acetylation is also thought toincrease gene transcription by establishing a platform forthe recruitment of other chromatin remodelling processessuch as the SWI/SNF complex [69].

The other well understood histone epigenetic changescan be transcription activating or repressing based uponthe location and type of amino acid residue undergoingthe covalent modification. Histone phosphorylation canoccur at serine or threonine amino acids and usually acti-vates gene transcription through neutralising the positivehistone charge of chromatin and relaxing the nucleosome[18]. Phosphorylation of particular serine residues (e.g. ser-ine-10) also appears to increase the efficacy of HATs fur-ther increasing transcription, and leading tophosphoacetylation of histones [72]. Phosphorylation iscatalysed by protein kinase enzymes and reversed by phos-phatases, though much is still unknown about their preciseaction.

The addition of between one and three methyl groupsfrom the cofactor S-adenosyl methionine to the –NH+

group of certain lysine residues and the addition of oneor two methyl groups to arginine residues maintains thepositive charge of DNA and condenses chromatin[54,122]. Several residues on the tails of H3 and H4 are par-ticularly prone to this process of histone methylation whichis catalysed by amino acid specific histone methyltransfer-


Fig. 1. Schematic of chromatin modifications. (A) Each nucleosome consists of DNA (blue lines) wrapped around eight core histone proteins (two copieseach of H2A, H2B, H3, H4—peach cylinders). Chromatin consists of several of these nucleosomes condensed together within the cell nucleus. (B)Epigenetic modifications at each nucleosome regulate gene transcription through altering the accessibility of the transcriptional machinery such as RNApolymerase to the DNA. The direct binding of methyl groups (large red circles) to DNA typically silences gene expression. Alternatively, various covalentmedications at histone tails (brown lines) which protrude from each histone protein can alter the structure of chromatin and gene transcription. These mayinclude, but are not limited to, acetylation, methylation, phosphorylation, ubiquitination, and sumoylation and occur at specific amino acid residues(depicted by multi-coloured spheres) on each tail. Environmental experiences such as maternal care, social interactions, drugs and aversive stimuli havebeen shown to induce stable long-term and dynamic short-term changes in DNA methylation and to specific amino acid residues on particular histone tailsof individual genes that are expressed in the brain (see text).

E.B. Keverne, J.P. Curley / Frontiers in Neuroendocrinology xxx (2008) xxx–xxx 3

ARTICLE IN PRESS

ases (HMTs) and reversed by the recently discovered his-tone lysine demethylases (HDMs), though histone argininedemethylases have yet to be identified [89,90]. Methylationis not always associated with transcriptional silencing how-ever, as methylation at certain residues (e.g. trimethylationat H3K4, H3K36, H3K79) is associated with transcrip-tional activation and may be critical in creating stableeuchromatin across extensive chromosomal regions [2,88].

There are numerous interactions between these pro-cesses with the methylation of histones preventing furtheracetylation and phosphorylation, histone acetylation inhib-iting the methylation of histones and phosphorylation acti-vating acetylation [34]. There are also a number of otherless well understood histone modifications. Histone ubiqui-tylation can occur at H1 and H3 and particularly H2A andH2B and involves the addition of the ubiquitin protein tothe –NH+ group of lysine residues [92]. The function ofubiquitin is modified depending on the number of its addi-tions or the lysine residue to which ubiquitin attaches. Themajor function of ubiquitin proteins is to tag other proteinsfor degradation by the proteosomes, but when attached tohistones they can alter gene transcription probably throughregulating the previously mentioned histone modifications[103]. Related to this is sumoylation, the addition of smallubiquitin-related modifier proteins to core histones whichthen associates with HDACs and other repressors to inhi-bit gene expression either directly or by antagonizing his-tone acetylation, phosphorylation or ubiquitination


[91,70]. Other covalent modifications of histones includeADP-ribosylation, biotinylation, deamination and prolineisomerisation, though their effects on gene transcriptionare still to be fully understood [39,23,51,111,71]. Neverthe-less, it is likely that all covalent histone modifications willhave the potential to activate or inhibit gene transcriptionunder different conditions. Histone acetylation, phosphor-ylation, ubiquitination, deamination and sumoylation areregarded as dynamic modifications being able to exert theireffects within minutes [70,52]. Conversely, histone methyla-tion is widely regarded as being the most stable, long-last-ing and difficult to reverse, though this is also beingchallenged with the identification of HDMs [3]. In additionto such covalent modifications, there are other changes tochromatin structure such as nucleosome sliding, thereplacement of core histones with other proteins and therepositioning of nucleosomes by ADP-dependent proteincomplexes (e.g. SWI/SNF) that can all exert tight controlof gene transcription [4,93,44].

In addition to modifications of histone proteins, thedirect methylation of DNA itself is associated with long-lasting stable changes to gene expression [79]. Indeed, thedegree of dinucleotide cytosine–guanine (CpG) methyla-tion at promoter regions inversely correlates with levelsof gene transcription. DNA methylation involves the trans-fer of a methyl group from S-adenosyl methionine to the 50

position of cytosine residues particularly those at the dinu-cleotide CpG sequences. De novo DNA methylation is



ARTICLE IN PRESS

catalysed by DNA-methyltransferases (DNMTs) 3a and3b, whereas maintenance of DNA methylation is catalysedby DNMT1 [81]. DNA methylation not only directly inter-feres with transcription binding but also leads to the bind-ing of CpG binding domain proteins (MDBs) such as(methyl CP-binding protein 2) MeCP2 that recruit chroma-tin modelling enzymes like HDACs and HMTs, leading todeacetylation, histone methylation and further transcrip-tional silencing [24]. Methylation can be extremely stable,though recent findings of proteins with DNA demethylaseactivity suggest that there may even be some reversibility ofDNA methylation [9]. Furthermore, there is bi-directional-ity in this relationship between histone modifications andDNA methylation [112]. For example, deacetylation andmethylation at H3 lysine residue 9 contributes to increasedDNA methylation probably via the recruitment ofDNMTs. Therefore, there exists an intricate synergybetween histone and DNA modifications within the nucle-osome such that gene transcription can be tightly epigenet-ically controlled through processes individually andthrough their effects of each other [34].

Detailed investigation of differentially methylatedregions (DMRs) of DNA for a number of imprinted genesin a variety of mammalian species have shown that tandemCpG repeats are necessary for maintaining the differentiallymethylated state [84]. Deleting the DMRs in mice results inloss of imprinting and biallelic gene expression [60]. Theposition of DMRs within human and mouse imprintedgenes is conserved but contains repeats of different unitlengths ranging from 425 to 740 nucleotides. However,these sequences do have the potential to fold into stablesecondary structures with significant similarities betweenmouse and human. This suggests that the precise numberof tandem CpG repeats is not important, but a subset thathas the capacity to form secondary DNA structures (hair-pins, tetraplex structures, G-quartets) which have featuresin common with telomeres and centromeres and may beimportant for specific protein binding [75].

3. Heritable epigenetic changes—genomic imprinting

Epigenetic effects have also been established that aresustained through germline transmission in the form of‘‘imprinted” genes. Imprinted genes are haploid dominantand expressed according to parent-of-origin such that asmall, but important set of key regulatory autosomal genesare only expressed when inherited from mother, and othersare only expressed when inherited from father [99]. Theactive process of gene silencing invariably involves DNAmethylation, but histone deacetylation, micro-RNAs,CpG repeat sequences, polycomb proteins and non-codinganti-sense RNAs, all play a part in the imprint mechanismsof different genes [78].

Imprinted genes usually occur in clusters with a differen-tially methylated region (DMR) that is methylated in thegermline and frequently regulates gene silencing acrossthe cluster of genes [115]. This DMR may involve a non-


coding RNA that ‘coats’ the inactivated region in cis, alter-ing the physical structure of chromatin and its availabilityfor transduction. In other cases, such as at the H19 pro-moter, chromatin organising proteins like the CCCTC-binding factor (CTCF) bind to chromatin and serve as abarrier that prevents interactions between remote enhanc-ers and promoters [77]. The insulin like growth factor(Igf2r) imprinted cluster is an example of anti-sense tran-scription mediating the silencing of neighbouring geneson the paternal chromosome [78]. The Kcnq1 (K+ channela subunit) cluster shows bi-directional paternal repressionof genes in the cluster which is dependent on a non-codingRNA and which enables recruitment of a repressive poly-comb protein complex [62]. The G-protein a subunit(Gnas) imprinting cluster is even more complex enablingpaternally, maternally and biallelic expression of tran-scripts in a tissue specific manner [115]. An important fac-tor that is emerging in the regulation of imprinted genes istheir chromosomal spatial organisation, which may involveregulatory elements more that 80 kb upstream [105] andothers may even be located on different chromosomes,invoking sub-nuclear functional compartmentalisation[119].

Many of the imprinted clusters contain both maternallyand paternally expressed genes. The Paternally expressedgene 3 (Peg3) gene cluster has an evolutionary conservedsequence in the first intron which serves as a binding sitefor the zinc finger YY1 protein. On the maternal allelethe YY1 binding site is methylated, specifically repressingbinding of this transcription factor [49] preventing mater-nal expression of Peg3. However YY1 is itself a memberof the polycomb group of proteins (PcG) which can recruitother PcG proteins (Eed, Ezhz) leading to histone modifi-cations (H3 K27 methylation) and transcription repression[114]. Because the YY1 binding site is positioned betweenmaternally and paternally expressed genes in the Peg3 clus-ter, the potential for insulator/boundary element shieldingof other genes in the cluster can have regulatory conse-quences for both maternal and paternal expressed genesin this cluster.

Unique to genomic imprinting is the ability to switch theepigenetic mark such that a paternally silenced allele onpassing through the maternal germline becomes expressed,and vice versa through the paternal germline for a mater-nally silenced allele. This germline reprogramming ofimprints requires dynamic changes in chromatin structureinvolving extensive DNA demethylation and erasure ofthe parental imprints [100,79]. This is an essential prelimin-ary to restoring totipotency for the future development ofdifferent cell types. Chromatin in the developing mouse pri-mary germ cells is initially found de-condensed and the era-sure of imprints occurs in a restricted period from E10.5–12.5 by active demethylation [28]. Not all genomic methyl-ation is removed at this stage and notably, regions contain-ing parasitic elements such as transposons, remainmethylated and silenced. Lymphoid specific helicase(LSH) is a major epigenetic regulator that is essential for



ARTICLE IN PRESS

transcriptional silencing of parasitic elements and main-taining genomic stability in meiosis [25]. The first wave ofoocyte growth is the time at which the next generation ofmaternal specific imprints are established in a specific locusby locus manner [61,73]. At this stage oocytes exhibit activetranscription which produces abundant cytoplasmic pro-teins that will serve as epigenetic modifiers of the futurepaternal genome, while others prevent maternal DNAmethylation after fertilisation.

Sperm are produced in great abundance and have a rel-atively short life while the total oocyte population is deter-mined in the foetal ovary and remains quiescent in meioticarrest for a substantial part of the female’s lifetime. It is notuntil the onset of puberty that the maternal LH surge acti-vates meiotic progression to the secondary oocyte stage[76]. The maternal and paternal genomes come togetherat fertilisation but in many respects they are not equivalent,and in order to participate in development the paternalgenome must also acquire an appropriate epigenetic state[100]. Maternally derived histone chaperones (Hira) andepigenetic modifiers of the polycomb group such as Ezh2and Eed are required to replace and modify histones afterthe removal of protamines in the highly condensed paternalchromatin [104,28]. While the maternal genome remainsmethylated, the paternal genome undergoes a genome-widedemethylation [64]. A maternal protein, Stella, has beenshown to be required for preventing the maternal demeth-ylation at this stage and its deletion also results in the lossof methylation from a number of maternally imprintedgenes [68]. Demethylation of the paternal genome post-fer-tilisation may account for the relatively few paternalimprints [82]. Thus the matriline appears to have greatercontrol over imprints of key developmental genes and theearliest post-fertilisation development is mainly regulatedby maternally inherited transcriptional and epigeneticmodifiers. This makes good biological sense for mammalswhere the mother also has the greatest time and energyinvestment in the developing foetus.

Heritable epigenetic changes such as those which occurat imprinted domains can be modified according to speciesand, within a species, according to tissue. Thus compara-tive analysis of the Peg3 domain shows that zinc finger pro-tein gene (Zfp264) has become biallelic in brain and testisof the cow, but is paternally expressed in the mouse brain[48]. In the mouse, Igf2 is imprinted but biallelic expressionoccurs in the choroid plexus, while its Igf2 receptor whichis normally maternally expressed shows biallelic expressionin the brain. Neuron specific relaxation of the Igf2 receptorexpression is associated with histone modifications in thedifferentially methylated regions leading to DNA hypome-thylation, acetylation of H3 and H4 and dimethylation ofH3K4 [120]. This ability to relax imprints applied in thegermline leading to tissue specific epigenetic modificationsappears to be regulated via the anti-sense transcriptsinvolved in imprinting. Frequent loss of imprinting alsooccurs with tumour development (e.g. wilms tumour,breast, and colo-rectal tumour) and may in part be respon-


sible for these aberrant developments. Imprinted genemethylation and gene expression has also been shown tobe altered as a consequence of in vitro fertilisation [59],the cloning of mouse embryos [74] and from completelyinbred mice derived from ES (embryonic stem) cells [26].Interestingly, the altered methylation pattern that arisesin the donor ES cells persists through differentiation ofthe foetus, indicating retention of ES cell epigenetic modi-fications. Thus, heritable germline derived imprints arethemselves capable of epigenetic modifications.

4. Genomic imprinting, growth and evolution

A number of theories have been proposed to account forthe evolution of genomic imprinting [97,37]. Some of theseare rather specialised, such as the protection of mammalsfrom parthenogenesis and the ‘‘placenta hypothesis”

which, because imprinted genes are not expressed in eithermonotreme, egg-laying mammals or birds, suggests thatviviparity was a driving force for genomic imprinting inplacental mammals [47]. A more widely accepted theoryis that of ‘‘genetic conflict” which proposes that paternallyexpressed genes promote embryonic and early post-natalgrowth by extracting nutritional resources from mother.This maximises the father’s reproductive success relativeto other males that subsequently mate and produce preg-nancies with the same female. Maternally expressed genes,in contrast, are theorised to resist maternal resources beingexhausted on a single pregnancy and ensure some reservesare withheld for subsequent pregnancies and future off-spring. The evolutionary advantage for a male is, therefore,to father larger, fitter progeny by the expression of growthpromoting alleles, a development which is considered to bein conflict with the maternal requirement to conserveresources across all pregnancies. Although this theorymatches well to a number of paternally expressed genesthat promote placental and embryonic growth [38], it is dif-ficult to reconcile with the mechanisms which underpingenomic imprinting at the molecular level. From a mecha-nistic viewpoint the control over imprinted genes that arepaternally expressed is often achieved in the female germ-line where the imprint is formed and leads to silencing ofthe maternal allele which results in only the paternal allelebeing expressed [83]. Hence, if genes which extractresources from mother achieve paternal expression by theactive process of maternal allele silencing, the questionarises as to how natural selection might have initially oper-ated at the maternal locus to effect the foetal–placental phe-notype which is disadvantageous to mothers.

Selective paternal expression of some imprinted genesoften requires more than simple maternal silencing. Takefor example Igf2, an important growth factor gene that isexpressed in the placenta and promotes placental growth,transport of nutrients and embryonic growth [80]. Thisgene is paternally expressed and together with its receptorIgf2r, which is maternally expressed, effects pre- andpost-natal growth. Igf2r is a mannose-6-phosphate recep-



ARTICLE IN PRESS

tor uncoupled from growth promoting properties and mayact as a sink for Igf2. This interactive process has beentheorised to represent and descriptively matches to a pater-nal/maternal form of genetic conflict [38]. However, a con-sideration of the mechanisms regulating Igf2 expressionreveals that an imprint control region (ICR) is methylatedand silenced on the paternal Igf2 chromosome. Thisrepressed state of the paternal ICR is produced by germlinemethylation in sperm, while the maternal allele escapesmethylation due to the binding of a maternally derivedCTCF protein to this region of maternal DNA [5,77].Mutations in the CTCF binding site on the maternalDNA results in the maternal ICR also acquiring methyla-tion due to failure of CTCF binding and hence providingthe barrier which protects the ICR from methylation[102]. Were it not for the maternal production of CTCFin the oocyte, both copies of H19 would be repressed andIgf2 would be expressed from both paternal and maternalalleles, the ancestral form of expression for this gene.

One of the receptors for Igf2 is maternally expressed andregulates the paternally expressed growth factor (Igf2) bybinding it and internalising it before transportation to thelysosomes for degradation. The Igf2r gene encodes tworeciprocally imprinted transcripts, which are methylateddifferentially at their DNA [118]. Differentially methylatedregion 1 (DMR1) includes the promoter for a sense Igf2receptor whereas DMR2, which is located at the secondintron of the gene, serves as a promoter for the anti-sensetranscript (Air), a long non-coding RNA transcript(108 kb). The paternally expressed Air is required for therepression of the paternal Igf2 receptor and two otherpaternally repressed genes in the cluster, whereas deletionof Air leads to loss of Igf2r imprinting and biallelic geneexpression in peripheral tissues [96].

There are clearly two ways of interpreting such a com-plex chain of interactions between maternal and paternalgenomes. One concerns the different parental germline ori-gins through which monoallelic expression is regulated,together with the possibility of differentially modifyinggene dosage in both a species and sometimes in a tissue spe-cific manner [116,33]. This might be interpreted as a mater-nal/paternal co-adaptive process. A second interpretationis that of genetic conflict since foetal growth promotion isassisted by the paternal allele of Igf2 while the maternallyexpressed receptor acts as a ‘‘sink” for Igf2 reducing itseffectiveness. However, it is difficult to comprehend, froman evolutionary viewpoint, why the maternal genomeactively relinquishes expression of a maternal allele andactively facilitates the expression of a paternal allele inthe case of Igf2, especially if this has the potential to hand-icap the female’s lifetime reproductive success. Involved inthe complex interactive regulation of Igf2 and its receptor,there are three maternal levels of control for this growthpromoting interaction (CTCF production, prevention ofmethylation of the ICR on the maternal chromosome,and silencing of maternal Air), and two paternal levels ofcontrol (repression of the paternal ICR by germline meth-


ylation in sperm; silencing paternal Igf2r by preventingmethylation of DMR2) [109]. While this could conceptu-ally be considered to reflect genomic conflict, it certainlydoes not lend itself, at a molecular genetic level, to an inter-pretation that exclusively favours paternal regulation ofgrowth promotion of embryos. On the contrary, most lev-els of control reside in the female genome.

The complexity of genomic imprinting in regulation ofpre-natal growth is further illustrated by the growth factorreceptor-bound protein 10 (Grb10) which is expressed fromthe maternal allele in the placenta and foetus, but not in thebrain, where notably in the hypothalamic neurons it ispaternally expressed [41]. This reciprocally imprintedexpression is thought to be regulated by DNA methylationand the polycomb group protein complex Eed/Ezh. Thehypothalamic specific transcript expressed from the pater-nal allele is highly expressed at the time of hypothalamicdevelopment (E10–12) and again in the peri-natal period.Although adult neurons metabolize glucose in an insulin-independent manner, deletion of Grb10 results in impairedinsulin signalling in the hypothalamus, which in turnresults in increased food intake, moderate obesity andhypogonadotrophic hypogonadism. These effects arethought to result from impaired signalling through thehypothalamic melanocortin pathway. Differential DNAmethylation of the promoter on the maternal chromosomeaccounts for the tissue specific silencing of the maternalallele which thereby promotes only paternal expression inthe hypothalamus. Analysis of histone modifications showthat methylation of H3K4 is associated with DNA methyl-ation of the maternal allele, whereas the H3K27 is regu-lated by Eed–Ezh2–Pcg complex and histone deacetylaseto correspond to specific silencing of the paternal alleleand maternal expression in foetal and placental tissue[121]. Maternally expressed Grb10 binds to the insulinreceptor and strongly influences growth of the foetus andplacenta [17]. Mutations of Grb10 support the role ofmaternal Grb10 in signal transmission from the insulinreceptor to IRS-1 [94]. Here we have an example of animprinted gene with maternal and paternal expressed tran-scripts that are co-adapted in the context of maternal feed-ing (paternal allele) and placental and foetal growth(maternal allele). The paternally expressed allele is promot-ing maternal feeding whereas the maternally expressedallele is regulating, in a restrictive way, growth of the pla-centa and foetus, which at a functional level matches betterto co-adaptation than to conflict.

As work progresses on the molecular genetic mecha-nisms of genomic imprinting, how this is differentially reg-ulated through the male and female germlines and howmonoallelic expression of certain genes may be modifiedacross tissues and across species, it is becoming clear thatno single theory can encompass the whole of genomicimprinting. However, it is also becoming clear that the roleof the matriline is emerging as the major player. Certainly,co-expression of imprinted genes in the hypothalamus andplacenta are functionally co-adapted between mother and



ARTICLE IN PRESS

foetus, and co-expression in the placenta and hypothala-mus may be developmentally co-adapted. This is an issuewe will consider in the next section.

5. Genomic imprinting and co-adaptive development of brain

and placenta

The success of placental mammals in sustaining lifeacross many environments is due to their ability to main-tain their ‘‘milieu interieur” reasonably intact. Maintainingconstantly high body temperature opened up many habi-tats that were closed to reptiles, enabling early ancestralmammals to feed at night and throughout winter, whileplacental evolution enabled their young to develop andgrow within the body cavity. Many of the mammalian phy-logenies evolved larger phenotypes, which is especiallyimportant in cold climates since it reduces the relative sur-face area to body weight thereby reducing heat loss. How-ever, larger mammals require greater maternalinvestment—i.e. prolonged pregnancy and extended mater-nal care and bonding. In this context the placenta hasplayed a pivotal role, primarily acting as an endocrineorgan regulating the secretion of hormones into maternalcirculation to influence maternal physiology, metabolismand behaviour. Hence, an efficient placenta can be seenas one which not only optimises pre-natal transfer ofmaternal resources but ensures provision of these resourcesby regulating maternal feeding, maternal behaviour andmetabolism. The placenta prepares for post-natal eventsby priming the mother’s brain for maternal behaviourand priming the mammary gland for milk production.

The placenta represents an important evolutionarydevelopment that is unique to mammals and differs fromother forms of vertebrate and invertebrate internal devel-opment where offspring may receive protection from pre-dators and dehydration, but little in the way of nutrientsis provided. The growth and development of the mamma-lian embryo is determined by the transfer capacity of theplacenta which is synchronised with the energetic require-ments of the embryo. This early growth is a complex pro-cess with some tissues, such as the brain, being moreenergetically demanding than others. Brain growth occurslate in embryonic development and in most mammals itscompletion is postponed to the post-natal period. Manyof the genes which regulate placental and embryonic

Table 1Imprinted genes expressed in brain and placenta

GnasX1 (P) Hypo/piNecdin (P) HypoDlk1 (P) Hypo/piSgce (P) HypoPeg3 (P) Hypo/piPeg1 (P) Hypo/piMagel2 (P) HypoSnrp (P) HypoNnat (P) Hypo


growth are imprinted, thereby coordinating foetal growthsynchronised with the growth and provisioning capacityof the placenta.

A subset of imprinted genes that are expressed in thedeveloping foetal placenta are also expressed in the hypo-thalamus, the functional outcome of which might be bestviewed as a parent–infant co-adaptive process. These genesare all paternally expressed, but assuming their ancestralstate to be biallelic expression then the regulatory processof silencing to enable only paternal expression, primarilyoccurs in the matriline [99]. Considering the relative spar-sity of imprinted genes [82] discovered (Website MRCMammalian Genetics Unit, Harwell, 2004) their expressionin certain tissues like the brain and placenta is greater thanmight be expected. Moreover, all of the genes so far iden-tified that are expressed in both placenta and brain, espe-cially the hypothalamus, are maternally imprinted (Table1). These distinct organ types, foetal placenta and maternalhypothalamus, function as one in the pregnant mother,although they are encoded by different genotypes. Recentdetailed studies on the phenotype of maternal hypotha-lamic versus foetal–placental expression of such imprintedgenes has been demonstrated for the maternally imprinted,paternally expressed genes, Peg1 [55,63] and Peg3 [58,40].Because imprinted genes are expressed according to par-ent-of-origin, a mother that carries a homozygous nullmutation when mated with a wild-type father will produceoffspring that are all wild-type normal, while a mutantfather mated with a wild-type mother will produce mutantoffspring that carry and express the mutant allele (Fig. 1).In this way the effects of the mutation on the maternal phe-notype and the infant phenotype can be investigated inde-pendently. The maternal consequences of expressing thistargeted deletion (Fig. 2) have much in common withlesions of the maternal hypothalamus, namely reducedfood intake, impaired maternal care, inability to maintainbody temperature under a cold challenge and a severeimpairment in milk letdown [58,40,21,22]. As a conse-quence, maternal weight gain and fat reserves during preg-nancy are impaired and pups suffer reduced pre- and post-natal growth, even though litter size is smaller [22]. Thesimilarity of the genetic lesions with neural lesions is notsurprising since Peg3 regulates apoptosis [85] and the neu-rons and receptors in relevant hypothalamic nuclei are sig-nificantly reduced in number [58].

t Nucleotide binding proteinNeuronal growth suppressor

t Negative regulator of notch 1Sarcoglycan family

t Bac transport–apoptosist b-Hydrolastfold family

MAGE like proteinSmall nuclear ribonucleo proteinPhosphorylates CREB " Ca2+ intra-cellular


Fig. 2. Peg3 inheritance. Crossing homozygous mutant Peg3 KO female mice (blue) with wild-type males (gray) produces all offspring as wild-type (gray)since the paternally expressed allele is silenced on inheritance from the mother. The converse is true when the father is homozygous with all offspringexpressing the mutation in the hypothalamus and placenta (blue).


ARTICLE IN PRESS

What is especially interesting about these maternallyimprinted alleles is that at the same time they are function-ing in the placenta to hormonally regulate the maternalhypothalamus, they are also actively engaged in developingthe foetal hypothalamus. Moreover, the effects of mutatingone such gene (Peg3) either in the maternal hypothalamusor separately and independently in the placenta and foetalhypothalamus produce remarkably similar phenotypes(Fig. 2). Since the expression and functional effects of thisgene in the maternal hypothalamus, foetal placenta andfoetal hypothalamus overlap in time (i.e. both genomes inthe pregnant mother) it is again difficult to reconcile itsactions with parental conflict. The placental and foetalhypothalamic expression of this gene results from imprint-ing of the very same maternal allele, while this gene in thematernal hypothalamus is expressed as a result of germlineimprinting in this mother’s mother. Expression runs pri-marily through matrilineal control (Fig. 3).

Fig. 3. Pre-natal and post-natal Peg3 effects. The maternally imprintedindependently in the maternal hypothalamus and the foetal placenta. The adevelopment.


A question of some merit in considering placental andhypothalamic evolution is how the adult hypothalamushas co-adapted to respond to the foetal placenta. Thematernal hypothalamus is fully developed and sexuallydimorphic at the stage when these hormonal interactionstake place. However, those imprinted genes involved indeveloping the placenta and hypothalamus (Table 1) aresimultaneously active in both developing structures at thesame time that one of them, the placenta, has to function-ally engage the adult hypothalamus. Herein lies the win-dow of opportunity for selection pressures to operate, awindow which may also be important in the context of‘‘foetal programming”.

Overall, imprinted genetic events are probably best con-ceptually envisaged in the context of co-adaptation, theevolutionary outcome of which is offspring that haveextracted adequate maternal provisioning and care bothpre- and post-natally are themselves both well adapted

gene, Peg3, produces functionally similar phenotypes when expressedction of Peg3 in bringing about these co-adaptive effects occurs during



ARTICLE IN PRESS

for and genetically predisposed toward this mothering stylewhen adult [21]. It is also noteworthy that these maternallyimprinted alleles are phylogenetically ancient genes thatwere once biallelically expressed. Such genes are mechanis-tically regulated by silencing of the maternal allele throughepigenetic ‘‘marks” applied in the female germline [87,61].Since early developmental mortality accounts for a majorpart of the variance in viability fitness of a gene, this matri-archal regulation of important components of the foetalgenome is optimally suited to safeguarding the motheragainst potential anomalies in development. Developmen-tal errors provide a risk not only for viability of offspring,but are also an ‘‘in-utero” danger to the mother’s life.However, developmental anomalies in the foetal genomeare more readily recognised through the placental interfacewith mother, and pregnancy may be terminated naturallywhen the same dysfunctional allele is also expressed inthe placenta.

Two important questions that arise from a mechanisticassessment of genomic imprinting in mammals are, whyrevert to the haploid condition of expression, and whyshould the matriline, by silencing maternal alleles, providefor paternal expression of genes that play an important rolein development of the foetal placenta and brain? Becausediploid organisms have twice the number of gene targetsfor mutations as do haploids, and because the incidenceof beneficial mutations is a rate limiting step in adaptation,such mutations are likely to confer increased fitness onorganisms that are diploid. However, haploid expressionhas the advantage of rapid fixation of a trait in the popu-lation since hemizygosity with parental specific alleleexpression carries the advantage for rapid efficient propa-gation [36]. Moreover, by providing a haploid expressionin a diploid organism, genomic imprinting confers the abil-ity to disseminate beneficial mutations in the population

Fig. 4. Co-adaptive evolution. As a consequence of genomic imprinting an allethe same time in the developing brain (red). Selection pressures operating atmaternal hypothalamus subsequently functions (black arrows). Placental expreswith the adult maternal hypothalamus (yellow). Co-adapting these functions (hypothalamus which again in the next generation interacts with the placenta. Cwith epigenetic flexibility through the matriline maximises advantages to offsprvia two epigenetic mechanisms.


while also providing a back-up silent copy of each geneshould mutations damage the functional copy [8].

Sex differences in mammalian reproductive and behav-ioural strategies mean that a male’s reproductive successdepends upon how many females he can mate, while femalereproductive success is restricted by the number of preg-nancies she can sustain together with the successful wean-ing of offspring. A favourable genetic event whichenhances maternal care through an action in the brainand also enhances offspring survival via an action in theplacenta, would have markedly different consequencesdepending on whether it were imprinted and dependingon which parent was expressing the allele. Unlike Mende-lian expression for an autosomal allele, genomic imprintingproduces haploid dominance and there is no weakening ofeffect in transmission through heterozygosity.

Because a male has potential for siring more offspringthan a female, paternal haploid expression of a maternallyimprinted gene has the greater potential for rapid fixationof a trait in the population. Availability of oestrus femalesis the rate limiting factor for male sexual behaviour andhence being able to discriminate oestrous females and tolearn about urine marks that signal oestrous conditionwould provide such males with a significant reproductiveadvantage. The maternally imprinted gene (Peg3) whichconfers advantage on female maternalism and placentalnurturing also confers olfactory advantages to males in thisregard [101]. Peg3 knockout males are unable to discrimi-nate between the odours of oestrous and di-oestrousfemales and are severely impaired in their reproductive per-formance. These phenotypic effects on sexual behaviour ofthe male would increase by more than 80:1 his likelihood offinding a fertile mate. Moreover, if the imprint results inmale expression rather than female expression, a further28-fold advantage to maternalism would be disseminated

le that is expressed in the foetal placenta (red) is also the allele expressed atthis early stage in development are fundamental to determining how thesion is, at this same time period, functionally constrained by its interactionorange) provides the selection pressures for development of the maternalombining genetic stability through a maternally imprinted gene alternatinging. These advantageous effects are therefore trans-generationally inherited



ARTICLE IN PRESS

within five generations. Such advantages would enhancereproductive skew, circumventing genetic bottlenecks,and rapidly establish homozygosity of this allele in thepopulation.

Findings such as these demonstrate the significance oftrans-generational interactions between hypothalamus,foetal placenta and foetal hypothalamus when they aregenetically regulated by the same imprinted gene, butinvolving two distinct genomes co-functioning in one indi-vidual, namely the mother. When this imprinted gene isinherited from father, both sons and daughters benefitfrom enhanced placental transfer of resources as well asthe effects of placental hormones on mother’s behaviour.Sons like their fathers also acquire enhanced mating advan-tages. When this gene is inherited from mother, it is silentin sons and daughters but both benefit from good maternalcare, increased maternal feeding and milk letdown throughan action of the gene in the maternal hypothalamus. Thesons of this generation in which the allele is silenced do,however, produce offspring in the next generation whichexpress the gene, thereby escalating the co-adaptive advan-tages to the following generation. Such inheritance pro-vides a template for mother–infant co-adaptive evolutionthrough heritable traits regulated by chromosomal epige-netic marks in the maternal germline [99] (Fig. 4).

6. Non-heritable epigenetic changes

Inherited epigenetic changes in the form of imprintedgenes are not the only form of non-sequence codingDNA and histone modifications that are important in theregulation of brain development and behaviour. Signifi-cantly, environmental variables are able to alter geneexpression via such epigenetic modifications [45]. Thesecan be long-term stable modifications akin to thosechanges in cellular phenotypes that are necessary for cellu-lar differentiation, but they can also confer short-termdynamic changes. Significantly, there is converging evi-dence that these short- and long-term changes appear tobe mediated by distinctive epigenetic alterations [56].

Contextual fear conditioning in rats is the associationmade between an environment (e.g. novel cage) and anaversive stimulus (e.g. foot-shock). Subjects that have beentrained to learn this association will eventually exhibit afearful response such as freezing even in the absence ofthe aversive stimulus and this response is maintained overlong periods. As with many learning processes, the acquisi-tion and retention of this memory requires distinctivechanges in gene transcription [65]. One hour after contex-tual fear conditioning large increases in H3 acetylation atthe lysine-14 residue, phosphorylation at the serine-10 res-idue and phosphoacetylation in CA1 of the hippocampusare observed [57,19]. Pre-treatment of subjects with HDACinhibitors such as Trichostatin A that prevent the deacety-lating ability of HDACs also increase the levels of H3 acet-ylation. These changes are dependent upon activation ofboth NMDA receptors and the ERK/MAPK signalling


pathways and blocking of either of these prevent theincrease in H3 acetylation [57,19]. The timing of these mod-ifications correlate with the onset of gene expressionchanges in the hippocampus following fear conditioning,but interestingly these epigenetic changes soon return tobaseline. Therefore H3 modifications are important forthe acquisition of long-term memories but other changesare required for the long-term storage. Other studies havealso implicated a role for histone modifications in learningand memory. For instance, mice lacking CBP (which hasintrinsic HAT activity) exhibit deficits in spatial memory,fear conditioning and novel object recognition [50,1,117].These deficits can be ameliorated following treatment withHDAC inhibitors, suggesting that it is the balance of HAT/HDAC activity in neural chromatin that may be significantfor the acquisition of certain long-term memories [56].

The role of acetylation on learning and memory pro-cesses has also been investigated in the CK-p25Tg trans-genic mouse [29,30]. In this mouse the expression of p25,a protein involved in various neurodegenerative diseases,is under the control of a dietary inducible calcium calmod-ulin kinase 2 (CamKII) promoter such that it can be turnedoff in adult mice leading to neurodegeneration. Mice thathave undergone such adult neurodegeneration becomeimpaired on fear conditioning and spatial water mazememory tasks [29], but these abilities can be preserved ifthey are exposed to environmental enrichment [30]. Envi-ronmentally enriched animals were found to have increasedlevels of marker proteins for synaptic integrity and plastic-ity and exhibit increased acetylation of H3 and H4 in thehippocampus, even as soon as 3 h post-enrichment. Inaddition, the repeated administration for four weeks ofHDAC inhibitors can lead to an increase in the numberof synapses and sprouting of dendrites and H3 and H4acetylation in post-mitotic hippocampal neurons.

Significantly, one recent study demonstrates thatdynamic changes in DNA methylation may also be criticalin the consolidation of contextual fear conditioned memo-ries. Immediately following fear conditioning, mRNA lev-els of DNA-methyltransferase 3 (DNMT3a) and DNMT3bare up-regulated in CA1 of the hippocampus and blockingDNMT activity prevents the normal memory consolidation[67]. This consolidation is reliant upon the suppression ofcertain genes such as memory suppressors e.g. PP1, andthe activation of other genes such as those involved in syn-aptic plasticity e.g. reelin. Peak changes in the methylationstatus of these genes occur in CA1 1 h following fear con-ditioning, with a return to baseline within 24 h. This studyprovides further evidence for the multitude of gene specificepigenetic modifications that are induced following expo-sure to an environmental cue such as a foot-shock leadingto both short- and long-term changes in gene transcription.It also suggests that changes in DNA methylation may notbe irreversible in post-mitotic neurons.

Drug addiction involves long-term changes in cellularphenotype and gene expression, particularly in striatalbrain areas including the nucleus accumbens [42]. Long



ARTICLE IN PRESS

after administration of an addictive substance, the drugand even other contextual cues can have powerful effectson behaviour. Some of these long-term changes in geneexpression and dependence are associated with epigeneticalterations. In rats, the chronic administration of cocaineleads to H3 acetylation of the promoters for FosB, BDNFand Cdk5 [53,108]. The product of FosB is a mediator ofreward in striatal neurons with its accumulation enhancingdrive for cocaine in animal models. Interestingly, the levelsof BDNF H3 acetylation increases even after drug with-drawal, demonstrating the pervasiveness of the effects ofthe drug on histone modifications [53]. Chronic cocaineadministration also induces an increase in MeCP2 andMBD1 in the adult brain [13]. Different epigenetic changesare observed when rats are given acute administration ofcocaine. Within 30 min there is an increase in H4 acetyla-tion at the cfos and fosB genes as well as H3 phosphoryla-tion at the cfos promoter only, though these modificationsreturn back to baseline within 3 h. The behavioural effectsof cocaine, as measured by a conditioned place preference,are dependent on this acetylation as they can be enhancedby treatment of subjects with HDAC inhibitors or inhib-ited when HDACs are over-expressed in vivo [53,20]. Like-wise, the administration of other drugs of abuse leads tochromatin remodelling. Chronic administration of amphet-amine leads to H4 hyperacetylation in the mouse striatum,and repeated treatment with HDAC inhibitors significantlyincreases amphetamine induced behavioural sensitisationin mice [46].

Therefore different histone modifications at specific genepromoters underlie both dynamic short-term and stablelong-term changes in gene transcription associated withdifferent drug exposures. Exposure to other environmentalchallenges also appears to lead to specific changes at partic-ular histones. For instance, following acute induced sei-zures, rats exhibit increased levels of H4 acetylation atthe P2 promoter of the brain derived neurotropic factor(BDNF) gene, as well as increased phosphoacetylation ofH3 of the cfos gene in the hippocampus [106]. However,following chronic seizures, there is acetylation of H3 ofthe BDNF gene. Therefore a pattern is emerging wherebytargetted regulation of gene transcription can occurthrough chromatin remodelling effects that are either stableand long-term (H3 acetylation) or dynamic (H4 acetyla-tion) [108]. Increased exposure to social stressors is a riskfactor for the development of depression. It is hypothesizedthat the long-lasting behavioural and mood changes associ-ated with depression may be regulated by histone modifica-tions. Although difficult to model in animals, the chronicsocial defeat paradigm is one of the few to be based uponsocial experiences. In this model, mouse subjects areexposed to an aggressive opponent on successive days ina test arena and are defeated [107]. After training, subjectsavoid contact, not only with aggressive individuals but withall other conspecifics. An interesting aspect of this para-digm is that anti-depressants (imipramine and fluoxetine)do reverse these behavioural effects but only after several


weeks of administration as is the case in the treatment ofdepression in humans. Following chronic anti-depressanttreatment, socially defeated subjects display an increasein BDNF expression in the hippocampus (which is down-regulated following social defeat) and a decrease in depres-sive-like symptoms, i.e. an increase in social contact [7,107].These changes in gene expression following chronic socialdefeat and anti-depressant treatment are associated withseparate histone modifications. Social defeat is associatedwith a 4-fold increase in H3 demethylation at the lysine-27 residue of the P3 and P4 promoters of the BDNF geneand a consequent down-regulation of BDNF in the hippo-campus [107]. These changes are long-lasting occurringeven one month after social defeat. Imipramine treatmentleads to a 2-fold increase in H3 acetylation at these samepromoters, but only in those animals that have beenstressed. Furthermore, when mice over-express HDAC5,the efficacy of imipramine is blocked, suggesting that imip-ramine, a tricyclic anti-depressant, may have its behav-ioural effects via a down-regulation of HDAC5 [107]. It iscurrently unknown through which epigenetic modificationsfluoxetine, a selective serotonin reuptake inhibitor, exertsits reversibility effects in this paradigm. However, it hasbeen shown to increase MBD1 and MeCP2 levels in areasof the adult rat brain with serotonergic receptors such asthe caudate putamen, frontal cortex and dentate gyrus.This requires chronic administration of fluoxetine implyinga role in altering the methylation and acetylation status ofgenes [13].

There is also accumulating evidence that exposure tonumerous stressors will lead to variations in gene expres-sion and transcription and that dynamic epigenetic modifi-cations may underlie these changes. For instance, inrodents exposure to a forced swim stress test, or a predatorleads to increases in which gene expression and H3 phos-phorylation at serine-10 residues of the dentate gyrus, par-ticularly in the granule layer. Indeed, there is a large degreeof correlation between H3 phosphorylation and the acqui-sition of immobility in the forced swim test [10]. Exposureof rats to a novel cage has been found to increase H3 phos-phoacetylation at serine-10 and lysine-14 residues andphosphorylation at serine-10 residues in the rat dentategyrus [16]. In the former studies, the alterations in histoneswere observed 8–24 h post-stressor exposure, but fell backto baseline after 48 h. In the latter study, the peak histonechanges were observed between 0.5 and 2 h after stressorexposure. Hence there exists a complex pattern of genetranscription involving numerous epigenetic modificationsoccurring over distinctive time courses in response to vari-ous environmental stimuli that lead to long-term changesin behaviour [86]. A major challenge for the future is todetermine how neural receptors signal to intra-cellularmediators of these effects to regulate epigenetic modifica-tions especially when those environmental changes arelong-lasting.

It is undoubtedly the case therefore that the mammalianbrain, even after neural fate has been determined and neu-



ARTICLE IN PRESS

rons exist in their post-mitotic state, is able to respond tothe environment in a plastic manner through the epigeneticmodifications of chromatin. These plastic changes mayoccur very late in life as the studies outlined above illus-trate, when adult animals are exposed to drugs, physicaland social stressors or aversive stimuli. Indeed, it has beendemonstrated that even monozygotic twins who share iden-tical DNA sequences differ with respect to epigenetic regu-lation of their genes, and that these differences increasewith age and appear to be related to the degree of discor-dance in environments experienced by the twins [32]. Sig-nificantly, however, they may also occur very early in lifeduring the period of extensive mother–infant interactionswhen the brain is developing most rapidly. An extensivebody of research by Michael Meaney and colleaguesreviewed elsewhere and in this volume demonstrates thatthe levels of tactile stimulation provided by rat dams inthe form of licking and grooming of pups leads to thosepups developing very different phenotypes [14,27]. Thosepups that receive low levels of stimulation developincreased stress responses, decreased response to reward,decreased cognitive ability and exhibit lower maternal carein comparison to pups that receive high levels of tactilestimulation. This increased stress response has been associ-ated with decreased expression of glucocorticoid receptors(GR) in the hippocampus due to a decrease in H3 acetyla-tion and increase in DNA methylation of the 1–7 exon ofthe GR promoter in the hippocampus that contains aNGFI-A binding site [113]. The decreased levels of mater-nal care that they themselves exhibit as dams is associatedwith a decrease in the number of oxytocin receptors andestrogen receptor a in the medial preoptic area that arerelated to the increased DNA methylation of the ERa pro-moter in this brain region [15]. Early life experiences canindeed have long-lasting profound effects upon adult phe-notype, and these may be mediated via epigenetic modifica-tions of gene promoters in a brain region specific manner.

7. Conclusion

Changes in chromatin structure that do not alter thenucleotide sequence of DNA have played a significant rolein all aspects of development. Waddington [110] recognised‘‘the causal interactions between genes and their productswhich bring the phenotype into being” and introduced theterm epigenetics to describe these genome and environmen-tal interactive events. Mammalian development is particu-larly noteworthy in this context because of in-utero growthand placentation which provides a living environment forthe two interacting genomes. This has required both co-adaptation and a specially tight control over gene dosageand transcriptional regulation to minimise any developmen-tal irregularities. Inherited monoallelically expressed,imprinted genes are epigenetically silenced, mainly undermatrilineal control and their haploid dominant expressionensures such tight control over internal development. Themammalian brain is specially notable in that it has under-


gone a phase shift in size relative to other vertebrates, andwithin mammals themselves there is a 20-fold increase in sizewhen controlling for body weight. These increases in sizecould also present a problem for viviparity, a problem whichhas been solved by postponing much of brain growth to thepost-partum period. This in turn has required extendedmaternal care which at the same time places the developingbrain in a unique mother–infant social context. For thedeveloping infant the mother provides the most significantenvironmental influence, shaping offspring brain develop-ment by producing long-term epigenetic modifications toneural and behavioural phenotypes. Mothers behaviour isable to influence the development of brain regions that areimportant in the future regulation of maternal care in theirdaughters and boldness in their sons. Thus non-heritable epi-genetic modifications enable long-term stable changes inneural and behavioural phenotypes in response to environ-mental experiences. Conceptually these experiences havemuch in common with learning and memory, but differ inthe timeframe whereby early life experiences may impactupon the behavioural phenotype at later periods in life.

In this chapter, we have considered at length the mech-anisms and phenotypic consequences of heritable andnon-heritable epigenetic changes in DNA and chromatin.However this represents only the tip of the iceberg, andthere is no doubt that epigenetic mechanisms will, in duecourse, be shown to contribute to many psychiatric disor-ders (schizophrenia, depression, obsessive compulsivebehaviour, traumatic stress, addiction) clinical conditions(obesity, cancer, neonatal programming in the context ofcardiovascular disease) and stem cell reprogramming. Formore than a decade the screening for genetic polymor-phisms has attempted to address these conditions, andthere have been many successes which have undoubtedlyincreased our understanding. However, these successescan only explain a very small percentage of the variance.Take for example, obesity, where several polymorphismshave advanced our knowledge of the endocrinology,metabolism and brain circuits for feeding, but these poly-morphisms account for only 4% of the variance. Similarfindings apply for schizophrenia where even in geneticallyidentical twins there is only 50% concordance. Determiningthe epigenetic contribution to such human disorders will becrucial though not necessarily easy as the epigenetic regula-tion of gene expression will be both developmentally andregionally specific within the brain. Therefore, in the nearfuture, we shall be reliant upon developing animal modelsthat extend our understanding of these processes.

References

[1] J.M. Alarcon, G. Malleret, K. Touzani, S. Vronskaya, S. Ishii, E.R.Kandel, A. Barco, Chromatin acetylation, memory, and LTP areimpaired in CBP+/� mice: a model for the cognitive deficit inRubinstein-Taybi syndrome and its amelioration, Neuron 42 (2004)947–959.

[2] A.J. Bannister, T. Kouzarides, Reversing histone methylation,Nature 436 (2005) 1103–1106.



ARTICLE IN PRESS

[3] A.J. Bannister, R. Schneider, T. Kouzarides, Histone methylation:dynamic or static? Cell 109 (2002) 801–806.

[4] P.B. Becker, The chromatin accessibility complex: chromatindynamics through nucleosome sliding, Cold Spring Harb. Symp.Quant. Biol. 69 (2004) 281–287.

[5] A.C. Bell, G. Felsenfield, Methylation of a CTCF-dependentboundary controls imprinted expression of the Igf2 gene, Nature405 (2000) 482–485.

[6] S.L. Berger, The complex language of chromatin regulation duringtranscription, Nature 447 (2007) 407–412.

[7] O. Berton, C.A. McClung, R.J. Dileone, V. Krishnan, W. Renthal,S.J. Russo, D. Graham, N.M. Tsankova, C.A. Bolanos, M. Rios,L.M. Monteggia, D.W. Self, E.J. Nestler, Essential role of BDNF inthe mesolimbic dopamine pathway in social defeat stress, Science311 (2006) 864–868.

[8] A.L. Beaudet, Y.-H. Jiang, A rheostat model for a rapid andreversible form of imprinting-dependent evolution, Am. J. Hum.Genet. 70 (2002) 1389–1397.

[9] S.K. Bhattacharya, S. Ramchandani, N. Cervoni, M. Szyf, Amammalian protein with specific demethylase activity for mCpGDNA, Nature 397 (1999) 579–583.

[10] A. Bilang-Bleuel, S. Ulbricht, Y. Chandramohan, S. De Carli, S.K.Droste, J.M. Reul, Psychological stress increases histone H3phosphorylation in adult dentate gyrus granule neurons: involve-ment in a glucocorticoid receptor-dependent behavioural response,Eur. J. Neurosci. 22 (2005) 1691–1700.

[11] A. Bird, Perceptions of epigenetics, Nature 447 (2007) 396–398.[12] A. Caspi, T.E. Moffitt, Gene-environment interactions in psychiatry:

joining forces with neuroscience, Nat. Rev. Neurosci. 7 (2006) 583–590.

[13] S. Cassel, D. Carouge, C. Gensburger, P. Anglard, C. Burgun, J.B.Dietrich, D. Aunis, J. Zwiller, Fluoxetine and cocaine induce theepigenetic factors MeCP2 and MBD1 in adult rat brain, Mol.Pharmacol. 70 (2006) 487–492.

[14] F.A. Champagne, D.D. Francis, A. Mar, M.J. Meaney, Variationsin maternal care in the rat as a mediating influence for the effects ofenvironment on development, Physiol. Behav. 79 (2003) 359–371.

[15] F.A. Champagne, I.C. Weaver, J. Diorio, S. Dymov, M. Szyf, M.J.Meaney, Maternal care associated with methylation of the estrogenreceptor-alpha1b promoter and estrogen receptor-alpha expressionin the medial preoptic area of female offspring, Endocrinology 147(2006) 2909–2915.

[16] Y. Chandramohan, S.K. Droste, J.M. Reul, Novelty stress inducesphospho-acetylation of histone H3 in rat dentate gyrus granuleneurons through coincident signalling via the N-methyl-D-aspartatereceptor and the glucocorticoid receptor: relevance for c-fos induc-tion, J. Neurochem. 101 (2007) 815–828.

[17] M. Charalambous, A. Ward, L.D. Hurst, Evidence for a primingeffect on maternal resource allocation: implications for interbroodcompetition, Proc. R. Soc. Lond. B 270 (2003) S100–S103.

[18] P. Cheung, C.D. Allis, P. Sassone-Corsi, Signaling to chromatinthrough histone modifications, Cell 103 (2000) 263–271.

[19] W.B. Chwang, K.J. O’Riordan, J.M. Levenson, J.D. Sweatt,ERK/MAPK regulates hippocampal histone phosphorylationfollowing contextual fear conditioning, Learn. Mem. 13 (2006)322–328.

[20] C.M. Colvis, J.D. Pollock, R.H. Goodman, S. Impey, J. Dunn, G.Mandel, F.A. Champagne, M. Mayford, E. Korzus, A. Kumar, W.Renthal, D.E. Theobald, E.J. Nestler, Epigenetic mechanisms andgene networks in the nervous system, J. Neurosci. 25 (2005) 10379–10389.

[21] J.P. Curley, S. Barton, A. Surani, E.B. Keverne, Coadaptation inmother and infant regulated by a paternally expressed imprintedgene, Proc. R. Soc. Lond. B 271 (2004) 1303–1309.

[22] J.P. Curley, S.B. Pinnock, S.L. Dickson, R. Thresher, N. Miyoshi,M.A. Surani, E.B. Keverne, Increased body fat in mice with a targetmutation of the paternally expressed imprinted gene Peg3, FASEB19 (2005) 1302–1304.


[23] G.L. Cuthbert, S. Daujat, A.W. Snowden, H. Erdjument-Bromage,T. Hagiwara, M. Yamada, R. Schneider, P.D. Gregory, P. Tempst,A.J. Bannister, T. Kouzarides, Histone deamination antagonizesarginine methylation, Cell 118 (2004) 545–553.

[24] A.C. D’Alessio, M. Szyf, Epigenetic tete-a-tete: the bilateralrelationship between chromatin modifications and DNA methyla-tion, Biochem. Cell Biol. 84 (2006) 463–476.

[25] R. De La Fuente, C. Baumann, T. Fan, A. Schmidtmann, I.Dobrinski, K. Meugge, Lsh is required for meiotic chromosomesynapsis and retrotransposon silencing in female germ cells, Nat.Cell Biol. 8 (2006) 1448–1454.

[26] T. Deng, Y. Kuang, D. Zhang, R. Sun, G. Xu, Z. Wang, J. Fei,Disruption of imprinting and aberrant embryo development incompletely inbred embryonic stem cell-derived mice, Dev. GrowthDiffer. 49 (2007) 603–610.

[27] J. Diorio, M.J. Meaney, Maternal programming of defensiveresponses through sustained effects on gene expression, J. PsychiatryNeurosci. 32 (2007) 275–284.

[28] S. Erhardt, I.-H. Su, R. Schneider, S. Barton, A.J. Bannister, L.Perez-Burgos, T. Jenuwein, T. Kouzarides, A. Tarakhovsky, M.A.Surani, Consequences of the depletion of zygotic and embryonicenhancer of zeste 2 during preimplantation mouse development,Development 130 (2003) 4235–4248.

[29] A. Fischer, F. Sananbenesi, P.T. Pang, B. Lu, L.H. Tsai, Opposingroles of transient and prolonged expression of p25 in synapticplasticity and hippocampus-dependent memory, Neuron 48 (2005)825–838.

[30] A. Fischer, F. Sananbenesi, X. Wang, M. Dobbin, L.H. Tsai,Recovery of learning and memory is associated with chromatinremodeling, Nature 447 (2007) 178–182.

[31] G. Felsenfeld, M. Groudine, Controlling the double helix, Nature421 (2003) 448–453.

[32] M.F. Fraga, E. Ballestar, M.F. Paz, S. Ropero, F. Satien, M.L.Ballestar, D. Heine-Suner, J.C. Cigudosa, M. Urioste, J. Benitez, M.Boix-Charnet, A. Sanchez-Aguilera, C. Ling, E. Carlsson, P.Poulsen, A. Vaag, Z. Stephan, T.D. Spector, Y.-Z. Wu, C. Plass,M. Esteller, Epigenetic differences arise during the lifetime ofmonozygotic twins, Proc. Natl. Acad. Sci. USA 102 (2005) 10604–10609.

[33] V.X. Fu, S.R. Schqarze, M.L. Kenowski, S. Leblanc, J. Svaren, D.F.Jarrard, A loss of insulin-like growth factor-2 (IGF2) imprinting ismodulated by CTFC downregulation at senescence in humanepithelial cells, J. Biol. Chem. 279 (2004) 52218–52226.

[34] F. Fuks, DNA methylation and histone modifications: teaming upto silence genes, Curr. Opin. Genet. Dev. 15 (2005) 490–495.

[35] P.A. Grant, A tale of histone modifications, Genome Biol. 2 (2001)0003.1–0003.6 (reviews).

[36] D. Greig, M. Travisano, Haploid superiority, Science 299 (2003)524–525.

[37] G. Haig, Genomic imprinting and kinship: how good is theevidence? Annu. Rev. Genet. 38 (2004) 583–585.

[38] D. Haig, C. Graham, Genomic imprinting and the strange case ofthe insulin-like growth factor II receptor, Cell 64 (1991) 1045–1046.

[39] P.O. Hassa, S.S. Haenni, M. Elser, M.O. Hottiger, Nuclear ADP-ribosylation reactions in mammalian cells: where are we today andwhere are we going? Microbiol. Mol. Biol. Rev. 70 (2006) 789–829.

[40] S.E. Hiby, M. Lough, E.B. Keverne, M.A. Surani, Y.W. Loke, A.King, Paternal monoallelic expression of PEG3 in the humanplacenta, Hum. Mol. Genet. 10 (2001) 1093–1100.

[41] T. Hitichi, T. Kohda, T.K. Ishino, F. Ishino, Imprinting regulationof the murine Meg1/Grb10 and human GRB10 genes; roles of brain-specific promoters and mouse-specific CTCF-binding sites, NucleicAcids Res. 31 (2003) 1398–1406.

[42] S.E. Hyman, R.C. Malenka, E.J. Nestler, Neural mechanisms ofaddiction: the role of reward-related learning and memory, Annu.Rev. Neurosci. 29 (2006) 565–598.

[43] T. Jenuwein, C.D. Allis, Translating the histone code, Science 293(2001) 1074–1080.



ARTICLE IN PRESS

[44] C. Jin, G. Felsenfeld, Nucleosome stability mediated by histonevariants H3.3 and H2A.Z, Genes Dev. 21 (2007) 1519–1529.

[45] R.L. Jirtle, M.K. Skinner, Environmental epigenomics and diseasesusceptibility, Nat. Rev. Genet. 8 (2007) 253–262.

[46] A. Kalda, L.T. Heidmets, H.Y. Shen, A. Zharkovsky, J.F. Chen,Histone deacetylase inhibitors modulates the induction and expres-sion of amphetamine-induced behavioral sensitization partiallythrough an associated learning of the environment in mice, Behav.Brain Res. 181 (2007) 76–84.

[47] T. Kaneko-Ishino, T. Kohda, F. Ishino, The regulation andbiological significance of genomic imprinting in mammals, J.Biochem. 133 (2003) 699–711.

[48] J. Kim, A. Bergmann, J.H. Choo, L. Stubbs, Genomic organizationand imprinting of the Peg3 domain in bovine, Genomics 90 (2007)85–92.

[49] J. Kim, A. Killhoff, A. Bergmann, L. Stubbs, Methylation-sensitivebinding of transcription factor YY1 to an insulator sequence withinthe paternally expressed imprinted gene, Peg, Hum. Mol. Genet. 12(2003) 233–245.

[50] E. Korzus, M.G. Rosenfeld, M. Mayford, CBP histone acetyltrans-ferase activity is a critical component of memory consolidation,Neuron 42 (2004) 961–972.

[51] N. Kothapalli, G. Camporeale, A. Kueh, Y.C. Chew, A.M.Oommen, J.B. Griffin, J. Zempleni, Biological functions of biotin-ylated histones, J. Nutr. Biochem. 16 (2005) 446–448.

[52] T. Kouzarides, Chromatin modifications and their function, Cell 128(2007) 693–705.

[53] A. Kumar, K.H. Choi, W. Renthal, N.M. Tsankova, D.E. Theo-bald, H.T. Truong, S.J. Russo, Q. Laplant, T.S. Sasaki, K.N.Whistler, R.L. Neve, D.W. Self, E.J. Nestler, Chromatin remodelingis a key mechanism underlying cocaine-induced plasticity in stria-tum, Neuron 48 (2005) 303–314.

[54] M. Lachner, T. Jenuwein, The many faces of histone lysinemethylation, Curr. Opin. Cell Biol. 14 (2002) 286–298.

[55] L. Lefebvre, S. Viville, S.C. Barton, F. Ishino, E.B. Keverne, M.A.Surani, Abnormal maternal behaviour and growth retardationassociated with loss of the imprinted gene Mest, Nat. Genet. 20(1998) 108–109.

[56] J.M. Levenson, J.D. Sweatt, Epigenetic mechanisms in memoryformation, Nat. Rev. Neurosci. 6 (2005) 108–118.

[57] J.M. Levenson, K.J. O’Riordan, K.D. Brown, M.A. Trinh, D.L.Molfese, J.D. Sweatt, Regulation of histone acetylation duringmemory formation in the hippocampus, J. Biol. Chem. 279 (2004)40545–40559.

[58] L.L. Li, E.B. Keverne, S.A. Aparicio, F. Ishino, S.C. Barton, M.A.Surani, Regulation of maternal behavior and offspring growth bypaternally expressed Peg3, Science 284 (1999) 330–333.

[59] T. Li, T.H. Vu, G.A. Ulaner, E. Littman, J.Q. Ling, H.L. Chen, J.F.Hu, B. Behr, L. Guidice, A.R. Hoffman, IVF results in de novoDNA methylation and histone methylation at an Igf2–H19 imprint-ing epigenetic switch, Mol. Hum. Reprod. 9 (2005) 631–640.

[60] S.-P. Lin, N. Youngson, S. Takada, H. Seitz, W. Reik, M. Paulsen,J. Cavaille, A. Ferguson-Smith, Asymmetric regulation of imprint-ing on maternal and paternal chromosomes at the Dlk1–Gtl2imprinted cluster on mouse chromosome 12, Nat. Genet. 35 (2003)97–102.

[61] D. Lucifero, M.R. Mann, M.S. Bartolemei, J.M. Trasler, Gene-specific timing and epigenetic memory in oocyte imprinting, Hum.Mol. Genet. 13 (2004) 839–849.

[62] D. Mancini-Dinardo, S.J. Steele, J.M. Levorse, R.S. Ingram, S.M.Tilghman, Elongation of the Kcnq1ot1 transcript is required forgenomic imprinting of neighboring genes, Genes Dev. 20 (2006)1268–1282.

[63] W. Mayer, M. Hemberger, H.G. Frank, R. Grummer, E. Winterh-ager, P. Kaufmann, R. Fundele, Expression of the imprinted genesMEST/Mest in human and murine placenta suggest a role inangiogenesis, Dev. Dyn. 217 (2000) 1–10.


[64] W. Mayer, A. Niveleau, J. Walter, R. Fundele, T. Haaf,Demethylation of the zygotic paternal genome, Nature 403(2000) 501–502.

[65] B. Mei, C. Li, S. Dong, C.H. Jiang, H. Wang, Y. Hu, Distinct geneexpression profiles in hippocampus and amygdala after fear condi-tioning, Brain Res. Bull. 67 (2005) 1–12.

[66] E.L. Mersfelder, M.R. Parthun, The tale beyond the tail: histonecore domain modifications and the regulation of chromatin struc-ture, Nucleic Acids Res. 34 (2006) 2653–2662.

[67] C.A. Miller, J.D. Sweatt, Covalent modification of DNA regulatesmemory formation, Neuron 53 (2007) 857–869.

[68] T. Nakamura, Y. Arai, H. Umehara, M. Mashuhara, T. Kimura, H.Taniguchi, T. Sekimoto, M. Ikawa, Y. Yoneda, M. Okabe, et al.,PgC7/Stella protects against DNA demethylation in early embryo-genesis, Nat. Cell Biol. 9 (2007) 64071.

[69] G.J. Narlikar, H.Y. Fan, R.E. Kingston, Cooperation betweencomplexes that regulate chromatin structure and transcription, Cell108 (2002) 475–487.

[70] D. Nathan, K. Ingvarsdottir, D.E. Sterner, G.R. Bylebyl, M.Dokmanovic, J.A. Dorsey, K.A. Whelan, M. Krsmanovic, W.S.Lane, P.B. Meluh, E.S. Johnson, S.L. Berger, Histone sumoylationis a negative regulator in Saccharomyces cerevisiae and showsdynamic interplay with positive-acting histone modifications, GenesDev. 20 (2006) 966–976.

[71] C.J. Nelson, H. Santos-Rosa, T. Kouzarides, Proline isomerizationof histone H3 regulates lysine methylation and gene expression, Cell126 (2006) 905–916.

[72] S.J. Nowak, V.G. Corces, Phosphorylation of histone H3: abalancing act between chromosome condensation and transcrip-tional activation, Trends Genet. 20 (2004) 214–220.

[73] Y. Obata, T. Kono, I. Hatada, Gene silencing: maturation of mousefetal germ cells in vitro, Nature 418 (2002) 497.

[74] H. Ogawa, Y. Ono, N. Shimozawa, Y. Sotomaru, Y. Katsuzawa, H.Huira‘, M. Ito, T. Kono, Disruption of imprinting in cloned mousefetuses from embryonic stem cells, Reproduction 126 (2003) 549–557.

[75] A. Paolini-Giacobino, L. D’Aiuto, M.C. Cirio, B. Reinhart, J.R.Chaillet, Conserved features of imprinted differentially methylateddomains, Gene 399 (2007) 33–45.

[76] D. Pepin, B.C. Vanderhyden, D.J. Picketts, B.D. Murphy, ISWIchromatin remodelling in ovarian somatic and germ cells: revenge ofthe NURFs, Trends Ecol. Metab. 18 (2007) 215–224.

[77] E. Rand, I. BenpOorath, I. Keshet, H. Cedar, CTDF elements directallele-specific undermethylation at the imprinting H19 locus, Curr.Biol. 14 (2004) 1002–1012.

[78] E. Rand, H. Cedar, Regulation of imprinting: a multi-tiered process,J. Cell. Biochem. 88 (2003) 400–407.

[79] W. Reik, Stability and flexibility of epigenetic gene regulation inmammalian development, Nature 447 (2007) 425–431.

[80] W. Reik, M. Constancia, A. Fowden, N. Anderson, W. Dean, A.Ferguson-Smith, B. Tycko§, C. Subley, Regulation of supply anddemand for maternal nutrients in mammals by imprinted genes, J.Physiol. 547 (2003) 35–44.

[81] W. Reik, W. Dean, J. Walter, Epigenetic reprogramming inmammalian development, Science 293 (2001) 1089–1093.

[82] W. Reik, J. Walter, Evolution of imprinting mechanisms: the battleof the sexes begins in the zygote, Nat. Genet. 27 (2001) 255–256.

[83] W. Reik, J. Walter, Genomic imprinting: parental influence on thegenome, Nat. Rev. 2 (2001) 21–32.

[84] B. Reinhart, A. Paoloni-Giacobino, J.R. Chaillet, Specific differen-tially methylated domain sequences direct the maintenance ofmethylation at imprinted genes, Mol. Cell. Biol. 26 (2006) 8347–8356.

[85] F. Relaix, X. Wei, W. Li, J. Pan, Y. Lui, D.D. Bowtell, D.A.Sassoon, X. Wu, Pw1/Peg3 is a potential cell death mediator andcooperates with Siah1a in p53-mediated apoptosis, Proc. Natl. Acad.Sci. USA 97 (2000) 2105–2110.



ARTICLE IN PRESS

[86] J.M. Reul, Y. Chandramohan, Epigenetic mechanisms in stress-related memory formation, Psychoneuroendocrinology 32 (2007)S21–S25.

[87] F. Santos, W. Dean, Epigenetic reprogramming during earlydevelopment in mammals, Reproduction 127 (2004) 643–651.

[88] H. Santos-Rosa, R. Schneider, A.J. Bannister, J. Sherriff, B.E.Bernstein, N.C. Emre, S.L. Schreiber, J. Mellor, T. Kouzarides,Active genes are tri-methylated at K4 of histone H3, Nature 419(2002) 407–411.

[89] Y. Shi, F. Lan, C. Matson, P. Mulligan, J.R. Whetstine, P.A. Cole,R.A. Casero, Y. Shi, Histone demethylation mediated by the nuclearamine oxidase homolog LSD1, Cell 119 (2004) 941–953.

[90] Y. Shi, J.R. Whetstine, Dynamic regulation of histone lysinemethylation by demethylases, Mol. Cell 25 (2007) 1–14.

[91] Y. Shiio, R.N. Eisenman, Histone sumoylation is associated withtranscriptional repression, Proc. Natl. Acad. Sci. USA 100 (2003)13225–13230.

[92] A. Shilatifard, Chromatin modifications by methylation and ubiq-uitination: implications in the regulation of gene expression, Annu.Rev. Biochem. 75 (2006) 243–269.

[93] C. Simone, SWI/SNF: the crossroads where extracellular signal-ing pathways meet chromatin, J. Cell. Physiol. 207 (2006) 309–314.

[94] F.M. Smith, L.J. Holt, A.S. Garfield, M. Charalambous, F.Koumanov, M. Perry, R. Bazzini, S.A. Sheardown, B.D. Hegarty,R.J. Lyons, G.J. Cooney, R.J. Daly, A. Ward, Mice with adisruption of the imprinted Grb10 gene exhibit altered bodycomposition, glucose homeostasis and insulin signalling duringpost-natal life, Mol. Cell. Biol. 16 (2007) 5871–5876.

[95] E.R. Sowell, P.M. Thompson, A.W. Toga, Mapping changes in thehuman cortex throughout the span of life, Neuroscientist 10 (2004)372–392.

[96] L. Spahn, D.P. Barlow, An ICE pattern crystallizes, Nat. Genet. 35(2003) 11–12.

[97] H.G. Spencer, Population genetics and evolution of genomicimprinting, Annu. Rev. Genet. 34 (2000) 457–477.

[98] D.E. Sterner, S.L. Berger, Acetylation of histones and transcription-related factors, Microbiol. Mol. Biol. Rev. 64 (2000) 435–459.

[99] M.A. Surani, Reprogramming of genome function through epige-netic inheritance, Nature 414 (2001) 122–128.

[100] A.M. Surani, K. Hayashi, P. Hajkova, Genetic and epigeneticregulators of pluripotency, Cell 128 (2007) 747–762.

[101] W.T. Swaney, J.P. Curley, F.A. Champagne, E.B. Keverne, Geno-mic imprinting mediates sexual experience-dependent olfactorylearning in male mice, Proc. Natl. Acad. Sci. USA 104 (2007)6084–6089.

[102] P.E. Szabo, S.H. Tang, F.J. Silva, W.M. Tsark, J.R. Mann, Role ofthe CTCF binding sites in the Igf2/H19 imprinting region, Mol. Cell.Biol. 24 (2004) 4791–4800.

[103] C. Tanny, H. Erdjument-Bromage, P. Tempst, C.D. Allis, Ubiqui-tylation of histone H2B controls RNA polymerase II transcriptionelongation independently of histone H3 methylation, Genes Dev. 21(2007) 835–847.

[104] M.E. Torres-Padilla, D.E. Parfitt, T. Kouzarides, M. Zernick-Goetz,Histone arginine methylation regulates pluripotency in the earlymouse embryo, Nature 445 (2007) 214–218.

[105] J.L. Thorvaldsen, D.L. Duran, M.S. Bartolemei, Deletion of theH19 differentially methylated domain results in loss of imprintedexpression of H19 and Igf2, Genes Dev. 12 (1998) 3693–3702.

[106] N.M. Tsankova, A. Kumar, E.J. Nestler, Histone modifica-tions at gene promoter regions in rat hippocampus after acute


and chronic electroconvulsive seizures, J. Neurosci. 24 (2004)5603–5610.

[107] N.M. Tsankova, O. Berton, W. Renthal, A. Kumar, R.L. Neve, E.J.Nestler, Sustained hippocampal chromatin regulation in a mousemodel of depression and antidepressant action, Nat. Neurosci. 9(2006) 519–525.

[108] N. Tsankova, W. Renthal, A. Kumar, E.J. Nestler, Epigeneticregulation in psychiatric disorders, Nat. Rev. Neurosci. 8 (2007)355–367.

[109] T.H. Vu, T. Li, A.R. Hoffman, Promoter-restricted histone code, notthe differentially methylated DNA regions or antisense transcripts,marks the imprinting status of IGF2R in human and mouse, Hum.Mol. Genet. 13 (2004) 2233–2245.

[110] C.H. Waddington, Canalization of development and the inheritanceof acquired characters, Nature 150 (1942) 563–565.

[111] R.A. Walterland, R.L. Jirtle, Transposable elements: targets forearly nutritional effects on epigenetic gene regulation, Mol. Cell.Biol. 15 (2003) 5293–5300.

[112] Y. Wang, J. Wysocka, J. Sayegh, Y.H. Lee, J.R. Perlin, L. Leonelli,L.S. Sonbuchner, C.H. McDonald, R.G. Cook, Y. Dou, et al.,Human PAD4 regulates histone arginine methylation levels viademethylimination, Science 306 (2004) 279–283.

[113] I.C. Weaver, N. Cervoni, F.A. Champagne, A.C. D’Alessio, S.Sharma, J.R. Seckl, S. Dymov, M. Szyf, M.J. Meaney, Epigeneticprogramming by maternal behavior, Nat. Neurosci. 7 (2004) 847–854.

[114] F.H. Wilkinson, K. Park, M.L. Atchison, Polycomb recruitment toDNA in vivo by the YY1 REPO domain, Proc. Natl. Acad. Sci.USA 103 (2006) 19296–19301.

[115] C.M. Williamson, M.D. Turner, S.T. Ball, W.T. Nottingham, P.Glenister, M. Fray, Z. Tymowska-Lalanne, A. Plagge, N. Powles-Glover, G. Kelsey, M. Maconochie, J. Peters, Identification of animprinting control region affecting the expression of all transcripts inthe Gnas cluster, Nat. Genet. 38 (2006) 350–355.

[116] J.B. Wolf, Gene interactions from maternal effects, Evolution 54(2000) 1882–1898.

[117] M.A. Wood, M.P. Kaplan, A. Park, E.J. Blanchard, T.L. Lombardi,T. Abel, Transgenic mice expressing a truncated form of CREB-binding protein (CBP) exhibit deficits in hippocampal synapticplasticity and memory storage, Learn. Mem. 12 (2005) 111–119.

[118] A. Wutz, O.W. Smrzka, N. Schweifer, K. Schellander, E.F. Wagner,D.P. Barlow, Imprinted expression of the Igf2r gene depends on anintronic CpG island, Nature 389 (1997) 745–749.

[119] Z. Xhoa et al., Circular organization of active and inactivechromatin domains uncovered by chromosome conformation cap-ture-on-chip (4C), Nat. Genet. 38 (2006) 1341–1347.

[120] Y. Yamasaki, T. Kayashima, H. Soejima, A. Kinoshita, K.Yoshiura, N. Matsumoto, T. Ohta, T. Urano, H. Masuzaki, T.Ishimaru, T. Mukai, N. Niikawa, T. Kishino, Neuron-specificrelaxation of Igf2r imprinting is associated with neuron-specifichistone modifications and lack of its antisense transcript Air, Hum.Mol. Genet. 14 (2005) 2511–2520.

[121] Y. Yamasaki-Ishizaki, T. Kayashima, C.K. Mapendano, H. Soej-ima, T. Ohta, H. Masuzaki, A. Kinoshita, T. Urano, K. Yoshiura,N. Matsumato, T. Ishimaru, T. Mikai, N. Niikawa, T. Kishino,Role of DNA methylation and histone H3 lysine 27 methylation intissue-specific imprinting of mouse Grb10, Mol. Cell. Biol. 27 (2007)732–742.

[122] Y. Zhang, D. Reinberg, Transcription regulation by histonemethylation: interplay between different covalent modifications ofthe core histone tails, Genes Dev. 15 (2006) 2343–2360.


epigenetics, brain evolution and behaviour · review epigenetics, brain evolution and behaviour...

Documents