news and views
Genomic imprinting
Silence across the border
Wolf Reik and Adele Murrell
I
mprinted genes in mammals are expressed
from only one chromosome: either the
gene inherited from the mother or that
from the father is silenced. Such genes
include Igf2 and other genes involved in
growth control; here, imprinting might
work to balance maternal and paternal
demands on the size of an embryo. We know
of several mechanisms by which imprinting
can be achieved. And now the work of several
groups1–4, including two whose reports
appear on pages 482 and 486 of this issue1,2,
allows us to add yet another weapon —
chromatin boundaries — to the imprinting
arsenal.
The differential labelling of cytosine–
guanine base pairs with methyl groups dis-
tinguishes the two copies of an imprinted
gene. This differential methylation usually
originates in the parental germ cells — in egg
or sperm5. It is generally assumed that
methylation of a gene ensures that it cannot
be transcribed. Indeed, in mice deficient
in methylation, some imprinted genes are
expressed from both parental chromo-
somes6. Surprisingly, however, other genes
then become silent on both chromosomes6.
Clearly, we do not know all there is to know
about the phenomenon of imprinting.
The four groups1–4 studied the H19 and
Igf2 genes, which are closely linked imprint-
ed genes (Fig. 1) that are part of an imprinted
gene cluster. They are oppositely imprinted:
on the maternal chromosome, H19 is turned
on and Igf2 is turned off; the opposite is true
for the paternal chromosome. Both genes
share a set of enhancers, or regions that
control gene expression, which are located
downstream of H19 (ref. 7). On the mater-
nal chromosome, the H19 gene uses the
enhancers; on the paternal chromosome, the
enhancers are used by Igf2 (Fig. 1).
But why are the maternal Igf2 and the
paternal H19 genes silenced? This is where
methylation comes in. Methylation can
occur at the promoter regions of genes, nec-
essary for the initiation of transcription. The
promoter of the paternal H19 copy is methyl-
ated, and this presumably leads to its silen-
cing. But the promoters of the maternal Igf2
gene are not methylated8, so how is this copy
silenced? A key insight came after experi-
mental deletion in mice of a paternally
methylated region called the imprinting-
control region (ICR), located a few kilobases
upstream of H19 (ref. 9; Fig. 1). Deletion
of this stretch of DNA from the maternal
chromosome (on which this region is
unmethylated) results in maternal H19 and
Igf2 genes both being active. These results
raised the possibility that, under normal
408
sites on the maternal H19 ICR1,2. Most
important, binding was abolished when the
ICR was methylated before the assay was car-
ried out. (Methylation-sensitive binding of
protein factors to the ICR has been found
previously11, but these proteins may be
circumstances, this region might act as an distinct from CTCF.) Binding of CTCF was
insulator or ‘chromatin boundary element’, also eliminated by mutation of the binding
shielding the maternal Igf2 promoters from sites, and such mutated fragments no
the H19 enhancers. longer showed enhancer-blocking activity.
Three of the new papers1–3 describe the These results establish a strong corre-
use of different DNA constructs containing lation between methylation and an open
the H19 ICR to test the effect of its position (unblocked) boundary, but functional evi-
on gene expression. All three groups show dence for this link has yet to be obtained.
that the ICR does indeed have boundary or This is not easy — preimposed methylation
insulator activity, but apparently no silenc- patterns may not be maintained in these
ing activity. Comparisons of the sequence of types of assay.
these ICRs from mouse, rat and human iden- Inheritable mechanisms that control
tified four sites that bind a protein called gene expression without affecting gene
CTCF1,2,4 (Box 1), which is required for sequence are said to be epigenetic, and the
boundary function in the chicken b-globin control of H19 and Igf2 expression by the
genetic locus10. ICR boundary region falls into this category.
In vitro, CTCF protein bound to all four But how do boundaries work, and what does
Box 1 Activator, repressor and boundary controller
The CTCF protein was
originally identified as a a Inhibition of initiation
ubiquitous repressor of the
transcription of chicken genes
encoding the oncoprotein CTCF
c-Myc and the enzyme
lysozyme14,15. Most of the b Boundary function
functions of CTCF lie in gene
repression, but it also has
roles in gene activation. It CTCF
binds to a long stretch of DNA
(up to 50 base pairs) by means
of its 11-zinc-finger domain. It
c Inhibition of elongation
can use different zinc-fingers
to bind to DNA, so the
CTCF-binding sites can show RNA
considerable sequence CTCF
divergence. Some of
these sites have many Enhancer
CpG (cytosine–guanine)
dinucleotides, whereas others being transcribed by removing Su(Hw) on the ‘gypsy’ DNA
have none. Methylation occurs acetyl groups from histone element mediate enhancer
only at CpG sites, so proteins in chromatin or by blocking in one direction.
methylation can interfere with inhibiting the transcription- This is also seen for the H19
the binding of CTCF to only a initiation complex (a). Finally, ICR in an assay based on
subset of sites — including the CTCF is found at sites at which autonomous, self-replicating
H19 ICR region1,2,4. the elongation stage of DNA in vertebrate cells3. The
Gene repression can be transcription is blocked (c). ICR region also confers gene
mediated by CTCF in several Different mechanisms of silencing in Drosophila, but
ways. These include boundary repression may be related; they bidirectionally16. The
functions (b in figure), such as may complement each other; Drosophila protein that binds
blocking the access of or they may synergize to result to this region has yet to be
enhancer regions to promoters, in full repression. identified.
as shown for H19 and Igf2 Another protein with many Rainer Renkawitz is at the
(refs 1–3). Another function is zinc-fingers, Drosophila Institute of Genetics, Justus
inhibition of the effects of Su(Hw), contains gene- Liebig University, 35392
chromatin-mediated silencing repression domains and is Giessen, Germany.
of translocated genes. CTCF involved in creating a boundary. e-mail: rainer.renkawitz@gen.
can also prevent genes from The clustered binding sites for bio.uni-giessen.de
© 2000 Macmillan Magazines Ltd NATURE | VOL 405 | 25 MAY 2000 | www.nature.com
 news and views

CTCF
Maternal
chromosome 100 YEARS AGO
Off On Possibly some of your photographer readers
may be glad to know that microphotography
of sorts is within the reach of all who
possess a microscope with suitable
substage-condenser and a camera… One of
my earliest attempts was to photograph fluid
inclusions in quartzes with ordinary sunlight,
and rock sections polarised. The only
Paternal difficulty was that the sun would not keep
chromosome
On Off still, and without a heliostat the work was
most troublesome, not to say aggravating. In
Igf 2 ICR H19 Enhancers
one case, a mere movement of the
condenser-diaphragm made the bubble in
Figure 1 Selective gene silencing by boundaries. H19 and Igf2 are genes that are expressed from only the inclusion fly backwards and forwards…
the maternal or the paternal chromosome. H19 is expressed from the maternal chromosome only; With a little device in the double lantern the
Igf2 is expressed only from the paternal chromosome. The two genes share an enhancer region, motion of bubbles in inclusions can be
located downstream of H19. The new papers1–4 show that the ICR (imprinting-control region) of shown on a nine-foot screen. These
H19 is a boundary element, controlled by DNA methylation. The CTCF protein binds to the negatives were taken with a 1/16th
unmethylated maternal ICR. This prevents the promoters located in the Igf2 gene from interacting immersion, the camera being extended with
with the enhancers downstream of the H19 gene, resulting in transcriptional silencing of Igf2. The a brown paper tube, and the extra apparatus
paternal ICR is methylated (filled circles), preventing binding of CTCF. This allows the enhancers to did not cost one shilling… While observing
contact the promoters of the paternal Igf2, allowing the gene to be transcribed. The paternal H19 the transit of Venus, I thought I would try a
gene is thought to be silenced by methylation. Differentially methylated regions are also found in photograph. I drilled a hole in the telescope
Igf2, but are not shown here. cap for diaphragm; took off the eye-piece
and stuffed the telescope into a common
CTCF do? First we need to take a step back the search for the regulators of H19 and Igf2? camera, with a red cloth to make it light-
and look at the way in which enhancers and Probably not — the fact that deletion of the tight; exposed six negatives with hand
promoters interact to control gene expres- ICR only partially activates the maternal Igf2 exposure on instantaneous plates. Result:
sion. These two control regions may work gene9 indicates that other sequences may four passable negatives and one good one.
together in one of two ways. One way be involved in keeping the silence. Indeed, This quite unlooked-for success was due to
involves the ‘tracking’ of transcription com- deletion of a silencer that is also controlled by some back volumes of Nature which
plexes from the enhancers, along the DNA, methylation and is upstream of Igf2 also propped up the camera.
to the promoter. Alternatively, the DNA may activates the silent Igf2 allele (M. Constancia From Nature 24 May 1900.
loop round such that the enhancers and pro- et al., unpublished observations). The
moters interact. The Igf2 promoters are imprinting arsenal includes promoters, 50 YEARS AGO
about 100 kilobases from the enhancers, so enhancers, antisense RNA transcripts13, Twilight in India
the looping model seems more attractive. On silencers and chromatin boundaries. What This book is well named, for the author has a
the maternal chromosome, either looping or is fascinating is that these sequences have very dim view of India. In spite of a good
the interaction between the promoters and come under epigenetic control. ■ deal of interesting and, on the whole, well-
the enhancers could be disrupted by CTCF Wolf Reik and Adele Murrell are in the Laboratory informed matter on south Indian castes, the
(and possibly other factors) binding to the of Developmental Genetics and Imprinting, The whole book is coloured by an obvious
ICR. In addition, binding of CTCF and other Babraham Institute, Cambridge CB2 4AT,UK. determination to view everything Hindu in
factors could have a role in opening the chro- e-mail: wolf.reik@bbsrc.ac.uk the darkest shadow, and all the less
matin of the H19 gene for transcription — 1. Bell, A. C. & Felsenfeld, G. Nature 405, 482–485 (2000). creditable aspects of Hinduism, particularly
CTCF has been implicated in gene activation 2. Hark, A. T. et al. Nature 405, 486–489 (2000). in regard to sex, are enlarged on at the
3. Kanduri, C. et al. Curr. Biol. 10, 449–457 (2000).
as well as repression (Box 1). 4. Szabo, P. E., Tang, S. H. E., Rentsendorj, A., Pfeifer, G. P. &
expense of its merits.
Indeed, the H19 ICR seems to have many Mann, J. Curr. Biol. 10, 607–610 (2000). There is obviously a good deal of
other functions as well as being a boundary. 5. Surani, M. A. Cell 93, 309–312 (1998). exaggeration in many of the statements
For example, it is needed for methylation of 6. Li, E., Beard, C. & Jaenisch, R. Nature 366, 362–365 (1993). made, for the meriah sacrifice is written of —
7. Leighton, P. A., Saam, J. R., Ingram, R. S., Stewart, C. L. &
the paternal H19 gene9, and for silencing of Tilghman, S. M. Genes Dev. 9, 2079–2089 (1995).
and that in 1949 — as if it continued as a
this gene independently of methylation12. The 8. Sasaki, H. et al. Genes Dev. 6, 1843–1856 (1992). routine ceremonial. Several of the illustrations
region seems to contain many functional ele- 9. Thorvaldsen, J. L., Duran, K. L. & Bartolomei, M. S. Genes Dev. are borrowed without acknowledgement from
12, 3693–3702 (1998).
ments that are involved in regional regulation Thurston; and, although misprints abound, the
10. Bell, A. C., West, A. G. & Felsenfeld, G. Cell 98, 387–396 (1999).
of methylation, in silencing and in insulating. 11. Ishihara, K. et al. Genome Res. 10, 664–671 (2000). fact that “tumeric” is repeatedly used for
Other functions may be revealed by further 12. Brenton, J. D. et al. Proc. Natl Acad. Sci. USA 96, 9242–9247 “turmeric” suggests that possibly it is not the
study of these elements and of the protein and (1999). printer who is to blame for “foistered” instead
13. Wutz, A. et al. Nature 389, 745–749 (1997).
chromatin factors that bind to them. 14. Baniahmad, A. et al. Cell 61, 505–514 (1990).
of “foisted” — or perhaps “fostered”.
Does the discovery of the epigenetic 15. Lobanenkov, V. V. et al. Oncogene 5, 1743–1753 (1990). From Nature 27 May 1950.
boundary upstream of H19 bring to an end 16. Lyko, F. et al. Nature Genet. 16, 171–173 (1997).

NATURE | VOL 405 | 25 MAY 2000 | www.nature.com © 2000 Macmillan Magazines Ltd 409