Generation of DNA interstrand cross-links by post-synthetic reductive amination

Article abstract of DOI:10.1021/ol802719a

(Chemical Equation Presented) DNA interstrand cross-links (ICLs) are the clinically most relevant adducts formed by many antitumor agents. To facilitate the study of biological responses triggered by ICLs, we developed a new approach toward the synthesis

Full text of DOI:10.1021/ol802719a

ORGANIC
LETTERS
2009
Vol. 11, No. 3
661-664
Generation of DNA Interstrand
Cross-Links by Post-Synthetic Reductive
Amination
Todor Angelov,† Angelo Guainazzi,‡ and Orlando D. Scha¨rer*,†,‡
Institute of Molecular Cancer Research, UniVersity of Zurich, Winterthurerstrasse 190,
8057 Zurich, Switzerland, and Departments of Pharmacological Sciences and
Chemistry, Chemistry 619, Stony Brook UniVersity, Stony Brook, NY11794-3400
orlando@pharm.stonybrook.edu
Received November 25, 2008
ABSTRACT
DNA interstrand cross-links (ICLs) are the clinically most relevant adducts formed by many antitumor agents. To facilitate the study of biological
responses triggered by ICLs, we developed a new approach toward the synthesis of mimics of nitrogen mustard ICLs. 7-Deazaguanine residues
bearing acetaldehyde groups were incorporated into complementary strands of DNA and cross-link formation induced by double reductive
amination. Our strategy enables the synthesis of major groove cross-links in high yields and purity.
A variety of bifunctional electrophilic agents have the ability
to covalently link two complementary strands of double-
stranded DNA, leading to the formation of DNA interstrand
cross-links (ICLs). Due to their ability to efficiently block
replication and transcription, ICLs are among the most
cytotoxic DNA lesions known. Their cytotoxic potential has
found application in anticancer chemotherapy, and cross-
linking agents such as the nitrogen mustards, platinum
complexes, chloro ethyl nitroso ureas, and mitomycin C are
among the most widely used antitumor agents today.1,2 Many
factors contribute to the resistance of tumor cells to treatment
with cross-linking agents, and the removal of ICLs from
DNA has emerged as one of the most important ones.3 Repair
pathways for ICLs have their evolutionary origin in the
necessity to counteract the threat posed by endogenous and
exogenous metabolites, for example, malonic dialdehyde and
formaldehyde.4 ICL repair is an inherently complex process
as the two strands need to be unhooked and repaired and no
intact template is available for repair synthesis.2 This
complexity provides one explanation why ICL repair path-
ways remain poorly understood.
A second problem in the study of ICL repair has been the
limited availability of defined ICL adducts.5 ICLs were
initially synthesized by the reaction of DNA with cross-
linking agents followed by isolation and purification of the
ICL.6-8 This approach yields mixtures of products (mono
(3) Panasci, L.; Paiement, J. P.; Christodoulopoulos, G.; Belenkov, A.;
Malapetsa, A.; Aloyz, R. Clin. Cancer Res. 2001, 7, 454–61. McHugh,
P. J.; Spanswick, V. J.; Hartley, J. A. Lancet Oncol. 2001, 2, 483–90.
(4) Chaw, Y. F.; Crane, L. E.; Lange, P.; Shapiro, R. Biochemistry 1980,
19, 5525–31. Stone, M. P.; Cho, Y. J.; Huang, H.; Kim, H. Y.; Kozekov,
I. D.; Kozekova, A.; Wang, H.; Minko, I. G.; Lloyd, R. S.; Harris, T. M.;
Rizzo, C. J. Acc. Chem. Res. 2008, 41, 793–804.
† University of Zurich.
‡ Stony Brook University.
(1) Lawley, P. D.; Phillips, D. H. Mutat. Res. 1996, 355, 13–40. Rajski,
S. R.; Williams, R. M. Chem. ReV. 1998, 98, 2723–2796
.
(2) Noll, D. M.; Mason, T. M.; Miller, P. S. Chem. ReV. 2006, 106,
(5) Noll, D. M.; Noronha, A. M.; Wilds, C. J.; Miller, P. S. Front. Biosci.
2004, 9, 421–37. Scha¨rer, O. D. ChemBioChem 2005, 6, 27–32.
277–301
.
10.1021/ol802719a CCC: $40.75
Published on Web 01/08/2009
2009 American Chemical Society

adducts, intra- and interstrand cross-links) and the desired
ICLs usually make up only a small fraction (typically 1-5%)
of all the products formed. More efficient approaches to the
chemical synthesis of ICLs have subsequently been devel-
oped. One was based on the cross-linking of nucleosides
outside of DNA followed by incorporation of the cross-linked
dimer into DNA using solid-phase synthesis.9 This approach
has been particularly successful in yielding ICLs that connect
two bases through their Watson-Crick faces, although the
solid-phase synthesis procedures are complicated by the
connected nucleoside dimers. A second approach consists
of the site-specific incorporation of postsynthetically modifi-
able cross-link precursors on one or two opposing strands
of DNA, annealing of the two single strands and subsequent
use of a specific coupling reaction to generate the ICL. This
concept has been used in the synthesis of ICLs containing
disulfide, psoralen, malondialdehyde, and alkyl linkages.10,11
With the exception of disulfide ICLs, there has been a lack
of efficient syntheses of ICLs formed in the major groove
of DNA, where adducts by the clinically most important ICL-
forming agents including the nitrogen mustards (NMs) are
formed.
NMs preferentially form ICLs between the N7 positions
of dG residues in a 5′-GNC sequence context and introduce
a slight distortion into the DNA.6,7,12,13 Following the initial
pioneering studies of NM ICLs by the Loechler and Hopkins
groups, no further efforts toward high-yielding synthesis of
these adducts have been reported.6
We designed a strategy for the synthesis of NM ICL
mimics 2 based on the incorporation of ICL precursor
nucleosides on opposing strands of DNA and the use of a
subsequent specific coupling reaction to establish the ICL
(Figure 1). We reasoned that alkylamine-containing cross-
links may be accessed by a double-amination reaction from
two aldehyde groups (3) on cDNA strands using an ap-
propriate amine. Since guanine bases alkylated at the 7
position in the native NM ICL 1 are prone to depurination
Figure 1. Structures of NM ICLs 1 and strategy for the synthesis
of NM ICL analog 2. NM ICLs (1) connect two complementary
DNA strands by connecting two guanine bases through the N(7)
positions. Our target NM ICL mimic 2 has the nitrogen at the 7
positions replaced with carbon to render the glycosidic bond of
the ICL stable for synthesis and functional studies (X ) amine-
containing compound). We envisioned that ICL 2 could be obtained
via double-reductive amination of an aldehyde 3 with an appropriate
amine, XH. 3 in turn could be derived from phosphoramidite 4,
where the aldehyde is masked as a protected diol.
due to the positive charge on nitrogen, we decided to pursue
the synthesis of the more stable, isosteric 7-deaza analogues
2. The increased hydrolytic stability of the 7-deaza com-
pounds should make it possible to incorporate them into
DNA using solid-phase synthesis and make them attractive
substrates for biological studies. We envisioned that the
aldehyde would be introduced into DNA masked as a
protected diol 4,14 using standard phosphoramidite chemistry
(Figure 1).
The synthesis of 4 (Scheme 1) started with 6-chloro-7-
deaza-3′,5′-di-O-p-toluoyl-2′-deoxyguanosine 5,15 which was
protected as an isobutyric amide at the N(2) position and
selectively iodinated at C(7).16 A Stille coupling reaction
was then used to introduce the allyl group in 6. Treatment
with pyrimidine-2-carboxaldoxime17 restored the deazagua-
nine core, and the toluoyl protecting groups in the carbo-
hydrate moiety were replaced with TBDMS. The allyl group
of 7 was then oxidized to corresponding diol (8) using
osmium tetroxide, and the newly generated hydroxyl groups
were protected as acetate esters. Finally, the TBDMS
protecting groups were removed, and the sugar moiety was
elaborated to the phosphoramidite 4 using standard proce-
dures.
(6) Ojwang, J. O.; Grueneberg, D. A.; Loechler, E. L. Cancer Res. 1989,
49, 6529–37. Millard, J. T.; Raucher, S.; Hopkins, P. B. J. Am. Chem. Soc.
1990, 112, 2459–2460.
(7) Rink, S. M.; Solomon, M. S.; Taylor, M. J.; Rjur, S.; McLaughlin,
L. W.; Hopkins, P. B. J. Am. Chem. Soc. 1993, 115, 2551–2557.
(8) Warren, A. J.; Hamilton, J. W. Chem. Res. Toxicol. 1996, 9, 1063–
71. Fischhaber, P. L.; Gall, A. S.; Duncan, J. A.; Hopkins, P. B. Cancer
Res. 1999, 59, 4363–8.
(9) Harwood, E. A.; Sigurdsson, S. T.; Edfeldt, N. B. F.; Reid, B. R.;
Hopkins, P. B. J. Am. Chem. Soc. 1999, 121, 5081–5082. Li, H. Y.; Qiu,
Y. L.; Moyroud, E.; Kishi, Y. Angew. Chem., Int. Ed. Engl. 2001, 40, 1471–
1475. Noll, D. M.; Noronha, A. M.; Miller, P. S. J. Am. Chem. Soc. 2001,
123, 3405–11. Wilds, C. J.; Noronha, A. M.; Robidoux, S.; Miller, P. S.
J. Am. Chem. Soc. 2004, 126, 9257–9265. Wilds, C. J.; Xu, F.; Noronha,
A. M. Chem. Res. Toxicol. 2008, 21, 686–95.
Using solid-phase synthesis, 4 was incorporated into two
cDNA strands as a part of a 5′-d(GNC) sequence (10a-d,
Figure 2A), which has been shown to be the preferred site
for NM ICL formation.7 The two single-stranded oligonucle-
otides were purified by solid phase extraction using TOP-
(10) Ferentz, A. E.; Keating, T. E.; Verdine, G. L. J. Am. Chem. Soc.
1993, 115, 9006–9014. Erlanson, D. A.; Chen, L.; Verdine, G. L. J. Am.
Chem. Soc. 1993, 115, 12583–12584. Kobertz, W. R.; Essigmann, J. M.
J. Am. Chem. Soc. 1997, 119, 5960–5961. Dooley, P. A.; Tsarouhtsis, D.;
Korbel, G. A.; Nechev, L. V.; Shearer, J.; Zegar, I. S.; Harris, C. M.; Stone,
M. P.; Harris, T. M. J. Am. Chem. Soc. 2001, 123, 1730–9. Hong, I. S.;
Greenberg, M. M. J. Am. Chem. Soc. 2005, 127, 10510–1. Hong, I. S.;
Greenberg, M. M. J. Am. Chem. Soc. 2005, 127, 3692–3
.
(14) Khullar, S.; Varaprasad, C. V.; Johnson, F. J. Med. Chem. 1999,
42, 947–50.
(11) Alzeer, J.; Scha¨rer, O. D. Nucleic Acids Res. 2006, 34, 4458–66
.
(12) Rink, S. M.; Hopkins, P. B. Biochemistry 1995, 34, 1439–45. Dong,
Q.; Barsky, D.; Colvin, M. E.; Melius, C. F.; Ludeman, S. M.; Moravek,
J. F.; Colvin, O. M.; Bigner, D. D.; Modrich, P.; Friedman, H. S. Proc.
(15) Seela, F.; Westermann, B.; Binding, U. J. Chem. Soc., Perkin Trans.
1 1988, 697–702.
(16) Ramzaeva, N.; Seela, F. HelV. Chim. Acta 1995, 78, 1083–1090.
(17) Buhr, C. A.; Wagner, R. W.; Grant, D.; Froehler, B. C. Nucleic
Acids Res. 1996, 24, 2974–80.
Natl. Acad. Sci. U.S.A. 1995, 92, 12170–4
.
(13) Fan, Y.-H.; Gold, B. J. Am. Chem. Soc. 1999, 121, 11942–11946
.
662
Org. Lett., Vol. 11, No. 3, 2009

Scheme 1. Synthesis of the ICL Precursor Phosphoramidite 4
cartridges (Varian), and the incorporation and integrity of
the diol was verified by ESI-MS (Table 1, Supporting
Information). Exploratory experiments revealed that the diol-
containing DNA could be oxidized to the aldehyde and
derivatized with an aldehyde-reactive semicarbazide fluo-
rescent dye (data not shown), validating our approach for
the generation of the aldehyde precursor.
Figure 2. Synthesis of DNA ICLs through reductive amination.
(A) Phosphoramidite 4 was incorporated into complementary
strands of DNA and deprotected (10a-d). Following annealing and
oxidization with NaIO4 to aldehydes 11a-d, reaction with am-
monium choride, methylamine, hydrazine, ethylenediamine, or
dimethylethylenediamine followed by reduction with NaBH3CN
was attempted to generate ICLs 12. (B) Structures of the target
ICLs 12a-h. (C) Denaturing PAGE analysis of the reductive
amination reaction of the two aldheyde containing 20-mers. The
amine group X resulting from the reaction, as well as the presence
or absence of NaIO4 and NaBH3CN in the reaction mixture, are
indicated above the gel. The positions of marker 20-mer (for
ssDNA) and 40-mer (for ICL containing DNA) oligomers are
indicated. The gel was stained with methylene blue.
With the aldehyde-containing single-stranded oligonucle-
otides in hand, we investigated the use of these building
blocks in ICL formation. To circumvent the need for lengthy
manipulations of the potentially unstable aldehyde function-
alities, the two diol-containing strands were first annealed
and subsequently oxidized with sodium periodate. Excess
oxidizing reagent was quenched by addition of sodium sulfite,
and the aldehyde-containing oligonucleotides were treated
with a variety of amines and NaBH3CN. Unexpectedly,
incubation with NH4OAc or methylamine did not lead to
any detectable formation of ICLs 12a or 12b, as only a band
corresponding to the 20-mer single stranded oligonucleotide
substrates was visible on a denaturing polyacrylamide gel
(Figure 2C, lanes 2 and 3). By contrast, treatment with
hydrazine, ethylenediamine, or N,N′-dimethylethylenedi-
amine resulted in the formation of a new major band with
mobility corresponding to a size roughly double that of the
starting 20-mers, indicating that the formation of ICLs 12c,
12d, and 12e had occurred (Figure 2B, C, lanes 4, 7, and
10). While the formation of ICLs by the ethylenediamines
was dependent on the presence of NaBH3CN, hydrazine ICLs
were also formed in the absence of any reducing agent, which
could be explained by the formation of a stable hydrazone
adduct (Figure 2C, compare lanes 4 and 5). Treatment with
excess hydrazine and formaldehyde readily reversed of the
nonreduced form of hydrazine ICL, but not the reduced form
of 12c (data not shown), indicating that the hydrazone linkage
was indeed reduced by NaBH3CN. The formation of an ICL
was dependent on the presence of aldehyde reactive groups
on both strands of DNA, as no slower migrating band was
formed if the aldehyde functionalized guanine residues were
absent or present only on one of the two strands (data not
shown). The ICL-containing oligonucleotides were purified
by reverse-phase HPLC and subjected to MS and nucleoside
composition analysis. ESI-MS analysis of the cross-linked
20-mer oligonucleotides revealed good agreement between
the calculated and measured mass, but were not accurate
enough to ascertain that the ICLs was present with two
reduced amine bonds (Table 1, Supporting Information). To
obtain more precise mass determinations, we repeated the
synthesis to obtain the shorter 11-mer ICL-containing oli-
gonucleotides 12f-h. In this case, the molecular weight
could be determined for all the ICL-containing oligonucle-
otides with a precision of ( one mass unit (Table 1,
Supporting Information) consistent with ICL formation and
reduction of the imine and hydrazone linkages. We further
Org. Lett., Vol. 11, No. 3, 2009
663

aldehyde groups with ammonia prior to ICL formation, we
also attempted to generate ICL 12a by reaction between a
7-(2-oxoethyl)-7-deazaguanine and a 7-(2-aminoethyl)-7-
deazaguanine residue on opposing strands in a d(GNC)
sequence. This approach also did not lead to formation of
ICL 12a (T.A. and O.D.S., data not shown), suggesting that
indeed distance and reactivity constraints were responsible
for the inability to complete the reductive amination with
ammonia and methylamine. The formation of hydrazine ICLs
(12c) and ethylenediamine ICLs (12d, 12e) on the other hand
would not require the introduction of a bend in the DNA
due to the longer linkage (8.9 Å for hydrazine, 11.7 Å for
ethylenediamine), thereby facilitating ICL formation. An
open question is how much the higher reactivity of hydrazine
compared to an amine contributes to successful ICL forma-
tion in the case of 12c. Upon reaction with the aldehyde,
hydrazine forms a hydrazone, which is much more stable
than an imine and more reactive toward the aldehyde in the
second reaction. Our observations that the reaction of the
dialdehyde can lead to efficient ICL formation with hydrazine
in the absence of a reducing agent supports this notion
(Figure 2C, lane 5).
In conclusion, we have developed synthetic methodology
for the preparation of pure oligonucleotides containing site-
specific ICLs in the major groove using a double reductive
amination reaction of aldehyde functionalities in two comple-
mentary DNA strands at a scale of >10 nmol. This method
should be readily scalable by a factor of 10 or 100 to generate
ICLs in quantities that will enable detailed structural studies.
These defined ICLs have already been proven to be valuable
tools for the study of ICL repair mechanisms.18
Figure 3. Structure of B-from DNA highlighting the distance linked
by a NM ICL. The distance between the two cross-linked N7 atoms
of guanine is marked by a red dotted line. The nominal distances
of an amine, hydrazine, and ethylenediamine ICLs are indicated in
the right panel.
attempted to analyze the ICL-containing oligonucleotides by
nucleoside composition analysis and incubated them with
phosphodiesterase I, exonuclease III and calf intestine
phosphatase, conditions we had previously employed in the
analysis of ICLs.11 HPLC analysis of the digestion of the
dimethylethylenediamine ICL 12e yielded the expected peaks
for the four native nucleosides dA, dC, dG, and T and an
additional slower eluting peak (Figure 1, Supporting Infor-
mation). This peak was identified by MS as the expected
cross-linked nucleoside dimer. Unexpectedly, digestion of
the hydrazine or ethylenediamine ICLs did not result in the
formation of a peak for the cross-linked dimers, as we
observed either incomplete digestion or decomposition at the
peak in the area of the dimethyl ethlylenediamine dimer (data
not shown). Nonetheless, the data from MS analyses and
digestion of the dimethyl ethlylenediamine ICL as well as
the difference of stability of reduced and nonreduced ICLs
and are only consistent with structures 12c-h.
Acknowledgment. We are grateful to Robert Rieger for
MS spectral analyses supported by Grant No. NIH/NCRR1
S10 RR023680-1 and to Arthur J. Campbell for discussions
and help with the figures. This work was supported by the
Swiss National Science Foundation (3130-054873.98), Swiss
Cancer League (OCS-01413-08-2003), and the New York
State Office of Science and Technology and Academic
Research (NYSTAR) (Grant No. C040069).
The lack of ICL formation in the reductive amination
reaction with ammonia and methylamine deserves some
comment. It has previously been shown that since the
distance between the N7-dG sites in 5′-d(GNC) sequences
(∼8.9 Å) is longer than the NM linkage (∼7.5 Å), the NM
ICLs must induce a bend into DNA (Figure 3).13 In our case,
the intrinsically reversible initial imine formation step of the
reductive amination reaction apparently does not provide
enough strength to lead to the formation of ICLs 12a and
12b with these two amines. To exclude that the inability to
form ICL 12a was due to reductive amination of both
Supporting Information Available: Supplementary Table
1, supplementary Figure 1, experimental procedures, 1H, 13C,
and 31P NMR spectra of all new compounds, and ESI-MS
spectra of oligonucleotides 10a-d and 12c-h. This material
is available free of charge via the Internet at http://pubs.acs.org.
OL802719A
(18) Raschle, M.; Knipsheer, P.; Enoiu, M.; Angelov, T.; Sun, J.; Griffith,
J. D.; Ellenberger, T. E.; Scha¨rer, O. D.; Walter, J. C. Cell 2008, 134, 969–
80.
664
Org. Lett., Vol. 11, No. 3, 2009