Demonstration of an alternative mechanism for G-to-G cross-link formation

Article abstract of DOI:10.1021/ja045108j

The cross-link dG-to-dG is an important product of DNA nitrosation. Its formation has commonly been attributed to nucleophilic substitution of N ₂ in a guaninediazonium ion by guanine, while recent studies suggest guanine addition to a cyanoami

Full text of DOI:10.1021/ja045108j

Published on Web 12/30/2004
Demonstration of an Alternative Mechanism for G-to-G
Cross-Link Formation
Ming Qian and Rainer Glaser*
Contribution from the Department of Chemistry, UniVersity of Missouri-Columbia,
605 South College AVenue, Columbia, Missouri 65211
Received August 13, 2004; E-mail: glaserr@missouri.edu
Abstract: The cross-link dG-to-dG is an important product of DNA nitrosation. Its formation has commonly
been attributed to nucleophilic substitution of N2 in a guaninediazonium ion by guanine, while recent studies
suggest guanine addition to a cyanoamine derivative formed after dediazoniation, deprotonation, and
pyrimidine ring-opening. The chemical viability of the latter mechanism is supported here by the experimental
demonstration of rG-to-aG formation via rG addition to a synthetic cyanoamine derivative. Thus, all known
products of nitrosative guanine deamination are consistent with the postulate of pyrimidine ring-opening.
This postulated mechanism not only explains what is already known but also suggests that other products
and other cross-links also might be formed in DNA deamination. The study suggests one possible new
product: the structure isomer aG(N1)-to-rG(C2) of the classical G(N2)-to-G(C2) cross-link. While the
formation of aG(N2)-to-rG(C2) has been established by chemical synthesis, the structure isomer aG(N1)-
to-rG(C2) has been assigned tentatively based on its MS/MS spectrum and because this assignment is
reasonable from a mechanistic perspective. Density functional calculations show preferences for the amide-
iminol tautomer of the classical cross-link G(N2)-to-G(C2) and the amide-amide tautomer of G(N1)-to-
G(C2). Moreover, the results suggest that both cross-links are of comparable thermodynamic stability, and
that there are no a priori energetic or structural reasons that would prevent the formation of the structure
isomer in the model reaction or in DNA.
these are thought to be involved in cross-link repair.15,16 Details
of the detection of the DNA damage, the signaling to initiate
repair, and the actual DNA repair are only now emerging.17
Deamination chemistry thus contributes to genomic instability
directly by causing DNA damage and indirectly by causing
damage to any part of the repair mechanisms.
Xanthosine is the main product of guanosine deamination,
and the formation of xanthine by deamination of guanine has
been known since the very discovery of guanine.18 The second
major product, oxanosine, was discovered only recently by
Suzuki et al.19 and described also by Shuker et al.20 and Dedon
et al.21 While oxanosine can be formed in yields of up to 20%,
its yield varies with conditions, and this might have been one
reason for its late discovery in 1996. Our particular interest here
lies with the interstrand cross-link dG-to-dG, which is a minor
product of major significance. As early as 1961, Geiduschek
discussed the possibility of interstrand cross-links formed by
nitrous acid.22,23 Initially, there was some discussion that the
Introduction
DNA oxidizing chemicals are of interest because of their
antibiotic, antitumor, carcinogenic, and mutagenic properties.1,2
An important class of DNA damaging agents includes nitrates,
nitrites, and nitrogen oxides because of the dietary and
environmental exposure of humans to these substances.3-5
Toxicological studies of deamination became more significant
in the past decade because it was recognized that endogenous
nitric oxide6,7 causes nitrosation8-11 and that this process is
accelerated by chronic inflammatory diseases.12-14 Several
human DNA repair genes have been identified, and some of
(1) Burrows, C. J.; Muller, J. G. Chem. ReV. 1998, 98, 1109-1152.
(2) Marnett, L. J.; Burcham, P. C. Chem. Res. Toxicol. 1993, 6, 771-785.
(3) The Health Effect of Nitrate, Nitrite, and N-Nitroso Compounds. Committee
on Nitrite and Alternative Curing Agents in Food; National Academy
Press: Washington, DC, 1981.
(4) Furia, T. E. Handbook of Food AdditiVes, 2nd ed.; CRC Press: Cleveland,
OH, 1972; pp 150-155.
(5) Wakabayashi, K. In Mutation and the EnVironment, Part E; Albertini, R.
J., Ed.; Wiley-Liss, Inc.: New York, 1990; pp 107-116.
(6) Culotta, E.; Koshland, D. E. Science 1991, 258, 1862-1865.
(7) Marnett, L. J. Chem. Res. Toxicol. 1996, 9, 807-808.
(8) Wink, D. A.; Kasprzak, K. S.; Maragos, C. M.; Elespuru, R. K.; Misra,
M.; Dunams, T. M.; Cebula, T. A.; Koch, W. H.; Andrews, A. W.; Allen,
J. S.; Keefer, L. K. Science 1991, 254, 1001-1003.
(15) Wood, R. D.; Mitchell, M.; Sgouros, J.; Lindahl, T. Science 2001, 291,
1284-1289.
(16) Scha¨rer, O. D. Angew. Chem., Int. Ed. 2003, 42, 2946-2974.
(17) Rouse, J.; Jackson, S. P. Science 2002, 297, 547-551.
(18) Strecker, A. Ann. 1861, 118, 151-177.
(9) Kosaka, H.; Wishnok, J. S.; Miwa, M.; Leaf, C. D.; Tannenbaum, S. R.
Carcinogenesis 1989, 10, 563-566.
(10) Davis, K. L.; Martin, E.; Turko, I. V.; Murad, F. Annu. ReV. Pharmacol.
Toxicol. 2001, 41, 203-236.
(19) Suzuki, T.; Yamaoka, R.; Nishi, M.; Ide, H.; Makino, K. J. Am. Chem.
Soc. 1996, 118, 2515-2516.
(11) Jackson, A. L.; Loeb, L. A. Mutat. Res. 2001, 477, 7-21.
(12) Ohshima, H.; Bartsch, H. Mutat. Res. 1994, 305, 253-264.
(13) deRojas-Walker, T.; Tamir, S.; Ji, H.; Wishnok, J. S.; Tannenbaum, S. R.
Chem. Res. Toxicol. 1995, 8, 473-477.
(20) Lucas, L. T.; Gatehouse, D.; Shuker, E. G. J. Biol. Chem. 1999, 274,
18319-18326.
(21) Dong, M.; Wang, C.; Deen, W. M.; Dedon, P. C. Chem. Res. Toxicol.
2003, 16, 1044-1055.
(14) Tamir, S.; Tannenbaum, S. R. Biochim. Biophys. Acta 1996, 1288, F31-
F36.
(22) Geiduschek, E. P. Proc. Natl. Acad. Sci. U.S.A. 1961, 47, 950-955.
(23) Geiduschek, E. P. J. Mol. Biol. 1963, 8, 377-391.
9
880
J. AM. CHEM. SOC. 2005, 127, 880-887
10.1021/ja045108j CCC: $30.25 © 2005 American Chemical Society

Alternative Method for G-to-G Cross-Link Formation
A R T I C L E S
cross-link formation occurs without depurination24 and would
be a derivative of xanthosine.25 The cross-link structures became
clear in 1977 when Shapiro et al. isolated and partially
characterized the cross-links dG-to-dG and dG-to-dA formed
in treatments of calf thymus DNA with nitric acid.26,27 The dG-
to-dG cross-link also can be formed by nitric oxide (NO).28
Hopkins et al. showed in the early 1990s that the dG-to-dG
interstrand cross-link is formed with sequence specificity in 5′-
CG over 5′-GC.29,30 In the last five years, chemical syntheses
of dG-to-dG have provided direct confirmation of the cross-
link structure.31-33
Scheme 1. Experimental Demonstration of Cross-Link rG-to-aG
Formation by Reaction of 4a (4, R ) CH2OCH2OH) with
Guanosine rG
The commonly discussed reaction mechanism for nitrosative
guanosine deamination postulates the guanosinediazonium ion
1d (1, R ) deoxyribose) as the key intermediate formed by the
nitrosation of the exocyclic primary amino group of guanosine.
This assumption explains the formation of xanthosine and of
the interstrand cross-links dG-to-dG and dG-to-dA, respec-
tively, by aromatic nucleophilic substitutions of N2 by water or
by the amino group of a guanosine or an adenosine of the
opposite stand, respectively. Recent theoretical studies revealed
that guaninediazonium ion 1 (R ) H) is much less stable than
had been thought, and that pyrimidine ring-opening occurs with
dediazoniation and deprotonation to form cyanoimines 2.34-36
Oxanosine formation requires such a pyrimidine ring-opening,
and recent labeling studies established the intermediacy of
5-cyanoimino-4-oxomethylene-4,5-dihydroimidazole and 5-cy-
anoamino-4-imidazolecarboxylic acid in nitrosative guanosine
deamination.37 On the other hand, the formations of xanthosine
and of the cross-links may proceed with or without pyrimidine
ring-opening or by a combination of these paths, and which of
these options is realized might depend on the environment
(solution, DNA, and pH). We postulate with the principle of
parsimony38 that all products are formed from the same common
intermediate, the cyanoimine 2, and it has been our goal to
explore whether this option is chemically viable.
function in the presence of a carboxamide and for a conceptual
reason (vide infra). We described the synthesis of 4a and
demonstrated its recyclization to guanosine aG and isoguanosine
aIG in analogy to the formations of xanthosine and oxanosine
from 3, respectively.41 Here, we report experiments that
demonstrate the formation of the cross-link rG-to-aG by
addition of guanosine rG (R′ ) ribose) to cyanoamine 4a.42
We are exploring the chemistry of the cyanoamines 4. The
CH2OCH2OH group is a simple glycoside mimic, and com-
pounds with this R group are designated by a (derivative of
acyclovir). Cyanoamines 3 and 4 are the products of the 1,4-
addition of H-XHn to cyanoimine 2, and this addition is very
fast because of the zwitterionic nature of cyanoimine 2.39,40 The
5-cyanoamino-4-imidazolecarboxamide 4a was targeted because
of practical advantages in the preparation of the cyanoamino
Results and Discussion
Strategy: Giving the Bimolecular Reaction a Chance in
the Absence of Templating. The most direct evidence for the
formation of dG-to-dG by guanosine addition to a cyanoamine
4 would involve the synthesis of oligonucleotides with 4d in
place of dG and the detection of dG-to-dG cross-link formation.
This experiment is impossible because 4d cannot be generated
in an oligonucleotide via a targeted synthesis. The next best
strategy involves the synthesis of a cyanoamine 4 and the study
of its bimolecular guanosine addition reaction. This strategy
could be improved by tethering the guanosine to the cyanoam-
ine, and we were contemplating such approaches for some time
before we gathered enough courage to even try the untethered
reaction, which to our surprise, worked quite well!
The recyclizations of cyanoamine 4a to isoguanine aIG and
guanine aG are very fast.41 The only chance to observe any
intermolecular reaction in solution requires high concentrations
of the starting materials (of rG in particular) to increase the
rate of the bimolecular addition. Due to the poor solubility of
guanosine rG in water, we therefore studied the reaction in
(24) Verly, W. G.; Lacroix, M. Biochim. Biophys. Acta 1975, 414, 185-192.
(25) Burnotte, J.; Verly, W. H. J. Biol. Chem. 1971, 246, 5914-5918.
(26) Shapiro, R.; Dubelman, S.; Feinberg, A. M.; Crain, P. F.; McCloskey, A.
M. J. Am. Chem. Soc. 1977, 99, 302-303.
(27) Dubelman, S.; Shapiro, R. Nucleic Acids Res. 1977, 4, 1815-1827.
(28) Caulfield, J. L.; Wishnok, J. S.; Tannenbaum, S. R. Chem. Res. Toxicol.
2003, 16, 571-574.
(29) Kirchner, J. J.; Hopkins, P. B. J. Am. Chem. Soc. 1991, 113, 4681-4682.
(30) Kirchner, J. J.; Sigurdsson, S. T.; Hopkins, P. B. J. Am. Chem. Soc. 1992,
114, 4021-4027.
(31) Harwood, E. A.; Sigurdsson, S. T.; Edfeldt, N. B. F.; Reid, B. R.; Hopkins,
P. B. J. Am. Chem. Soc. 1999, 121, 5081-5082.
(32) Harwood, E. A.; Hopkins, P. B.; Sigurdsson, S. T. J. Org. Chem. 2000,
65, 2959-2964.
(33) De Riccardis, F.; Johnson, F. Org. Lett. 2000, 2, 293-295.
(34) Glaser, R.; Son, M. S. J. Am. Chem. Soc. 1996, 118, 10942-10943.
(35) Glaser, R.; Rayat, S.; Lewis, M.; Son, M.-S.; Meyer, S. J. Am. Chem. Soc.
1999, 121, 6108-6119.
(36) Glaser, R.; Lewis, M. Org. Lett. 1999, 1, 273-276.
(37) Rayat, S.; Majumdar, P.; Tipton, P.; Glaser, R. J. Am. Chem. Soc. 2004,
126, 9960-9969.
(38) Wassermann, G. D. From Occam’s Razor to the Roots of Consciousness:
20 Essays on Philosophy, Philosophy of Science and Philosophy of Mind;
Avebury: Aldershot, U.K., 1997.
(41) Qian, M.; Glaser, R. J. Am. Chem. Soc. 2004, 126, 2274-2275.
(42) rG was used for economical reasons, and no significant differences would
be expected for dG.
(39) Rayat, S.; Glaser, R. J. Org. Chem. 2003, 68, 9882-9892.
(40) Rayat, S.; Wu, Z.; Glaser, R. Chem. Res. Toxicol. 2004, 17, 1157-1169.
9
J. AM. CHEM. SOC. VOL. 127, NO. 3, 2005 881

A R T I C L E S
Qian and Glaser
Scheme 2. Outline of the Synthesis of rG-to-aG
DMSO. All prior reports of cross-link formation of natural or
synthetic DNA and oligonucleotides with HNO2 or NO in
aqueous buffer solution were carried out with low DNA
concentrations because minimal structural distortions of the
duplex shape suffice to position the amino N atom of the
nucleophile dG in the proximity of the reactive intermediate
generated by dG deamination on the opposite strand. These
proximity effects are important and were cited to explain the
sequence specificity.29,30 Cross-link formation caused by nitrous
acid is preferred at 5′-CG positions compared to 5′-GC positions
because the B conformation of 5′-CG positions the amino group
of the nucleophile dG well, while a 3 Å base pair sliding is
required to achieve proximity in 5′-GC. But even with all the
advantages provided by the DNA environment, cross-link
formation in DNA still hardly can compete with hydrolyses to
form xanthine and oxanine products (1 in 4 deamination events).
The yield of dG-to-dG was ca. 0.06% in Shapiro’s HNO2
treatment of calf thymus DNA,26 and the Hopkins group reported
a cross-link yield of 2.4% in their synthetic oligodeoxynucleotide
with the 5′-CG sequence.29,30 Bimolecular reactions also are
accelerated by increases in temperature because the collision
frequency increases, and we obtained detectable amount of
products of the bimolecular reaction at about 80 °C. In DNA
work, of course, no such temperature increases are needed
because proximity is established by the environment. Finally,
we employed different R groups on the cyanoamine 4a and the
guanosine rG to ensure that any detected cross-link was indeed
the product of bimolecular addition rather than some sort of
dimerization.
Figure 1. HPLC chromatogram of the product of the reaction between
cyanoamine 4a and guanosine rG demonstrates rG-to-aG formation. The
cross-link was identified by the addition of synthetic and authenticated cross-
link rG-to-aG (increased peak, in red, with retention time of 16.0 min).
rG and of the OH group of aG were carried out by reaction of
tert-butyldimethylsilyl chloride (TBDMSCl) in the presence of
imidazole in DMF solution. The O6 atoms of 6 and 9 were
subsequently protected by conversion into the benzyl ethers 7
and 10, respectively. Then, 7 was brominated by diazotization
with tert-butylnitrite in methylene bromide and treatment with
SbBr3 to afford 8. The coupling reaction of 8 to 10 was catalyzed
by 10 mol % palladium acetate, 15 mol % (()-BINAP, and the
presence of Cs2CO3 and gave 11 in 75% yield. The silyl
protecting groups were removed with tetrabutylammonium
fluoride (TBAF), and the benzyl protecting groups were
removed by hydrogenation. The product rG-to-aG was fully
characterized by 1H and 13C NMR spectroscopy and mass
spectrometry. Ours is the first unsymmetrical (R * R′) G-to-G
cross-link, and the 1H NMR spectrum of rG-to-aG shows two
aromatic nonexchangeable hydrogens (C8 positions).
Synthesis of rG-to-aG Cross-Link. The reliable detection
of tiny amounts of cross-link in large amounts of products of
unimolecular cyclization poses an analytical challenge. We
synthesized and fully characterized rG-to-aG and thereby made
possible the tuning of the mass spectrometer to optimal
sensitivity for rG-to-aG. The synthesis of the cross-link rG-
to-aG is outlined in Scheme 2 and employs chemistry developed
by the group of Hopkins31,32 and by De Riccardis and Johnson.33
The protections of the 2′-, 3′-, and 5′-OH groups in ribose of
Cross-Link Formation via Cyanoamine. The cross-link rG-
9
882 J. AM. CHEM. SOC. VOL. 127, NO. 3, 2005

Alternative Method for G-to-G Cross-Link Formation
A R T I C L E S
Figure 2. LC-MS/MS spectra of pure synthetic rG-to-aG and cross-links formed by the bimolecular reaction of cyanoamine 4a and guanosine rG. In the
top graph is shown the MS/MS spectrum of the parent ion [(rG-to-aG) + H]+ at m/z ) 492 of the pure synthetic cross-link, with a retention time in the mass
trace of about 10 min (mass trace shown below). For the bimolecular cross-link formation, two MS/MS spectra are shown for two parent ions with that same
mass m/z ) 492, but retention times of about 8 and 10 min.
to-aG was formed by rG addition to cyanoamine 4a in DMSO
solution at 80 °C in the course of 120 min. The HPLC
chromatogram of the reaction mixture is shown in Figure 1.
The three large peaks correspond to unreacted cyanoamine 4a,
the products of intramolecular cyclization [isoguanine aIG and
guanine aG (overlapping)], and unreacted guanosine rG. The
inset shows three minor peaks, and the intensity of the central
one increased upon addition of synthetic rG-to-aG. Hence, rG-
to-aG was formed in the bimolecular reaction.
The LC-MS/MS of the [(rG-to-aG) + H]+ parent ion, m/z
) 492, of synthetic rG-to-aG is shown in the top graphs of
Figure 2. The concentration of this solution was 0.02 mM, so
that the intensity of the rG-to-aG peak in the LC chromatogram
was about the same as that for the corresponding peak in the
bimolecular reaction mixture. The fragmentation pattern is
described in Scheme 3. The m/z ) 286 typically occurs in MS/
MS spectra of dG-to-dG obtained by treating DNA with HNO2
and subsequent digestion.26,29
The LC-MS/MS analysis of the bimolecular cross-link
formation shows that two structure isomeric cross-links are
formed; there are two parent ions with m/z ) 492 from products
with different retention times. The MS/MS analysis of the parent
ions with m/z ) 492 shows that only the central one of the
three minor peaks in the HPLC chromatogram of Figure 1 causes
a significant ion concentration. The cross-link isomer detected
in the mass trace at a retention time of about 8 min had escaped
UV detection at 254 nm in LC chromatograms (e.g., Figure 1).
In Figure 8 of the Supporting Information, we provide UV and
mass traces, which show that the peak at 8 min does not
correspond to any of the peaks around 16 min in Figure 1.
The MS/MS spectrum of the reaction product with the
retention time of about 10 min (below center, Figure 2) has
essentially the same retention time as that of synthetic rG-to-
aG, and its fragmentation pattern is identical to the MS/MS
spectrum of pure synthetic rG-to-aG (top, Figure 2). There are
intensity differences in the MS/MS spectra of the pure cross-
9
J. AM. CHEM. SOC. VOL. 127, NO. 3, 2005 883

A R T I C L E S
Qian and Glaser
Scheme 3. Fragmentation of the [(rG-to-aG) + H]+ Parent Ion with m/z ) 492
Scheme 4. Formations of Cross-Links G(N2)-to-G(C2) and G(N1)-to-G(C2)
Scheme 5. Positions of the “Intact Guanine” (green) and of the
“Deaminated Guanine” (red) in the G-to-G Cross-Links,
G(N2)-to-G(C2) and G(N1)-to-G(C2)
link and the cross-link in the reaction mixture (Figure 2), and
the cause of these intensity differences is not clear. We have
recorded the MS/MS spectrum of the reaction mixture contain-
ing the cross-links formed by the bimolecular reaction of 4a
and rG after addition of some pure synthetic rG-to-aG, and
this spectrum (Figure 8 in the Supporting Information) dem-
onstrates beyond any doubt that the classical cross-link is
formed. We also recorded LC-MS/MS spectra in selective
reaction monitoring (SRM) mode, monitoring the parent ion
m/z ) 492 and the product ion m/z ) 360 to further corroborate
the identity of the synthetic rG-to-aG and the cross-link made
by the addition chemistry.
The other MS/MS spectrum corresponds to the peak with a
mass trace retention time of about 8 min. The proposed structure
of the second cross-link is shown in Scheme 4 along with a
proposal for its formation. Both amino N atoms of guanidine 5
can attack the carboxamido group, leading to either the classical
or the new rG-to-aG cross-link, depending on whether the
guanidine is in the (E)- or (Z)-configuration about the CdN
double bond, respectively. When this possibility for structure
isomerism of the cross-links is recognized, it becomes im-
mediately obvious that their fragmentation patterns are alike as
well because they lead to structure isomeric ions.
There is a need to differentiate clearly between these structure
isomers, and we suggest the following formalism. Both cross-
links formally link a guanine fragment deaminated at C2 to
either the N2 or the N1 atom of a complete guanine. Hence, we
refer to the classical cross-link as G(N2)-to-G(C2) and to its
isomer as G(N1)-to-G(C2). This is a formal nomenclature
indeed as it is clear that G(N1)-to-G(C2) is not formed by
alkylation at N1! This important point is illustrated in Scheme
5, where color is used to differentiate between the guanines
that enter the cross-link either intact or deaminated. The formal
N1 alkylation and the actual mechanisms involving addition and
cyclization result in G(N1)-to-G(C2) cross-links in which the
deaminated guanine occupy different positions. In other words,
9
884 J. AM. CHEM. SOC. VOL. 127, NO. 3, 2005

Alternative Method for G-to-G Cross-Link Formation
A R T I C L E S
Table 1. Total and Relative Energies of G-to-G Cross-Link
Structures
Relative Energy (kcal/mol)
IPCM, DMSO
B3LYP/6-31 G**//
structure isomer,
tautomer
B3LYP/6-31
+
G**//
+
B3LYP/6-31G*
B3LYP/6-31G*
B3LYP/6-31G*)
G(N2)-to-G(C2),
amide-amide
0.22
1.47
0.12
G(N1)-to-G(C2),
amide-amide
8.77
0.00
10.95
0.00
-8.72
0.00
G(N2)-to-G(C2),
amide-iminol
G(N1)-to-G(C2),
amide-iminol
14.58
14.03
11.56
Figure 3. Molecular models of the B3LYP/6-31G* structures are shown
of the classical cross-link G(N2)-to-G(C2), amide-amide (top) and amide-
iminol (bottom) tautomers, and of the structure isomer, G(N1)-to-G(C2),
again amide-amide (top) and amide-iminol (bottom) tautomers. Dipole
moments computed at B3LYP/6-31G* and B3LYP/6-31+G**//B3LYP/6-
31G* levels are given in that order.
the positions of R and R′ differ in G(N1)-to-G(C2), depending
on the mechanism of the cross-link’s formation. While there is
only one G(N2)-to-G(C2) cross-link, rG(N2)-to-aG(C2) is the
same as aG(N2)-to-rG(C2); the cross-links rG(N1)-to-aG(C2)
and aG(N1)-to-rG(C2) are isomers, and only aG(N1)-to-rG-
(C2) is formed.
classical structure. The amide-iminol tautomer is clearly
disadvantaged; while the N-H‚‚‚N3 hydrogen bond (219.2 pm)
is maintained, the distance between the carbonyl-O and the
hydroxyl group is too long (430.1 pm) for a second hydrogen
bond.
Two Reaction Pathways to G(N1)-to-G(C2). At the begin-
ning, we stated that we are interested in the cyanoamines 3 (X
) OH) and 4 (X ) NH2). Cyanoamines 3 are water adducts of
the cyanoimines 2, and cyanoamines 4 are their amine adducts.
We are studying 4 for practical reasons (vide supra) and for an
important conceptual reason which is now becoming clear:
cyanoamine 4 is the simplest model of the DNA base adduct
13! Since 4a cyclizes to guanosine41 and because of the G-
(N1)-to-G(C2) formation shown here, we have established that
there are two possible reaction channels leading to G(N1)-to-
G(C2) in DNA. Similar logic would suggest that 14, formed
by addition of guanine to the cyanoimine function of 2, might
provide a path to the classical cross-link G(N2)-to-G(C2).
However, the electronic structure of cyanoimines suggests that
nucleophilic addition to the ketene moiety is faster39 and it also
leads to aromatization. Any classical cross-link G(N2)-to-G(C2)
formed via pyrimidine ring-opening most likely passed through
a guanidine (E)-5.
Structures of G-to-G Cross-Links. Density functional theory
at the B3LYP/6-31G* level43 was employed to determine the
structures of the G-to-G cross-links. The amide-amide and
amide-iminol tautomers were considered for both structure
isomers. Relative energies are given in Table 1, and the
structures are shown in Figure 3.
The commonly drawn amide-amide tautomer of G(N2)-to-
G(C2) greatly suffers from carbonyl dipole alignment and (N1,
N1′) lone pair repulsion, and rotation about the C2-N bond
leads to the conformation with an N1-H‚‚‚N3 hydrogen bond
(197.0 pm). While this rotation is possible for the free molecule,
it is not possible in DNA, and G(N2)-to-G(C2) should greatly
prefer the amide-iminol tautomer in DNA. This tautomer
features an N1-H‚‚‚N1 hydrogen bond (198.8 pm) and a
distance of 975.1 pm between the R groups at N9. The amide-
amide tautomer of structure isomer G(N1)-to-G(C2) features
two hydrogen bonds, N-H‚‚‚N3 (194.5 pm) and N1-H‚‚‚O6
(191.5 pm), some propeller distortion, and a distance of 917.3
pm between the R groups at N9, and this tautomer is only 8.8
kcal/mol less stable than the amide-iminol tautomer of the
Structures are usually well reproduced at the B3LYP/6-31G*
level. Relative energies, on the other hand, are more dependent
on the theoretical level, and they also are solvent dependent.
Hence, we also computed relative energies with the fully
polarized and diffuse function augmented 6-31+G** basis set
for the isolated structures, B3LYP/6-31+G**//B3LYP/6-31G*,
and with implicit solvation44 effects included via the isodensity
polarizable continuum model,45 IPCM (B3LYP/6-31+G**//
B3LYP/6-31G*). In particular, these additional basis functions
allow for improved descriptions of the hydrogen bonds because
the sets of diffuse s- and p-functions improve the hydrogen bond
donor atoms O and N, and the sets of p-polarization function
should be beneficial for the H-atoms involved in hydrogen
bonding. The data in Table 1 show only modest changes of the
relative energies in the gas phase (up to 2 kcal/mol), while a
remarkable solvent effect is found on the relative stability of
the amide-amide tautomer of G(N1)-to-G(C2). These dipole
moments of the cross-links vary significantly (Figure 3), and
the stabilization by solvent roughly parallels molecular polarity.
Yet, the solvation effect on the amide-amide tautomer of
G(N1)-to-G(C2) is remarkable even in light of the fact that
this structure has the largest dipole moment among all the
isomers. The number of hydrogen bond donors and hydrogen
bond acceptors available for any specific solvation also differs
in the isomers, and in particular, the amide-amide tautomer of
G(N1)-to-G(C2) has the least number of available hydrogen
bond donors.
The major conclusions from the theoretical study are as
follows. (1) The amide-amide and the amide-iminol tautomers
of the classical G-to-G cross-link are almost isoenergetic. (2)
In DNA, the amide-iminol tautomers of the classical G-to-G
cross-link are greatly favored because the accommodation of
the amide-amide tautomer would require a 180° rotation about
(44) Cramer, C. J.; Truhlar, D. G. Chem. ReV. 1999, 99, 2161.
(45) Foresman, J. B.; Keith, T. A.; Wiberg, K. B.; Snoonian, J.; Frisch, M. J. J.
Phys. Chem. 1996, 100, 16098. (b) Pomelli, C. S.; Tomasi, J. J. Phys. Chem.
A 1997, 101, 3561.
(43) Cramer, C. J. Computational Chemistry; John Wiley & Sons: New York,
2001.
9
J. AM. CHEM. SOC. VOL. 127, NO. 3, 2005 885

A R T I C L E S
Qian and Glaser
HPLC Analysis. The HPLC analysis was carried out with a
Shimadzu system equipped with an SPD-M10Avp photodiode array
detector. For system 1, we used Waters XTerra Prep MS C18, 10 µm
10 × 250 mm column, monitoring at 254 nm, and elution at a flow
rate of 6.0 mL/min with a gradient from 3% CH3CN in 10 mM
ammonium acetate (pH 7) to 30% CH3CN for the first 17 min, followed
by an increase to 60% CH3CN over 5 min, then an increase to 70%
CH3CN in the next 5 min, then reversed to 3% CH3CN over 5 min,
and finally eluting at 3% CH3CN over the last 2 min. This system was
used for collecting the synthetic cross-link rG-to-aG. For system 2,
we used the same column, flow rate, and detection as with system 1,
but an elution gradient from 1 to 60% aqueous CH3CN over the course
of 30 min. System 2 was used for desalting. For system 3, we employed
Waters XTerra C18, 5 µm, 4.5 × 250 mm analytical column, and flow
rate of 1 mL/min, with the same gradient as that used in system 1.
System 3 was used for the analysis of the mixture of the cross-link-
forming reaction and the LC-MS analysis.
the C2-NH bond, and as discussed above, that would cause
disruption of hydrogen bonding and dipole-dipole repulsion.
Apparently, this tautomer has never been discussed in the more
than 40 year history of the cross-link, and there is every reason
to expect this amide-iminol tautomer to occur in DNA. (3)
The relative stabilities of the G-to-G structure isomers show
that there is no thermodynamic reason for the preferred
formation of the classical cross-link, and the product formation
must be under kinetic control. (4) The distances between the
N9 and N9′ atoms in the structures of the structure isomeric
cross-links are such that the formation of either cross-link would
not be impeded by constraints on the distances between the
sugars in ds-DNA. For the cross-link formation to occur in
DNA, many additional conditions need to be met.
Conclusion
Mass Spectrometric Analysis. LC-MS studies were performed
with a Thermo Finnigan TSQ7000 mass spectrometer, equipped with
an electrospray ionization (ESI) source and operated in the positive
ion mode, LC system with P4000 pump, AS3000 auto sampler, and
UV6000 photodiode array detector. To tune the mass spectrometer,
the solution of the synthetic rG-to-aG (0.02 mM, flowing at a rate of
20 µL/min) was continuously mixed with a solution of 27% aqueous
CH3CN (flowing at a rate of 1 mL/min), and the mixture was infused
into the electrospray source to determine the optimum instrument
settings. The LC-MS and LC-MS/MS analyses were carried out with
analytical system 3 (vide supra) using both full scan and selective
reaction monitoring (SRM) detection modes.
We have demonstrated the chemical viability of rG-to-aG
formation via rG addition to a synthetic cyanoamine derivative
4a. Hence, we have established an alternative to the common
hypothesis that assumes G-to-G cross-link formation to occur
by direct nucleophilic aromatic substitution of a guanosinedia-
zonium ion by a guanosine. This alternative hypothesis has the
attractive feature that all known products of nitrosative guanine
deamination can be consistently explained with the postulate
of pyrimidine ring-opening. This postulated mechanism not only
explains what is already known but also suggests that other
products and other cross-links also might be formed in DNA
deamination.
Our study suggests one possible “other product”: the structure
isomer G(N1)-to-G(C2) of the classical G(N2)-to-G(C2) cross-
link. While the formation of aG(N2)-to-rG(C2) has been
established by chemical synthesis, the structure isomer aG(N1)-
to-rG(C2) has been assigned based on its MS/MS spectrum
because this assignment is reasonable from a mechanistic
perspective and because the computational study does not
suggest any reason that would prevent its formation. Both cross-
links are structurally possible in DNA with similar distances
between the N9 sugars. Since G(N1)-to-G(C2) has not been
detected to date, this cross-link apparently is formed in low yield
in DNA deamination, if it is formed at all. Prior to the present
study, there was no reason to expect the formation of the
structure isomer and there has not been any such discussion.
Because of the results of the present study, there now is a
compelling reason to synthesize the new G-to-G cross-link, both
to confirm the tentative assignment and to begin the search for
this cross-link in DNA deamination. This case exemplifies well
the significant role of studies of reaction mechanisms to guide
searches for new kinds of chemical DNA damage, and this role
is particularly important when very small amounts of damage
are to be detected.
5′-O,3′-O,2′-O-Tris(tert-butyldimethylethylsilyl)-6-O-benzylgua-
nosine, 7. Triphenylphosphine (3.315 g, 12.64 mmol) and 5′-O,3′-O,2′-
O-tris(tert-butyldimethylethylsilyl)guanosine (3.956 g, 12.64 mmol)
were placed in a dried flask under vacuum for 15 min. After the flask
was refilled with argon, dry dioxane (109 mL) and benzyl alcohol (1.294
g, 12.64 mmol) were added via syringe. Diethylazodicarboxylate
(DEAD, 5.8 mL, 12.64 mmol) was added to the resulting suspension
slowly via syringe. After the solution became homogeneous, the reaction
mixture was stirred for an additional 2 h. Removal of the solvent in
vacuo generated a yellow oil, which was redissolved in diethyl ether
(37 mL) and left to stand at -20 °C for 1 h. The precipitated white
solid was removed by filtration and washed with cold ether. The filtrate
was concentrated in vacuo, and the residue was purified by flash
chromatography on silica gel (6% ethyl acetate/hexane) to yield 2.89
1
g of pure 7 (64%) as a white solid. H NMR (250 MHz, d6-DMSO):
δ 8.06 (s, 1H), 7.50-7.36 (m, 5H), 6.47 (s, 2H), 5.83 (d, J ) 6.9 Hz,
1H), 5.48 (s, 2H), 4.69 (dd, J ) 6.9, 4.6 Hz, 1H), 4.20 (d, J ) 3.65
Hz, 1H), 3.95-3.84 (m, 2H), 3.70 (dd, J ) 11.0, 3.7 Hz, 1H), 0.91 (s,
9H), 0.89 (s, 9 H), 0.71 (s, 9H), 0.10 (d, J ) 4.6 Hz, 6H), 0.08 (d, J
) 0.8 Hz, 6H), -0.10 (s, 3H), -0.31 (s, 3H). 13C NMR (500 MHz,
d6-DMSO): δ 160.0, 159.8, 154.6, 137.5, 136.5, 128.5, 128.4, 128.1,
113.7, 85.53, 85.48, 74.7, 72.7, 66.9, 62.8, 25.8, 25.7, 25.5, 18.0, 17.8,
17.5, -4.7, -4.78, -4.8, -5.4, -5.5. ESI-HRMS: m/e 738.38386
(M + Na+) (calcd 738.387269).
5′-O,3′-O,2′-O-Tris(tert-butyldimethylethylsilyl)-6-O-benzyl-2-
bromoguanosine, 8. Amine 7 (583 mg, 0.814 mmol) was dried by
dissolving it in anhydrous toluene (6.5 mL) and by removing the solvent
in vacuo. To the flask containing the dried amine rG was added
antimony(III) bromide (413 mg, 1.143 mmol), and the flask was then
purged with argon. Anhydrous methylene bromide (9.55 mL), cooled
to 0 °C, was added to the reaction flask which was cooled to 0 °C, as
well. After the mixture was further cooled to -10 °C, tert-butylnitrite
(376 µL, 2.85 mmol) was added slowly via syringe. After completion
of the nitrite addition, the reaction mixture was stirred for 1 h at -10
°C and then poured into 32.5 mL of a mixture of crushed ice, water,
and 1.32 g of sodium bicarbonate. The reaction product was filtered,
and the filtrate was extracted with methylene chloride. The combined
Experimental Section
General Methods. 13C and 1H NMR spectra were collected on DRX
500 and ARX 250 MHz spectrometers, using d6-DMSO and CD3Cl as
solvents. High-resolution MS studies were performed by the Mass
Spectrometry Laboratory of the Department of Chemistry, The Ohio
State University.
Materials. 1-[(2-Hydroxyethoxy)methyl]-5-cyanoamine-4-imida-
zolecarboxamide 4a was prepared as described previously by us
elsewhere.41 Other starting materials and solvents were commercially
available and were used without further purification.
9
886 J. AM. CHEM. SOC. VOL. 127, NO. 3, 2005

Alternative Method for G-to-G Cross-Link Formation
A R T I C L E S
organic extracts were dried over sodium sulfate. After filtration, the
solvent was removed in vacuo, and a yellow oil was obtained. Flash
chromatography with 3% ethyl acetate/hexane as the eluent afforded
392 mg of 8 as a white solid (61.7%). 1H NMR (500 MHz,
d6-DMSO): δ 8.59 (s, 1H), 7.52 (d, J ) 6.5 Hz, 2H), 7.43-7.33 (m,
3H), 5.92 (d, J ) 5.5 Hz, 1H), 5.59 (d, J ) 2.5 Hz, 2H), 4.88 (t, J )
5.0 Hz, 1H), 4.37 (t, J ) 3.5 Hz, 1H), 4.01-4.00 (m, 2H), 3.73 (dd, J
) 10.5, 7.5 Hz, 1H), 0.91 (s, 9H), 0.86 (s, 9H), 0.74 (s, 9H), 0.13 (s,
3H), 0.10 (s, 3H), 0.06 (d, J ) 3.0 Hz, 6H), -0.08 (s, 3H), -0.29 (s,
3H). 13C NMR (500 MHz, d6-DMSO): δ 159.8, 152.7, 143.2, 142.1,
135.4, 128.8, 128.5, 120.9, 88.1, 85.0, 73.9, 71.6, 69.1, 61.9, 25.74,
25.68, 25.4, 18.0, 17.7, 17.5, -4.6, -4.9, -5.4, -5.5, -5.6. ESI-
HRMS: m/e 801.28435 (M + Na+) (calcd 801.286881).
153.7, 153.4, 153.3, 140.7, 139.5, 136.23, 136.20, 128.7, 128.4, 128.3,
128.1, 128.0, 117.5, 117.2, 87.8, 85.4, 76.3, 73.0, 72.1, 71.7, 70.6, 68.41,
68.36, 62.6, 62.4, 26.1, 25.9, 25.7, 18.6, 18.3, 18.1, 17.9, -4.4, -4.6,
-4.7, -5.0, -5.1, -5.38, -5.40. ESI-HRMS: m/e 1150.58255 (M
+ Na+) (calcd 1150.580337).
2-N-[2-(2′-O-tert-Butyldimethylethylsilylacycloinosyl)]-5′-O,3′-
O,2′-O-tris (tert-butyldimethylethylsilyl)guanosine, 12. Compound
11 (427.3 mg, 0.378 mmol) was dissolved in 9.4 mL of THF, and 0.1
M tetrabutylammonium fluoride (TBAF) in THF solution (1.882 mL,
1.89 mmol) was added. The reaction solution was stirred for 1 h at 25
°C, then concentrated in vacuo, and the resulting residue was purified
by flash chromatography (silica gel, 8% methanol/methene chloride)
1
to give 30.7 mg product 8 (57.2%) as a white solid. H NMR (500
2′-O-(tert-Butyldimethylethylsilyl)acycloguanosine, 9. Acyclogua-
nosine (200 mg, 0.889 mmol) was added to a mixture of TBDMSCl
(294.8 mg, 1.956 mmol) and imidazole (266.19 mg, 3.91 mmol) in 4.5
mL of DMF. The resulting solution was stirred at room temperature
overnight. The solvent was removed, and the residue was purified by
flash chromatography with 7-10% MeOH/CH2Cl2 as eluent to give
332.4 mg of white product. Recrystallization from MeOH/H2O gave
MHz, d6-DMSO): δ 8.07 (s, 1H), 8.02 (s, 1H), 7.43 (d, J ) 7.5 Hz,
4H), 7.28-7.22 (m, J ) 6.1 Hz, 6H), 5.93 (d, J ) 6.5 Hz, 1H), 5.68
(d, J ) 3.0 Hz, 2H), 5.61 (d, J ) 3.5 Hz, 2H), 5.58 (d, J ) 3.0 Hz,
2H), 4.63 (t, J ) 5.2 Hz, 2H), 4.52 (s, 1H), 4.27 (q, J ) 2.1 Hz, 1H),
4.12 (d, J ) 2.5 Hz, 1H), 3.70 (d, J ) 2.8 Hz, 2H), 3.68 (d, J ) 3.0
Hz, 1H), 3.65 (s, 1H), 3.60 (d, J ) 3.0 Hz, 1H), 3.5 (s, 4H). 13C NMR
(500 MHz, d6-DMSO): δ 159.5, 153.0, 152.4, 140.9, 139.7, 135.6,
128.0, 127.9, 127.7, 127.6, 86.4, 77.5, 77.3, 77.0, 74.6, 72.6, 71.0, 70.4,
67.8, 60.4. ESI-HRMS: m/e 694.23341 (M + Na+) (calcd 694.234429).
2-N-(2-Acycloinosyl)guanosine, rG-to-aG. Ten percent palladium
on activated carbon catalyst (104.5 mg) was added to a solution of 12
(321.4 mg, 0.479 mmol) in methanol (104.5 mL). The system was
flushed with hydrogen gas three times before the reaction mixture was
hydrogenated for 24 h at 60 psi with stirring. After completion of the
reaction, the catalyst was removed by filtration through a pad of Celite.
The filtrate was concentrated in vacuo; the crude product was purified
by HPLC using analytical method A, and then desalted by HPLC
employing analytical method B to yield 110.2 mg (87%) of rG-to-aG
as a white solid. 1H NMR (500 MHz, d6-DMSO): δ 8.15 (s, 1H), 8.09
(s, 1H), 5.78 (d, J ) 5.5 Hz, 1H), 5.52 (s, 2H), 5.05 (s, 1H), 4.70 (s,
1H), 4.50 (s, 1H), 3.92 (q, J ) 4.5 Hz, 1H), 3.72 (d, J ) 10.0 Hz, 1H),
3.59 (t, J ) 4.5 Hz, 2H), 3.55 (dd, J ) 11.5, 3.5 Hz, 1H), 3.50 (s, 2H),
3.07 (q, J ) 7.0 Hz, 1H). 13C NMR (500 MHz, d6-DMSO): δ 149.0,
139.4, 137.8, 119.5, 119.0, 85.4, 73.7, 72.8, 71.0, 70.1, 61.4, 60.0. ESI-
HRMS: m/e 514.14206 (M + Na+) (calcd 514.140529).
rG-to-aG Formation by Guanosine Addition to Cyanoamine 4a.
1-[(2-Hydroxyethoxy)methyl]-5-cyanoamine-4-imidazolecarboxam-
ide (4a, 15 mg, 0.06 mmol) and guanosine (11 mg, 0.04 mmol) were
added to 250 µL of DMSO, and the solution was stirred at 80 °C for
2 h. Then, 10 µL of the resulting light-yellow solution were diluted
with 250 µL of DMSO for HPLC and LC-MS analyses using the
analytical system 3 (vide supra) and an injection volume of 10 µL.
The HPLC chromatogram of the reaction mixture showed the same
peak that was observed for synthetic rG-to-aG, and the retention time
was 16.0 min. To tune the mass spectrometer to optimal sensitivity,
the solution of synthetic rG-to-aG was diluted to 0.02 mM, and at
this concentration, the peaks of the synthetic cross-link and of the cross-
link formed by the rG addition also agreed with regard to their intensity.
1
263 mg of analytically pure product 9 in 87% yield. H NMR (500
MHz, d6-DMSO): δ 10.6 (s, 1H), 7.8 (s, 1H), 6.46 (s, 2H), 5.3 (s,
2H), 3.6 (t, J ) 4.5 Hz, 2H), 3.5 (t, J ) 5.0 Hz, 2H), 0.81 (s, 9H),
-0.023 (s, 6H). 13C NMR (500 MHz, d6-DMSO): δ 156.8, 153.8,
151.4, 137.6, 116.5, 72.1, 70.1, 61.8, 25.7, 17.9, -5.3. ESI-HRMS:
m/e 362.16214 (M + Na+) (calcd 362.161885).
2′-O-tert-Butyldimethylethylsilyl-6-O-benzylacycloguanosine, 10.
Compound 10 was prepared in analogy to the preparation of 7,
employing 9 (172 mg, 0.505 mmol), triphenylphosphine (266 mg,
1.0147 mmol), dry dioxane (8.7 mL), benzyl alcohol (105 µL, 1.0147
mmol), and diethylazodicarboxylate (464 µL, 1.0147 mmol). The crude
product was purified by preparative TLC (1.5:4 hexane/ethyl acetate)
1
to give 124.5 mg of white solid 10 (60% yield). H NMR (500 MHz,
d6-DMSO): δ 7.71 (s, 1H), 7.48 (d, J ) 7.1 Hz, 2H), 7.33-7.24 (m,
3H), 5.56 (s, 2H), 5.49 (s, 2H), 4.92 (br, 2H), 3.71 (t, J ) 4.7 Hz, 2H),
3.57 (t, J ) 5.3 Hz, 2H), 0.86 (s, 9H), 0.027 (s, 6H). 13C NMR (500
MHz, d6-DMSO): δ 161.2, 159.5, 154.3, 139.4, 136.4, 129.3, 128.4,
128.2, 128.0, 115.5, 72.9, 70.8, 68.1, 62.5, 25.9, 18.3, -5.3. ESI-
HRMS: m/e 452.20677 (M + Na+) (calcd 452.208835).
2-N-[2-(2′-O-tert-Butyldimethylethylsilyl-6-O-benzylacycloinosyl)]-
5′-O,3′-O,2′-O-tris(tert-butyldimethylethylsilyl)-6-O-benzylgua-
nosine, 11. In a dry flask, amine 10 (226.5 mg, 0.528 mmol), cesium
carbonate (240.85 mg, 0.739 mmol, 1.4 equiv), palladium acetate (11.85
mg, 0.0528 mmol, 0.1 equiv), BINAP (49.32 mg, 0.0792 mmol, 0.15
equiv), bromide 8 (494.2 mg, 0.634 mmol, 1.2 equiv), and toluene (4.6
mL) were mixed under argon at room temperature for 30 min and then
heated to 90 °C for 16 h. After the completion of the reaction, the
reaction solution was diluted with ethyl acetate. The precipitate was
removed by centrifugation, and the supernatant liquid was concentrated
in vacuo. The residue was purified by flash chromatography (silica
gel, 20-22% ethyl acetate/hexane) and afforded product 11 as a light-
1
yellow oil (427.3 mg, 71.8% yield). H NMR (500 MHz, CDCl3): δ
Acknowledgment. We are grateful for support by NIH Grant
GM61027.
7.88 (s, 1H), 7.78 (s, 1H), 7.50 (m, 4H), 7.33-7.27 (m, 6H), 6.05 (d,
J ) 5.0 Hz, 1H), 5.77 (s, 2H), 5.68 (dd, J ) 47, 12.5 Hz, 3H), 5.57
(dd, J ) 13, 10.5 Hz, 2H), 4.52 (t, J ) 4.5 Hz, 1H), 4.29 (t, J ) 3.5
Hz, 1H), 4.08 (q, J ) 2.5 Hz, 1H), 3.97 (dd, J ) 11.5, 4.0 Hz, 1H),
3.77 (dd, J ) 11.5, 2.5 Hz, 1H), 3.62 (t, J ) 4.5 Hz, 2H), 3.52 (t, J )
5.5 Hz, 2H), 0.92 (d, J ) 3.0 Hz, 18H), 0.82 (s, 9H), 0.78 (s, 9H),
0.11 (d, J ) 3.5 Hz, 6H), 0.01 (s, 6H), -0.02 (s, 6H), -0.05 (s, 3H),
-0.23 (s, 3H). 13C NMR (500 MHz, d6-DMSO): δ 160.7, 160.6, 153.8,
Supporting Information Available: 1H and 13C NMR spectra
of 7-12 and rG-to-aG, and the results of the computational
study. This material is available free of charge via the Internet
at http://pubs.acs.org.
JA045108J
9
J. AM. CHEM. SOC. VOL. 127, NO. 3, 2005 887