Deamination, oxidation, and C-C bond cleavage reactivity of 5-hydroxymethylcytosine, 5-formylcytosine, and 5-carboxycytosine

Article abstract of DOI:10.1021/ja403229y

Three new cytosine derived DNA modifications, 5-hydroxymethyl-2′- deoxycytidine (hmdC), 5-formyl-2′-deoxycytidine (fdC) and 5-carboxy-2′-deoxycytidine (cadC) were recently discovered in mammalian DNA, particularly in stem cell DNA. Their function is currently not clear, but it is assumed that in stem cells they might be intermediates of an active demethylation process. This process may involve base excision repair, C-C bond cleaving reactions or deamination of hmdC to 5-hydroxymethyl-2′- deoxyuridine (hmdU). Here we report chemical studies that enlighten the chemical reactivity of the new cytosine nucleobases. We investigated their sensitivity toward oxidation and deamination and we studied the C-C bond cleaving reactivity of hmdC, fdC, and cadC in the absence and presence of thiols as biologically relevant (organo)catalysts. We show that hmdC is in comparison to mdC rapidly oxidized to fdC already in the presence of air. In contrast, deamination reactions were found to occur only to a minor extent. The C-C bond cleavage reactions require the presence of high concentration of thiols and are acid catalyzed. While hmdC dehydroxymethylates very slowly, fdC and especially cadC react considerably faster to dC. Thiols are active site residues in many DNA modifiying enzymes indicating that such enzymes could play a role in an alternative active DNA demethylation mechanism via deformylation of fdC or decarboxylation of cadC. Quantum-chemical calculations support the catalytic influence of a thiol on the C-C bond cleavage.

Full text of DOI:10.1021/ja403229y

Article
pubs.acs.org/JACS
Deamination, Oxidation, and C−C Bond Cleavage Reactivity of
5‑Hydroxymethylcytosine, 5‑Formylcytosine, and 5‑Carboxycytosine
Stefan Schiesser,†,⊥ Toni Pfaffeneder,†,⊥ Keyarash Sadeghian,†,‡ Benjamin Hackner,†
Barbara Steigenberger,† Arne S. Schroder,† Jessica Steinbacher,† Gengo Kashiwazaki,† Georg Hofner,§
̈
̈
Klaus T. Wanner,§ Christian Ochsenfeld,†,‡ and Thomas Carell*,†
‡
†Center for Integrated Protein Science (CiPSM) at the Department of Chemistry, Chair for Theoretical Chemistry at the
§
Department of Chemistry, Center for Drug Research at the Department of Pharmacy, Ludwig-Maximilians-Universitat Munchen,
̈
̈
Butenandtstrasse 5-13, 81377 Munich, Germany
S
* Supporting Information
ABSTRACT: Three new cytosine derived DNA modifica-
tions, 5-hydroxymethyl-2′-deoxycytidine (hmdC), 5-formyl-2′-
deoxycytidine (fdC) and 5-carboxy-2′-deoxycytidine (cadC)
were recently discovered in mammalian DNA, particularly in
stem cell DNA. Their function is currently not clear, but it is
assumed that in stem cells they might be intermediates of an
active demethylation process. This process may involve base
excision repair, C−C bond cleaving reactions or deamination
of hmdC to 5-hydroxymethyl-2′-deoxyuridine (hmdU). Here
we report chemical studies that enlighten the chemical reactivity of the new cytosine nucleobases. We investigated their
sensitivity toward oxidation and deamination and we studied the C−C bond cleaving reactivity of hmdC, fdC, and cadC in the
absence and presence of thiols as biologically relevant (organo)catalysts. We show that hmdC is in comparison to mdC rapidly
oxidized to fdC already in the presence of air. In contrast, deamination reactions were found to occur only to a minor extent. The
C−C bond cleavage reactions require the presence of high concentration of thiols and are acid catalyzed. While hmdC
dehydroxymethylates very slowly, fdC and especially cadC react considerably faster to dC. Thiols are active site residues in many
DNA modifiying enzymes indicating that such enzymes could play a role in an alternative active DNA demethylation mechanism
via deformylation of fdC or decarboxylation of cadC. Quantum-chemical calculations support the catalytic influence of a thiol on
the C−C bond cleavage.
known oxidative damage, one has to consider the possibility
that they are formed by nonenzymatic oxidation processes
INTRODUCTION
■
In addition to the canonical nucleosides dA, dC, dG, and dT,
mammalian DNA contains 5-methyl-2′-deoxycytidine (mdC)
and further dC derivatives that are generated from mdC by
oxidation of the methyl group. The oxidizing enzymes are TET
(ten eleven translocation) enzymes. These are α-ketoglutarate
dependent oxygenases, which oxidize 5-methyl-2′-deoxycyti-
dine to 5-hydroxymethyl-2′-deoxycytidine (hmdC) and further
to 5-formyl-2′-deoxycytidine (fdC) and 5-carboxy-2′-deoxycy-
tidine (cadC).1−5 The fate of these nucleosides and their
function are currently controversially discussed. For hmdC for
example it was postulated that the nucleoside may be
deaminated in vivo to give 5-hydroxymethyl-2′-deoxyuridine
(hmdU) by the action of special deaminases such as the AID
(activation-induced deaminase)/APOBEC (apolipoprotein B
mRNA-editing enzyme complex) protein family.6,7 However,
newer in vitro data suggest that deamination of hmdC is
unlikely to occur enzymatically,8 raising the question of
whether the occurrence of hmdU may result from non-
enzymatic spontaneous deamination of hmdC. Furthermore,
since hmdC and fdC are known oxidative lesions of mdC9−11
and because the levels of fdC and in particular cadC3−5,12 are in
the range of those reported for 8-oxo-dG,13 which is a well-
Scheme 1. The dC Derivatives hmdC, fdC, and cadC Could
Either Deaminate to dU Derivatives (Left), Undergo C−C
bond Cleavage to dC (Right), or Oxidize
during DNA isolation and analysis.13,14 Finally, while there is a
possibility that enzymatic dehydroxymethylation of hmdC,
deformylation of fdC, and decarboxylation of cadC might
occur,15−19 it cannot be ruled out that these processes occur
Received: April 11, 2013
© XXXX American Chemical Society
A
dx.doi.org/10.1021/ja403229y | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
already to a significant extent without the help of an
enzyme.18,20
To differentiate between potential nonenzymatic background
reactivities and enzyme-catalyzed processes, it is therefore
essential to understand the intrinsic reactivity profile of the new
nucleosides hmdC, fdC and cadC (Scheme 1).
hairpin ODNs in buffer and in the presence of thiols. We
investigated the reactivity with thiols in order to simulate
potential enzymatic reactions, which often start by initial
nucleophilic attack of the thiol at the C6 position of the
pyrimidine. This then activates the corresponding C5 position
(numbering see Scheme 1).18 Examples are the DNA
methyltransferases (DNMT), which convert dC to mdC.19
We first studied the reactivity of the different xdC-
nucleosides at pH 7.4 in buffer at different temperatures. We
simultaneously quantified the various oxidation, deamination,
and C−C bond cleavage reactions. Moreover, to elucidate how
thiol-mediated C5 activation would influence the reactions, we
performed studies with a systematic increase of β-mercapto-
ethanol (β-ME, 0 to 12 M). High thiol concentrations (12 M)
were chosen to simulate the high effective molarity of, for
example, the reactive cysteine moiety in the active sites of
enzymes.17,18 In all cases we performed product analysis using a
UHPLC-MS/MS method with a triple quadrupole mass
spectrometer (Scheme 2; for method development see
Supporting Information [SI]). The developed method allows
sensitive and accurate quantification of the whole product
spectrum in one single analysis (12 min total run time). The
exact quantification of the reaction products was conducted
using the stable isotope dilution technique as described by
others and us.12,13,26,27 For the synthesis of the used
isotopologues [15N2]-dC, [D2,15N2]-hmdC, [15N2]-fdC,
[15N2]-cadC, [D3]-dT, [15N2]-dU, [D2]-hmdU, and [15N2]-
fdU see SI.
RESULTS AND DISCUSSION
■
Recently, others and us reported the synthesis of phosphor-
amidite and triphosphate building blocks for hmdC, fdC, and
cadC and their incorporation into oligonucleotides using either
solid phase phosphoramidite chemistry or by polymerase chain
reaction (PCR).21−25 For this study we synthesized hmdC, fdC,
and cadC nucleosides and phosphoramidites and incorporated
the latter into 20mer hairpin-oligodeoxynucleotides (ODN)
with either hmdC (ODN1 and ODN4), fdC (ODN2 and
ODN5), or cadC (ODN3 and ODN6) in a double-stranded
xCpG or xCpNpG context using solid phase phosphoramidite
chemistry (Scheme 2).
To investigate deamination-, oxidation-, and C−C bond
cleavage reactions, the reactivities of the new epigenetic
modifications were studied on the nucleoside level and in
Scheme 2. Depiction of the Hairpins Used in This Study and
a
of the Experimental Workflow
Oxidation Processes. Our first question was how quickly
air would oxidize mdC to hmdC, hmdC to fdC, and fdC to
cadC. We dissolved the respective nucleosides in a phosphate
buffer at pH = 7.4 and incubated the solution exposed to air for
up to 60 h at 60.0, 67.5, 75.0, and 80.0 °C (Figure 1 and SI).
Figure 1. Oxidation kinetics of mdC to hmdC (cyan), hmdC to fdC
(blue), and fdC to cadC (green) at 60.0 °C at pH 7.4. While hmdC is
efficiently oxidized to fdC, fdC gives only a little cadC. Depicted are
the means of triplicate experiments; error bars reflect the standard
deviations. For details of the kinetic measurements at 67.5, 75.0, and
80.0 °C and linear regression analyses see SI.
For data analysis we assumed a pseudo-first order kinetic profile
in which the oxygen concentration is a not rate limiting factor.
Under these conditions we observed that mdC is oxidized only
to a small extent to hmdC. In contrast, hmdC reacts efficiently
to form fdC (Figure 1, blue line). Oxidation of fdC to cadC
(Figure 1, green dots) is slow.
After 60 h at 60 °C, the yield of cadC was only 0.03%. Table
1 summarizes the determined pseudo-first-order rate constants.
The data show that the formation of fdC by oxidation of hmdC
is more than 2 orders of magnitude faster than the oxidation of
mdC to hmdC. For the fdC to cadC oxidation, a rate constant
a
After incubation of the xdC-containing hairpins in β-mercaptoethanol
(β-ME)/imidazole, the DNA was isolated. The corresponding
isotopologues were added (omitted for clarity in the chromatogram),
and the DNA was digested. Quantification of the reaction products
was performed by UHPLC-MS/MS. Im = imidazole.
B
dx.doi.org/10.1021/ja403229y | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Table 1. Rate Constants k for the Oxidation of mdC to
Scheme 3. Depiction of the Thiol-Catalyzed Oxidation of
hmdC to fdC and the Proposed Mechanism of the C−C
Bond Cleavage of hmdC to dC. R-SH = β-Mercaptoethanol
R = CH2CH2OH
hmdC and hmdC to fdC at pH 7.4. For the Rate Constants
a
at Higher Temperatures see SI
k37 °C [s−1
n.d.
1.3 0.1 × 10−8
]
k60 °C [s−1
4.7 0.6 × 10−11
2.2 0.2 × 10−8
]
Ea [kJ mol−1
]
mdC→hmdC
hmdC→fdC
n.d.
20.8 2.2
a
n.d. = not determined.
could not be calculated, since the data deviated substantially
from linearity (see Figure 1 and SI, Figure S2, Figure S11).
We next studied the oxidation reactions of hmdC and fdC in
the hairpin-ODNs (Scheme 2). The reactions were performed
without and with increasing amounts of β-ME (and imidazole
at pH 5.0) in solution for 48 h at 50 °C (melting temperatures
of hairpins: ≥70 °C; see SI, Table S1). The samples were
desalted with a 0.025 μm filter, spiked with the labeled internal
standards, then fully digested with nuclease S1, snake venom
phosphodiesterase, and antarctic phosphatase and analyzed by
LC−MS/MS (Scheme 2). The amounts of the nucleoside
products were normalized to the amount of dT [%]. Despite
the shielding effect of the duplex environment, which could
limit the reaction of the thiol at the C6 position, we measured
increasing amounts of fdC with increasing concentrations of β-
ME in ODN1 (hmdC). The fdC compound reached
surprisingly high levels of up to 20% at 12 M β-ME (80% v/
v β-ME; Figure 2, blue curve), showing that the C5−C6
comparatively slow (Figure 2, green curve). The formation of a
dithioacetal of fdC was not detected (see SI). It should be
noted that concomitant deformylation of fdC or decarbox-
ylation of the oxidation product cadC to form dC in the
presensce of β-ME (see last section) may cause a slight
underestimation of the oxidation rate.
In summary, experiments on the nucleoside and duplex level
reveal that the oxidation of hmdC to fdC is a relatively fast
process that is furthermore catalyzed by thiols. This result has
to be taken into account when biological samples are
investigated regarding the fdC levels.
Deamination Reactions. We next incubated the different
nucleosides at pH 7.4 in water to investigate the deamination of
dC, mdC, hmdC, and fdC to dU, dT, hmdU, and fdU (5-
formyl-2′-deoxyuridine), respectively. The data are depicted in
Figure 3. Clearly evident is that dC, mdC, hmdC, and fdC are
deaminated under these conditions by about the same extent.
Deamination of cadC was not detected. To obtain kinetic data
at 37 °C, we determined the deamination rate constants at four
Figure 2. Investigation of the oxidation reactions of either hmdC or
fdC in a hairpin-oligonucleotide (CpG-ODN 1, 2) with increasing
concentration of β-mercaptoethanol/imidazole (pH 5.0, 50 °C, 48 h).
Reaction yields (normalized to dT [%]) are plotted against the
concentration of β-mercaptoethanol [% v/v]. The yellow data points
show the intensity of the mass signal of the thiol adduct 5-((2″-
hydroxyethyl)thio)methyl-dC (see top structure of Scheme 3), which
was scaled to the right ordinate.
saturation by the thiol has a dramatic influence on the event of
oxidation. In the absence of thiols we detected fdC at about
0.2% in ODN1 (48 h, 37 °C, pH 5.0). Other typical pyrimidine
oxidation products were not detected regardless of the reaction
conditions. We also monitored the levels of the well-established
dG oxidation product 8-oxo-dG and noted here no significant
level change (Figure 2, black curve), arguing for a thiol-
catalyzed oxidation of hmdC to fdC.
Analyzing the mixture by mass spectrometry in more detail
revealed the presence of 5-((2″-hydroxyethyl)thio)methyl-dC
(Figure 2, yellow curve; top structure in Scheme 3), which
shows that a 5-methylene intermediate may be formed during
the reaction, which was first described by the Klimasauskas
group.28
Figure 3. Deamination kinetics of dC to dU (black), mdC to dT
(cyan), hmdC to hmdU (blue), and fdC to fdU (green) at 60.0 °C, pH
7.4. Depicted are the means of triplicate experiments; error bars reflect
the standard deviations. For details of the kinetic measurements at
67.5, 75.0, and 82.5 °C and linear regression analyses see SI.
Treatment of ODN2 (fdC) at 80% v/v β-ME gave rise to the
formation of the oxidized product cadC, but the reaction is
C
dx.doi.org/10.1021/ja403229y | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
different temperatures (see SI) and extrapolated the pseudo-
first-order deamination rates to 37 °C. The data are compiled
in Table 2.
Table 2. Rate Constants k and Activation Energies Ea for the
Deamination of dC, mdC, hmdC, and fdC at pH 7.4
k37 °C [s−1
]
Ea [kJ mol−1
]
dC→dU
9.4 0.5 × 10−10
7.8 0.3 × 10−10
5.8 0.8 × 10−10
1.2 0.2 × 10−9
108.7 1.9
105.0 2.5
104.8 3.9
102.2 2.4
mdC→dT
hmdC→hmdU
fdC→fdU
Figure 4. Investigation of the C−C bond cleavage reaction of cadC to
dC in a hairpin-oligonucleotide (CpG-ODN 3) in a β-mercapto-
ethanol/imidazole mixture. Depicted are the reaction yields
(normalized to dT [%]) depending on the concentration of β-
mercaptoethanol [% v/v] at pH = 5.0 or 7.4 and at 37 or 50 °C after
48 h. The gray area reflects the limit of quantification.
At 37 °C, the deamination rates of dC, mdC, hmdC, and fdC
to form their corresponding 2′-deoxyuridine derivatives are
approximately the same with (6−12) × 10−10 s−1 on the
nucleoside level. The determined rate constant and activation
energy for the deamination of dC are in good agreement with
those reported for single-stranded DNA.29 Bearing in mind that
the rates are more than 2 orders of magnitude lower in double-
stranded DNA,29,30 our data argue that spontaneous deam-
ination of hmdC, fdC, and cadC should be a negligible
background reaction in comparison to the oxidation of hmdC
to form fdC (k37 °C = 1.3 0.1 × 10−8 s −1; Ea = 20.8 2.2 kJ
mol−1). If, consequently, significant amounts of deaminated
compounds are detected, we conclude that these are likely
derived from an enzymatic process.
C−C bond Cleaving Reactions. To investigate the
reaction of hmdC, fdC, and cadC to form dC we heated the
nucleosides for 60 h in water at pH 7.4 at 60 and 80 °C. At
60 °C we detected only traces of dC and the obtained reaction
rates at 80 °C were slow (SI, Figure S9 and Table 3), showing
that uncatalyzed C−C bond cleavage reactions can be
neglected.
Figure 5. Investigation of the C−C bond cleavage reactions of either
hmdC, fdC, or cadC in a hairpin-oligonucleotide (CpG-ODN 1, 2, 3)
in a β-mercaptoethanol/imidazole mixture (pH 5.0, 50 °C, 48 h).
Depicted are the reaction yields (normalized to dT [%]) of hmdC
(blue), fdC (green), and cadC (red) to dC depending on the
concentration of β-mercaptoethanol [% v/v]. The gray area in the
inset shows the limit of quantification.
This picture changes in the presence of thiols, which we
recently reported to catalyze the decarboxylation of cadC.18 To
investigate the conditions of the thiol-mediated decarboxylation
reaction in more detail, we treated the cadC-containing ODN3
with increasing concentrations of β-ME for 48 h at different
pH-values and different temperatures in the presence of
imidazole.18 The oligonucleotides were isolated and analyzed
by LC−MS/MS as outlined in Scheme 2. The results of the
experiments are depicted in Figure 4. Clearly evident is that
under the investigated conditions (pH 5.0 and 50 °C)
decarboxylation of cadC is a relatively efficient reaction. The
yield of dC increased along with the concentration of β-ME up
to 28%. A lower temperature of 37 °C and a higher pH-value of
7.4 resulted in a strong reduction of the dC yield, which shows
that the decarboxylation reaction is an activated proton
catalyzed reaction. We next performed analogous experiments
with hmdC and fdC (pH 5.0, 50 °C, 48 h). Figure 5 compares
the C−C bond cleaving yields of all three nucleosides
embedded in the hairpin duplex structures.
group, who used photohydration conditions.16,17,20 For fdC, in
contrast, we observed considerable deformylation and obtained
yields of up to 2.5%. Decarboxylation of cadC is most efficient
with an obtained yield of up to 28%.
We next studied how different sequences would affect the
C−C bond cleaving and deamination reactions and repeated
the experiments with ODNs 4−6 which feature the xdC-
derivatives in a non-CpG context (Figure 6). We observed only
a small reactivity difference in these sequences compared to the
CpG-ODNs. The yields of deamination products were found to
be lower than 0.4% under these conditions.
The data show that both fdC and cadC can undergo C−C
bond cleavage reactions mediated by thiols. Dehydroxymethy-
lation of hmdC is in contrast a considerably slower process.
To gain deeper insights into the thiol catalysis of the C−C
bond cleavage reactions, we finally computed the reaction
energies using quantum-chemical methods. Computational
details are described in the SI. Carboxylated nucleobases
(caC and caU) were capped at the N1 position with a methyl
We noticed that hmdC dehydroxymethylates to give dC.
However, the obtained yield of dC was very small (0.5%). Far
higher yields were reported by the Klimasauskas group, who
applied a mutated DNA methyltransferase and by the Sowers
Table 3. Rate Constants for the Non-Thiol-Mediated C−C Bond Cleavage of hmdC, fdC, and cadC to dC at pH 7.4 at 80 °C
hmdC→dC
fdC→dC
cadC→dC
k80 °C [s−1
]
1.3 0.2 × 10−9
7.3 1.1 × 10−9
6.9 0.7 × 10−9
D
dx.doi.org/10.1021/ja403229y | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 6. The C−C-bond cleavage of hmdC, fdC, and cadC to dC is
almost independent of the sequence context. Investigation of the C−C
bond cleavage and deamination reactions of either hmdC, fdC, or
cadC in a hairpin-oligonucleotide (CpG-ODN 1, 2, 3; CpNpG-ODN
4, 5, 6) in a 80% (v/v) β-mercaptoethanol/imidazole mixture (pH 5.0,
50 °C, 48 h). Depicted are the reaction yields (normalized to dT [%])
of hmdC (blue), fdC (green), and cadC (red) to dC as well as the
corresponding deamination products hmdU and fdU. Depicted are the
means of triplicate experiments; error bars reflect the standard
deviations.
Figure 7. Schematic representation of different decarboxylation
pathways. Depicted are the reaction energies obtained from quantum
chemical calculations.
group. To describe the explicit solvent−solute hydrogen bonds,
five water molecules were included in the study. Ideally, one
would like to describe more of the long-range electrostatic
solute−solvent interactions by including more water molecules
in the calculations. This would, however, mean a computational
effort which is beyond the scope of this work. A crude way to
approximate the influence of the continuum is to use an
implicit solvent model. Here, we have performed calculations
with an implicit solvent cavity using the COSMO-model31
(data shown in SI). Although the energetics are clearly affected,
the overall trend remains the same. Triple-zeta basis sets32 were
used throughout the calculations. RI-MP232,33 reaction
energies, obtained using the DFT/B3LYP-D334−36 energy
optimized structures, are depicted in Figure 7. No transition
state search was carried out as the reaction rates are already
obtained from the experimental data presented above.
We first computed the direct decarboxylation of isoorotate
(caU), a reaction that is catalyzed by the enzyme isoorotate
decarboxylase (IDCase).37,38 We assumed in our study a direct
decarboxylation via formation of a vinyl anion type
intermediate. Similar mechanistic ideas were the basis of a
recent detailed mechanistic and structural study of the
IDCase.39 We obtained by our calculations a rather high
energy of +34 kcal mol−1 for the vinyl anion of U (gray
intermediate in Figure 7). In agreement with our results from
the thiol-free reaction conditions, our data show that a direct
decarboxylation mechanism for caC is unlikely. In comparison
to the vinyl anion of U, the energy of the vinyl anion of C is
with +47 kcal mol−1 significantly higher. This may in part
explain the observed weak activity39 of IDCase to decarboxylate
5caC to C. In contrast, the energetics of the thiol addition at
the C6 position of caC and the subsequent decarboxylation are
much more favorable: Calculations predict an only slightly
endothermic reaction energy of +6 kcal mol−1 for the
decarboxylation of the thiol-reacted anionic intermediate.
Overall, the reaction of the C6 position with the thiol reduces
the energy of the corresponding anionic intermediate by more
than 40 kcal mol−1 (Figure 7, intermediates shown in red). This
explains why the reaction of cadC with a thiol leads to fast
decarboxylation. We also noted during the computational study
that simultaneous decarboxylation and thiol elimination is an
even more favorable process. This reaction pathway is
symbolized with dotted lines in Figure 7 and illustrated in
Scheme 4. In summary, the calculations support the influence
of the thiol catalysis.
CONCLUSIONS
■
We show that the new nucleosides hmdC, fdC, and cadC
possess an increased reactivity compared to mdC and dC. First,
the intrinsic deamination rates of hmdC and fdC are
comparable to those of mdC and dC. Second, hmdC is more
susceptible to oxidation. It reacts surprisingly quickly to fdC if
exposed to atmospheric oxygen, which is a problem that needs
to be considered when DNA is isolated from biological material
for the determination of fdC. In addition, the reaction of hmdC
to fdC is accelerated in the presence of thiols. If fdC is exposed
to atmospheric oxygen further oxidation to cadC is a rather
inefficient process.
Third and importantly, fdC and cadC can undergo thiol-
mediated and acid-catalyzed C−C bond cleavage reactions to
form dC under release of formic acid and CO2, respectively
(see Scheme 4). Here decarboxylation is by a factor of 11 more
efficient than the deformylation of fdC. If we consider that
DNA demethylation requires stepwise oxidation of hmdC to
fdC and cadC, both deformylation of fdC and decarboxylation
of cadC could take place via alternative active demethylation
mechanisms.
E
dx.doi.org/10.1021/ja403229y | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Scheme 4. Proposed Mechanisms of the Thiol-Mediated and Acid-Catalyzed C−C Bond Cleavage Reactions of fdC and cadC to
a
dC
a
The reactions are thought to proceed via a covalent enamine intermediate. The deformylation reaction of fdC further requires the addition of a
nucleophile (water or thiol) to the aldehyde before the release of formic acid (or the thiol ester) can proceed. In principal, the release of formic acid
from fdC or carbon dioxide from cadC and thiol elimination could occur simultaneously or in two steps (shown by gray or black arrows,
respectively). R-SH = β-mercaptoethanol; Im = imidazole.
(6) Guo, J. U.; Su, Y.; Zhong, C.; Ming, G.-l.; Song, H. Cell 2011,
145, 423.
ASSOCIATED CONTENT
* Supporting Information
■
S
(7) Cortellino, S.; Xu, J.; Sannai, M.; Moore, R.; Caretti, E.; Cigliano,
Experimental procedures and spectroscopic data of standards
A.; Le Coz, M.; Devarajan, K.; Wessels, A.; Soprano, D.; Abramowitz,
for the UHPLC-MS/MS analysis, detailed procedures for mass
L. K.; Bartolomei, M. S.; Rambow, F.; Bassi, M. R.; Bruno, T.;
spectrometric measurements. and method development. This
material is available free of charge via the Internet at http://
pubs.acs.org.
Fanciulli, M.; Renner, C.; Klein-Szanto, A. J.; Matsumoto, Y.; Kobi, D.;
Davidson, I.; Alberti, C.; Larue, L.; Bellacosa, A. Cell 2011, 146, 67.
(8) Nabel, S. N.; Jia, H.; Ye, Y.; Shen, L.; Goldschmidt, H. L.; Stivers,
J. T.; Zhang, Y.; Kohli, R. M. Nat. Chem. Biol. 2012, 8, 751.
(9) Bienvenu, C.; Wagner, J. R.; Cadet, J. J. Am. Chem. Soc. 1996,
118, 11406.
(10) Cao, H.; Wang, Y. Nucleic Acids Res. 2007, 35, 4833.
(11) Wagner, J. R.; Cadet, J. Acc. Chem. Res. 2010, 43, 564.
(12) Liu, S.; Wang, J.; Su, Y.; Guerrero, C.; Zeng, Y.; Mitra, D.;
Brooks, P. J.; Fisher, D. E.; Song, H.; Wang, Y. Nucleic Acids Res. 2013,
41, 6421.
(13) Taghizadeh, K.; McFaline, J. L.; Pang, B.; Sullivan, M.; Dong,
M.; Plummer, E.; Dedon, P. C. Nat. Protoc. 2008, 3, 1287.
(14) Ravanat, J.-L.; Douki, T.; Duez, P.; Gremaud, E.; Herbert, K.;
Hofer, T.; Lasserre, L.; Saint-Pierre, C.; Favier, A.; Cadet, J.
Carcinogenesis 2002, 23, 1911.
AUTHOR INFORMATION
Corresponding Author
Thomas.Carell@cup.uni-muenchen.de
■
Author Contributions
⊥S.S. and T.P. contributed equally.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
This work was supported by the Deutsche Forschungsgemein-
schaft via SFB749 (TPA4/C7) and the Excellence cluster
EXC114 (CiPSM). S.S., T.P., and A.S.S. thank the Fonds der
Chemischen Industrie for predoctoral fellowships. K.S. and
C.O. thank Dr. Denis Flaig for helpful discussions. G.K. thanks
the Japan Society for the Promotion of Science (JSPS) as a
JSPS Postdoctoral Fellow for Research Abroad.
(15) Globisch, D.; Munzel, M.; Muller, M.; Michalakis, S.; Wagner,
̈
̈
M.; Koch, S.; Bruckl, T.; Biel, M.; Carell, T. PLoS One 2010, 5, e15367
̈
DOI: 10.1371/journal.pone.0015367.
(16) Hamm, S.; Just, G.; Lacoste, N.; Moitessier, N.; Szyf, M.;
Mamer, O. Bioorg. Med. Chem. Lett. 2008, 18, 1046.
(17) Liutkeviciute, Z.; Lukinavicius, G.; Masevicius, V.; Daujotyte, D.;
Klimasauskas, S. Nat. Chem. Biol. 2009, 5, 400.
(18) Schiesser, S.; Hackner, B.; Pfaffeneder, T.; Muller, M.;
Hagemeier, C.; Truss, M.; Carell, T. Angew. Chem., Int. Ed. 2012,
51, 6516.
(19) Wu, S. C.; Zhang, Y. Nat. Rev. Mol. Cell Biol. 2010, 11, 607.
(20) Privat, E.; Sowers, L. C. Chem. Res. Toxicol. 1996, 9, 745.
(21) Tardy-Planechaud, S.; Fujimoto, J.; Lin, S. S.; Sowers, L. C.
Nucleic Acids Res. 1997, 25, 553.
̈
REFERENCES
■
(1) Kriaucionis, S.; Heintz, N. Science 2009, 324, 929.
(2) Tahiliani, M.; Koh, K. P.; Shen, Y.; Pastor, W. A.; Bandukwala,
H.; Brudno, Y.; Agarwal, S.; Iyer, L. M.; Liu, D. R.; Aravind, L.; Rao, A.
Science 2009, 324, 930.
(3) Pfaffeneder, T.; Hackner, B.; Truß, M.; Munzel, M.; Muller, M.;
Deiml, C. A.; Hagemeier, C.; Carell, T. Angew. Chem., Int. Ed. 2011,
̈
̈
(22) Dai, Q.; He, C. Org. Lett. 2011, 13, 3446.
(23) Munzel, M.; Lischke, U.; Stathis, D.; Pfaffeneder, T.; Gnerlich,
50, 7008.
̈
(4) He, Y.-F.; Li, B.-Z.; Li, Z.; Liu, P.; Wang, Y.; Tang, Q.; Ding, J.;
Jia, Y.; Chen, Z.; Li, L.; Sun, Y.; Li, X.; Dai, Q.; Song, C.-X.; Zhang, K.;
He, C.; Xu, G.-L. Science 2011, 333, 1303.
(5) Ito, S.; Shen, L.; Dai, Q.; Wu, S. C.; Collins, L. B.; Swenberg, J.
A.; He, C.; Zhang, Y. Science 2011, 333, 1300.
F. A.; Deiml, C. A.; Koch, S. C.; Karaghiosoff, K.; Carell, T. Chem.
Eur. J. 2011, 17, 13782.
(24) Steigenberger, B.; Schiesser, S.; Hackner, B.; Brandmayr, C.;
Laube, S. K.; Steinbacher, J.; Pfaffeneder, T.; Carell, T. Org. Lett. 2013,
15, 366.
F
dx.doi.org/10.1021/ja403229y | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(25) Munzel, M.; Globisch, D.; Trindler, C.; Carell, T. Org. Lett.
̈
2010, 12, 5671.
(26) Munzel, M.; Globisch, D.; Bruckl, T.; Wagner, M.; Welzmiller,
̈
̈
V.; Michalakis, S.; Muller, M.; Biel, M.; Carell, T. Angew. Chem., Int.
̈
Ed. 2010, 49, 5375.
(27) Spruijt, Cornelia G.; Gnerlich, F.; Smits, Arne H.; Pfaffeneder,
T.; Jansen, Pascal W. T. C.; Bauer, C.; Munzel, M.; Wagner, M.;
̈
Muller, M.; Khan, F.; Eberl, H. C.; Mensinga, A.; Brinkman, Arie B.;
̈
Lephikov, K.; Muller, U.; Walter, J.; Boelens, R.; van Ingen, H.;
Leonhardt, H.; Carell, T.; Vermeulen, M. Cell 2013, 152, 1146.
̈
̌ ̌
(28) Liutkeviciute, Z.; Kriukiene, E.; Grigaityte, I.; Masevicius, V.;
̇ ̇ ̇
̅
̌
Klimasauskas, S. Angew. Chem., Int. Ed. 2011, 123, 2138.
(29) Frederico, L. A.; Kunkel, T. A.; Shaw, B. R. Biochemistry 1990,
29, 2532.
(30) Shen, J.-C.; Rideout, W. M., III; Jones, P. A. Nucleic Acids Res.
1994, 22, 972.
(31) Klamt, A.; Schuurmann, G. J. Chem. Soc., Perkin Trans. 2 1993, 5,
̈
̈
799.
(32) Weigend, F.; Haser, M.; Patzelt, H.; Ahlrichs, R. Chem. Phys.
̈
Lett. 1998, 294, 143.
(33) Feyereisen, M.; Fitzgerald, G.; Komornicki, A. Chem. Phys. Lett.
1993, 208, 359.
(34) Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B 1998, 37, 785.
(35) Becke, A. D. J. Chem. Phys. 1993, 98, 5648.
(36) Grimme, S.; Ehrlich, S.; Goerigk, L. J. Comput. Chem. 2011, 32,
1456.
(37) Smiley, J. A.; Angelot, J. M.; Cannon, R. C.; Marshall, E. M.;
Asch, D. K. Anal. Biochem. 1999, 266, 85.
(38) Palmatier, R. D.; McCroskey, R. P.; Abbott, M. T. J. Biol. Chem.
1970, 245, 6706.
(39) Xu, S.; Li, W.; Zhu, J.; Wang, R.; Li, Z.; Xu, G.-L.; Ding, J. Cell
Res. 2013, DOI: 10.1038/cr.2013.107.
G
dx.doi.org/10.1021/ja403229y | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX