, 1229 (2011);
, et al.Christopher S. Nabel
Demystifying DNA Demethylation
This copy is for your personal, non-commercial use only.
clicking here.colleagues, clients, or customers by
, you can order high-quality copies for your
If you wish to distribute this article to others
The following resources related to this article are available online at
here.following the guidelines
can be obtained by
Permission to republish or repurpose articles or portions of articles
Updated information and services,
): September 1, 2011 www.sciencemag.org (this infomation is current as of
version of this article at:
including high-resolution figures, can be found in the online
, 4 of which can be accessed free:
cites 15 articles
This article appears in the following
registered trademark of AAAS.
is a Science2011 by the American Association for the Advancement of Science; all rights reserved. The title
CopyrightAmerican Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005.
(print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by theScience
on September 1, 2011
www.sciencemag.org SCIENCE VOL 333 2 SEPTEMBER 2011
Demystifying DNA Demethylation
Christopher S. Nabel and Rahul M. Kohli
the challenges of multicellu-
lar life. To address this need,
nature has evolved a sub-
stantial enzymatic toolbox
for altering cytosine within
the genome. Methylation of
the nucleotide cytosine (C)
at the 5-position of the base
has profound impacts on
gene expression and cellu-
lar identity. The reverse of this process, DNA
demethylation, is equally important for clean-
ing the genomic slate during embryogenesis
or achieving rapid reactivation of previously
silenced genes. Although the mechanism of
DNA methylation has been rigorously estab-
lished, active DNA demethylation in mam-
mals has remained enigmatic, as disparate
observations have failed to coalesce into a
consistent model. Cytosine deamination,
oxidation, and base excision repair enzymes
have been proposed in a dizzying variety of
combinations ( 1). Against this backdrop, two
reports in this issue, by Ito et al. ( 2) on page
1300 and He et al. on page 1303 ( 3), help
bring new clarity to the mechanistic model
for DNA demethylation.
The studies by Ito et al. and He et al.
expand on the recent discovery that 5-methyl-
cytosine (mC) can be oxidized to 5-hydroxy-
DNA modifying and repair enzymes make
a new connection in the mechanism of DNA
CREDIT: B. STRAUCH/SCIENCE
istic single-photon sources ( 4– 7), quantum
memories ( 8), or quantum switches with pho-
tons ( 9). In the present approach, an ensemble
rather than a single atom was used, which led
to a substantial reduction of the characteris-
tic time scales by a factor N1/2 for an N-atom
cloud. Even more important, and in contrast
to previous proposals, the input and output of
the probe fi eld was not mediated by the high-
quality cavity but rather by the atomic ensem-
ble. This approach allowed for longer opera-
tional times without sacrifi cing operational
speed and also reduced the effects of losses.
Finally, this experimental achievement is also
quite different from recent experiments that
demonstrated EIT in strongly coupling cavi-
ties with individual atoms ( 10, 11). In those
experiments, the control field was rather
strong (essentially a classical external laser
and closer to an EIT experiment) and not
the fi eld of a single photon scattered from
the probe fi eld into the cavity that would be
needed for quantum switching.
The work by Tanji-Suzuki et al. has shown
that transparency of an opaque medium and
a substantial time delay can be induced by a
few photons. Their demonstration suggests
that the switching of light by light on the sin-
gle-quantum level—an all-optical quantum
gate—may soon be in reach.
References and Notes
1. H. Tanji-Suzuki et al., Science 333, 1266 (2011);
2. M. Fleischhauer, A. Imamoglu, J. P. Marangos, Rev. Mod.
Phys. 77, 633 (2005).
3. J. E. Field, Phys. Rev. A 47, 5064 (1993).
4. A. Kuhn, M. Hennrich, G. Rempe, Phys. Rev. Lett. 89,
5. J. McKeever et al., Science 303, 1992 (2004).
6. M. Hijlkema et al., Nat. Phys. 3, 253 (2007).
7. A. D. Boozer, A. Boca, R. Miller, T. E. Northup, H. J.
Kimble, Phys. Rev. Lett. 98, 193601 (2007).
8. H. P. Specht et al., Nature 473, 190 (2011).
9. K. M. Birnbaum et al., Nature 436, 87 (2005).
10. T. Kampschulte et al., Phys. Rev. Lett. 105, 153603
11. M. Mücke et al., Nature 465, 755 (2010).
ariability and adapt-
ability are neces-
sary for overcoming
methylcytosine (hmC) by TET enzymes,
members of the α-ketoglutarate–dependent
oxygenase family ( 4). Although hmC exists
in low quantities—less than 1% of all cyto-
sines ( 5)—the base has become a commod-
ity in the epigenetics fi eld, particularly given
studies implicating TET in global and locus-
specific DNA demethylation ( 6– 8). One
notable proposal posits that iterative oxida-
tion by TET could yield 5-formylcytosine
(fC) and 5-carboxylcytosine (caC) ( 1). Given
the precedent of a decarboxylase in pyrimi-
dine salvage, a similar enzyme could ulti-
mately regenerate cytosine.
In evaluating this proposal, Ito et al. and
He et al. both demonstrate that TET enzymes
are capable of iterative oxidation of mC. Puri-
fi ed TETs converted hmC to fC and caC in
oligonucleotides. In mouse embryonic stem
(ES) cells, both fC and caC were detected in
the genome by mass spectrometry, albeit at
low amounts, confi rming the recent detection
of fC in ES cells ( 9). Accumulation of higher
oxidation products depended on TET, as the
absence of a TET iso-
type decreased amounts
of the modified bases
in ES cells. The discov-
ery of these seventh and
eighth bases, after the
sixth base hmC, pro-
vides clear evidence that
we have underestimated
the dynamic nature of
the genome ( 10).
How can higher oxi-
dation products of mC
revert to cytosine? Although Ito et al. suggest
an unknown decarboxylase, He et al. pro-
pose direct removal of the entire caC nucleo-
base by thymine-DNA glycosylase (TDG).
Subsequent repair of the resulting abasic site
would restore unmodified cytosine. TDG
has been previously implicated in demethyl-
ation, as its absence is embryonic lethal and
perturbs DNA methylation patterns ( 11, 12).
A requirement for TDG has fostered the pre-
vailing assumption that demethylation must
involve deamination, as the canonical TDG
substrates are mispairs between thymine and
guanine nucleotides that result from deamina-
tion of genomic mC. However, TDG has some
activity against cytosine analogs, particularly
when substituents weaken the N-glycosidic
bond between the nucleobase and the sugar
( 13), as would likely be the case for fC and
caC. Indeed, He et al. show that ES cell lysates
contain glycosylase activity against caC-
containing oligonucleotides—activity that
is lost when TDG is depleted. Furthermore,
TDG overexpression decreases genomic
Department of Medicine and Department of Biochemis-
try and Biophysics, Perelman School of Medicine, Univer-
sity of Pennsylvania, Philadelphia, PA 19104, USA. E-mail:
X = CHO or COO-
DNA demethylation. TET enzymes are proposed to oxidize 5-methylcytosine (mC) to 5-hydroxy-
methylcytosine (hmC) and subsequently to generate the higher oxidation substituents 5-form-
ylcytosine (fC) and 5-carboxylcytosine (caC) (shown as the structure with the 5-X substituent).
Unmodifi ed cytosine (C) is on the far right. Base excision repair, initiated by thymine-DNA glyco-
sylase (TDG), releases and replaces the entire modifi ed oxidized base with unmodifi ed C.
Published by AAAS
on September 1, 2011
2 SEPTEMBER 2011 VOL 333 SCIENCE www.sciencemag.org
caC, whereas caC accumulates when TDG
is depleted. These observations overturn the
assumption associating TDG solely with
deamination-mediated demethylation; TDG
activity on the higher oxidation products
of mC links two proposed players in DNA
demethylation—oxidation and base excision
repair—in a new and plausible manner.
As an important point of discrepancy,
Ito et al. fi nd that fC accumulates relative
to caC, whereas He et al. report that hmC
is effi ciently converted to caC without any
accumulation of fC. This raises the ques-
tion of the identity of the penultimate cyto-
sine oxidation product prior to the action of
base excision repair. Mechanistically, it is
feasible that fC could be the better substrate
for TDG ( 13), given impacts on N-glycosidic
bond stability. Higher relative amounts of fC
in ES cells would support this possibility, and
caC accumulation may refl ect altered steady-
state amounts given perturbed TDG or TET
expression. A kinetic appraisal of the relative
rates of formation of hmC, fC, and caC by
TETs should resolve these possibilities.
The factors regulating the extent of TET-
mediated oxidation must also be explored.
Why does TET sometimes oxidize to hmC
and at other times iteratively to fC or caC?
Do fC and caC have roles in shaping the
genome other than as intermediates in
demethylation? Further, Ito et al. and He et
al. demonstrate a viable demethylation path-
way by examining oligonucleotides or by
assessing global amounts of nucleotides.
Neither shows that caC is specifi cally pres-
ent in promoters undergoing demethylation.
Additional experiments will need to confi rm
the coupling of iterative oxidation and base
excision repair at activated promoters for
this model to gain acceptance as a bona fi de
mechanism of DNA demethylation.
Although iterative oxidation coupled
to base excision repair provides a plausi-
ble demethylation pathway, what becomes
of observations that favor other pathways?
Rather than viewing these mechanisms as
mutually exclusive, it is possible that they
may assume accessory roles. Pathways such
as those involving deamination by activa-
tion-induced cytidine deaminase ( 6, 14, 15)
may serve as bypass routes to cytosine in
specifi c physiological settings such as pri-
mordial germ cells. Given the multitude of
ways to manipulate cytosine, it is fortunate
that a framework for demystifying demethyl-
ation is now at hand.
1. S. C. Wu, Y. Zhang, Nat. Rev. Mol. Cell Biol. 11, 607
2. S. Ito et al., Science 333, 1300 (2011).
3. Y. F. He et al, Science 333, 1303 (2011).
4. M. Tahiliani et al., Science 324, 930 (2009).
5. D. Globisch et al., PLoS ONE 5, e15367 (2010).
6. J. U. Guo, Y. Su, C. Zhong, G. L. Ming, H. Song, Cell 145,
7. S. Ito et al., Nature 466, 1129 (2010).
8. K. Iqbal, S. G. Jin, G. P. Pfeifer, P. E. Szabó, Proc. Natl.
Acad. Sci. U.S.A. 108, 3642 (2011).
9. T. Pfaffeneder et al., Angew. Chem. Int. Ed. 50, 7008
10. R. M. Kohli, Nat. Chem. Biol. 6, 866 (2010).
11. S. Cortellino et al., Cell 146, 67 (2011).
12. D. Cortázar et al., Nature 470, 419 (2011).
13. M. T. Bennett et al., J. Am. Chem. Soc. 128, 12510 (2006).
14. K. Rai et al., Cell 135, 1201 (2008).
15. N. Bhutani et al., Nature 463, 1042 (2010).
Through Thick and Thin
Eric Brown 1 and Heinrich M. Jaeger 2
nian fl uid like water, the viscosity is constant.
Such simple behavior can change drastically,
however, when small particles are suspended
in the liquid. In some instances, the viscos-
ity decreases with increasing shear rate and
the fl uid is said to exhibit shear thinning. For
applications such as paints, this is desirable
because it keeps suspended pigments on the
painted surface at rest but lets them fl ow eas-
ily when brushed. There is the opposite pos-
sibility of shear thickening whereby the vis-
cosity increases with shear rate. For some
suspensions, such as cornstarch in water,
this effect can be so dramatic that a person
can run across the surface of a pool fi lled
with the suspension, but sinks when standing
still. Such non-Newtonian fl ow behaviors are
thought to be caused by changes in the par-
ticle arrangements under shear. To investigate
New experimental results probe
the relationship between rheology
and particle-scale structure of suspensions.
he ratio of shear stress to shear rate in
a flowing fluid defines its viscosity,
or resistance to shear. For a Newto-
this, Cheng et al. ( 1), on page 1276 of this
issue, report direct measurements of particle
arrangements while moving between regimes
of shear thinning and thickening.
A textbook example of the role of particle
arrangement in driving the behavior of sus-
pensions is shear thinning resulting from the
organization of particles into layers oriented
along the direction of fl ow in which they can
slide over each other more easily than if they
were randomly distributed ( 2). Similarly, it
was predicted that shear thickening could
result from the formation of particle clusters
( 3, 4). These “hydroclusters” form when the
particles are pushed together by shear and
can bunch together transiently as a result of
large viscous drag forces in the thin lubrica-
tion layers between the particles, which slow
To test these ideas of suspension rheol-
ogy, Cheng et al. developed a particularly
fast and sensitive confocal rheometer that
allows them to track the three-dimensional
locations of individual, 1-µm-diameter par-
ticles suspended in a liquid while simultane-
ously shearing the sample and measuring the
stresses. They can quantify subtle changes in
the local particle arrangements as the sam-
ple transitions between shear-thinning, New-
tonian, and shear-thickening regimes. Their
data lead to two important results. First, the
viscosity decrease during shear thinning can
be quantitatively characterized as the sum
of two contributions: a constant, Newtonian
portion resulting from viscous stresses, and
an entropic contribution that comes from
the pressure produced by random collisions
of particles under thermal motion, which
decreases with shear rate. Second is the fi rst
direct experimental verifi cation of clusters
of nearly touching particles that grow as the
suspension thickens, consistent with earlier
predictions ( 3, 4).
Not all suspensions show all the flow
regimes seen in the model system investi-
gated. For example, some suspensions do
not shear thicken, some do not have any
appreciable intervening Newtonian regime,
and others do not exhibit entropic effects.
However, the particle-scale detail revealed
by the elegant Cornell experiments informs
the broader problem in rheology of how to
attribute suspension properties to particle
interactions on the one hand and structural
1School of Natural Sciences, University of California,
Merced, CA 95343, USA. 2James Franck Institute, Univer-
sity of Chicago, Chicago, IL 60637, USA. E-mail: h-jaeger@
Published by AAAS
on September 1, 2011