Furan-oxidation-triggered inducible DNA cross-linking: Acyclic versus cyclic furan-containing building blocks - On the benefit of restoring the cyclic sugar backbone

Article abstract of DOI:10.1002/chem.201100067

Oligodeoxynucleotides incorporating a reactive functionality can cause irreversible cross-linking to the target sequence and have been widely studied for their potential in inhibition of gene expression or development of diagnostic probes for gene analysi

Full text of DOI:10.1002/chem.201100067

DOI: 10.1002/chem.201100067
Furan-Oxidation-Triggered Inducible DNA Cross-Linking: Acyclic Versus
Cyclic Furan-Containing Building Blocks—On the Benefit of Restoring the
Cyclic Sugar Backbone
Kristof Stevens,[a] Diederica D. Claeys,[b] Saron Catak,[b] Sara Figaroli,[a] Michal Hocek,[c]
Jan M. Tromp,[d] Stefan Schꢀrch,[d] Veronique Van Speybroeck,[b] and
Annemieke Madder*[a]
6940
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 6940 – 6953

FULL PAPER
Abstract: Oligodeoxynucleotides incor-
porating a reactive functionality can
cause irreversible cross-linking to the
target sequence and have been widely
studied for their potential in inhibition
of gene expression or development of
diagnostic probes for gene analysis. Re-
active oligonucleotides further show
potential in a supramolecular context
for the construction of nanometer-sized
DNA-based objects. Inspired by the cy-
tochrome P450 catalyzed transforma-
tion of furan into a reactive enal spe-
cies, we recently introduced a furan-ox-
idation-based methodology for cross-
linking of nucleic acids. Previous ex-
periments using a simple acyclic build-
ing block equipped with a furan moiety
for incorporation into oligodeoxynu-
cleotides have shown that cross-linking
occurs in a very fast and efficient way
and that substantial amounts of stable,
site-selectively cross-linked species can
be isolated. Given the destabilization
of duplexes observed upon introduc-
tion of the initially designed furan-
modified building block into DNA du-
plexes, we explore here the potential
benefits of two new building blocks
featuring an extended aromatic system
and a restored cyclic backbone. Thor-
ough experimental analysis of cross-
linking reactions in a series of contexts,
combined with theoretical calculations,
permit structural characterization of
the formed species and allow assess-
ment of the origin of the enhanced
cross-link selectivity. Our experiments
clearly show that the modular nature
of the furan-modified building blocks
used in the current cross-linking strat-
egy allow for fine tuning of both yield
and selectivity of the interstrand cross-
linking reaction.
Keywords: bioorganic chemistry ·
DNA · molecular modeling · nu-
cleosides · oxidation
Introduction
(+)-Aspicillin.[9] The principle was further applied in a com-
binatorial process for the generation of a large library of dis-
tinct molecular skeletons starting from a common 2-alkoxy-
furan building block.[10]
Furan is a chemical that has been found to be specifically
hepatotoxic.[1–3] The mechanism of action involved in this
toxicity rests on the cytochrome P450 mediated oxidation of
the stable furan aromatic moiety into the reactive (Z)-2-bu-
tenedial, which is readily attacked by biological nucleophiles
leading to a variety of deleterious adducts.[4–6] The study of
its genotoxic effect and the underlying mechanism thereof
continues to be an active area of interest and still needs to
be completely understood.[7] Notwithstanding these toxic ef-
fects, furan has been identified as representing a caged reac-
tive functionality and seems to be an ideal template for the
mild and selective generation of a reactive intermediate on
demand. This furan-ring oxidation principle and the idea of
using an easily incorporated and stable aromatic furan
moiety as a substitute for a reactive unit has been elegantly
used in the total synthesis of macrosphelides A and B[8] and
Despite a few examples of its application both in total
synthesis and in diversity-oriented synthesis, this furan oxi-
dation strategy has never been exploited in a biological con-
text before. Nevertheless, the furan moiety represents the
ideal masked reactive functionality, offering the possibility
for selective and on-demand triggering of its reactivity
through oxidation, which can be achieved both in a chemical
or an enzymatic way. We have therefore recently exploited
this principle in the development of a novel furan-oxidation-
based method for selective DNA interstrand cross-linking
(ICL).[11,12]
Besides DNA interstrand cross-link repair, nucleotide ex-
cision repair, mismatch repair, base excision repair, and
double-strand break repair are processes that are of para-
mount importance for the maintenance of our genome. The
genomic material is constantly challenged and an average of
50000 DNA lesions per cell can be estimated to be pro-
duced on a daily basis. Detailed understanding of the natu-
ral defense mechanisms against DNA damage that is con-
stantly being produced through exposure to, for example,
sunlight, is necessary and hampered by the fact that among
the above-cited repair processes, the mechanism of DNA
cross-link repair still is considered to be a black box in our
knowledge.[13] Indeed, the DNA damage induced by cross-
linking the strands represents a formidable challenge to cell
survival. As opposing strands are involved in the damage,
these lesions are also very complex to process.[14] Human
diseases such as Fanconi Anemia and Nijmegen breakage
syndrome are characterized by a high sensitivity to DNA in-
terstrand cross-links and predisposition to malignance and
the basis of the basic defect in ICL repair remains un-
clear.[15] Next to the importance of a complete understand-
[a] Dr. K. Stevens, S. Figaroli, Prof. Dr. A. Madder
Laboratory for Organic and Biomimetic Chemistry
Department of Organic Chemistry, Ghent University
Krijgslaan 281, S4, 9000 Gent (Belgium)
Fax : (+32)9-264-49-98
E-mail: Annemieke.Madder@UGent.be
[b] Dr. D. D. Claeys, Dr. S. Catak, Prof. Dr. V. V. Speybroeck
Center for Molecular Modeling, Ghent University
Technologiepark 903, 9052 Zwijnaarde (Belgium)
[c] Prof. Dr. M. Hocek
Institute of Organic Chemistry and Biochemistry
Academy of Sciences of the Czech Republic
Gilead Sciences & IOCB Research Center
Flemingovo nam. 2, 16610 Prague 6 (Czech Republic)
[d] Dr. J. M. Tromp, Prof. Dr. S. Schꢁrch
Department of Chemistry and Biochemistry
University of Bern, Freiestrasse 3, 3012 Bern (Switzerland)
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201100067.
Chem. Eur. J. 2011, 17, 6940 – 6953
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6941

A. Madder et al.
ing of these mechanisms for the comprehension of natural
repair processes, this knowledge is essential in the fight
against resistance occurring in cancer treatments. Indeed,
cross-link repair is known to be one of the major processes
lying at the origin of resistance to anticancer chemothera-
peutics and is still poorly understood.
derstanding of the cytotoxic and mutagenic potential of spe-
cific ICLs. We therefore found it of importance to expand
our previous studies towards furan-containing building
blocks with a different chemical structure. Previous experi-
ments were carried out using a building block connecting
the furan moiety through an amido or ureido linker to the
2’-position of the sugar backbone.[11] In further studies we
focussed our attention on an acyclic building block with the
furan serving as base substitute.[12] Though significant influ-
ence on the modified duplex stability was observed, cross-
link yields remained high. In view of this rather strong de-
stabilization observed when incorporating the acyclic cross-
linking building block into a DNA sequence, we now report
on two new building blocks featuring an extended aromatic
system for enhanced stacking interactions and considered
both an acyclic and a cyclic sugar backbone. We here de-
scribe in detail a comparative study in which the benefits of
an extended aromatic system were evaluated against the in-
clusion of a building block with restored cyclic sugar moiety.
To fully understand the relation between the structure of
the furan building blocks and the DNA ICL, in addition to
the extensive experimental characterization of the formed
species, various computational molecular modeling tech-
niques have been applied.
The interest of ICL oligonucleotides lies in the possibility
to generate substrates for a detailed study of cross-link
repair in order to elucidate the mechanistic background for
this phenomenon. Indeed, isolation of defined ICL adducts
from natural sources in sufficient quantities for mechanistic
studies remains a major problem[16] and treatment of DNA
with external cross-linking reagents[17–22] typically yields
complex mixtures of mono-adducts, and inter- and intra-
strand cross-links with the desired ICLs are usually repre-
sented as minor fractions.[23] A variety of applications in the
chemical biology area further justify the ongoing and in-
tense research for efficient and highly sequence-specific
cross-linked oligonucleotides. Next to increased antisense
activity through irreversible binding to the target for specific
knock down of particular mRNAs or miRNAs,[24] the inter-
est of synthesis methods for the preparation of selectively
cross-linked, small, double-stranded DNA sequences can
further be found in the sequence-specific regulation of tran-
scription factors by so-called decoy ODNs. ICLs have been
shown to enhance decoy stability[25,26] and have the possibili-
ty to constrain the decoys into a more active conformation
towards the target transcription factor.[27] Finally cross-link-
ing of DNA can serve various purposes in structural biology
and nanotechnology. Cross-linked DNA with an altered con-
formational behavior offers opportunities for understanding
structure-specific DNA recognition in biological systems
and in the development of nucleic acid based therapeu-
tics.[28] A wealth of exotic and unfavorable DNA conforma-
tions has been restrained by cross-linking strategic parts. In
addition to R-shaped DNA, n-, h-, and H-type DNA have
been constructed in this way. Furthermore cross-linking hair-
pin termini can be used to continuously constrain the struc-
ture in the hairpin conformation.[29–31]
Bearing the problems associated with alkylating or reac-
tive oligonucleotides and their instability in physiological
media in mind, our recently developed cross-linking meth-
odology is based on the use of a furan moiety as a stable
precursor for reactive aldehyde functionality.[11] Relying on
the inducible reactivity principle,[32] the furan moiety is oxi-
dized selectively, after duplex formation, generating the
cross-linking moiety in an environment in which the target
nucleophiles are forced into proximity so as to minimize
side reactions with non-target nucleophiles. Others have re-
cently picked up this methodology and based on our furan-
oxidation strategy have shown that such type of furan-con-
taining probes can be used for the diagnosis of adenine-re-
lated DNA mutations.[33]
Results and Discussion
Furan-oxidation exploited in a biological context—selective
DNA interstrand cross-linking: Recently, we have described
the application of the furan-oxidation principle in a chemi-
cal biology context rather than the above-mentioned total
synthesis. Oxidation of oligonucleotides incorporating an
acyclic, furan-modified, building block results in the forma-
tion of an enal species that shows high reactivity towards
cross-linking with the opposing base in the complementary
strand (see Scheme 1 for illustration of the furan-oxidation-
triggered cross-linking principle).[12]
Proximate nucleophilic functionality on the complementa-
ry DNA effectively quenches the reactive intermediate oxo–
enal–oligonucleotide conjugate that is formed. Cross-linked
duplexes can be isolated in high yields (up to 70% of isolat-
ed and purified cross-linked duplex) and were shown to be
stable over extended periods of time.
It has been recently illustrated that the exact chemical
structure of an ICL can play a significant role in the repair
processing of the cross-link.[34] Understanding how structur-
ally different ICLs are repaired may provide a deeper un-
Scheme 1. Furan-oxidation triggered cross-linking of complementary
oligodeoxyribonucleotides.
6942
www.chemeurj.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 6940 – 6953

Bioorganic Chemistry
FULL PAPER
Design considerations for improved building blocks: For the
incorporation of a furan unit into DNA duplex sequences,
building block 1 (shown here) has been used in previous
Scheme 2. Synthesis of phosphoramidite 7 required for incorporation of
building block 2. Reagents and conditions: a) KOH, toluene, Dean–
Stark, 93% ; b) i) furan (10 equiv), nBuLi (9.5 equiv), ZnCl2 (9.5 equiv),
À78 to 08C; ii) [Pd
08C, 97%; d) DMTCl (0.9 equiv), pyridine, 08C to RT, 73%; e)
(iPr2N)2(NCCH2CH2O)PCl, diisopropylammonium tetrazolide
(1.5 equiv), CH2Cl2, 08C to RT, 79%
ACHTUNGTRNEN(UNG PPh3)4] (0.03 equiv), 508C, 70%; c) HCl (4m), THF,
studies.[12] Interstrand cross-links are formed upon N-bromo-
succinimide (NBS) oxidation of the incorporated furan
moiety and it was observed that only the base which is com-
plementary to the introduced modified nucleoside is in-
volved in the covalent-bond formation. Strong selectivity for
cross-linking to either a complementary adenine base or a
complementary cytosine base has been demonstrated. The
formed covalent interstrand link is stable and cross-linked
duplexes can be stored for prolonged periods of time with-
out degradation. However, duplexes incorporating modified
building block 1 were between 17 and 288C less stable than
fications, use of the standard b-cyanoethyl chlorophosphora-
midite does not result in formation of the phosphoramidite.
The less reactive b-cyanoethyl N,N,N’,N’-tetraisopropylphos-
phoramidite in the presence of diisopropylammonium tetra-
zolide as an activating species is used because of the mild-
ness and chemoselectivity of this phosphitylation reaction.
Building block 7 was thus obtained ready for incorporation
into the desired oligonucleotide sequences.
The synthetic route for building block 3 was based on the
modular approach[35–40] consisting in synthesis of halogenat-
ed aryl-C-nucleoside intermediate and its follow-up cross-
coupling (Scheme 3). Protected bromophenyl-C-nucleoside
9 was prepared by a previously reported procedure[41,42]
from halogenose 8 and 4-bromophenylmagnesium bromide.
The Stille cross-coupling of 9 with 2-Bu3Sn-furan gave the
protected biaryl nucleoside 10 in excellent 88% yield. Stan-
dard deprotection under the Zemplen conditions gave the
the corresponding unmodified sequences. For antiACTHNUGRTNEUNGsense ap-
plications, this observed destabilization represents an impor-
tant disadvantage. Moreover, in studies of the enzymatic
mechanisms involved in cross-link repair, stability and struc-
ture of the duplex can play an important role for recognition
of the synthetic substrate by the studied enzymes.
In an attempt to reduce destabilization of the duplex
upon introduction of the furan modified residue, building
blocks 2 and 3 were synthesized. We expect building block 2
to show enhanced stacking interactions and therefore influ-
ence duplex stability less, while reconstitution of the cyclic
sugar moiety in 3 is expected to induce additional restriction
of backbone flexibility and consequently cause less duplex
destabilization.
Synthesis of new furan modified building blocks: Synthesis
of building block 2 is outlined in Scheme 2. The first step in-
volves SN2 substitution of the 4-bromobenzyl bromide with
commercially
available
(2,2-dimethyl-1,3-dioxolan-4-
yl)methanol ((S)-solketal). Negishi cross-coupling of the ar-
ACHTUNGTRENNUNG
omatic bromide with furylzinc chloride completes the skele-
ton (2). Subsequent protection of the primary alcohol with
diACHTUNGTRENNUNGmethoxytrityl chloride should be carried out with care, as
only a small difference in reactivity between the primary
and secondary alcohol was observed. This results in forma-
tion of doubly dimethoxytritylated product, even when not
all starting material is consumed. The highest selectivity
could be obtained with a slight excess of the diol at low tem-
perature. For the synthesis of phosphoramidite 7, it was
shown that in the case of similar glycerol nucleic acid modi-
Scheme 3. Synthesis of phosphoramidite 12 required for incorporation of
building block 3. Reagents and conditions: a) Ref. [35]; b) 2-Bu3Sn-furan,
[PdCl
DMTCl
(iPr2N)(NCCH2CH2O)PCl, N,N-diisopropylethylamine, CH2Cl2, RT, 59%
(dppf)], DMF, 1008C, 16 h, 88%; c) NaOMe, MeOH, RT, 85%; d)
(1.5 equiv), pyridine, 08C to RT, 66%; e)
Chem. Eur. J. 2011, 17, 6940 – 6953
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6943

A. Madder et al.
desired furylphenyl C-nucleoside 3. Transformation into the
protected phosphoramidite 12 was in this case achieved with
the standard b-cyanoethyl chlorophosphoramidite and fur-
nished the correct building block for automated DNA syn-
thesis.
Oligonucleotide synthesis and purification: To allow de-
tailed evaluation of cross-link selectivity and influence of
the sequence context, building blocks 2 and 3 were incorpo-
rated into a series of four different sequences varying the
neighboring bases (see Table 1, below). Incorporation of 2
into oligonucleotides was carried out by automated DNA
synthesis in accordance with methods previously de-
scribed.[12] 4,5-Dicyanoimidazole in acetonitrile was added
as activator.[43] Coupling of the modified nucleoside was per-
formed manually with a prolonged coupling time of 10 min.
The coupling yields under these conditions were estimated
to be more than 99%. DMT-groups were left on the termi-
nal 5’-position in order to allow purification by Sep-pak fil-
tration. As for sequences incorporating modified building
block 3, again manual couplings and 4,5-dicyanoimidazole
activation were applied. However, coupling yields using 12
were much lower as compared to the acyclic building block.
In this case, the estimated coupling yield resided around
80% based on trityl group release. Further extension of the
chain was not hampered and synthesis could be continued
with standard good coupling yields. Analysis data for all syn-
thesized oligonucleotides can be found in Supporting Infor-
mation.
Figure 1. Comparative graph of melting temperatures for duplexes incor-
porating building blocks 1, 2, and 3. (Example shown for 5’-CTGAC-
GAXATGC-3’:3’-GACTGCTNTACG-5’, with X=1, 2, 3 or the comple-
ment to N (from left to right in each block) and N=A, C, T or G).
porating 3 were, in most cases, even somewhat lower in sta-
bility. Whereas building block 2 benefits from enhanced
stacking interactions resulting from the increased aromatic
surface (when compared to 1), restriction of conformational
flexibility and restoration of the backbone in building block
3 does not result in an increased duplex stability.
To obtain a more profound understanding of these rela-
tions between the structure of the various building blocks
and the overall stability, molecular dynamic simulations with
implicit solvent were performed in AMBER 9.[45,46] The
computational details are given in the Supporting Informa-
tion. Modeling of the ON1 and ON5 series shows a very
similar behavior for both biaryl modifications, irrespective
of the complementary base. Incorporation of the phenyl–
furan nucleoside into the duplex opposite to a purine or pyr-
imidine nucleobase causes local disruption of the stacking.
Most of the time, the phenyl–furan moiety is forced into the
major groove, but has an inner helical position that allows
the modification to stack with
Thermal analysis of modified duplexes: To allow detailed
evaluation of cross-link selectivity each of the oligonucleo-
tides was paired with four different complements varying
the opposing base. Table 1 summarizes melting point results
for duplexes incorporating the modified building blocks.
its 5’ base. The modified du-
Table 1. Melting temperatures [in 8C] of duplexes modified with nucleoside building blocks 2 and 3 (DTm
values between brackets refer to fully complementary reference duplexes without 2 or 3). Recorded at
260 nm, 2 mm duplex concentration, 10 mm phosphate buffer pH 7 and 100 mm NaCl.
plexes rarely form interstrand
stacked zip-like arrangements.
The modification is too large to
be accommodated in one plane
with its complement and, appa-
rently, to form an intercalating
structure, the aromatic part of
the modification is too small to
Modified strand\
A
C
T
G
Complementary base
Tm (DTm)
Tm (DTm)
Tm (DTm)
Tm (DTm)
ON1: 5’-CTG ACG G2G TGC-3’
ON2: 5’-CTG ACG C2C TGC-3’
ON3: 5’-CTG ACG A2A TGC-3’
ON4: 5’-CTG ACG T2T TGC-3’
ON5: 5’-CTG ACG G3G TGC-3’
ON6: 5’-CTG ACG C3C TGC-3’
ON7: 5’-CTG ACG A3A TGC-3’
ON8: 5’-CTG ACG T3T TGC-3’
45.2 (À11.5)
41.9 (À11.8)
36.9 (À11.2)
35.0 (À14.7)
44.4 (À12.3)
40.4 (À13.3)
36.7 (À11.4)
34.7 (À15.0)
48.9 (À10.4)
41.3 (À20.3)
38.3 (À13.7)
35.6 (À17.9)
47.1 (À12.1)
39.9 (À21.7)
35.4 (À16.6)
32.6 (À20.9)
45.1 (À8.0)
47.6 (À8.7)
38.4 (À11.3)
34.5 (À12.7)
40.1 (À13.0)
41.4 (À14.9)
34.0 (À15.7)
33.8 (À13.4)
45.6 (À16.8)
46.2 (À12.5)
37.9 (À15.5)
36.4 (À15.3)
45.3 (À17.1)
43.4 (À15.3)
37.8 (À15.6)
37.2 (À14.5)
provide
a stable interstrand
stacked pair.
This is illustrated in Figure 2
which shows two representative
snapshots from the molecular
Graphical representation of this melting point analysis il-
lustrates more clearly that with respect to the previous gen-
eration of duplexes incorporating building block 1, stabiliza-
tion can be observed when incorporating an extra phenyl
unit as in building block 2 (Figure 1).[44]
dynamics simulations performed on D2a and D3a (see
Table 2) with a complementary adenine base. Similar results
have been reported in literature for a biphenyl modification
flanked at both sides by a C·G base pair[47] and have been
deduced from an experimental fluorescence study for the bi-
phenyl modification flanked by an A·T and a C·G base
pair.[48]
As illustrated, the extra stabilization expected for a re-
stored entirely cyclic backbone was absent. Duplexes incor-
6944
www.chemeurj.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 6940 – 6953

Bioorganic Chemistry
FULL PAPER
Interstrand cross-link formation and selectivity: We next in-
vestigated the cross-link performance and selectivity within
this new series of oligonucleotides incorporating the new
furan-modified building blocks (see Table 2). Cross-link ex-
periments were carried out as previously described[12] by
mixing equimolar amounts of the furan-modified and com-
plementary non-modified oligonucleotides in a 1:1 ratio, fol-
lowed by controlled addition of NBS. In contrast to the re-
cently developed methodology of Schꢂrer,[23] but similar to
cross-linking occurring at abasic sites,[49] no extra reductive
amination step is necessary to ensure stable cross-link for-
mation. The mere NBS treatment of the mixture of modi-
fied and complementary oligonucleotide in the current
study results in the formation of cross-linked products that
are stable and can be analyzed by using both HPLC and
MALDI-TOF methods.
Figure 2. Representative samples of the D2a (core G7 2G9:C18 A17 C16)
and D3a (core G7 3G9:C18 A17 C16) simulations show the modification (in
green) that rarely forms an interstrand stacked pair, but resides in the
major groove of the locally distorted double helix. The complementary
adenine base (in orange) shows p-stacking with its 5’ base.
Cross-linking occurs instantaneously after NBS addition,
and upon analysis of the reaction mixture two new species
can be observed in the reverse-phase HPLC chromatogram.
Figure 4 shows reverse-phase HPLC analysis data for the
specific case of duplex D2a cross-linking, before (Fig-
Simulations of D2i and D3i show that duplex D3i with a
restored cyclic backbone has more often a zip-like confor-
mation, while modification 2 rarely forms an intercalated
structure. An example of this intercalating conformation
with the cyclic modification 3 is given in Figure 3.
Also the duplexes D3 f and
D3e have more often an inter-
calated zip-like conformation,
which correlates with the
number of purine bases at posi-
tions 16, 17, and 18, as these
have a more extended aromatic
system. Again, this is not ob-
served for modification 2. It is
clear that an optimal intercalat-
ing alignment of the extended
aromatic phenyl–furan modifi-
cation in building blocks 2 and
3 is hampered. Only a p-stack-
ing interaction, mainly with the
Watson 5’ base can be formed.
If modification 3 is flanked by
base pairs that form only two
hydrogen bridges, zip-like inter-
Figure 3. Example of an inter-
calating zip-like structure of
modification 3 (in green) be-
calating structures can be
formed at the expense of
duplex stabilization by hydro-
gen bonds and probably also p-
stacking (see Table S3 in Sup-
porting Information for details).
This explains the lack of en-
tween its complementary base
(in orange) and the Crick 3’
base taken from the simulation
of
D3i
(core
A7 3A9:T18 A17 T16).
Figure 4. Reverse-phase HPLC chromatogram for cross-linking of duplex
D2a (ON1+ON9, core G7 2G9:C18 A17 C16) and D3a (ON5+ON9, core
G7 3G9:C18 A17 C16) before addition of NBS (chromatogram A and C re-
spectively) and after addition of four equivalents of NBS (B and D re-
spectively). All ON sequences are mixed in a 1:1 ratio. D2aICL and
D3aICL designate the formed cross-linked species.
hancement of the duplex stability by restoration of the
cyclic backbone in modification 3.
Chem. Eur. J. 2011, 17, 6940 – 6953
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6945

A. Madder et al.
Table 2. Sequences of modified duplexes incorporating building blocks 2 and 3. Modified nucleosides are des-
ignated by X, variable nucleosides are designated by N. Specific identity of the core sequences is specified for
each duplex incorporating 2 (left column) and 3 (right column).
the remaining unmodified start-
ing strand, a cross-linked prod-
uct was characterized and again
a small percentage of the bro-
minated modified strand ON5
was detected.
To get an overview of cross-
link selectivity, the systematic
variation of flanking and oppos-
ing bases resulted in cross-link-
ing experiments with a series of
16 duplexes each for building
blocks 2 and 3. Gel electropho-
resis analysis gives a quick and
clear idea of the overall selec-
tivity of the cross-link reaction
in terms of the influence of the
identity of the opposing and
neigboring bases on the cross-
link outcome (Figure 5). In the
case of building block 2, cross-
link formation occurs to com-
plementary C or A bases, while
in case of complementary G or
T, no major cross-linked species
are formed. This behavior is
analogous to the cross-link per-
formance of the first-generation
building block 1.
Duplex N7
2
N9 Duplex N7
2
N9
Duplex N7
3
N9 Duplex N7
3
N9
N18 N17 N16
N18 N17 N16
N18 N17 N16
N18 N17 N16
D2a
ON1
ON9
D2i
G9 ON3
D3a
ON5
ON9
D3i
G9 ON7
G7
2
A7
2
A9
G7
3
A7
3
A9
C18 A17 C16 ON17
T18 A17 T16
C18 A17 C16 ON17
T18 A17 T16
D2b
D2j
D3b
D3j
ON1
ON10
G7
2
G9 ON3
A7
2
A9
ON5
ON10
G7
3
G9 ON7
A7
3
A9
C18 C17 C16 ON18
T18 C17 T16
C18 C17 C16 ON18
T18 C17 T16
D2c
D2k
D3c
D3k
ON1
ON11
G7
2
G9 ON3
A7
2
A9
ON5
ON11
G7
3
G9 ON7
A7
3
A9
C18 T17 C16 ON19
T18 T17 T16
C18 T17 C16 ON19
T18 T17 T16
D2d
D2l
D3d
D3l
ON1
ON12
G7
2
G9 ON3
A7
2
A9
ON5
ON12
G7
3
G9 ON7
A7
3
A9
C18 G17 C16 ON20
T18 G17 T16
C18 G17 C16 ON20
T18 G17 T16
D2e
D2m
D3e
D3m
ON2
ON13
C7
2
C9
ON4
T7
2
T9
ON6
ON13
C7
3
C9
ON8
T7
3
T9
G18 A17 G16 ON21
A18 A17 A16
G18 A17 G16 ON21
A18 A17 A16
However, for building block
3, a much more pronounced C
selectivity can be observed.
Clear cross-linking to both A
and C is only observed (lane 2
and 4) when 3 is flanked by two
D2 f
D2n
D3 f
D3n
ON2
ON14
C7
2
C9
ON4
T7
2
T9
ON6
ON14
C7
3
C9
ON8
T7
3
T9
G18 C17 G16 ON22
A18 C17 A16
G18 C17 G16 ON22
A18 C17 A16
D2g
D2o
D3g
D3o
ON2
ON15
C7
2
C9
ON4
T7
2
T9
ON6
ON15
C7
3
C9
ON8
T7
3
T9
G
residues (duplexes D3a–
G18 T17 G16 ON23
A18 T17 A16
G18 T17 G16 ON23
A18 T17 A16
D3d); in the other flanking
contexts, only in case of a C
base opposite 3 can a clear
cross-linked species be visual-
ized (see lanes 9, 14, and 19 for
a clear spot related to ICL
D2h
ON2
ON16
D2p
ON4
D3h
ON6
ON16
D3p
ON8
C7
2
C9
T7
2
T9
C7
3
C9
T7
3
T9
G18 G17 G16 ON24
A18 G17 A16
G18 G17 G16 ON24
A18 G17 A16
ure 4A) and after (Figure 4B) complete oxidation of modi-
fied ON1. In addition to the characterization of remaining
unmodified strand (eluting first in both cases), MALDI-
TOF MS analysis confirmed the cross-linked nature of the
compound eluting at 15.175 min (see Supporting Informa-
tion for detailed MS analysis of all cross-link reaction mix-
tures). As described earlier, furan-based cross-linking within
a duplex can give rise to two different cross-linked species
that co-elute. In addition to a mass corresponding to the
mere sum of ON1oxidized +ON9, a second mass corresponding
to a dehydration product of this cross-linked compound can
be deduced. The last compound eluting at 16.133 min was
shown to correspond to a brominated derivative of ON1. A
similar behavior during cross-link reactions was observed in
case of cyclic building block 3 modified duplexes. Besides
product; lanes 7, 12 and 17 only show a minor trace of
cross-linked product).
This more pronounced C preference over A was also in-
vestigated by molecular dynamic simulations of D2m, D2n,
D3m, and D3n. Although a furan moiety was used instead
of the reacting 4-oxobutenal, these simulations give a first
indication to explain the selectivity, as stacking occurs very
often through the phenyl ring of the modification. With T·A
neighboring base pairs, the hydrogen bridges are more
easily broken to accommodate the modification in a zip-like
structure. During the simulation of modification 3 and a
complementary A base, the phenyl–furan moiety had a zip-
like intercalating conformation (Figure 6A) or showed p-
stacking with its 5’ base (Figure 6B). This more defined po-
sition of the furan ring could hamper correct positioning of
6946
www.chemeurj.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 6940 – 6953

Bioorganic Chemistry
FULL PAPER
occur if the phenyl–furan
moiety adopts a non-stacking
inner helical position or when it
stacks with its 5’ base, as illus-
trated in Figure 6D. Another
explanation could be that the
NH2 group on the pyrimidine
base has a better cross-linking
position than on the purine
base.
Reverse-phase HPLC analy-
sis of all cross-linking reactions
revealed the same trends in se-
lectivity for complementary
bases. Both cross-linking to A
and C is consistently observed
for duplexes incorporating 2.
Similar to observations at the
gel electrophoresis stage, for
duplexes incorporating 3,
a
higher selectivity is con-
C
firmed and only in case of
duplex D3a (G7 3G9:C18 A17 C16
Figure 5. Denaturing PAGE results of the cross-link reaction with 2 and 3. Furan-containing ON1–8 were an-
nealed to the different complements in a 1:1 ratio to obtain duplexes D2a–D2d (A, lanes 2–5), D2e–D2h (A,
lanes 7–10), D2i–D2l (A, lanes 12–15), D2m–D2p (A, lanes 17–20), and D3a–D3d (B, lanes 2–5), D3e–D3h
(B, lanes 7–10), D3i–D3l (B, lanes 12–15) and D3m–D3p (B, lanes 17–20), after which NBS was added in por-
tions. ICL formation was analyzed by 20% denaturing PAGE. Lanes 1, 6, 11, and 16 refer to single-strand ref-
erences.
sequence context),
a
small
amount of duplex cross-linked
to A can be observed and char-
acterized. All cross-linked spe-
cies were isolated and analyzed
by UV–denaturation experi-
ments to determine their stabil-
ity. It was shown that all cross-
linked duplexes are stable when
stored over longer time periods
and display a melting point of
around 70–808C (see Support-
ing Information for detailed
data and corresponding curves).
Isolated yields of purified cross-
linked duplexes range from 10–
30% (see Supporting Informa-
tion for details). Considering
that only one equivalent of the
opposing strands is used, the
isolated yields of specifically
Figure 6. Snapshots of the simulation of D3m and D2m showing the more rigid position of modification 3 and
the more flexible modification 2. Complementary bases are represented in orange, furan-modified building
blocks in yellow: A) duplex D3m (core T7 3T9:A18 A17 A16) showing intercalated 3 and distant from NH2 of
A17; B) duplex D3m (core T7 3T9:A18 A17 A16) showing 3 stacked with its 5’ neighbor and distant from NH2 of
A17; C) duplex D2m (core T7 2T9:A18 A17 A16) showing 2 residing in the major groove in good position for
attack by NH2 of A17; D) duplex D3n (core T7 3T9:A18 C17 A16) showing 3 stacked with its 5’ neighbor and in cross-linked duplex are still im-
good position for attack by NH2 of C17.
pressive when comparing to
previously published proce-
dures.
Moreover
increased
the complement for attack to the generated oxo–enal func-
tionality and inhibit cross-linking to A, whereas the more
flexible modified building block 2 allows cross-link forma-
tion (Figure 6C). The smaller pyrimidine C complement
shows less p-stacking interaction with the phenyl–furan
moiety. For about 50% of the simulation time, modification
3 stacks with its 5’ base or forms zip-like structures; during
the other half it resides in the major groove. This less-de-
fined position allows cross-link formation. Cross-linking can
equivalents of either modified or target strand should easily
allow the isolation of larger amounts of specifically cross-
linked species for ICL repair studies. It should be noted that
these yields are lower than our previously reported yields
for cross-link formation using 1. This can clearly be attribut-
ed to the formation of the brominated side product. As no
brominated species were observed by using cross-link
moiety 1 and all brominated species were observed as single
strands, apparently the bromination of the modified oligonu-
Chem. Eur. J. 2011, 17, 6940 – 6953
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6947

A. Madder et al.
cleotides excludes further cross-link formation. We therefore
embarked on the detailed characterization of the brominat-
ed species.
ing from the 5’-side of the modified residue are observed,
and identical observations are made for the w ions originat-
ing from the 3’-side of the modification. Consequently, the
bromine atom must reside on the modified residue exclu-
sively. Incorporation of the modified building block into the
single strand promotes backbone cleavage at alternative po-
sitions, resulting in additional fragment ions that provide
more information about the bromination site. In the product
ion spectrum of the brominated 7-mer single chain, the pres-
Characterization of brominated side products: For the pur-
pose of facilitating identification of the brominated species,
cross-link experiments were performed on a shorter model
D2q (shown in Figure 7, left) and the brominated single-
stranded species (Figure 7, right) was isolated and analyzed.
2À
ence of bromine-containing a42À, b42À, y42À, and z4 ions due
to backbone cleavage adjacent to the modified residue is ad-
ditional unambiguous evidence for the bromination site
(Figure 7).
The most susceptible site for furan bromination in the
modified residue is C2 (Scheme 4). Initial attack of the elec-
trophilic bromine generates an oxonium species, which can
Scheme 4. Proposed mechanism for bromination studied on minimized
structures I and II with methyl and 4-methylphenyl substituted furan.
be attacked by an H2O molecule and further oxidized by ex-
pelling the bromine (pathway a).[51] If water attack on C5 is
disfavored, a side reaction that was previously proposed by
Kishihara et al.[52] may occur; that is, abstraction of the C2
proton by a neighboring base. This will generate the bromi-
nated furan moiety, which remains insensitive to further oxi-
dative ring-opening, thus interfering with the cross-linking
process (pathway b). As brominated species were observed
in the case of building blocks 2 and 3, but not for 1, theoreti-
cal calculations were performed as to elucidate the role of
the phenyl moiety in the reaction with NBS. As this part of
the simulations involves the localization of reactive path-
ways, density functional theory based calculations were per-
formed as explained in detail in the Supporting Information.
Bromination could be attributed to the increased stability of
the phenyl conjugated furan oxonium species, for which
water attack on C5 is disfavored, since it will disrupt extend-
ed conjugation. An increased barrier for furan oxidation
would then allow the side reaction a competitive chance and
result in brominated side products.
To rationalize the formation of the brominated side prod-
uct, the difference in ease of furan oxidation has been inves-
tigated by density functional theory for structures I and II,
in which methyl and 4-methylphenyl side chains, respective-
ly, were used to model the rate-determining step.[51] These
theoretical calculations have further validated the difference
in ease of water attack on the oxonium species in the non-
conjugated (I) versus phenyl-conjugated systems (II), show-
ing significantly higher barriers for the latter. Details of the
Figure 7. Indicative fragment ions for determination of the bromination
site within the 7-mer single strand ON25Br (2195.3 Da). The [MÀ4H]4À
ion (m/z 547.8) was selected as the precursor for collision-induced disso-
ciation on a hybrid quadrupole time-of-flight mass spectrometer (Applied
Biosystems QStar Pulsar). Fragment ions, which do not incorporate the
modified building block (w1À, w2À, w3À, [a1-B]À, [a2-B]À, and [a3-B]2À
)
give no evidence for the presence of any bromine atom. Fragment ions
incorporating the modified residue (a42À, b42À, w42À, y42À, and z42À) all
show the bromine-typical isotopic pattern and mass shift.
Localization of the bromination site was performed by
nanoelectrospray tandem mass spectrometry on a hybrid
quadrupole-time-of-flight instrument. Multiply deprotonated
single strands were selected as precursor ions for subsequent
collision-induced dissociation. The corresponding product
ion spectra provide complete sequence information of the
single strands due to the presence of abundant [a-B]- and w-
ion series, which are generated by cleavage of the oligonu-
À
cleotide backbone at the 3’-C O bonds. Such dissociation
behavior is in agreement with the generally accepted gas-
phase dissociation mechanism of oligodeoxynucleotides.[50]
Bromine-containing fragment ions are easily recognized
by their characteristic isotopic pattern and the mass shift
caused by the halogen substituent. For all brominated single
strands analyzed, no bromine-containing [a-B] ions originat-
6948
www.chemeurj.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 6940 – 6953

Bioorganic Chemistry
FULL PAPER
computational methodology can be found in Supporting In-
formation. More specifically, energy profiles and transition-
state geometries for water attack on the furan oxonium are
depicted in Figure 8. TS-Me and TS-Tol are both “product-
accomplished, we now turned our attention to the actual
species of interest, the cross-linked duplexes. Enzymatic
degradation studies using exonuclease III (EXO III) on iso-
lated cross-linked species were employed to allow the locali-
zation of the exact position of the cross-linking bond. The
observed cleavage pattern indicating inhibition of the
EXO III by the presence of the cross-link is consistent with
earlier results showing that EXO III activity is inhibited by
ethenoadenine up to and beyond the position of its occur-
rence in ds DNA.[53]
A typical reverse-phase HPLC outcome of cross-linked
duplex digestion by EXO III is shown in Figure 9 for the
specific case of duplex D2a. As illustrated by this analysis,
digestion occurs up to two bases before the position of the
modified building block in the modified strand. Digestion is
further hampered in the non-modified complementary
strand, as in addition to fragments containing most of the
modified strand covalently connected to an A residue (Fig-
ure 9a), cross-linked fragments containing a 5’-tail can also
be isolated (Figure 9b).
Figure 9. Reverse-phase HPLC chromatogram after EXO III digestion of
duplex D2a d(CTGACGG2GTGC:GACTGCCACACG). Enzymatic
degradation was performed using 200 units of EXOIII for purified cross-
linked duplex (0.6 nmol). Digestions were carried out at 378C for 50 min.
Collected fractions were analyzed by MALDI-TOF MS.
Figure 8. Reaction profile (MP2/6-31+GACTHUNRTGENNUG(d,p)//B3LYP/6-31+GACHTUNGTRNE(NUGN d,p)) and
transition states for the hydrolysis of the intermediates arising from the
bromination of I (above) and II (below) with two water molecules.
like” and have similar critical distances; however, they
differ remarkably (approximately 40 kJmolÀ1) with respect
to their activation barriers. This is due to the extra stability
of the phenyl-conjugated furan; hence, the barrier to attack
is higher since conjugation is disturbed in the transition
It was further confirmed that in all cases, cross-linking
with 2 preferably occurs to the opposing base leaving the
neighboring bases in the complementary strand mainly un-
touched.[54]
state (TS-Tol). As
a
consequence, the side reaction
As for cross-link reactions through building block 3,
Figure 10 shows analysis data for the representative example
of duplex D3b. Incorporation of the cyclic furan-containing
building block 3 apparently causes less severe duplex distor-
tion as indicated by the complete absence of the tailed com-
pounds (cfr Figure 9), indicating a smoother recognition and
degradation of the duplex. Moreover EXO III can approach
the modified residue more closely as indicated by the frag-
ment represented in Figure 10b. The additional isolation of
the fragment shown in Figure 10c further shows that the nu-
(Scheme 4, pathway b), which restores aromaticity through
proton abstraction at C2 becomes competitive and bromi-
nated side products may form. An alternative mechanism
leading to brominated species via a bromohydrin intermedi-
ate is also possible (see Supporting Information for more
details).
Localization of the cross-linked site—enzymatic digestion
studies: With the characterization of the major side product
Chem. Eur. J. 2011, 17, 6940 – 6953
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6949

A. Madder et al.
Figure 11. Reverse-phase HPLC (top) and ESI-MS (bottom) data of puri-
fied dinucleoside 13.
Figure 10. Reverse-phase HPLC chromatogram after EXO III digestion
of duplex D3b d(CTGACGG3GTGC:GACTGCCCCACG). Enzymatic
degradation was performed using 200 units of EXOIII for purified cross-
linked duplex (0.6 nmol). Digestions were carried out at 378C for 50 min.
Collected fractions were analyzed by MALDI-TOF MS.
clease can degrade even beyond the modified residue. As
such, comparison of digestion fragments between D2 and
D3 duplexes can be interpreted in terms of a better recogni-
tion of duplexes containing cyclic building block 3 by the
natural enzyme.
Detailed structural characterization of the cross-linked spe-
cies: Although the described experiments yield information
about the exact location of the cross-link and the selectivity
of the specific furan-modified building block involved, no
direct structural information as to the exact chemical struc-
ture of the cross-linked species can be deducted. Similar to
our previous studies,[12] a solution experiment on the nucleo-
side level was carried out involving NBS oxidation of pro-
tected furan analogue 5 in the presence of deoxycytidine.
The obtained crude dinucleoside was purified by flash
chromatography and the resulting compound was analyzed
by reverse-phase HPLC/MS and NMR spectroscopy (see
Supporting Information for experimental procedures and
NMR data). Isolation of the product resulted in clean re-
verse-phase HPLC and MS spectra (Figure 11). The signal
at m/z 532 corresponds to the protonated [M+H]+ cross-
linked product 13 and m/z 416 corresponds to a compound
in which the glycosidic bond has been cleaved and in which
only the cytosine base is attached to the oxidized cross-link
residue.
Scheme 5. Structural characterization of the cross-linked species through
generation of a cross-linked dinucleoside.
dized abasic site b-elimination product isolated by Ravanat,
Cadet, and co-workers.[56]
Conclusion
The exact nature of the furan-containing building block
used in the current furan-oxidation cross-linking method has
a considerable influence on the yield and chemoselectivity
of the cross-link formation through the generated enal func-
tionality. A delicate interplay between electronic and steric
factors can be observed. Although the enlargement of the
aromatic system within the phenyl–furan building blocks has
a beneficial influence on duplex stability, the altered elec-
tronic nature of the furan moiety promotes a bromination
side reaction that negatively influences the yield. For a more
in depth comprehension of the experimental results theoret-
ical calculations were carried out. As the duplex structure is
important in the study of ICL repair studies, the presented
molecular modeling studies shed light on the structural con-
1
The obtained H NMR spectrum, indicating the presence
of a mixture of stereoisomers, was compared with literature
data[55] for the earlier characterized natural adduct of cis-bu-
tenedial and dC. The proposed pathway for cross-link for-
mation thus is again suggested to involve the initial attack
of the exocyclic amine on the aldehyde of the generated
enal functionality, followed by a rearrangement leading to
Michael addition on the remainder of the oxidized furan
moiety (Scheme 5). As such the isolated adduct proves to be
structurally similar to the dC adduct of the formal C4’-oxi-
6950
www.chemeurj.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 6940 – 6953

Bioorganic Chemistry
FULL PAPER
enex Luna C18 column (250ꢃ4.6 mm, 5 m at 358C). The used solvent
system was 5 mm NH4OAC in water (A) and MeCN (B). The used gradi-
ent went from 0 to 100% B in 15 min. Reverse-phase HPLC: Agilent
1100 system equipped with a Phenomenex Luna C18 column (250ꢃ
4.6 mm, 5 m, at 508C) with 0.1m TEAA (containing 8% MeOH) and
MeOH as mobile phase (linear gradient: 0–30% MeOH in 15 min, 30–
100% MeOH in 3 min).
sequences of introducing cyclic or acyclic building blocks.
Rather than duplex stability, the precise orientation of the
modified building block seems to play an important role in
achieving selective cross-linking to one specific complemen-
tary base, as the observed localization of the furan moiety
coincides well with the observed cross-link selectivity. Fur-
thermore, calculations have illustrated the role of the phenyl
moiety in the bromination side reaction. The structural ex-
ploration conducted to date, with a gradually increasing
building block complexity, allows us to define some of the
parameters important for achieving acceptable yields and
cross-link selectivity.
In terms of a homogeneous diagnosis of a cytosine-related
one-point variation within DNA, the observed C selectivity
represents a remarkable advantage of cross-linking by cyclic
furan derivative 3. Whereas all previous furan-decorated
moieties in general form covalent bonds with A or C com-
plements, cyclic derivative 3 exhibits a rather pronounced
selectivity for cross-linking to one specific complementary
base. The method described herein constitutes one of the
few methodologies for site specific ICL formation in the
major groove. Furthermore, as indicated by enzymatic diges-
tion studies, cross-linking through 3 results in a less distorted
duplex, which possibly is an important factor for recognition
of cross-linked model duplexes by natural enzymes.
It can be concluded from the current work that restricted
mobility, though causing duplex destabilization, can have a
beneficial influence on the selectivity of the cross-link reac-
tion. Although the phenyl–furan moiety utilized here does
give rise to an undesired side reaction, a clear stabilizing in-
fluence on the duplex can be noted when comparing to the
acyclic furan-based system 1. Duplexes incorporating either
the building block 2 or 3 have been shown to be more
stable, in the range 5–158C, than the ones incorporating the
acyclic furan-containing counterpart.
In conclusion, the cyclic phenyl–furan-containing building
block shows a somewhat lower but still very appreciable iso-
lated yield of cross-linked duplex; however, its selectivity to-
wards cross-linking with C is considerably higher than in the
previous series. Given the modular nature of our building
block synthesis and the commercial availability of various
furan derivatives, the current study demonstrates that the
furan-oxidation cross-link strategy allows fine-tuning of se-
lectivity by the precise choice and conformational behavior
of the modified furan-containing moiety.
Synthesis of nucleoside building blocks
(S)-1-[Bis(4-methoxyphenyl)phenylmethoxy]-3-(4-furan-2-ylbenzyloxy)-
propan-2-ol (6): See Supporting Information for detailed procedures,
analysis data, and NMR spectra:
Diisopropylphosphoramidous acid 2-[bis(4-methoxyphenyl)methoxy]-1-
(4-furan-2-ylbenzyloxymethyl)ethyl ester 2-cyanoethyl ester (7): Com-
pound 6 (100 mg, 0.18 mmol) was dissolved in fresh distilled CH2Cl2
(6 mL) under argon atmosphere. N,N-Diisopropylammoniumtetrazolide
(46 mg, 0.27 mmol, 1.5 equiv) was added and the resulting mixture was
stirred at room temperature. 2-Cyanoethyltetraisopropylphosphoramidite
(60 mg, 0.2 mmol, 1.1 equiv) was added, while cooling on an ice-bath and
the reaction mixture was allowed to stir overnight at room temperature.
When no further conversion occurred, the reaction mixture was
quenched with MeOH (4 mL). The solvent was partially evaporated and
water was added. The product was extracted with CH2Cl2, dried on anhy-
drous Na2SO4, and the solvent was removed under reduced pressure. The
residue was purified using Merck silica gel (2:8 EtOAc/petroleum ether
with 1% Et3N) to yield 7 (106 mg, 79%) (Rf =0.32 (2:7 EtOAc/petrole-
um ether)). This product was used in the oligonucleotide synthesis. IR
(KBr-film): n˜ =2928, 1608, 1508, 1302, 1176, 829 cmÀ1; UV: lmax =285 nm
(H2O/ACN); LRMS (ESI-MS): m/z calcd for C44H51N2O7P: 750.3, found:
773.7 [M+Na]+; 31P NMR (121 MHz, CDCl3): d=149.44, 149.62 ppm.
2-[Bis-(4-methoxyphenyl)phenylmethoxymethyl]-5-(4-furan-2-ylphenyl)-
tetrahydrofuran-3-ol (11): See Supporting Information for detailed proce-
dures, analysis data and NMR spectra.
Diisopropylphosphoramidous acid 2-[bis-(4-methoxyphenyl)phenylmeth-
AHCTUNGTRENNUNG
(iPr2N)(NCCH2CH2O)PCl (95 mg, 0.4 mmol, 2.5 equiv) was added in one
portion and the resulting mixture was stirred at room temperature. After
2 h the reaction was quenched by addition of MeOH (2 mL), stirred for
another 15 min, followed by addition of saturated NaHCO3 solution
(3 mL). This mixture was extracted three times with CH2Cl2. The organic
layer was dried over Na2SO4 and evaporated under reduced pressure.
The reaction mixture was purified by column chromatography with 4:1
isooctane:EtOAc (1%TEA). The product was collected as a yellow oil
(70 mg, 0.091 mmol, 59%). Rf =0.30 (7:2 petroleum ether/EtOAc).
31P NMR (121 MHz, CDCl3): d=147.98, 148.16 ppm.
Materials and methods for oligonucleotides: Reagents for DNA synthesis
were obtained from Glen Research. All reverse-phase HPLC experi-
ments with oligonucleotides were recorded on an Agilent 1100 system
equipped with a Phenomenex Clarity column (250ꢃ4.6 mm, 5 m) at 508C
with 0.1m TEAA (with 5% MeCN) and MeCN as mobile phase (linear
gradient: 0–30% MeCN in 15 min). When stated the same conditions
were used with a Jupiter 300 ꢄ column (250ꢃ4.6 mm, 5 mm). The chro-
matograms were analyzed at 260 nm. Enzymatic digestion reaction mix-
tures were analyzed on a Phenomenex Luna C18 column (250ꢃ4.6 mm,
5 m) using the same method. HPLC-purified samples were subjected to
MALDI-TOF analysis. MALDI-TOF spectra were recorded in positive
mode on an Applied Biosystems voyager DE-STR biospectrometry
workstation. The matrix was a mixture of 0.7m 3-hydroxypicolinic acid
and 0.07m ammoniumcitrate for adduct suppression. The samples were
desalted by addition of small amounts DOWEX beads which were thor-
oughly rinsed with water before use. This was done in the sample tube
and not on the sample plate as in certain cases (if NaCNBH3 was pres-
ent) the sample plate corroded and presence of DOWEX beads resulted
in amorphous drying instead of crystallization.
Experimental Section
Materials and methods for chemical synthesis: All solvents and chemical
reagents were purchased from Sigma–Aldrich. 1H NMR and 13C NMR
spectra were recorded on a Bruker Avance 300 or a Bruker DRX 500
spectrometer operating at room temperature. Chemical shifts are report-
ed in parts per million (d) relative to the residual solvent peak. Multiplic-
ities are reported as singlet (s), doublet (d), doublet of doublets (dd),
triplet (t) or multiplet (m). Purities of small compounds were checked by
reverse-phase HPLC-MS or by reverse-phase HPLC. Reverse-phase
HPLC-MS: Agilent-1100 series LC/MS system equipped with a Phenom-
Synthesis of modified DNA: All oligonucleotides were synthesized
DMT-on using an ABI 394 DNA-synthesizer at 1 mmol scales. A standard
synthesis protocol was used except for coupling of the modified residues.
Chem. Eur. J. 2011, 17, 6940 – 6953
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6951

A. Madder et al.
The synthesis column was removed from the DNA-synthesizer for the in-
troduction of the modified residues. For these manual couplings, a 0.06m
solution of the phosphoramidite in acetonitrile and a 0.1m solution of di-
cyanoimidazole (DCI) in acetonitrile were dried on molecular sieves for
about 20 min. Then small portions of the phosphoramidite solution
(0.4 mL) and the DCI solution (0.5 mL) were alternately pushed through
the reaction column. The column was flushed with acetonitrile (1 mL)
and a mixture of Cap A (0.5 mL) and Cap B (0.5 mL) was pushed
through the column, after which it was again flushed with acetonitrile
(1 mL). The column was reinstalled on the DNA-synthesizer and auto-
mated synthesis was resumed.
[1] G. J. Moser, J. Foley, M. Burnett, T. L. Goldsworthy, R. Maronpot,
Exp. Toxicol. Pathol. 2009, 61, 101–111.
[2] A. E. Sirica, Toxicol. Pathol. 1996, 24, 90–99.
[3] R. Fransson-steen, T. L. Goldsworthy, G. L. Kedderis, R. Maronpot,
Toxicology 1997, 118, 195–204.
[4] D. Lu, L. A. Peterson, Chem. Res. Toxicol. 2010, 23, 142–151.
[5] D. Lu, M. M. Sullivan, M. B. Phillips, L. A. Peterson, Chem. Res.
Toxicol. 2009, 22, 997–1007.
[6] L. A. Peterson, M. E. Cummings, J. Y. Chan, C. C. Vu, B. A. Matter,
Chem. Res. Toxicol. 2006, 19, 1138–1141.
[7] K. C. Hickling, J. M. Hitchcock, V. Oreffo, A. Mally, T. G. Ham-
mond, J. G. Evans, J. K. Chipman, Toxicol. Pathol. 2010, 38, 230–
243.
[8] Y. Kobayashi, G. B. Kumar, T. Kurachi, H. P. Acharya, T. Yamazaki,
T. Kitazume, J. Org. Chem. 2001, 66, 2011–2018.
[9] Y. Kobayashi, M. Nakano, H. Okui, Tetrahedron Lett. 1997, 38,
8883–8886.
[10] M. D. Burke, E. M. Berger, S. L. Schreiber, Science 2003, 302, 613–
618.
[11] a) S. Halila, T. Velasco, P. De Clercq, A. Madder, Chem. Commun.
2005, 936–938; b) M. Op de Beeck, A. Madder, J. Am. Chem. Soc.
2011, 133, 796–807.
[12] K. Stevens, A. Madder, Nucleic Acids Res. 2009, 37, 1555–1565.
[13] S. S. Lange, M. C. Reddy, K. M. Vasquez, DNA Repair 2009, 8, 865–
872.
[14] K. M. Mccabe, S. B. Olson, R. E. Moses, J. Cell. Physiol. 2009, 220,
569–573.
[15] K. Tsuchida, K. Komatsu, Cancer Sci. 2008, 99, 2238–2243.
[16] O. D. Schꢂrer, ChemBioChem 2005, 6, 27–32.
[17] J. T. Millard, S. Raucher, P. B. Hopkins, J. Am. Chem. Soc. 1990,
112, 2459–2460.
[18] P. L. Fischhaber, A. S. Gall, J. A. Duncan, P. B. Hopkins, Cancer Res.
1999, 59, 4363–4368.
[19] A. J. Warren, J. W. Hamilton, Chem. Res. Toxicol. 1996, 9, 1063–
1071.
[20] D. M. Noll, A. M. Noronha, P. S. Miller, J. Am. Chem. Soc. 2001,
123, 3405–3411.
[21] C. J. Wilds, A. M. Noronha, S. Robidoux, P. S. Miller, J. Am. Chem.
Soc. 2004, 126, 9257–9265.
[22] C. J. Wilds, F. Xu, A. M. Noronha, Chem. Res. Toxicol. 2008, 21,
686–695.
[23] T. Angelov, A. Guainazzi, O. D. Scharer, Org. Lett. 2009, 11, 661–
664.
[24] Y. Taniguchi, Y. Kurose, T. Nishioka, F. Nagatsugi, S. Sasaki, Bioorg.
Med. Chem. 2010, 18, 2894–2901.
[25] A. Murakami, Y. Yamamoto, M. Namba, R. Iwase, T. Yamaoka,
Bioorg. Chem. 2001, 29, 223–233.
[26] M. Nakane, S. Ichikawa, A. Matsuda, J. Org. Chem. 2008, 73, 1842–
1851.
Oligonucleotides were cleaved from solid support and deprotected by in-
cubation at 558C overnight in concentrated aqueous ammonia. The syn-
thesized DMTr-ON oligonucleotides were deprotected and purified on
Sep-pak C18 cartridges (obtained from Waters).
Thermal denaturation experiments: All UV experiments were recorded
on a Varian Cary 300 Bio equipped with a six-cell thermostatted cell
holder. Concentrations were measured at 260 nm for all oligonucleotides
at room temperature. Extinction coefficients (e) were calculated accord-
ing to the nearest neighbour method. Melting curves were monitored at
260 nm with a heating rate of 0.38CminÀ1. The buffer contained 100 mm
NaCl and 10 mm phosphate buffer (pH 7). Oligonucleotide concentration
was 2 mm for each strand. Melting temperatures were calculated from the
first derivative of the heating curves using the Cary 300 Bio software.
Cross-linking reactions: The modified strands were mixed with their com-
plements in equimolar amounts at 0.02 mm concentration in 10 mm phos-
phate buffer (pH 7) and 100 mm NaCl before annealing from 958C to
room temperature. The final concentration of oligonucleotide duplex was
20 mm. Temperatures during the whole reaction were kept constant in an
Eppendorf thermomixer comfort at 208C. A stock solution of NBS in
H2O (1 equiv/2 mL) was freshly prepared and to start the reaction, one
equivalent NBS was added. This was repeated every 15 min until com-
plete disappearance of the modified oligonucleotide. The reactions were
monitored by RP-HPLC.
Gel-electrophoresis experiments: A 20% polyacrylamide (acrylamide:
bisacrylamide 19:1) with 1X TBE buffer and 7m urea was used for all
analyses. The temperature of the gel was stabilized with a Julabo F12 at
258C. The power supply used for gel-electrophoresis was a consort
EV202. Gels were stained with GelRed (VWR) and pictures were taken
with an Autochemi imaging system (UVP).
Enzymatic digestion: Exonuclease III was purchased from GE healthcare
or Sigma. For the enzymatic degradations the cross-linked duplex
(0.6 nmol) was dissolved in 1X reaction buffer (66 or 88 mL) supplied
with the enzyme. Then, 200 units of the enzyme were added and the mix-
ture was shaken at 378C for 55 min. The reaction mixture was injected
immediately onto reverse-phase HPLC for purification and the peaks
eluting after 7 min were collected. Those eluting faster correspond to
mononucleotides from the digestion.
[27] M. Comb, N. Mermod, S. E. Hyman, J. Pearlberg, M. E. Ross, H. M.
Goodman, EMBO J. 1988, 7, 3793–3805.
[28] S. Murata, Y. Mizumura, K. Hino, Y. S. Ichikawa, A. Matsuda, J.
Am. Chem. Soc. 2007, 129, 10300–10301.
Acknowledgements
[29] M. Endo, T. Majima, Org. Biomol. Chem. 2005, 3, 3476–3478.
[30] M. Ogino, K. Fujimoto, Angew. Chem. 2006, 118, 7381–7384;
Angew. Chem. Int. Ed. 2006, 45, 7223–7226.
K.S. is indebted to the Special Research Fund of Ghent University
(BOF-GOA2007, BOF-BAS 01B04405). This research project has further
been supported by a Marie Curie Early Stage Research Training Fellow-
ship of the European Communityꢅs Sixth Framework Programme under
contract number MEST-CT-2005-020643. M.H. is indebted to the Acade-
my of Sciences of the Czech Republic (Z4 055 905), to the Ministry of
Education of the Czech Republic (LC 512), and to the Grant Agency of
the ASCR (IAA400550902). A.M. further thanks the FWO Vlaanderen
for financial support (FWO-KAN 1.5.137.09N, FWO-KAN 1.5.186-03).
Computational resources and services used in this work were provided
by Ghent University.
[31] H. Y. Li, Y. L. Qiu, E. Moyroud, Y. Kishi, Angew. Chem. 2001, 113,
1519–1523; Angew. Chem. Int. Ed. 2001, 40, 1471–1475.
[32] a) Q. B. Zhou, S. E. Rokita, Proc. Natl. Acad. Sci. USA 2003, 100,
15452–15457; b) J. T. Sczepanski, A. C. Jacobs, M. M. Greenberg, J.
Am. Chem. Soc. 2008, 130, 9646–9647; c) X. Peng, I. S. Hong, H. Li,
M. M. Seidman, M. M. Greenberg, J. Am. Chem. Soc. 2008, 130,
10299–10306; d) T. Kawasaki, F. Nagatsugi, A. Monsur, M. Maeda,
K. Sugiyama, K. Hori, S. Sasaki, J. Org. Chem. 2005, 70, 14–23.
[33] A. Kobori, J. Morita, M. Ikeda, A. Yamayoshi, A. Murakami,
Bioorg. Med. Chem. Lett. 2009, 19, 3657–3660.
[34] M. B. Smeaton, E. M. Hlavin, A. M. Noronha, S. P. Murphy, C. J.
Wilds, P. S. Miller, Chem. Res. Toxicol. 2009, 22, 1285–1297.
6952
www.chemeurj.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 6940 – 6953

Bioorganic Chemistry
ˇ
FULL PAPER
ˇ
ˇ
[35] M. Stefko, R. Pohl, B. Klepetꢆrovꢆ, M. Hocek, Eur. J. Org. Chem.
[47] L. Zendlovꢆ, D. reha, M. Hocek, P. Hobza, Chem. Eur. J. 2009, 15,
2008, 73, 1689–1704.
7601–7610.
ˇ
[36] M. Urban, R. Pohl, B. Klepetꢆrovꢆ, M. Hocek, J. Org. Chem. 2006,
[48] I. Singh, W. Hecker, A. K. Prasad, V. S. Parmar, O. Seitz, Chem.
Commun. 2002, 500–501.
71, 7322–7328.
ˇ
[37] N. Joubert, R. Pohl, B. Klepetꢆrovꢆ, M. Hocek, J. Org. Chem. 2007,
[49] J. T. Sczepanski, A. C. Jacobs, A. Majumdar, M. M. Greenberg, J.
Am. Chem. Soc. 2009, 131, 11132–11139.
72, 6797–6805.
ˇ
[38] J. Bꢆrta, R. Pohl, B. Klepetꢆrovꢆ, N. P. Ernsting, M. Hocek, J. Org.
[50] Z. Wang, K. X. Wan, R. Ramanathan, J. S. Taylor, M. L. Gross, J.
Am. Soc. Mass Spectrom. 1998, 9, 683–691.
[51] Mechanistic details of furan oxidation with NBS in water.
Chem. 2008, 73, 3798–3806.
ˇ
[39] M. Stefko, R. Pohl, M. Hocek, Tetrahedron 2009, 65, 4471–4483.
[40] M. Stefko, L. Slavetꢇnskꢆ, B. Klepetꢆrovꢆ, M. Hocek, J. Org. Chem.
ˇ
ˇ
ˇ
2010, 75, 442–449.
ˇ
[41] M. Hocek, R. Pohl, B. Klepetꢆrovꢆ, Eur. J. Org. Chem. 2005, 4525–
4528.
[42] N. Joubert, M. Urban, R. Pohl, M. Hocek, Synthesis 2008, 1918–
1932.
[43] C. Vargeese, J. Carter, J. Yegge, S. Krivjanski, A. Settle, E. Kropp,
K. Peterson, W. Pieken, Nucleic Acids Res. 1998, 26, 1046–1050.
[44] Figure 1 graphically summarizes Tm results for duplexes incorporat-
ing two A residues neighboring the modified residue. Similar results
and conclusions apply for duplexes incorporating any of the other
three residues at the neighboring positions (see data in Table 1).
[45] D. A. Case, T. A. Darden, T. E. Cheatham, III, C. L. Simmerling, J.
Wang, R. E. Duke, R. Luo, K. M. Merz, D. A. Pearlman, M. Crow-
ley, R. C. Walker, W. Zhang, B. Wang, S. Hayik, A. Roitberg, G.
Seabra, K. F. Wong, F. Paesani, X. Wu, S. Brozell, V. Tsui, H.
Gohlke, L. Yang, C. Tan, J. Mongan, V. Hornak, G. Cui, P. Beroza,
D. H. Mathews, C. Schafmeister, W. S. Ross, P. A. Kollman,
AMBER 9, University of California, San Francisco, 2006.
[46] D. A. Case, T. Cheatham, T. Darden, H. Gohlke, R. Luo, K. M.
Merz, Jr., A. Onufriev, C. Simmerling, B. Wang, R. Woods, J.
Comput. Chem. 2005, 26, 1668–1688.
[52] Y. Kobayashi, M. Nakano, G. B. Kumar, K. Kishihara, J. Org. Chem.
1998, 63, 7505–7515.
[53] A. Machwe, R. Ganunis, V. A. Bohr, D. K. Orren, Nucleic Acids
Res. 2000, 28, 2762–2770.
[54] Only in case of duplex D2b (G7 2G9:C18 C17 C16 sequence context),
MALDI-TOF analysis revealed the presence of a minor amount of
cross-linked product containing a covalent bond to the neighboring
C16 through the isolation of the corresponding tailed compound.
[55] L. Gingipalli, P. C. Dedon, J. Am. Chem. Soc. 2001, 123, 2664–2665.
[56] P. Regulus, B. Duroux, P. A. Baylet, A. Favier, J. Cadet, J. L. Rava-
nat, Proc. Natl. Acad. Sci. USA 2007, 104, 14032–14037.
Received: January 7, 2011
Published online: May 20, 2011
Chem. Eur. J. 2011, 17, 6940 – 6953
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6953