Quantification of the sixth DNA base hydroxymethylcytosine in the brain

Article abstract of DOI:10.1002/anie.201002033

Mind over matter: LC-MS has allowed the amount of the post-replicatively formed DNA base 5hydroxymethylcytosine (see picture; left) to be quantified in brain tissue. The nucleoside is most abundant in areas that are associated with higher cognitive functions, and its content in mouse hippocampi seems to increase with age. The new method enables hydroxymethylcytosine to be quantified with unprecedented accuracy. (Figure Presented)

Full text of DOI:10.1002/anie.201002033

Angewandte
Chemie
DOI: 10.1002/anie.201002033
Hydroxymethylcytosine
Quantification of the Sixth DNA Base Hydroxymethylcytosine in the
Brain**
Martin Mꢀnzel, Daniel Globisch, Tobias Brꢀckl, Mirko Wagner, Veronika Welzmiller,
Stylianos Michalakis, Markus Mꢀller, Martin Biel, and Thomas Carell*
The genetic code is established by the sequence of the four
canonical DNA nucleosides dA, dC, dG, and dT.[1] Of these
four bases, only the dC base is chemically modified inside cells
of higher organisms to control transcriptional activity.[2]
Special methyltransferases replace the H atom at position 5
by a methyl group to form methylcytosine (5-MedC).[3]
Methylation occurs only in CpG sequences and is mostly
responsible for the silencing of genes.[4] In two recent
publications 5-hydroxymethylcytosine (5-HOMedC) was estab-
lished as a new post-replicatively formed DNA nucleoside
(Scheme 1). Kriaucionis and Heintz detected 5-HOMedC in
cerebellar purkinje neurons.[5] Tahiliani et al. reported the
presence of traces of 5-HOMedC (ca. 0.032% of all nucleosides)
in mouse embryonic stem cells and observed the new base in
CpG sequences in human embryonic kidney (HEK) cells
when they overexpressed the hydroxylating 2-oxoglutarate
and FeII-dependent enzyme TET in these cells.[6] It was
additionally shown that these TET enzymes are able to
oxidize the 5-methyl group of 5-MedC to give the hydroxy-
methyl group in vitro. The new base was detected using thin-
layer chromatography after radioactive labeling of the
nucleotides. The function of 5-HOMedC is currently unclear,
but it is speculated that it may establish another level of
transcriptional control or that it is an intermediate of a
putative oxidative demethylation mechanism.[7]
We developed a quantitative LC-MS method to inves-
tigate the distribution of 5-HOMedC in mammal brains and to
determine the relative quantities of 5-HOMedC and 5-MedC. To
this end, we synthesized both nucleosides in natural and
isotope-labeled forms[8] (Scheme 2) and quantified their
amounts in different mouse brain tissues.
5-HOMedC and 5-MedC were labeled as 18O and CD3
derivatives, respectively. The synthesis of 5-HOMedC started
Scheme 1. Structure of the four canonical nucleosides and of the post-
replicatively formed bases 5-MedC and 5-HOMedC.
[*] Dipl.-Chem. M. Mꢀnzel,[+] Dipl.-Chem. D. Globisch,[+]
Dipl.-Chem. T. Brꢀckl, Dipl.-Chem. M. Wagner,
Dipl.-Chem. V. Welzmiller, Dr. M. Mꢀller, Prof. Dr. T. Carell
Center for Integrated Protein Science (CiPSM) at the Department of
Chemistry, Ludwig-Maximilians-University
Butenandtstrasse 5–13, 81377 Munich (Germany)
E-mail: thomas.carell@cup.uni-muenchen.de
Dr. S. Michalakis, Prof. Dr. M. Biel
Center for Integrated Protein Science (CiPSM) at the Department of
Pharmacy, Ludwig-Maximilians-University
Butenandtstrasse 5–13, 81377 Munich (Germany)
Scheme 2. Syntheses of the isotope-labeled nucleosides 5-HOMedC,
[18O]5-HOMedC, 5-MedC, and [D3]5-MedC. a) H2O, DIPEA, 72%, b) TBSCl,
imid., 46%, c) NaH, TPSCl, d) NH3/MeOH, 76% over two steps,
e) 3HF·NEt3, 52%, f) NaH, TPSCl, g) NH3/MeOH, 81% over two
steps, h) HF·pyr, 88%, i) CD3MgI, CuCl, [Pd(PPh3)4], 90% inseparable
mixture of 6 and TBSdC, j) HF·pyr, 84% (based on pure 6).[18O]5-
HOMedC was synthesized by the same route as 5-HOMedC. The asterisk
(*) indicates the 18O label. DIPEA=N,N-diisopropylethylamine, imi-
d.=imidazole, pyr=pyridine, TBS=tert-butyldimethylsilyl, TPS=2,4,6-
triisopropylbenzenesulfonyl.
[+] These authors contributed equally to this work.
[**] We thank the excellence cluster CiPSM, SFB 646, and SFB 749 for
generous support. M. Mꢀnzel is grateful to the Fonds of the
Chemical Industry for a Kekulꢁ fellowship. We thank V. Hammel-
mann for support during this project.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201002033.
Angew. Chem. Int. Ed. 2010, 49, 5375 –5377
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
5375

Communications
with the TBS-protected bromo-dT
derivative 1.[9] SN2 substitution with
H2O followed by protection with a
TBS group furnished compound 2.
H218O was used for the labeled com-
pound. Subsequent conversion of 2
into the corresponding dC derivative
by activation as a sulfonate[10] and
subsequent replacement with NH3
yielded 5-HOMedC after final deprotec-
tion. For 5-MedC we used TBS-pro-
tected dT and converted it into TBS-
protected 5-MedC by using the strategy
described above. The isotope-labeled
6 was prepared from TBS-protected
iodocytosine 5[11] by palladium-cata-
lyzed methylation at C5. Final depro-
tection furnished [D3]5-MedC. These
nucleosides were used to generate
mass spectrometry calibration curves
to enable precise LC-MS quantifica-
tion (see the Supporting Informa-
tion).[12] The obtained areas of the
ion currents at the exact masses of
each nucleoside were plotted against
different concentration ratios of natu-
ral to labeled nucleosides. In both
cases linear plots were obtained with
R2 values of 0.999.
We next used the calibration curve
equations to analyze the content of
5-HOMedC and 5-MedC in various mouse
brain tissues (Figure 1a). To this end,
we collected the hypothalamus,
cortex, hippocampus, olfactory bulb,
brainstem, cerebellum, and retina
from four 90 day old mice. The tissues
were homogenized and the DNA iso-
lated by phenol/chloroform extraction
(see the Supporting Information).
From the obtained DNA, 4–10 mg
were completely digested to the
nucleosides in a two-step procedure
by incubating first with nuclease S1 at
378C for 3 h. In the second incubation
step phosphodiesterase I and antarctic
Figure 1. a) Sagittal section of a mouse brain. The brain areas highlighted in color were studied
here. b) Ratio of 5-HOMedC and 5-MedC to dG in the different tissues in percent. dG was chosen as a
reference, because it forms base pairs with dC, 5-HOMedC, and 5-MedC in DNA. I–III: see text.
c) Ratio of 5-HOMedC and 5-MedC to dG in the hippocampus of 1 day and 90 day old mice in percent.
d) Ratio of 5-HOMedC and 5-MedC to dG in a neuronal (Neuro-2a) and a glial (U-87 MG) cell line in
phosphatase were used at 378C for 3 h percent.
to enable total digestion of the DNA.
We then added the isotope-labeled
compounds and analyzed the nucleoside mixture by HPLC
and using a high-resolution mass spectrometer as detector
(Thermo Finnigan LTQ Orbitrap XL). In all experiments
5-HOMedC and 5-MedC eluted with retention times of 12.3 min
and 18.5 min, respectively. One signal for the natural (light)
and one for the synthetic (heavy) compound was detected in
each experiment. Quantification was performed by compar-
ing the integrals of the ion current of the natural compound
(determined amount) with their corresponding heavy atom
labeled derivatives (known amounts) by using the calibration
curve equations. Repetition of the experiments allowed us to
determine an average error for the quantification data of
Æ 5% for 5-MedC and 5-HOMedC. All the data for each mouse are
listed in the Supporting Information. Figure 1 shows the
averaged value.
The first discovery (Figure 1b) is that the new base
5-HOMedC is clearly detected in all brain tissues. Between 0.3%
and 0.7% of all the dC nucleosides are hydroxymethylated at
position 5, which clearly establishes 5-HOMedC as a new post-
replicatively formed nucleoside in the brain. The second
5376
www.angewandte.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2010, 49, 5375 –5377

Angewandte
Chemie
surprising result is that the amount of 5-HOMedC varies
significantly in the investigated tissues. The previous study[5]
found 5-HOMedC predominantly in purkinje neurons, which are
found in the cerebellum. We now see that the amount of the
base is much larger in the cortex and hippocampus, where
purkinje cells are not present. Analysis of brain cancer cell
lines (Figure 1d) revealed that the new base is absent in
neuroblastome and glial cell lines. We found, however, and in
agreement with the literature, reduced 5-MedC content (from a
typical 4.5% to 3%) in these cancer cell lines.[13]
nucleoside. We were able to show that the base is widely
distributed in the brain and that the nucleoside is particularly
prominent in those brain tissues which are involved in higher
cognitive functions. Finally, we obtained initial data that show
that the 5-HOMedC nucleoside is present in new born mice at
reduced levels.
Received: April 6, 2010
Revised: May 3, 2010
Published online: June 25, 2010
By using our data we can roughly divide the mouse brain
into three different areas (Figure 1b). Most 5-HOMedC is found
in the hippocampus and cortex (I), which are the brain areas
that have higher cognitive functions. The brainstem and
olfactory bulb form a second category, which possess inter-
mediate 5-HOMedC levels (II). The cerebellum and retina
contain significantly less 5-HOMedC (III). Surprisingly, initial
data also show a relatively high level of 5-HOMedC in the
hypothalamus, which is part of the endocrine system that
controls hormone-based processes. The distribution of
5-HOMedC was confirmed by immunolocalization (see the
Supporting Information).
Keywords: demethylation · DNA methylation ·
DNA modification · epigenetics · hydroxymethylcytosine
.
[1] J. D. Watson, F. H. C. Crick, Nature 1953, 171, 737 – 738.
[2] a) A. Bird, Nature 2007, 447, 396 – 398; b) P. A. Jones, D. Takai,
Science 2001, 293, 1068 – 1070.
[3] M. G. Goll, T. H. Bestor, Annu. Rev. Biochem. 2005, 74, 481 –
514.
[4] a) A. P. Feinberg, Nature 2007, 447, 433 – 440; b) W. Reik, Nature
2007, 447, 425 – 432.
[5] S. Kriaucionis, N. Heintz, Science 2009, 324, 929 – 930.
[6] M. Tahiliani, K. P. Koh, Y. Shen, W. A. Pastor, H. Bandukwala,
Y. Brudno, S. Agarwal, L. M. Iyer, D. R. Liu, L. Aravind, A.
Rao, Science 2009, 324, 930 – 935.
The relative amounts of 5-HOMedC and 5-MedC are also
interesting. We are unable to correlate the amount of 5-HOMedC
and 5-MedC in these tissues, which one would expect if 5-HOMedC
is exclusively generated from 5-MedC as a precursor. The 5-MedC
values are significantly lower in the hypothalamus and here
we indeed detect high 5-HOMedC values. In contrast, and in
accord with the literature,[14] we found that the 5-MedC values
are stable at a typical value of around 4.5% in all other
tissues, while the 5-HOMedC values vary significantly. This
suggests that 5-HOMedC has a function that is not correlated
with the 5-MedC value.
[7] a) C. Loenarz, C. J. Schofield, Chem. Biol. 2009, 16, 580 – 583;
ˇ
ˇ
ˇ
˙
b) Z. Liutkeviciu¯te, G. Lukinavicius, V. Masevicius, D. Daujo-
ˇ
˙
tyte, S. Klimasauskas, Nat. Chem. Biol. 2009, 5, 400 – 402; c) J.-K.
Zhu, Annu. Rev. Genet. 2009, 43, 143 – 166.
[8] For an alternative synthesis, see C. J. LaFrancois, J. Fujimoto,
L. C. Sowers, Chem. Res. Toxicol. 1998, 11, 75 – 83.
[9] a) R. K. Grover, S. J. K. Pond, Q. Cui, P. Subramaniam, D. A.
Case, D. P. Millar, P. Wentworth, Angew. Chem. 2007, 119, 2897 –
2901; Angew. Chem. Int. Ed. 2007, 46, 2839 – 2843.
[10] N. Bischofberger, M. D. Matteucci, J. Am. Chem. Soc. 1989, 111,
3041 – 3046.
Finally, we asked the question how the age of the animal
influences the 5-MedC and 5-HOMedC values. To this end, we
analyzed the hippocampus tissue of one-day old mice (Fig-
ure 1c). We indeed detected a significantly lower level of both
5-MedC and 5-HOMedC. The 5-MedC value increases with age from
(3.5 Æ 0.1)% to (4.3 Æ 0.3)%. More significantly, the amount
of 5-HOMedC is raised in 90 day old mice by approximately 75%
from (0.34 Æ 0.02)% to (0.59 Æ 0.04)%. To exclude that
5-HOMedC accumulates because of oxidative stress in older
animals we also analyzed the damaged base 8-oxodG, which is
a typical oxidative stress marker. Here, however, a difference
could not be detected, which shows that 5-HOMedC in the brain
does not result from an accumulation of oxidative DNA
damage.[15]
[11] a) C. H. Hwang, J. S. Park, J. H. Won, J. N. Kim, E. K. Ryu, Arch.
Pharmacal Res. 1992, 15, 69 – 72; b) M. Kuwahara, J.-I. Naga-
shima, M. Hasegawa, T. Tamura, R. Kitagata, K. Hanawa, S.-I.
Hososhima, T. Kasamatsu, H. Ozaki, H. Sawai, Nucleic Acids
Res. 2006, 34, 5383 – 5394.
[12] a) T. Brꢀckl, D. Globisch, M. Wagner, M. Mꢀller, T. Carell,
Angew. Chem. 2009, 121, 8074 – 8077; Angew. Chem. Int. Ed.
2009, 48, 7932 – 7934; b) E. P. Quinlivan, J. F. Gregory, Nucleic
Acids Res. 2008, 36, e119; c) Y. Yang, D. Nikolic, S. M. Swanson,
R. B. van Breemen, Anal. Chem. 2002, 74, 5376 – 5382.
[13] M. Esteller, N. Engl. J. Med. 2008, 358, 1148 – 1159.
[14] a) M. A. Gama-Sosa, R. M. Midgett, V. A. Slagel, S. Githens,
K. C. Kuo, C. W. Gehrke, M. Ehrlich, Biochim. Biophys. Acta
Gene Struct. Expression 1983, 740, 212 – 219; b) B. F. Vanyushin,
N. G. Lopatina, C. K. Wise, F. R. Fullerton, L. A. Poirier, Eur. J.
Biochem. 1998, 256, 518 – 527.
In summary, we have quantified the amount of 5-MedC and
5-HOMedC in brain tissues with excellent accuracy and we
confirmed 5-HOMedC as a new post-replicatively formed
[15] C. Bienvenu, J. R. Wagner, J. Cadet, J. Am. Chem. Soc. 1996, 118,
11406 – 11411.
Angew. Chem. Int. Ed. 2010, 49, 5375 –5377
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
5377