Influencing Epigenetic Information with a Hydrolytically Stable Carbocyclic 5-Aza-2′-deoxycytidine

Article abstract of DOI:10.1002/anie.201904794

5-Aza-2′-deoxycytidine (AzadC) is an antimetabolite in clinical use, which reduces the level of the epigenetic modification 5-methyl-2′-deoxycytidine (mdC). AzadC is incorporated into the genome of proliferating cells, where it inhibits DNA methyltransferases (DNMTs), leading to a reduction of mdC. The loss of mdC, which is a transcriptional silencer in the promoter region found upstream of genes, leads to the reactivation of the corresponding gene, including tumor-suppressor genes, which elicits a beneficial effect. The problem associated with AzadC is that the compound is hydrolytically unstable. It decomposes during treatment to a variety of poorly characterized hydrolysis products. After its incorporation into the genome, this hydrolytic instability generates abasic sites. It is consequently difficult to dissect whether the activity of the compound is caused by DNMT inhibition or more generally by DNA lesion formation. We now discovered that a disarmed version of AzadC, in which the ribose oxygen was replaced by a CH₂ group, is surprisingly stable under a variety of pH values while keeping activity against the DNMTs.

Full text of DOI:10.1002/anie.201904794

A Journal of the Gesellschaft Deutscher Chemiker
Angewandte
Chemie
International Edition
www.angewandte.org
Accepted Article
Title: Influencing epigenetic information with a hydrolytically stable
carbocyclic 5 aza-2’-deoxycytidine
Authors: Thomas Carell, Thomas Wildenhof, Sarah Schiffers,
Franziska Regine Traube, and Peter Mayer
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Angew. Chem. Int. Ed. 10.1002/anie.201904794
Angew. Chem. 10.1002/ange.201904794
Link to VoR: http://dx.doi.org/10.1002/anie.201904794
http://dx.doi.org/10.1002/ange.201904794

10.1002/anie.201904794
Angewandte Chemie International Edition
COMMUNICATION
Influencing epigenetic information with a hydrolytically stable
carbocyclic 5-aza-2’-deoxycytidine
Thomas M. Wildenhof,[a] Sarah Schiffers,[a] Franziska R. Traube,[a] Peter Mayer[a] and Thomas Carell*[a]
Dedicated to Prof. J. Rebek, Jr. on the occasion of his 75th birthday
Abstract: 5-Aza-2’-deoxycytidine (AzadC) is an antimetabolite in
clinical use, which reduces the level of the epigenetic modification
5-methyl-2’-deoxycytidine (mdC). AzadC is incorporated into the
genome of proliferating cells, where it inhibits the DNA
methyltransferases (DNMTs) in a suicide process leading to a
reduction of mdC. The loss of mdC, which is a transcriptional silencer
in promoters, leads to the reactivation of genes including tumor
suppressor genes, which elicits a beneficial effect. The problem
associated with AzadC is that the compound is hydrolytically unstable.
It decomposes during treatment to a variety of poorly characterized
hydrolysis products. After its incorporation into the genome, this
hydrolytic instability generates abasic sites. It is consequently difficult
to dissect if the activity of the compound is caused by DNMT inhibition
or more generally by DNA lesion formation. We now discovered that
a disarmed version of AzadC, in which the ribose oxygen was
replaced by a CH2-group, is surprisingly stable under a variety of pH
values while keeping the epigenetic activity against the DNMTs.
use as one of the first pharmaceuticals that operates at the
epigenetic level for the treatment of myelodysplastic syndromes
(MDS)[2] and for acute myeloid leukemia (AML)[4]. Clinically, it is
administered in several cycles, with each cycle involving one
week of treatment and three weeks of pausing.
The problem associated with AzadC is that the compound
hydrolyses in aqueous solution following the path depicted in
Fig. 1b. This hydrolysis compromises the activity of AzadC,
particularly over the long treatment times. In order to circumvent
this problem, it is necessary to generate an AzadC compound that
can demethylate (and hence react with an S-nucleophile), while
hydrolysis (reaction with an O-nucleophile) should be blocked.
Such a compound may allow to dissect how demethylation and
lesion formation contribute to the anti-cancer activity, which is an
information needed for the design of new epigenetically acting
antimetabolites.
Here, we report that replacing the oxygen of the ribose by a CH2-
group has a surprisingly large remote effect on the reactivity of
the heterocycle. The created carbocyclic version of AzadC
(cAzadC, 1) still inhibits DNMTs but is hydrolytically stable
(Fig. 1c).
5-Aza-2’-deoxycytidine (decitabine, AzadC) is a nucleoside
analogue that is able to manipulate epigenetic information.[1-5]
Epigenetic information in DNA is associated with the formation of
5-methyl-2’-deoxycytidine (mdC) from 2’-deoxycytidine (dC) with
the help of DNA methyltransferases (DNMTs) and
S-adenosylmethionine (SAM) as the methylating cofactor.[6-7, 4]
Methylation of dC to mdC in promoter regions is typically
associated with transcriptional silencing of genes.[8-9] AzadC is a
prodrug that is inside cells converted into the corresponding active
triphosphate and subsequently incorporated into the genome
during cell division. The mode of action of AzadC involves
reaction of its electrophilic C6 positions with a DNMT active site
thiol nucleophile (Fig. 1a).[10-11] This generates
a covalent
intermediate that is methylated by the SAM cofactor as depicted
in Fig 1a. Due to the N-atom at position 5 of the triazine
heterocycle, the final -elimination reaction, which would usually
release mdC from the DNMT enzyme, is not possible anymore.
The consequence is the formation of a covalent DNA-DNMT
crosslink. As a result of administering AzadC, a large drop of the
mdC levels (hypomethylating effect) is observed, which leads to
the reactivation of silenced tumor suppressor genes in cancer
cells.[1] This epigenetic effect is hoped to re-differentiate cancer
cells back into normally proliferating cells. AzadC is currently in
Figure 1. Depiction of 5-aza-2’-deoxycytidine (decitabine, AzadC) together with
its mode of action. a) Active site thiol reacts with the C6-position of AzadC. b)
Hydrolytic degradation pathway that goes in hand with reaction of a water
molecule with the C6-position (O-reactivity) of AzadC. This leads to a final base
loss and formation of an abasic site. c) Boxed, depiction of the carbocyclic
version cAzadC 1.
[a]
M. Sc. T. M. Wildenhof, M. Sc. S. Schiffers, M. Sc. F. R. Traube, Dr.
P. Mayer, Prof. Dr. T. Carell
Department of Chemistry
Ludwig-Maximilians-Universität, Munich
Butenandtstrasse 5-13
E-mail: Thomas.carell@lmu.de
The synthesis of cAzadC 1 is depicted in Scheme 1. It starts with
the Boc-protected aminocyclopentane derivative 2 that we used
previously to synthesize DNA lesion analogues.[12-15] Compound
2 was first benzyl-protected to 3, Boc-deprotected to 4, and then
reacted with carbimidazole 5, which was prepared in two steps
from isomethylurea 6 after generation of the free base 7 with
potassium hydroxide and reaction of 7 with carbonyldiimidazole.
Supporting information for this article is given via a link at the end of
the document
For internal use, please do not delete. Submitted_Manuscript
This article is protected by copyright. All rights reserved.

10.1002/anie.201904794
Angewandte Chemie International Edition
COMMUNICATION
We next investigated the stability of cAzadC 1 in direct
comparison to the pharmaceutical AzadC (Fig. 3). Since one
treatment cycle goes over four weeks we decided to measure the
stability at a time point related to a half cycle (14 d). We dissolved
AzadC and cAzadC 1 at a concentration of 100 mM in a
phosphate buffer (100 mM) at three different pH values (7.4, 5.5
and 8.5) and measured NMR spectra after keeping the solutions
at r.t. Since tumour cells often provide an acidic micro-
environment,[16] the stability under slightly acidic pH is particularly
informative. As evident from the data shown in Fig. 3, the
pharmaceutical AzadC strongly degraded within these 14 d.
Importantly, at pH = 5.5 and at pH = 8.5, intact AzadC was only
hardly detectable anymore. At physiological pH (7.4), AzadC was
still present after 14 d but the level of degradation is dramatic. In
contrast to these results, we observed for cAzadC 1 surprisingly
no degradation at all tested pH values, including pH = 5.5. This
result led to the surprising discovery that the simple O → CH2
exchange causes a strong remote disarming effect that seems to
change the properties of the triazine ring so that reaction with
water is stopped.
This provides the carbamoylurea-cyclopentane nucleoside
analogue 8. Cyclization to the triazine base 9 was subsequently
performed with triethylorthoformate. Reaction of 9 with NH3 in
methanol and deprotection of the benzyl groups with BCl3 in
dichloromethane furnished the final compound cAzadC 1 as the
free nucleoside.
Scheme 1. Synthesis of the carbocyclic 5-aza-2’-deoxycytidine (cAzadC, 1). a)
NaH, BnBr, DMF, 0 °C, 1.5 h and stirred for additional 2 h at r.t.; b) TFA (30%),
CH2Cl2 then Na2CO3, 10 min r.t.; c) CH3CN, reflux, 2 h d) HC(OEt)3, TFA cat.,
reflux, 3 h; e) NH3 (7 N, MeOH), 3 h, r.t., then H2O ;f) CH2Cl2, -78 °C, BCl3, 1 h,
then  r.t., 2 h, MeOH, 20 min. g) KOH, Et2O:H2O (39:1), -15 °C, 30 min, h)
carbonyldiimidazole, THF, r.t., 3 h; R = Me or Et.
Recrystallization of compound cAzadC 1 from hot methanol gave
colourless needles, which allowed us to solve the crystal structure
that is depicted in Fig. 2. Interesting is the observation that
cAzadC 1 exists with two different cyclopentane conformations in
the crystal (Fig. 2; Fig.SI 1). One conformer adopts a C6’-endo
(P = 88.2 °, max = 47.8 °) conformation (Fig. 2a), while the second
exists as the C2’-endo-C3’-exo (South, P = 150.8 °, max = 45.4 °)
conformer (Fig. 2b). The latter conformation is typical for
2’-deoxynucleosides in DNA. This shows that the cAzadC 1
nucleoside can adopt the correct DNA-type conformation, fueling
hope that the analogue has the potential to get phosphorylated
and integrated into the genome.
Figure 3. HPLC-based stability measurements showing (a) severe hydrolytic
decomposition of 5-aza-2’-deoxycytidine (AzadC) solutions at different pH
values, while (b) the carbocyclic compound cAzadC 1 was stable at all three pH
values. The inset table in (a) shows the chromatogram between t1 = 10 min and
t2 = 20 min for AzadC. The AzadC signal is depicted in red.
We next investigated if this disarming effect would influence the
biological functions. We used for this purpose mouse embryonic
stem cells (mESC) that were primed in serum/LIF as a model
system, since mdC levels increase from naïve to primed state.[17]
We added cAzadC 1 in two different concentrations (1 µM and
5 µM) to mESC that have been primed for 48 h and allowed the
cells to further proliferate under priming conditions in the presence
of cAzadC 1 for additional 72 h. After the 72 h, we harvested the
cells, isolated the DNA and digested the DNA down to the
nucleoside level using our described protocol.[18] The levels of
mdC were finally precisely quantified using isotope dilution
UHPLC-MS2. To this end, isotopically labelled standards of the
18]
nucleosides were spiked in for exact quantification.[19,
In
addition to mdC, we quantified the levels of 5-hydroxymethyl-
2’-deoxycytidine (hmdC), which is formed from mdC by the action
of TET enzymes.[20-21] The absolute levels of hmdC are in mESC
Figure 2. Crystal structure of carbocyclic 5-aza-2’-deoxycytidine (cAzadC 1)
showing the molecule in the observed C6’-endo conformation (a) and the C2’-
endo-C3’-exo conformation (2T1) (b).
For internal use, please do not delete. Submitted_Manuscript
This article is protected by copyright. All rights reserved.

10.1002/anie.201904794
Angewandte Chemie International Edition
COMMUNICATION
more than ten times lower than the mdC levels[20, 22]. The
consequence is that even after a substantial reduction of mdC,
there should be sufficient mdC to keep the hmdC levels constant.
The question if and by how much the hmdC level is affected can
therefore inform us about how epigenetic reprogramming is
organized. Parallel to the quantification of mdC and hmdC we also
quantified to which extent cAzadC 1 itself was incorporated into
the genome of the mESC. Detection of AzadC in the genome of
treated cells is only possible after treatment of the DNA with
NaBH4. Application of NaBH4 reduces the C(5)=C(6) double bond,
triphosphate. The slower kinetics of cAzadC 1, however, is not
necessarily a disadvantage given the long treatment times that
are applied in the clinic.
Very interesting is also the discovery that the hmdC levels were
reduced to about 50% already in the 1 M experiment. At 5 M,
we were even unable to detect hmdC above background levels
using 0.5 µg of genomic DNA. The result shows that the hmdC
level dropped even faster than the mdC levels, although hmdC is
more than ten times less abundant in the genome. This result is
interesting. It indicates that hmdC might be potentially
predominantly generated in the mdC maintenance process during
cell division. We see here that compound cAzadC 1 is a perfect
tool molecule that now allows to gain further insight into the
interplay between methylation of dC to mdC and oxidation of mdC
to hmdC. With the new compound cAzadC 1 in hand we can now
begin to clearly correlate demethylation of the genome with the
corresponding cellular effects without compromising DNA
damaging effects. Finally, cAzadC 1 may not only be a valuable
tool compound but potentially even a next generation epigenetic
pharmaceutical.
which stabilizes the compound so that its quantification becomes
19]
possible.[23,
To our delight, we noted that the stability of
cAzadC 1 allowed its quantification without this pre-treatment. We
also noted that the applied enzymatic digestion protocol allowed
to digest genomic DNA (gDNA) even in the presence of large
amounts of cAzadC 1. Taken together, quantification of cAzadC 1
by UHPLC-MS2 using an external calibration curve (Fig. SI2) was
possible in parallel to quantification of canonical and epigenetic
bases.
In summary, we show that the replacement of the in-ring O-atom
by a CH2-unit stabilizes the pharmaceutical so that its nucleophilic
reaction with water is stopped. The new nucleoside cAzadC 1 is
accepted by the phosphorylating enzymes in cells and the
corresponding cAzadC-triphosphates are efficiently incorporated
into the genome. cAzadC 1 is incorporated in the genome with
several million nucleotides and it causes the mdC level to
decrease to 70% relative to the control levels.
Experimental Section
Figure 4. Depiction of the quantification data of DNA modifications of
carbocyclic 5-aza-2’-deoxycytidine-treated (cAzadC 1) mouse embryonic stem
cells (mESC) obtained by UHPLC-MS2. For each condition, three biological
replicates were measured in technical triplicates. For each technical replicate,
0.5 µg of DNA were digested. Bar graphs represent mean, error bars represent
standard deviation. LLOQ indicates the lower limit of quantification.
Synthesis
All synthetic procedures are described in detail in the supplementary
material.
Cell culture of mESC for cAzadC treatment
Feeder independent wt J1 (strain 129S4/SvJae)[24] cells were cultured
in the presence of serum and LIF as previously described[25]. They
were routinely maintained on gelatinized plates in 2i/L medium. For
priming experiments, 2i cultures were passaged when applicable in
DMEM supplemented with FBS and LIF as above but lacking the
inhibitors. For drug treatment, cells were moved into the primed state
by removing 2i from the medium. Cells were incubated 2 d in DMEM
supplemented with FBS and LIF in 6-well plates (VWR). After splitting,
2x105 cells were transferred into a 6-well plate culture dish and
supplemented with either 1 µM (in 0.01% DMSO) or 5 µM cAzadC (in
0.05% DMSO) and treated for 72 h. After removal of the medium and
washing the cells with DPBS, they were directly lysed with RLT+ buffer
as described in a previous publication[18]
At 1 M cAzadC 1 concentration, we detected a cAzadC 1 level
of 5x10-4 cAzadC per dN (Fig. 4a). This amounts to almost
3 million cAzadC nucleotides integrated into the genome. At the
higher concentration of 5 µM cAzadC 1, the level increased 3-fold
to 1.7x10-3 cAzadC per dN and consequently to more than
8 million integrated cAzadCs per genome. Compared to the
incorporation of AzadC, which reaches 1.2x10-3 AzadC per dN,
when applied with 1 µM[19], the levels of cAzadC 1 reaches about
a third of this level. The data clearly show that the carbocyclic
version of AzadC (cAzadC 1) is incorporated and that it reaches
in the genome finally comparable levels at 5 µM concentration.
Importantly, after exposing the mESC for 72 h at 1 M cAzadC in
the medium, we detected a reduction of the mdC values by almost
30% (Fig. 4b). At 5 M concentration in the medium, the mdC
levels dropped even to about 50% of the original value. A
decrease to 50% is observed for AzadC as well. Here, however,
the 50%-reduction is reached faster (24 h) and already with lower
AzadC concentration (1 µM).[19] The data show that the
carbocyclic version cAzadC 1 needs simply more time to affect
the mdC levels by the same amount. We believe that this effect is
caused by a potentially slower conversion of cAzadC 1 into the
gDNA isolation, total enzymatic digest and UHPLC-MS2
The gDNA was isolated as described previously[18]. Due to the higher
stability of cAzadC 1, a hydrogenation procedure was not necessary
and the gDNA was directly subjected to a total enzymatic digest and
analyzed using UHPLC-MS2 as described in a previous publication.[19]
For internal use, please do not delete. Submitted_Manuscript
This article is protected by copyright. All rights reserved.

10.1002/anie.201904794
Angewandte Chemie International Edition
COMMUNICATION
[21]
S. Ito, L. Shen, Q. Dai, S. C. Wu, L. B. Collins, J. A.
Swenberg, C. He, Y. Zhang, Science 2011, 333, 1300-
1303.
Acknowledgements
We thank the Deutsche Forschungsgemeinschaft (DFG) for
financial support via the programs SFB1309 (TP-A4), SFB1361,
SPP1784, and GRK2338/1 (P12). This project has received
funding from the European Research Council (ERC) under the
European Union's Horizon 2020 research and innovation
programme (grant agreement n° EPiR 741912). F.R.T. thanks the
Boehringer Ingelheim Fonds for a PhD fellowship.
[22]
[23]
M. Münzel, D. Globisch, T. Carell, Angew. Chem. Int.
Ed. Engl. 2011, 50, 6460-6468.
A. Unnikrishnan, A. N. Q. Vo, R. Pickford, M. J.
Raftery, A. C. Nunez, A. Verma, L. B. Hesson, J. E.
Pimanda, Leukemia 2018, 32, 900-910.
E. Li, T. H. Bestor, R. Jaenisch, Cell 1992, 69, 915-
926.
T. Pfaffeneder, B. Hackner, M. Truss, M. Münzel, M.
Müller, C. A. Deiml, C. Hagemeier, T. Carell, Angew.
Chem. Int. Ed. Engl. 2011, 50, 7008-7012.
[24]
[25]
Keywords: epigenetics • 5-methyl-2’-deoxycytidine • DNA
methylation • antimetabolite • decitabine
[1]
[2]
M. Daskalakis, T. T. Nguyen, C. Nguyen, P. Guldberg,
G. Kohler, P. Wijermans, P. A. Jones, M. Lubbert,
Blood 2002, 100, 2957-2964.
H. Kantarjian, J. P. Issa, C. S. Rosenfeld, J. M.
Bennett, M. Albitar, J. DiPersio, V. Klimek, J. Slack, C.
de Castro, F. Ravandi, R. Helmer, 3rd, L. Shen, S. D.
Nimer, R. Leavitt, A. Raza, H. Saba, Cancer 2006,
106, 1794-1803.
[3]
[4]
[5]
C. Stresemann, F. Lyko, Int. J. Cancer 2008, 123, 8-
13.
Y. Koh, Y. A. Kim, K. Kim, J.-A. Sim, S.-S. Yoon, S. M.
Park, Y. H. Yun, Blood 2016, 128, 2381-2381.
M. Nieto, P. Demolis, E. Béhanzin, A. Moreau, I.
Hudson, B. Flores, H. Stemplewski, T. Salmonson, C.
Gisselbrecht, D. Bowen, F. Pignatti, Oncologist 2016,
21, 692-700.
[6]
[7]
S. S. Smith, B. E. Kaplan, L. C. Sowers, E. M.
Newman, Proc. Natl. Acad. Sci. 1992, 89, 4744-4748.
G. G. Wilson, R. J. Roberts, S. Kumar, J. Posfai, M.
Sha, S. Klimasauskas, X. Cheng, Nucleic Acids Res.
1994, 22, 1-10.
[8]
[9]
A. Bird, Genes Dev. 2002, 16, 6-21.
R. J. Klose, A. P. Bird, Trends Biochem. Sci. 2006, 31,
89-97.
[10]
[11]
[12]
[13]
[14]
R. Jüttermann, E. Li, R. Jaenisch, Proc. Natl. Acad.
Sci. 1994, 91, 11797-11801.
S. Gabbara, A. S. Bhagwat, Biochem. J. 1995, 307,
87-92.
M. Ober, H. Müller, C. Pieck, J. Gierlich, T. Carell, J.
Am. Chem. Soc. 2005, 127, 18143-18149.
H. Müller, T. Carell, Eur. J. Org. Chem. 2007, 2007,
1438-1445.
F. Büsch, J. C. Pieck, M. Ober, J. Gierlich, G. W. Hsu,
L. S. Beese, T. Carell, Chem. Eur. J. 2008, 14, 2125-
2132.
[15]
[16]
T. H. Gehrke, U. Lischke, K. L. Gasteiger, S.
Schneider, S. Arnold, H. C. Müller, D. S. Stephenson,
H. Zipse, T. Carell, Nat. Chem. Bio. 2013, 9, 455.
Y. Kato, S. Ozawa, C. Miyamoto, Y. Maehata, A.
Suzuki, T. Maeda, Y. Baba, Cancer Cell Int. 2013, 13,
89.
[17]
[18]
S. Takahashi, S. Kobayashi, I. Hiratani, Cell. Mol. Life
Sci. 2018, 75, 1191-1203.
F. R. Traube, S. Schiffers, K. Iwan, S. Kellner, F.
Spada, M. Müller, T. Carell, Nat. Protoc. 2019, 14,
283-312.
[19]
[20]
S. Schiffers, T. M. Wildenhof, K. Iwan, M. Stadlmeier,
M. Müller, T. Carell, Helv. Chim. Acta 2019, 102,
e1800229.
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.
For internal use, please do not delete. Submitted_Manuscript
This article is protected by copyright. All rights reserved.

10.1002/anie.201904794
Angewandte Chemie International Edition
COMMUNICATION
Entry for the Table of Contents (Please choose one layout)
Layout 1:
COMMUNICATION
A
carbocyclic analogue of the
Thomas M. Wildenhof, Sarah Schiffers,
Franziska R. Traube, Peter Mayer,
Thomas Carell*
epigenetic drug 5-aza-2’-deoxycytidine
is hydrolytically stable and shows
excellent reduction of the methylation
levels in stem cells
Page No. – Page No.
Influencing epigenetic information
with a disarmed carbocyclic
azacytidine
For internal use, please do not delete. Submitted_Manuscript
This article is protected by copyright. All rights reserved.