Opinion TRENDS in Neurosciences Vol.28 No.

9 September 2005

Maternal care as a model for

experience-dependent chromatin
plasticity?
Michael J. Meaney1 and Moshe Szyf 2
1
Douglas Hospital Research Center, 6875 LaSalle Boulevard, Montreal, Quebec H4H 1R3, Canada
2
Department of Pharmacology and Therapeutics, McGill University, 3655 Sir William Oslar Promenade, Montreal,
Quebec H3G 1Y6, Canada

It is widely acknowledged that the nature of the first week postpartum (high-LG mothers) show reduced
maternal care a child receives can have long-term reper- plasma adrenocorticotropin hormone (ACTH) and cortico-
cussions, and that children raised in deprived environ- sterone responses to acute stress compared with adult
ments can have severe cognitive and behavioural offspring of low-LG mothers [12,14]. These effects are, in
difficulties that last into adulthood. The mechanisms part, mediated by the influence of maternal care on gene
underlying these effects are not understood, but recent expression. The offspring of high-LG mothers show
data from rodents provide insight into a potential
molecular mechanism. Like humans, rodent maternal
behaviour towards offspring can effect long-term Hippocampus
changes in responses of the offspring to stress through-
out the rest of their lives. Remarkably, these changes
reflect permanently altered gene expression, so-called
environmental programming, and its downstream
effects on the hypothalamicpituitaryadrenal axis.
This review discusses the nature of this environmental
programming the mechanism by which it occurs in
rats, its long-term implications, and opportunities for its
reversal in rodents and ultimately in humans.

Introduction
The quality of early family life influences health through- Hypothalamus
out life [1]. In part, such influences appear to be mediated
by effects of parentoffspring interactions on the develop-
ment of neural systems, including those that mediate CRF
responses to stress [26]. There is evidence in humans
that parental care can affect endocrine and autonomic
Pituitary
responses to stress that endure into adulthood. Dramatic
examples derive from studies on the effects of sexual
and/or physical abuse on stress responses [7]. These
ACTH
findings represent environmental programming instances
where exposure to an environmental event in develop-
ment stably alters phenotype in adulthood. A model for
such programming effects lies in older studies on hand-
ling of newborn rats. Such handling decreases the mag-
nitude of hypothalamicpituitaryadrenal (HPA; Figure 1) Adrenal
responses to stress in adulthood [810]. Interestingly, glands Glucocorticoids
these effects are mediated by changes in maternal care.
Handling increases pup licking and grooming (LG) by the TRENDS in Neurosciences
rat mother [1113]. Predictably, adult offspring of mothers
Figure 1. In response to stress, CRF, and co-secretagogues such as vasopressin, are
that naturally (i.e. in the absence of any experimental released from paraventricular nucleus of the hypothalamus (dark oval) into the
manipulation) show increased levels of pup LG over the portal blood supply of the anterior pituitary, where they stimulate the synthesis and
release of ACTH. Pituitary ACTH, in turn, causes release of glucocorticoids from the
Corresponding authors: Meaney, M.J. ([email protected]), Szyf, M. adrenal gland. CRF synthesis and release is subsequently inhibited through a
([email protected]). glucocorticoid negative-feedback system mediated by mineralocorticoid and
Available online 27 July 2005 glucocorticoid receptors in several brain regions, including the hippocampus.

www.sciencedirect.com 0166-2236/$ - see front matter Q 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.tins.2005.07.006
 Opinion TRENDS in Neurosciences Vol.28 No.9 September 2005 457

Tactile stimulation
(maternal LG) 5-HT

5-HT7 receptor

cAMP
PKA CBP

NGFI-A AP-2

GR gene

TRENDS in Neurosciences

Figure 2. A summary of in vivo studies using hippocampal tissue from neonates and in vitro studies using primary hippocampal cell cultures. In vivo, both handling and
increased maternal licking and grooming (LG) increase hippocampal 5-HT turnover. This increases activation of a 5-HT7 receptor positively coupled to cAMP, PKA and the
CREB-binding protein CBP. In vivo, both handling and increased maternal LG increase hippocampal cAMP concentrations and activation of PKA in pups [1720]. Postnatal-
day 6 offspring of high-LG mothers or pups of the same age exposed to handling show increased hippocampal expression of factors such as NGFI-A (also known as zif-268,
krox-24, egr-1 and zenk) and the adaptor protein AP-2 [20]. In vitro, 5-HT increases glucocorticoid receptor (GR) expression in cultured hippocampal neurons; the effect of 5-HT
on GR expression is mimicked with 8-bromo-cAMP and completely blocked by concurrent treatment using 5-HT7 receptor antagonists [17], a PKA inhibitor or an antisense
oligonucleotide directed at the NGFI-A mRNA [22]. Grey shading of the GR gene represents exons 17 and 29 (see Figure 3).

significantly increased hippocampal glucocorticoid recep- or 5-HT) turnover and in hippocampal expression of the
tor (GR) mRNA and protein expression, enhanced gluco- transcription factor nerve-growth-factor-inducible factor
corticoid negative feedback sensitivity and decreased A (NGFI-A, also known as zif-268, krox-24, egr-1 and
hypothalamic corticotrophin release factor (CRF) mRNA zenk; Figure 2). In vitro, 5-HT increases expression of both
levels (Figure 1). Cross-fostering the biological offspring of GR and NGFI-A in cultured rat hippocampal neurons, and
high-LG and low-LG mothers reverses the phenotype, the effect of 5-HT is blocked by concurrent treatment with
suggesting a direct relationship between variations in an antisense oligonucleotide directed to NGFI-A mRNA
maternal care and development of HPA responses to stress [22]. The 5 0 non-coding variable exon 1 region of the rat
[15]. Finally, reversing the effect on hippocampal GR levels hippocampal GR gene (Figure 3) contains multiple alter-
eliminates the influence of early experience on HPA nate sequences, including the exon 17 sequence, which
responses to stress [16]. appears to be a brain-specific promoter [23] (see [24,25] for
comparable sequences in the mouse). In adult rats, hippo-
Molecular mechanisms for maternal effects on HPA campal expression of GR mRNA splice variants contain-
responses to stress ing exon 17 is increased by postnatal handling [23] or
Results of in vitro and in vivo studies [10,1721] suggest maternal LG. Exon 17 contains an NGFI-A-binding
that maternal effects on hippocampal GR expression are consensus sequence (GCGGGGGCG) [26]. Maternal LG
mediated by increases in 5-hydroxytryptamine (serotonin increases hippocampal NGFI-A expression, and chromatin

11 14 15 16 1718 19 110 111 29

GR promoter 17 sequence

1681 ccc
1741 ctctgctagt gtgacacact t1cg2cgcaact c 3cgcagttgg4cggg 5cg 6cgga ccacccctg7c
1801 ggctctgc8cg gctggctgtc accct9cgggg gctctggctg c10cgaccca11cg ggg12cgggct
1861 c13cgag14cggtt ccaagcct15cg gagtggg16cg gggg17cgggag ggagcctggg agaa
NGFI-A

0
Figure 3. The 5 non-coding variable exon 1 region of the rat hippocampal glucocorticoid receptor gene contains multiple alternate exon 1 sequences [23], four of which
(exons 11, 15, 19 and 110) correspond to alternative exon 1 sequences previously identified in mice [24,25]. Transfection studies in various cell types show that activity of
constructs of the different exons fused to a luciferase reporter is similar, with one notable exception the exon 17 promoter has higher activity than the other promoter
constructs [23]. CpG dinucleotides in the 17 promoter sequence are numbered, and the NGFI-A-recognition element [26] is encircled; its 5 0 and 3 0 ends are CpG dinucleotide
sites 16 and 17. Exons 29 represent the coding region of the GR gene.

www.sciencedirect.com
458 Opinion TRENDS in Neurosciences Vol.28 No.9 September 2005

immunoprecipitation (ChIP) assays with hippocampal the fifth position in the cytosine ring. The carboncarbon
samples from postnatal day (P)6 pups reveal that NGFI-A bond between the methyl group and cytosine residue is a
binding to the exon 17 promoter is dramatically increased stable, enduring epigenetic mark. DNA methylation
in the offspring of high-LG mothers compared with the patterns are commonly thought to be established during
offspring of low-LG mothers (I.C.G. Weaver et al., early embryonic development and then faithfully maintained
unpublished). Co-transfection of a NGFI-A vector and an through life by DNA methyltransferases [3032,38,39].
exon 17luciferase construct into human embryonic The only way methyl residues were thought to be lost
kidney (HEK) cells increases luciferase activity, reflecting was through passive demethylation (i.e. replication in the
NGFI-A-induced activation of transcription through the absence of DNA methyltransferase) and such processes
exon 17 promoter. were not considered applicable to postmitotic cells, includ-
These findings suggest that NGFI-A might increase ing neurons. Methylation is involved in gene imprinting
GR expression in hippocampal neurons and provide a and in early phases of cellular differentiation [38]. This
mechanism for the effect of maternal care over the first raises the questions of whether cellular programming
week of life. However, although there are striking through alterations in DNA methylation is unique to this
differences in NGFI-A expression in the neonatal offspring period, and whether variations in methylation marks
of high-LG and low-LG mothers, there is no effect of occur solely through passive demethylation.
maternal care on hippocampal NGFI-A expression in the DNA methylation promotes gene silencing through
adult. Therefore, we asked whether the increased NGFI-A effects on chromatin structure [3032,4042]. In one
exon 17 interaction in the pups of high-LG mothers might silencing mechanism, binding of a transcription factor to
result in an epigenetic modification of the exon 17 its recognition element is interfered with directly, by a
sequence that alters NGFI-A binding and maintains the methylated cytosineguanine pair (CpG) within the
maternal effect into adulthood. element such a mechanism inhibits binding of c-Myc to
its recognition element [27]. Essentially, the methylated
The epigenome: chromatin structure and DNA cytosine serves as a mutation of the recognition element.
methylation A second silencing mechanism is indirect, and links DNA
Most DNA is tightly packaged into nucleosomes and methylation to inactive chromatin structure. A region of
wrapped around a core of histone proteins [27]. The methylated DNA attracts different members of a family of
histoneDNA configuration is maintained by electrostatic methylated DNA-binding proteins, such as MeCP2, which
bonds between positively-charged histones and negatively- recruit HDACs [43,44] and histone methyltransferases
charged DNA, and regulates gene expression [28]. This [44] to methylated genes [32,44], blocking histone acetyl-
closed chromatin structure commonly precludes tran- ation and inactivating chromatin. MeCP2 also facilitates
scription-factor binding to DNA and underscores the chromatin inactivation through chromatin looping [45]
importance of enzymes that modify histoneDNA inter- and recruitment of Brahma (Brm), a catalytic component
actions. One class of such proteins, which includes many of the SWI/SNF-related chromatin-remodelling complex
known transcriptional cofactors, are histone acetyltrans- [46,47]. Whereas the first mechanism refers to a discrete
ferases (HATs) [29]. HATs catalyze the acetylation of methylation pattern, the second depends on the general
selected positively-charged amino acids (e.g. lysine and density of methylated cytosine residues.
arginine) on the protruding histone tails, most commonly The model positioning DNA methylation as driving
histone (H)3 or H4. Histone acetylation of the lysine (K) chromatin inactivation is well established. Nevertheless,
residue on H3 neutralizes the positively-charged histone, current data suggest that chromatin structure can also
opening chromatin and facilitating transcription-factor determine DNA methylation, and that chromatin can
binding to DNA. Thus, H3-K9 acetylation is a marker of affect DNA methylation in both directions, triggering
active gene transcription. Histone acetylation is regulated either de novo DNA methylation [48] or demethylation
by histone deacetylases (HDACs), which block histone [4951]. These findings indicate that DNA methylation,
acetylation and suppress gene expression. Thus, intra- although extremely stable, can be altered later in life
cellular signals can regulate gene expression through when there is a sufficiently stable and consistent signal
effects on HATs or HDACs [3033] (or on factors involved for chromatin activation. This relationship between
in other histone modifications, such as methylation, phos- chromatin state and DNA methylation forms a molecular
phorylation, ribosylation and ubiquitination) that alter link through which environmental signals might alter
chromatin structure. Indeed, some environmentally regu- DNA methylation in specific genes in postmitotic neurons.
lated alterations of histone acetylation in specific pro- According to this model, environmental signals trigger
moter sequences following seizures [34,35] or learning [36] cellular signalling pathways, which activate trans-acting
are likely to be caused by neurotransmitter activation of factors that recruit HATs, resulting in histone acetylation,
multiple signalling pathways [37]. However, such histone chromatin opening and increased accessibility to DNA
modifications are transient and cannot directly explain demethylating agents. This mechanism would enable
enduring early environmental programming effects. reversal of the methylation mark by a similar intense
A more likely, highly stable candidate could involve change in chromatin structure later in life [52].
modification of the genome itself.
DNA is covalently modified by DNA methyltrans- Epigenetic programming of HPA stress responses
ferases that transfer the methyl group from the methyl We initially found evidence that methylation across the
donor S-adenosyl methionine (SAM) to the carbon atom at GR exon 17 promoter sequence in the hippocampus was
www.sciencedirect.com
 Opinion TRENDS in Neurosciences Vol.28 No.9 September 2005 459

greater in adult offspring of low-LG mothers than in those this approach could reverse the epigenetic state of GR in
of high-LG mothers [14]. These findings suggested the adult offspring of low-LG mothers. Central infusion
maternal effects on DNA methylation patterns in off- of adult rats with the HDAC-inhibitor trichostatin A
spring. We then examined the methylation status of (TSA) [55] significantly increased H3 acetylation, cytosine
individual CpGs in the exon 17 sequence using sodium demethylation and NGFI-A binding at the exon 17 site in
bisulfite mapping [53], focusing on the NGFI-A consensus the offspring of low-LG mothers, to levels comparable with
sequence (Figure 3). The results reveal significant those observed in the offspring of high-LG mothers.
differences in cytosine methylation within the 5 0 CpG Expression profiling of TSA-treated rats reveal specific
dinucleotide of the NGFI-A consensus sequence. This site effects of TSA on the hippocampal transcriptome (I.C.G.
is always methylated in the offspring low-LG mothers, and Weaver et al., unpublished). The enhanced NGFI-A bind-
rarely methylated in those of high-LG dams. Cross- ing to the exon 17 promoter is associated with increased
fostering reverses the differences in the methylation of hippocampal GR expression in the offspring of low-LG
the 5 0 CpG site and suggests a direct relationship between mothers, to levels that are indistinguishable from those
maternal behaviour and changes in DNA methylation of the high-LG offspring. Most importantly, TSA infusion
within the GR exon 17 promoter [14]. The effect of also eliminates the effect of maternal care on HPA
maternal care is remarkably specific, with highly signifi- responses to acute stress. Because there is considerable
cant alterations in the methylation status of the 5 0 CpG H3 acetylation and NGFI-A binding at the exon 17
site but no effect on the 3 0 site. sequence in the untreated offspring of high-LG mothers,
The difference in methylation within the 5 0 CpG TSA is without effect in these animals. These results
dinucleotide of the NGFI-A-response element suggests suggest causal relationships between maternal care,
alteration of NGFI-A binding. In vitro, binding of increas- histone acetylation, DNA methylation of the GR exon 17
ing concentrations of purified recombinant NGFI-A promoter, GR expression and HPA responses to stress.
protein [54] to its response element was examined [21] These findings, and others discussed later in this
using an electrophilic mobility shift assay (EMSA), with review, suggest that DNA methylation patterns are
32
P-labelled synthetic oligonucleotide sequences bearing dynamic and potentially reversible even in adult neurons,
the NGFI-A-binding site differentially methylated at the which presumably bear the machinery required to
5 0 or 3 0 CpG sites. The results indicate that methylation demethylate genes. If DNA methylating and demethylat-
of the cytosine within the 5 0 CpG dinucleotide in the ing enzymes dynamically maintain the DNA methylation
NGFI-A-response element of the GR exon 17 promoter pattern in adult neurons, it should also be possible to
inhibits NGFI-A binding, a finding consistent with the role reverse the demethylated state of the GR exon 17 pro-
for cytosine methylation. moter. Indeed, chronic central infusion of adult offspring
Differences in NGFI-A expression between the off- of high-LG or low-LG mothers with the methyl-donor
spring of high-LG and low-LG mothers in early postnatal amino acid methionine increases DNA methylation at the
life are no longer apparent in adulthood. Instead, it NGFI-A-binding site and reduces NGFI-A binding to the
appears that the methylation of the NGFI-A consensus exon 17 promoter selectively in the offspring of high-LG
sequence interferes with NGFI-A binding to the GR exon mothers. These effects eliminate group differences in
17 promoter in the offspring of low-LG mothers. Indeed, both hippocampal GR expression and HPA responses to
ChIP assays indicate that binding of NGFI-A protein to stress (I.C.G. Weaver et al., unpublished). Methionine
the hippocampal GR exon 17 promoter in adults is increases the levels of SAM and DNA methylation [56],
threefold greater in offspring of high-LG mothers than and SAM could increase DNA methylation either by
offspring of low-LG mothers [14]. Studies using the same stimulating DNA methylation enzymes [57] or by inhibit-
tissue samples and an antibody against the acetylated ing demethylases [51].
form of H3 show increased association of acetylated H3
with the GR exon 17 promoter in offspring of the high-LG How does maternal care alter cytosine methylation?
mothers. Because histone acetylation is associated with Developmental studies provide clear evidence for an active
active states of gene expression, these findings are con- process of demethylation driven by maternal care. High-LG
sistent with the idea of increased NGFI-A binding to the and low-LG mothers differ in the frequency of pup LG only
exon 17 promoter, and increased transcriptional acti- during the first postnatal week [12,13,48,58]. This period
vation. Moreover, transient transfection studies show corresponds to appearance of the difference in DNA
that DNA methylation inhibits the ability of NGFI-A to methylation of the NGFI-A-response element of the GR
activate the exon 17 promoter in HEK cells (I.C.G. Weaver exon 17 promoter. On embryonic-day 20 (2428 h before
et al., unpublished). Taken together these findings suggest birth), the entire exon 17 region is completely unmethy-
that an epimutation at a single cytosine residue within lated. On P1, both the 5 0 and 3 0 CpGs of the NGFI-A site
the NGFI-A consensus sequence alters NGFI-A binding are heavily methylated in both high-LG and low-LG
and might explain the sustained effect of maternal care on groups, suggesting a comparable postnatal wave of de novo
hippocampal GR expression and HPA responses to stress. methylation. The differences in methylation emerge
between P1 and P6, which is precisely when differences
Reversal of maternal effects on GR expression and HPA in the maternal behaviour of high-LG and low-LG dams
stress responses are apparent. By P6, the 5 0 , but not the 3 0 , CpG dinucleo-
HDAC inhibitors can trigger active, replication-indepen- tide of the NGFI-A-response element is demethylated in
dent DNA demethylation [49,50]. We tested [14] whether the high-LG but not in the low-LG group, and the
www.sciencedirect.com
460 Opinion TRENDS in Neurosciences Vol.28 No.9 September 2005

maternal effect persists through to adulthood. These find- NGFI-A binding at the exon 17 promoter; predictably, the
ings suggest that the group difference in DNA methylation treatment also results in demethylation of the 5 0 CpG site
involves a process of demethylation. in the absence of cell replication [14]. In HEK cells tran-
These data beg the question of how maternal care siently transfected using a methylated GR exon 17
might activate demethylation of the exon 17 promoter. promoterluciferase vector, NGFI-A overexpression ulti-
Note that on P1 the CpG sites of the NGFI-A consensus mately leads to demethylation of the 5 0 CpG in the NGFI-A-
sequence in the exon 17 promoter are heavily methylated binding site within the exon 17 construct. Because the
and, thus, in a low binding affinity state in the offspring non-integrated plasmid does not bear an origin of repli-
of both high-LG and low-LG mothers. We suggest that cation and does not replicate in HEK cells, the assay
increased pup LG increases NGFI-A transcription by measures the effects of NGFI-A on active, replication-
activating 5-HT7 receptor signalling through cAMP and independent demethylation. To demonstrate that this
protein kinase A (PKA; Figure 4). Increased NGFI-A DNA demethylation requires direct contact between
levels enhance NGFI-A binding to the methylated GR NGFI-A and its recognition element, we performed site-
exon 17 promoter, recruitment of HATs, histone acetyl- directed mutagenesis of the two CpGs of the NGFI-A-
ation and accessibility of the GR exon 17 promoter to recognition element. Preliminary results suggest that
demethylase, resulting in DNA demethylation. The rela- these manipulations abolish the ability of NGFI-A to
tionship between environmental stimulation, transcription- activate the GR exon 17 promoter (I.C.G. Weaver et al.,
factor activation and histone acetylation has been unpublished). Finally, hippocampal cell cultures treated
established using hippocampal neurons [37]. However, with either 5-HT or 8-bromo-cAMP, a stable cAMP ana-
and more controversially, this model also assumes a logue, show increased GR expression and hypomethyla-
steady state of DNA methylation and demethylation, the tion of the 5 0 CpG dinucleotide of the NGFI-A consensus
direction of which is determined by the state of chromatin sequence within the GR exon 17 promoter, with no effect at
structure [39,40]. The hypothesis predicts that both DNA the 3 0 site (I.C.G. Weaver et al. unpublished) once again,
methyltransferases and demethylases are present in in the absence of cell replication. Cultures maintained
neurons and that alterations of the chromatin state by under control conditions show complete methylation of
persistent physiological signals can alter the methylation both the 5 0 and 3 0 CpG sites of the NGFI-A consensus
of a gene in postmitotic tissue such as hippocampal sequence. Bromodeoxyuridine labelling, which marks
neurons. Pharmacological activation of chromatin struc- newly generated cells, reveals little or no cell replication
ture by HDAC inhibitors, such as TSA, can trigger in the cultures at the time of 5-HT treatment; indeed, the
replication-independent, active demethylation of DNA cultures are treated with mitotic inhibitors to prevent glial
[4951]. As already described, TSA treatment of adult proliferation. How does NGFI-A trigger demethylation
offspring of low-LG mothers increases H3 acetylation and upon binding to specific sequences? One possibility is that

5-HT

cAMP

Demethylase
PKA
HAT/CBP

NGFI-A

Ac
NGFI-A
Ac
X
CH3 CH3

High LG Low LG
DNMT
TRENDS in Neurosciences

Figure 4. A proposed model for maternal programming of GR epigenetic states. Increased maternal LG activates NGFI-A expression (via 5-HT, cAMP and PKA; Figure 2).
Although the affinity of NGFI-A to its recognition sequence on GR is reduced by the methylation of this element on postnatal-day 1, binding occurs through the increased
levels of NGFI-A associated with enhanced tactile stimulation derived from maternal LG. Binding of NGFI-A results in recruitment of histone acetyltransferases (HATs) such as
CBP; these increase histone acetylation (Ac), which in turn facilitates the access of demethylase and demethylation of the GR promoter, and possibly the recruitment of
methylated DNA-binding proteins [59]. In the absence of increased LG and NGFI-A expression, the promoter remains methylated. The unmethylated promoter will maintain
high affinity to NGFI-A throughout adulthood, resulting in greater NGFI-A-induced activity of the GR promoter in adult offspring of high-LG mothers, whereas the methylated
GR promoter exhibits reduced affinity for NGFI-A, resulting in low activity of the GR in adult offspring of low-LG mothers. Abbreviation: DNMT, DNA methyltransferase.

www.sciencedirect.com
 Opinion TRENDS in Neurosciences Vol.28 No.9 September 2005 461

NGFI-A directly recruits a demethylase to the gene; In the context of the present model (Figure 5), further
another, as already proposed here, is that NGFI-A recruits studies are required to determine precisely how maternal
a HAT, which increases acetylation, thus increasing behaviour alters epigenomic status of the GR exon 17
accessibility to demethylase. NGFI-A can actively target promoter. The data suggest an active process of demethy-
methylated-DNA-binding proteins to genomic targets lation. Once considered heretical, the findings discussed
[59]. Interestingly, 5-HT increases expression of cAMP- here, along with a recent paper [61] on altered expression
response-element-binding protein (CREB)-binding pro- of interleukin-2 expression by T lymphocytes following
tein (CBP) in hippocampal cell cultures (and in Aplysia activation, implicate an active process of demethylation in
during conditioning [36]), and CBP binding to the GR exon non-transformed somatic cells. Active demethylation now
17 promoter is increased in the neonatal offspring of appears to be an accepted event [62]. And there is evidence
high-LG mothers. CBP is a HAT [60]. for existence of a demethylase [63,64], although the exact
identity of the demethylase at work in these studies
remains to be clarified. Nevertheless, these findings
Conclusions: experience-dependent chromatin
provide a possible mechanism for the environmental
plasticity?
programming of gene expression and function during
A defining question for scientists interested in develop-
development and beyond. Studies on the reversal of
ment concerns the mechanism by which environmental
effects, including social interactions such as parenting, in maternal effects on DNA methylation using either TSA
early development are biologically embedded and thus or methionine suggest that neurons express the enzymatic
sustained into adulthood. The challenge is to define the machinery necessary for methylation and demethylation
exact nature of the geneenvironment interactions that in adulthood. DNA methylation, although a stable epi-
mediate such environmental programming. The differen- genetic mark maintained through carboncarbon bonds,
tial epigenetic status of the GR exon 17 promoter in the can be altered through sustained alterations of chromatin
offspring of high-LG mothers is a possible mechanism for structure such as histone acetylation. These findings thus
the maternal effect on hippocampal GR expression and raise the fascinating question of the degree to which such
HPA responses to stress. However, these findings are processes might remain sensitive to environmental
restricted to the study of a single promoter in only one regulation throughout life.
gene in one brain region; at this time, these results might It is important to note that maternal effects on the
be best thought of as a proof of principle. The degree to expression of defensive responses, such as increased HPA
which such results generalize to other instances of activity, are a common theme in biology [65]. Such effects
environmental programming remains to be determined, could reflect environmental experience of the mother
and could reveal alternative mechanisms for program- being translated, through an epigenetic mechanism of
ming of the epigenome. inheritance, into phenotypic variation in the offspring.

HDAC TSA

SAM DNMT

Ac
NGFI-A
Ac
X
CH3 CH3

High LG Low LG

Demethylase SAM

Methionine
TRENDS in Neurosciences

Figure 5. Reversibility of maternal-programmed GR expression and stress responses. DNA methylation patterns appear subject to a sustained modification through
alterations in chromatin structure. Inhibition of histone deacetylases (HDACs) by trichostatin A (TSA) in adult animals increases histone acetylation, facilitating access of
demethylase and demethylation of GR [14]. TSA eliminates the effect of maternal care on GR methylation and expression. Conversely, preliminary data reveal that central
methionine treatment in adult animals increases S-adenosyl methionine (SAM) concentrations and results in the hypermethylation of GR and decreased GR expression,
possibly through the inhibition of demethylase and the stimulation of DNA methyltransferase (DNMT). Although currently speculative, we propose that these mechanisms
form the basis for environment-regulated chromatin plasticity.

www.sciencedirect.com
462 Opinion TRENDS in Neurosciences Vol.28 No.9 September 2005

Indeed, maternal effects could result in the transmission 20 Meaney, M.J. et al. (2000) Postnatal handling increases the expression
of adaptive responses across generations [10]. In humans, of cAMP-inducible transcription factors in the rat hippocampus: the
effects of thyroid hormones and serotonin. J. Neurosci. 20, 39263935
such effects might contribute to the familial transmission
21 Weaver, I.C.G. et al. (2004) Early environmental regulation of hippo-
of risk and resilience. Finally, it is interesting to consider campal glucocorticoid receptor gene expression: characterization of
the possibility that epigenetic changes could be an intracellular mediators and potential genomic target sites. Ann. N. Y.
intermediate process that imprints dynamic environ- Acad. Sci. 1024, 182212
mental experiences on the fixed genome, resulting in 22 Hellstrom, I. et al. (2004) Ex vivo characterization of maternal
programming of stress responses through modulation of the epigenome.
stable alterations in phenotype a process of environ-
Program number 662.11. In 2004 Abstract Viewer and Itinerary Planner,
ment-dependent chromatin plasticity. Society for Neuroscience Online (http://sfn.scholarone.com/)
23 McCormick, J.A. et al. (2000) 5 0 -heterogeneity of glucocorticoid
receptor messenger RNA is tissue specific: differential regulation of
References variant transcripts by early-life events. Mol. Endocrinol. 14, 506517
1 Repetti, R.L. et al. (2002) Risky families: family social environments 24 Chen, F. et al. (1999) Multiple glucocorticoid receptor transcripts in
and the mental and physical health of offspring. Psychol. Bull. 128, membrane glucocorticoid receptor-enriched S-49 mouse lymphoma
330366 cells. J. Cell. Biochem. 74, 418429
2 Higley, J.D. et al. (1991) Nonhuman primate model of alcohol abuse: 25 Strahle, U. et al. (1992) At least three promoters direct expression of
effects of early experience, personality and stress on alcohol the mouse glucocorticoid receptor gene. Proc. Natl. Acad. Sci. U. S. A.
consumption. Proc. Natl. Acad. Sci. U. S. A. 88, 72617265 89, 67316735
3 Heim, C. et al. (2002) The role of early adverse experience and 26 Crosby, S.D. et al. (1991) The early response gene NGFI-C encodes
adulthood stress in the prediction of neuroendocrine stress reactivity a zinc finger transcriptional activator and is a member of the
in women: a multiple regression analysis. Depress. Anxiety 15, GCGGGGGCG (GSG) element-binding protein family. Mol. Cell.
117125 Biol. 11, 38353841
4 Pollak, S.D. and Kistler, D.J. (2002) Early experience is associated 27 Turner, B. (2001) Chromatin Structure and the Regulation of Gene
with the development of categorical representations for facial Expression, Blackwell
expressions of emotion. Proc. Natl. Acad. Sci. U. S. A. 99, 90729076 28 Grunstein, M. (1997) Histone acetylation in chromatin structure and
5 Pruessner, J.L. et al. (2004) Dopamine release in response to a transcription. Nature 389, 349352
psychological stress in humans and its relationship to early life 29 Roth, S.Y. et al. (2001) Histone acetyltransferases. Annu. Rev.
maternal care: a positron emission tomography study using [11C] Biochem. 70, 81120
raclopride. J. Neurosci. 24, 28252831 30 Bird, A. (2001) Methylation talk between histones and DNA. Science
6 Luecken, L.J. and Lemery, K.S. (2004) Early caregiving and 294, 21132115
physiological stress responses. Clin. Psychol. Rev. 24, 171191 31 Bird, A.P. and Wolffe, A.P. (1999) Methylation-induced repression
7 Heim, C. et al. (2000) Pituitaryadrenal and autonomic responses to belts, braces, and chromatin. Cell 99, 451454
stress in women after sexual and physical abuse in childhood. J. Am. 32 Li, E. (2002) Chromatin modification and epigenetic reprogramming
Med. Soc. 284, 592597 in mammalian development. Nat. Rev. Genet. 3, 662673
8 Levine, S. (1962) Plasma-free corticosteroid response to electric shock 33 Razin, A. and Szyf, M. (1984) DNA methylation patterns. Formation
in rats stimulated in infancy. Science 135, 795796 and function. Biochim. Biophys. Acta 782, 331342
9 Denenberg, V.H. (1964) Critical periods, stimulus input, and emo- 34 Huang, Y. et al. (2002) Altered histone acetylation at glutamate
tional reactivity: a theory of infantile stimulation. Psychol. Rev. 71, receptor 2 and brain-derived neurotrophic factor genes is an early
335351 event triggered by status epilepticus. J. Neurosci. 22, 84228428
10 Meaney, M.J. (2001) The development of individual differences in 35 Tsankova, N.M. et al. (2004) Histone modifications at gene promoter
behavioral and endocrine responses to stress. Annu. Rev. Neurosci. 24, regions in rat hippocampus after acute and chronic electroconvulsive
11611192 seizures. J. Neurosci. 24, 56035610
11 Lee, M.H.S. and Williams, D.I. (1975) Long term changes in nest
36 Guan, Z. et al. (2002) Integration of long-term-memory-related
condition and pup grouping following handling of rat litters. Dev.
synaptic plasticity involves bidirectional regulation of gene expression
Psychobiol. 8, 9195
and chromatin structure. Cell 111, 483493
12 Liu, D. et al. (1997) Maternal care, hippocampal glucocorticoid recep-
37 Crosio, C. et al. (2003) Chromatin remodeling and neuronal response:
tor gene expression and hypothalamicpituitaryadrenal responses to
multiple signaling pathways induce specific histone H3 modifications
stress. Science 277, 16591662
and early gene expression in hippocampal neurons. J. Cell Sci. 116,
13 Caldji, C. et al. (1998) Maternal care during infancy regulates the
49054914
development of neural systems mediating the expression of behavioral
38 Reik, W. et al. (2001) Epigenetic reprogramming in mammalian
fearfulness in adulthood in the rat. Proc. Natl. Acad. Sci. U. S. A. 95,
development. Science 293, 10891093
53355340
39 Razin, A. and Cedar, H. (1977) Distribution of 5-methylcytosine in
14 Weaver, I.C.G. et al. (2004) Epigenetic programming through
maternal behavior. Nat. Neurosci. 7, 847854 chromatin. Proc. Natl. Acad. Sci. U. S. A. 74, 27252728
15 Francis, D. et al. (1999) Nongenomic transmission across generations 40 Kadonaga, J.T. (1998) Eukaryotic transcription: an interlaced net-
in maternal behavior and stress responses in the rat. Science 286, work of transcription factors and chromatin-modifying machines. Cell
11551158 92, 307313
16 Meaney, M.J. et al. (1989) Neonatal handling alters adrenocortical 41 Hashimshony, T. et al. (2003) The role of DNA methylation in setting
negative feedback sensitivity and hippocampal type II glucocorticoid up chromatin structure during development. Nat. Genet. 34, 187192
receptor binding in the rat. Neuroendocrinology 50, 597604 42 Nan, X. et al. (1998) Transcriptional repression by the methyl-CpG-
17 Weaver, I.C.G. et al. (2001) Early environmental regulation of binding protein MeCP2 involves a histone deacetylase complex.
hippocampal glucocorticoid receptor gene expression: characteriz- Nature 393, 386389
ation of intracellular mediators and potential genomic target sites. 43 Jones, P.L. et al. (1998) 1998, Methylated DNA and MeCP2 recruit
Mol. Cell. Endocrinol. 185, 205218 histone deacetylase to repress transcription. Nat. Genet. 19, 187191
18 Mitchell, J.B. et al. (1990) Serotonin regulates type II, corticosteroid 44 Razin, A. (1998) CpG methylation, chromatin structure and gene
receptor binding in cultured hippocampal cells. J. Neurosci. 10, silencing-a three-way connection. EMBO J. 17, 49054908
17451752 45 Horike, S-I. et al. (2005) Loss of silent-chromatin looping and impaired
19 Mitchell, J.B. et al. (1992) Serotonergic regulation of type II imprinting of DLX5 in Rett syndrome. Nat. Genet. 37, 3140
corticosteroid receptor binding in cultured hippocampal cells: The 46 Harikrishnan, K.N. et al. (2005) Brahma links the SWI/SNF
role of serotonin-induced increases in cAMP levels. Neuroscience 48, chromatin-remodeling complex with MeCP2-dependent transcrip-
631639 tional silencing. Nat. Genet. 37, 254264
www.sciencedirect.com
 Opinion TRENDS in Neurosciences Vol.28 No.9 September 2005 463

47 Bachman, K.E. et al. (2003) Histone modifications and silencing prior and behavioral neuropathologies related to schizophrenia vulner-
to DNA methylation of a tumor suppressor gene. Cancer Cell 3, 8995 ability. Proc. Natl. Acad. Sci. U. S. A. 99, 1709517100
48 Cervoni, N. et al. (2002) The oncoprotein Set/TAF-1b, an inhibitor 57 Pascale, R.M. et al. (1991) Comparative effects of L-methionine,
of histone acetyltransferase, inhibits active demethylation of DNA, S-adenosyl-L-methionine and 5 0 -methylthioadenosine on the growth
integrating DNA methylation and transcriptional silencing. J. Biol. of preneoplastic lesions and DNA methylation in rat liver during the
Chem. 277, 2502625031 early stages of hepatocarcinogenesis. Anticancer Res. 11, 16171624
49 Cervoni, N. and Szyf, M. (2001) Demethylase is directed by histone 58 Champagne, F.A. et al. (2003) Variations in maternal care in the rat as
acetylation. J. Biol. Chem. 276, 4077840787 a mediating influence for the effects of environment on development.
50 Detich, N. et al. (2003) The methyl donor S-adenosylmethionine Physiol. Behav. 79, 359371
inhibits active demethylation of DNA: a candidate novel mechanism 59 Carvin, C.D. et al. (2003) Site-selective in vivo targeting of cytosine 5
for the pharmacological effects of S-adenosylmethionine. J. Biol. DNA methylation by zinc-finger proteins. Nucleic Acids Res. 31,
Chem. 278, 2081220820 64936501
51 Szyf, M. et al. (2004) DNA methylation and breast cancer. Biochem. 60 Bannister, A.J. and Kouzaridea, T. (1996) CBP co-activator is a
Pharmacol. 68, 11871197 histone acetyltransferase. Nature 384, 641643
52 Szyf, M. (2001) Towards a pharmacology of DNA methylation. Trends 61 Bruniquel, D. and Schwartz, R.H. (2003) Selective, stable demethyla-
Pharmacol. Sci. 22, 350354 tion of the interleukin-2 gene enhances transcription by an active
53 Frommer, M. et al. (1992) A genomic sequencing protocol that yields a process. Nat. Immunol. 4, 235240
positive display of 5-methylcytosine residues in individual DNA 62 Bird, A.P. (2002) IL-2 transcription unleashed by active DNA
strands. Proc. Natl. Acad. Sci. U. S. A. 89, 18271831 demethylation. Nat. Immunol. 4, 208209
54 Milbrandt, J. (1987) A nerve growth factor-induced gene encodes a 63 Bhattacharya, S.K. et al. (1999) A mammalian protein with specific
possible transcriptional regulatory factor. Science 238, 797799 demethylase activity for mCpG DNA. Nature 397, 579583
55 Yoshida, M. et al. (1990) Potent and specific inhibition of mammalian 64 Ramchandani, S. et al. (1999) DNA methylation is a reversible
histone deacetylase both in vivo and in vitro by trichostatin A. J. Biol. biological signal. Proc. Natl. Acad. Sci. U. S. A. 96, 61076112
Chem. 265, 1717417179 65 Mousseau, T.A. and Fox, C.W. (1998) The adaptive significance of
56 Tremolizzo, L. et al. (2002) An epigenetic mouse model for molecular maternal effects. Trends Ecol. Evol. 13, 403407

Articles of interest in Current Opinion journals

Pax genes in eye development and evolution
Zbynek Kozmik
Current Opinion in Genetics & Development DOI: 10.1016/j.gde.2005.05.001
From sparks to spikes: information processing in the electrosensory systems of fish
Nathaniel B. Sawtell, Alan Williams and Curtis C. Bell
Current Opinion in Neurobiology DOI: 10.1016/j.conb.2005.06.006
Insect chemoreception
Anupama Dahankar, Elissa A. Hallem and John R. Carlson
Current Opinion in Neurobiology DOI: 10.1016/j.conb.2005.06.001
Magnetoreception and its use in bird navigation
Henrik Mouritsen and Thorsten Ritz
Current Opinion in Neurobiology DOI: 10.1016/j.conb.2005.06.003
Have we found the tip link, transduction channel, and gating spring of the hair cell?
Peter G. Gillespie, Rachel A. Dumont and Bechara Kachar
Current Opinion in Neurobiology DOI: 10.1016/j.conb.2005.06.007
Learning-induced plasticity in animal and human auditory cortex
Frank W. Ohl and Henning Scheich
Current Opinion in Neurobiology DOI: 10.1016/j.conb.2005.07.002
Cochlear implants: the view from the brain
John C. Middlebrooks, Julie Arenberg Bierer and Russell L. Snyder
Current Opinion in Neurobiology DOI: 10.1016/j.conb.2005.06.004

www.sciencedirect.com