Contiguous metal-mediated base pairs comprising two AgI ions

Article abstract of DOI:10.1002/chem.201002944

The incorporation of transition-metal ions into nucleic acids by using metal-mediated base pairs has proved to be a promising strategy for the site-specific functionalization of these biomolecules. We report herein the formation of Ag⁺-mediated Hoogsteen-type base pairs comprising 1,3-dideaza-2′-deoxyadenosine and thymidine. By defunctionalizing the Watson-Crick edge of adenine, the formation of regular base pairs is prohibited. The additional substitution of the N3 nitrogen atom of adenine by a methine moiety increases the basicity of the exocyclic amino group. Hence, 1,3-dideazaadenine and thymine are able to incorporate two Ag⁺ ions into their Hoogsteen-type base pair (as compared with one Ag⁺ ion in base pairs with 1-deazaadenine and thymine). We show by using a combination of experimental techniques (UV and circular dichroism (CD) spectroscopies, dynamic light scattering, and mass spectrometry) that this type of base pair is compatible with different sequence contexts and can be used contiguously in DNA double helices. The most stable duplexes were observed when using a sequence containing alternating purine and pyrimidine nucleosides. Dispersion-corrected density functional theory calculations have been performed to provide insight into the structure, formation and stabilization of the twofold metalated base pair. They revealed that the metal ions within a base pair are separated by an Ag...Ag distance of about 2.88 A. The Ag-Ag interaction contributes some 16 kcal mol^-1 to the overall stability of the doubly metal-mediated base pair, with the dominant contribution to the Ag-Ag bonding resulting from a donor-acceptor interaction between silver 4d-type and 4s orbitals. These Hoogsteen-type base pairs enable a higher functionalization of nucleic acids with metal ions than previously reported metal-mediated base pairs, thereby increasing the potential of DNA-based nanotechnology. Copyright

Full text of DOI:10.1002/chem.201002944

FULL PAPER
DOI: 10.1002/chem.201002944
Contiguous Metal-Mediated Base Pairs Comprising Two AgI Ions
Dominik A. Megger,[a] Cꢀlia Fonseca Guerra,[b] Jan Hoffmann,[c] Bernhard Brutschy,[c]
F. Matthias Bickelhaupt,*[b] and Jens Mꢁller*[a]
Dedicated to Professor Bernhard Lippert on the occasion of his 65th birthday
Abstract: The incorporation of transi-
tion-metal ions into nucleic acids by
using metal-mediated base pairs has
proved to be a promising strategy for
the site-specific functionalization of
these biomolecules. We report herein
the formation of Ag+-mediated Hoog-
steen-type base pairs comprising 1,3-di-
deaza-2’-deoxyadenosine and thymi-
dine. By defunctionalizing the Watson–
Crick edge of adenine, the formation
of regular base pairs is prohibited. The
additional substitution of the N3 nitro-
gen atom of adenine by a methine
moiety increases the basicity of the
exocyclic amino group. Hence, 1,3-di-
deazaadenine and thymine are able to
incorporate two Ag+ ions into their
Hoogsteen-type base pair (as compared
with one Ag+ ion in base pairs with 1-
deazaadenine and thymine). We show
by using a combination of experimental
techniques (UV and circular dichroism
(CD) spectroscopies, dynamic light
scattering, and mass spectrometry) that
this type of base pair is compatible
with different sequence contexts and
can be used contiguously in DNA
double helices. The most stable duplex-
es were observed when using a se-
quence containing alternating purine
and pyrimidine nucleosides. Disper-
structure, formation and stabilization
of the twofold metalated base pair.
They revealed that the metal ions
within a base pair are separated by an
Ag···Ag distance of about 2.88 ꢀ. The
Ag–Ag interaction contributes some
16 kcalmolꢀ1 to the overall stability of
the doubly metal-mediated base pair,
with the dominant contribution to the
Ag–Ag bonding resulting from
a
donor–acceptor interaction between
silver 4d-type and 4s orbitals. These
Hoogsteen-type base pairs enable a
higher functionalization of nucleic
acids with metal ions than previously
reported metal-mediated base pairs,
thereby increasing the potential of
DNA-based nanotechnology.
sion-corrected
density
functional
theory calculations have been per-
formed to provide insight into the
Keywords: bioinorganic chemistry ·
DNA
·
nucleobases
·
silver
·
supramolecular chemistry
Introduction
ing interactions, but with artificial nucleobases.[1] Moreover,
numerous examples exist in which the base pairs are held
together by hydrophobic interactions[2] or coordinative
bonds to a central metal ion.[3] The latter example, that is,
the generation of metal-mediated base pairs, represents a
particularly interesting type of functionalization. It has been
envisaged that by using this method, metal-based properties,
such as redox chemistry, magnetism, conductivity, or catalyt-
ic activity, might be passed on to the modified nucleic
acid.[3b,4] Hence, this method provides an opportunity to en-
hance the scope of DNA nanotechnology even further.
Almost all metal-mediated base pairs reported to date
comprise two (either like or unlike) nucleobases coordinat-
ed to one central metal ion. Metal ions incorporated into
The incorporation of artificial base pairs into nucleic acid
double helices represents a versatile method for the func-
tionalization of these self-assembling biomolecules. The base
pairing in these surrogates may still rely on hydrogen-bond-
[a] Dr. D. A. Megger, Prof. Dr. J. Mꢁller
Institute for Inorganic and Analytical Chemistry
University of Muenster
Corrensstr. 28/30, 48149 Mꢁnster (Germany)
Fax : (+49)251-83-36007
E-mail: mueller.j@uni-muenster.de
[b] Dr. C. Fonseca Guerra, Prof. Dr. F. M. Bickelhaupt
Department of Theoretical Chemistry and
Amsterdam Center for Multiscale Modeling (ACMM)
Scheikundig Laboratorium der Vrije Universiteit
De Boelelaan 1083, 1081 HV Amsterdam (The Netherlands)
Fax : (+31)20-59-87629
DNA oligonucleotide double helices comprising such base
[5–10]
pairs include Ag+, Cu2+, Mn3+, Ni2+, Fe2+, and Hg2+
.
By carefully choosing the appropriate nucleosides, even two
different metal ions (Cu2+ and Hg2+) can be site-specifically
inserted into a DNA duplex.[11] Likewise, one metal-mediat-
ed base pair created from 5-fluorouracil has recently been
reported to contain two Ag+ ions within just one base
pair.[10b] In addition to DNA, metal-mediated base pairs
have been successfully included in other nucleic acids and
nucleic acid derivatives such as RNA,[12] PNA,[13] and
E-mail: f.m.bickelhaupt@vu.nl
[c] Dipl.-Phys. J. Hoffmann, Prof. Dr. B. Brutschy
Institut fꢁr Physikalische and Theoretische Chemie
Goethe-Universitꢂt Frankfurt
Max-von-Laue-Str. 7, 60438 Frankfurt (Germany)
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201002944.
Chem. Eur. J. 2011, 17, 6533 – 6544
ꢃ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
6533

GNA.[9a,14] In total, three metal-containing duplexes of this
type have been structurally characterized to date. The DNA
structures differ significantly and appear to depend not only
on the type of artificial base pair but also on the surround-
ing sequence of natural nucleobases. For example, the incor-
poration of two pyridine-2,6-dicarboxylate···Cu2+···pyridine
base pairs into a DNA sequence of alternating pyrimidine
and purine bases resulted in the formation of a Z-DNA
structure.[9d] In contrast, a DNA duplex comprising three
contiguous imidazole···Ag+···imidazole base pairs surround-
ed by natural A···T base pairs was shown to adopt a regular
B-DNA conformation with only minor structural devia-
tions.[5a] The structure adopted by a GNA duplex with hy-
droxypyridonate···Cu2+···hydroxypyridonate base pairs is de-
fined mainly by the acyclic backbone and appears to be only
slightly influenced by the artificial base pair.[14]
sents a major advancement compared with the rather tedi-
ous seven-step synthesis of the previously investigated 1-
deaza-2’-deoxyadenosine by which the nucleoside was ob-
tained in an overall yield of only 5%.[5d] In principle, both 1-
deazaadenine and 1,3-dideazaadenine should form exclu-
sively Hoogsteen base pairs because of the lack of an endo-
cyclic hydrogen bond acceptor in the six-membered ring.
However, different base-pairing properties have previously
been reported in the absence of Ag+. Whereas 1-deazaade-
nine forms stable Hoogsteen-type duplexes in the absence
of Ag+,[16] 1,3-dideazaadenine does not do so despite an
identical sequence.[17] It was therefore highly interesting to
study 1,3-dideazaadenine in the context of metal-mediated
base pairs and to investigate whether the different base-pair-
ing properties would also be manifested in the presence of
Ag+.
The diversity of the duplex conformations observed to
date for nucleic acids comprising metal-mediated base pairs
is very promising with respect to the development of new
base pairs of this type, because it shows a wide structural
flexibility of the metal-modified nucleic acids. For example,
we have recently shown that oligonucleotides comprising a
natural pyrimidine nucleobase (namely thymine) and a
purine-derived artificial nucleobase (namely 1-deazaade-
nine) can adopt a stable duplex structure in the presence of
Ag+ ions.[5d] Geometrical considerations suggest that these
Ag+-mediated base pairs are of the Hoogsteen type. The de-
velopment of additional metal-mediated Hoogsteen-type
base pairs represents an interesting goal, particularly in the
context of expanding the genetic code. Indeed, the natural
nucleobases cytosine and guanine have already been shown
to be able to engage in metal-mediated Hoogsteen base
pairing.[15] However, no suitable artificial purine derivative
has yet been established that selectively forms metal-medi-
ated base pairs. Attempts to use cytosine in combination
with 6-nitro-1,3-dideazapurine resulted only in the formation
of Ag+-mediated cytosine self pairs.[5b] Herein, we report on
the use of 1,3-dideaza-2’-deoxyadenosine 5 (Scheme 1) as a
nucleoside that is complementary to thymidine. This nucleo-
side has been obtained by a five-step synthesis in an overall
yield of 32% (see the Experimental Section), which repre-
Results and Discussion
Synthesis of 1,3-dideaza-2’-deoxyadenosine: The artificial
nucleoside 1,3-dideaza-2’-deoxyadenosine 5 was synthesized
from 2,6-dinitroaniline by a five-step synthesis. The first four
steps were carried out according to literature proce-
dures.[5b,18,19] The final reduction of 4 was achieved using
Raney nickel with hydrazine hydrate in methanol (see the
Supporting Information). Scheme 1 gives an overview of the
synthetic procedure. To obtain a monomer suitable for auto-
mated DNA synthesis according to the phosphoramidite
method, three additional steps were performed. The exocy-
clic amino group of 5 was FMOC-protected, and DMT and
CEDIP groups were attached to the 5’- and 3’-OH groups,
respectively.[17]
Based on compound 8, two oligonucleotide sequences I
and II were synthesized (Table 1). These sequences contain
Table 1. Oligonucleotide sequences used in this study.
Oligonucleotide
Sequence[a]
I
II
5’-d(XTX TXT XTX TXT XTX TXT)-3’
5’-d(XXX XXX XXX TTT TTT TTT)-3’
[a] X=1,3-dideazaadenine, T=thymine.
Scheme 1. Synthesis of 1,3-dideaza-2’-deoxyadenosine 5. a) Na2S, NaHCO3, H2O;[18] b) formic acid;[19] c) i) NaH, ii) Hofferꢄs chloro sugar, CH3CN;[5b]
d) NH3 (aq), CH3OH; e) Raney nickel, hydrazine hydrate, CH3OH; f) TMS-Cl, FMOC-Cl, pyridine;[17] g) DMT-Cl, pyridine;[17] h) CEDIP-Cl, CH2Cl2.[17]
FMOC=fluorenylmethoxycarbonyl, DMT=4,4’-dimethoxytrityl, CEDIP=cyanoethyl-N,N-diisopropyl phosphoramidite.
6534
www.chemeurj.org
ꢃ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 6533 – 6544

Base Pairs Mediated by Two AgI Ions
FULL PAPER
exclusively 1,3-dideazaadenine (X) and thymine (T). Se-
quence I should be able to adopt either an antiparallel- or a
parallel-stranded alignment. In the latter case, the duplex
would contain single-nucleotide overhangs at the 5’- and 3’-
ends. In contrast, sequence II is self-complementary only if
an antiparallel-stranded duplex is formed. The Ag+-binding
behavior of oligonucleotides I and II was characterized by
means of UV and circular dichroism (CD) spectroscopies,
dynamic light scattering, and mass spectrometry.
mation and dissociation of a hairpin are unimolecular and
therefore concentration-independent processes. Compared
with that of I, Tm is lower by 208C. This significant differ-
ence in duplex stability is in good agreement with the fact
that Hoogsteen-type double helices are preferentially
formed from alternating AT sequences.[20]
These UV melting studies of I and II are at variance with
previous observations for 1-deazaadenine-containing oligo-
nucleotides, for which Tm remained constant in the presence
of excess Ag+.[5d] However, a further increase of Tm upon
the addition of excess metal salt is not unprecedented and
has been observed for other metal-mediated base pairs.[5b,c]
UV melting curves: For oligonucleotide I, no cooperative
melting was observed in the absence of Ag+, which is in
agreement with earlier reports.[17] However, distinct melting
curves were obtained in the presence of Ag+ (Figure 1a),
UV titrations: Since the data presented in Figure 1 do not
allow a clear determination of the number of Ag+ ions
bound per base pair, additional titrations were performed
and monitored by UV spectroscopy. In the case of I, signifi-
cant changes in the UV spectrum were observed upon the
addition of Ag+ (Figure 2a). As these changes occurred in a
Figure 1. a) and c) UV melting curves of I and II, respectively, in the
presence of various amounts of AgNO3 (0 equiv (c), 1 equiv (b),
2 equiv (g)); b) and d) changes of Tm upon the addition of Ag+. Con-
ditions: 1 mm oligonucleotide, 150 mm NaClO4, 5 mm MOPS (pH 6.8).
Figure 2. a) UV spectra of I in the presence of various amounts of
AgNO3 (0 equiv (c), 0.5 equiv (b), 1 equiv (g), 2 equiv (d));
b) bathochromic shifts of the absorption minimum (originally at 240 nm
with the melting temperature Tm increasing smoothly upon
the addition of AgNO3 (Figure 1b). In the presence of two
equivalents of Ag+, the melting temperature of the duplex
was 828C. In this context and throughout this discussion,
one equivalent of metal ions corresponds to one metal ion
per potential metal-mediated base pair. Therefore, in the
case of an 18-mer oligonucleotide in which each base pair is
mediated by one metal ion, one equivalent corresponds to
18 metal ions.
Similar observations were also made for oligonucleotide
II (Figure 1c and d). Tm was seen to increase significantly up
to the presence of about two equivalents of AgNO3. Addi-
tion of more Ag+ led to a slightly less steep increase in Tm.
The melting temperatures of II in the presence of two equiv-
alents of Ag+ were 628C (c=1 mm) and 838C (c=3 mm).
Based on this concentration dependence of Tm, the forma-
tion of a hairpin structure can be excluded, because the for-
^
(N)) and maximum (originally at 262 nm ( )) of I; c) absorption changes
^
at 240 (N) and 262 nm ( ), respectively, upon the addition of AgNO3. In
b) and c), one and two equivalents of Ag+ are marked by broken and
dotted lines, respectively. Conditions: 5 mm oligonucleotide, 150 mm
NaClO4, 5 mm MOPS (pH 6.8).
wavelength region in which the absorption is attributed to
the nucleobases,[21] one can assume Ag+ binding to these po-
sitions rather than the phosphate backbone. Plots of the ab-
sorption changes and of the bathochromic shifts of the ab-
sorption maximum and minimum clearly demonstrate major
spectral changes upon the addition of the first two equiva-
lents of Ag+ (Figure 2b, c). Larger amounts of Ag+ led to
hardly any additional changes.
These results indicate that each metal-mediated base pair
contains two Ag+ ions. The second metal ion probably coor-
Chem. Eur. J. 2011, 17, 6533 – 6544
ꢃ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6535

J. Mꢁller, F. M. Bickelhaupt et al.
Hoogsteen-type duplexes comprising alternating purine–pyr-
imidine sequences are particularly stable.[20] Hence, the
duplex formed from sequence I probably adopted a slightly
different conformation to that formed from sequence II. It
is tempting to speculate that in the latter duplex the second
Ag+ ion is more exposed to the solvent and therefore disso-
ciates more easily than in the duplex of oligonucleotide I
(for further evidence, see the discussion of the mass spec-
tra).
Scheme 2. Potential twofold-metalated base pairs between deprotonated
thymine and a) 1,3-dideazaadenine or b) deprotonated 1,3-dideazaade-
nine (R, R’=2’-deoxyribose). Metal binding sites are numbered accord-
ing to the purine and pyrimidine nomenclature.
CD titrations: The oligonucleotides were investigated by
CD spectroscopy to gather information on any conforma-
tional changes occurring during the Ag+-mediated duplex
formation. The CD spectrum of I displays two negative
bands at 210 and 253 nm, as well as two positive ones at 227
and 280 nm. Upon the addition of Ag+, the entire spectrum
is redshifted (Figure 3a). Specifically, in the presence of two
dinates to the exocyclic functional groups of thymine and
1,3-dideazaadenine (Scheme 2). Such a twofold metalation
might be facilitated by the fact that the free electron pair of
the amino group of 1,3-dideazaadenine is not delocalized
into the aromatic ring and can therefore act as a metal-bind-
ing site. Indeed, metalation of this site has been observed
for the N9-methylated model nucleobase.[22]
Based on the experimental data, a two-step process for
base-pair formation can be anticipated. Initially, base pairs
mediated by a single Ag+ are formed. Subsequently, a
second, more weakly bound Ag+ is inserted into each
metal-mediated base pair. Such a mechanism is supported
by DFT calculations (see below). Moreover, it is evidenced
in the UV spectra by the observation that at certain wave-
lengths the coordination of the second metal ion results in
less distinct spectral changes compared with the first one
(Figure 2c, 240 nm). These different changes arise because
coordination by the exocyclic groups of the nucleobases in-
fluences the electron-absorption spectrum to a different
extent than coordination by endocyclic donor groups. Fur-
thermore, the first metalation requires the deprotonation of
thymine, which could contribute to the more significant
changes observed upon the addition of the first equivalent
of Ag+. In principle, a neutral base pair (with the exocyclic
amino group being deprotonated upon metalation) or a base
pair with a single positive charge could be formed. Howev-
er, formation of the former species appears less likely, as the
exocyclic amino group of 1,3-dideazaadenine is expected to
be relatively basic.[22] In fact, the pKa values for protonations
of N7 and N6 of 1,3-dideaza-2’-deoxyadenosine were deter-
Figure 3. a) CD spectra of I in the presence of various amounts of
AgNO3 (0 equiv (c), 1 equiv (b), 2 equiv (g), 3 equiv (d));
b) bathochromic shifts of the minimum originally at 252 nm (N) and the
+
^
maximum originally at 227 nm ( ) upon the addition of Ag ; c) change
of the CD signal at 275 nm upon the addition of Ag+; d) Job plot based
on the CD signal at 275 nm, showing a 1:1 stoichiometry between nucleo-
bases and Ag+ ions (corresponding to two Ag+ ions per base pair). For
details regarding the Job plot, see the Experimental Section. In b)–d),
one and two equivalents of Ag+ per base pair are marked by broken and
dotted lines, respectively. Conditions: 1 mm oligonucleotide (Job plot:
0.83–8.3 mm), 150 mm NaClO4, 5 mm MOPS (pH 6.8).
1
mined by H NMR spectroscopy to be 4.12ꢁ0.03 and 0.8ꢁ
0.2, respectively (Figure S6 in the Supporting Information).
Similar observations were made in the case of oligonu-
cleotide II. Here, however, the binding of the first and
second Ag+ ions per base pair resulted in slightly different
changes in the UV spectrum. Comparably large changes
were observed upon the addition of the first equivalent of
Ag+ (Figure S1a in the Supporting Information), whereas
only minor but still significant changes were detectable
upon the addition of the second equivalent (Figure S1b in
the Supporting Information). Plots of the absorption
changes and bathochromic shifts versus the equivalents of
Ag+ added showed the same results (Figure S1c and d in
the Supporting Information). The less distinct changes in
the UV spectra upon the addition of the second equivalent
of Ag+ can be attributed to the above-mentioned fact that
equivalents of Ag+, the minima are located at 212 and
275 nm, and the maxima at 233 and 306 nm. Interestingly,
the positive band at 227 nm was only redshifted up to the
addition of one equivalent of Ag+, whereas the negative
band at 252 nm shifted to longer wavelengths until two
equivalents of Ag+ had been added (Figure 3b). This is fur-
ther evidence for a consecutive (rather than concomitant)
incorporation of the two metal ions into the base pair. It is
tempting to speculate that the changes at around 227 nm in-
6536
www.chemeurj.org
ꢃ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 6533 – 6544

Base Pairs Mediated by Two AgI Ions
FULL PAPER
dicate a conformational change from random coil to duplex
upon the addition of the first equivalent of Ag+. The second
equivalent of Ag+ mainly influences the base-pair structure,
as indicated by the continuing spectral changes at around
252 nm. Upon the addition of more than two equivalents of
Ag+, only minor additional changes were observed (Fig-
ure 3c), confirming the proposed stoichiometry of the two-
fold-metalated base pairs. Similar behavior was observed for
oligonucleotide II (Figure S2 in the Supporting Informa-
tion).
Job plots were generated to provide additional proof of
the presence of two metal ions per base pair. Figure 3d
shows the Job plot based on the CD signal at 275 nm. Addi-
tional Job plots (including those for oligonucleotide II)
based on the bathochromic shifts in the CD spectrum and
the UV spectrum are given in Figure S8 (in the Supporting
Information). All of the Job plots clearly show a 1:1 stoichi-
ometry (dotted line) between the nucleobases and the Ag+
ions rather than a 2:1 stoichiometry (broken line). Hence,
with one Ag+ ion per nucleobase, two Ag+ ions must be
present per base pair, again indicating the formation of two-
fold-metalated base pairs. It is interesting to note that the
Job plot for oligonucleotide II shown in Figure S2d (in the
Supporting Information) shows two points of discontinuity,
corresponding to the incorporation of one and two Ag+ ions
per base pair, respectively. In good agreement with all of
our other experiments, this suggests that in II the singly and
doubly metalated base pairs are of similar stability, and
hence that the second Ag+ ion is weakly bound. For details
regarding the generation of the Job plots, see the Experi-
mental Section.
ing only one Ag+ per pair. To calculate theoretically the hy-
drodynamic radius for an 18-mer oligonucleotide duplex,[12]
we assumed a helical rise of 0.34 nm and a helix diameter of
2 nm, both based on the ideal B-DNA geometry. Such an
approximation is feasible because Hoogsteen-type double
helices and B-DNA have very similar geometries.[20,23] The
resulting theoretical value of 1.86 nm perfectly reproduced
the experimentally determined radii of (2.0ꢁ0.2) and (1.8ꢁ
0.1) nm, respectively. A significant increase in rH was ob-
served upon the addition of the second equivalent of Ag+,
whereas excess Ag+ did not influence rH any further. The
large increase in rH upon the addition of the second Ag+ ion
could be indicative of a change in the hydration of the DNA
duplex.[24a] For example, the Ag+ ions bound to the exocy-
clic groups of the nucleobases are expected to occupy some
of the space formerly taken by hydrogen-bonded water mol-
ecules.[25] The release of structured water molecules upon
the formation of a metal-mediated base pair has recently
been reported for the thyminate···Hg2+···thyminate base
pair.[10a] A loose association of these water molecules would
result in a larger apparent particle size.[24a] Moreover, where-
as the first Ag+ ion formally substitutes the thymine imino
proton and hence does not change the overall charge of the
base pair, the incorporation of the second Ag+ ion is most
probably not accompanied by deprotonation (see below).
Hence, positive charge accumulates along the double helix,
leading to the release of Na+ ions. In analogy to the water
molecules, the Na+ ions are expected to remain in the vicin-
ity of the oligonucleotide for electrostatic reasons, again
providing a possible explanation for the increase in rH. This
suggestion is in agreement with previous observations that
an accumulation of positive charge along the DNA helix
leads to a significantly larger hydrodynamic radius.[24b]
Dynamic light scattering: Dynamic light scattering (DLS)
was used to further investigate the proposed formation of a
double helix with metal-mediated base pairs. The results of
the DLS measurements are given in Table 2. With a polydis-
persity not exceeding 30%, all samples were mono- to
medium-disperse.
Mass spectrometry: LILBID (laser-induced liquid bead ion
desorption) mass spectrometric studies[26] were performed to
finally prove the formation of metal-containing double heli-
ces. By virtue of its low consumption of analyte and soft
laser-induced ion desorption, this recently developed tech-
nique is an adequate method for the investigation of bio-
molecules even in the presence of buffer.[27] In contrast to
the UV, CD, and DLS experiments described above, in
which only stoichiometric amounts of AgNO3 were necessa-
ry to detect duplex formation, an excess of AgNO3 (20 mm
oligonucleotide and 1 mm AgNO3) was applied during the
incubation of the samples for analysis by mass spectrometry
(for further details, see the Experimental Section).
Figure 4a shows LILBID mass spectra of oligonucleotide
I recorded under soft conditions (i.e., with low laser power).
They clearly verify that in the absence of Ag+ no duplex
was formed, since the major peak A corresponds to the
single-stranded oligonucleotide (m/zꢂ5.5 kDa). In the pres-
ence of Ag+, a different peak C was observed for the single-
stranded oligonucleotide, shifted by about 0.7 kDa with re-
spect to A. This peak can be assigned to a metalated single
strand. Furthermore, a distinct peak D at m/zꢂ13.7 kDa ap-
peared, corresponding to the mass of a metalated duplex
Table 2. Hydrodynamic radii rH (nm) and polydispersities (%) of oligo-
nucleotides I and II (c=0.3 mm) in the presence of different equivalents
of AgNO3.[a]
Equiv of Ag+
rH [nm] (polydispersity)
I
II
0
1
2
5
1.4ꢁ0.1 (18)
2.0ꢁ0.2 (19)
6.6ꢁ1.4 (22)
6.4ꢁ1.4 (26)
1.4ꢁ0.1 (13)
1.8ꢁ0.1 (24)
5.4ꢁ1.1 (30)
5.6ꢁ0.5 (26)
[a] Values are the mean averages of at least five individual measure-
ments. Error limits correspond to one standard deviation.
Within error limits of 3s, the hydrodynamic radii of the
oligonucleotides were identical in the absence and presence
of the first equivalent of Ag+. Nevertheless, when consider-
ing error limits of 1s, the small increase of about 0.5 nm ap-
pears to become significant. It could be due to the formation
of a double helix with metal-mediated base pairs incorporat-
Chem. Eur. J. 2011, 17, 6533 – 6544
ꢃ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6537

J. Mꢁller, F. M. Bickelhaupt et al.
companied by a much smaller peak B at m/zꢂ12 kDa,
which we tentatively assign to an accidentally formed binary
adduct of two single strands rather than a duplex. Moreover,
metalated single strand C and double helix D were also ob-
served in the presence of Ag+. However, the ratio between
the double and single strands was significantly lower com-
pared with that gleaned from the mass spectrum of I. Fur-
ther aggregation leading to higher aggregates was not appar-
ent from the mass spectra of II. Spectra recorded at higher
laser intensity, that is, under harsher conditions, indicated
that the Ag+–DNA adducts formed by I were more stable
than those formed by II, as the former remained detectable,
whereas in the case of the latter the Ag+ ions were stripped
off (Figure S3 in the Supporting Information). This observa-
tion is in good agreement with the higher stability of I indi-
cated by the UV melting experiments. Moreover, a compari-
son of different laser energies in the case of II (Figure S3 in
the Supporting Information) clearly showed that peak D
indeed corresponded to a duplex based on metal-mediated
base pairs. Upon removal of the Ag+ ions by applying in-
creased laser intensity, the duplex loses its structural integri-
ty and dissociates into single strands rather than forming a
nonmetalated duplex. Hence, the Ag+ ions are an integral
part of the double helix. In summary, even though the mass
spectrometric studies do not resolve precisely how many
Ag+ ions are bonded to the oligonucleotide, they clearly
verify that both I and II form double helices only in the
presence of Ag+.
Figure 4. LILBID mass spectra of a) I and b) II. For an explanation of
the letters A–E, see text. Black bars indicate the theoretically expected
mass of the single strand and the resulting masses of hypothetical dimeric
and trimeric aggregates, respectively. Dotted lines represent the peak
maxima of the nonmetalated species used to determine the metal-in-
duced mass shift. All spectra were recorded using the same laser energy.
DFT calculations: The Hoogsteen-type base pairs between
thymine and 1,3-dideazaadenine were also examined from a
theoretical point of view. In particular, dispersion-corrected
DFT calculations were performed to determine whether a
base pair containing one, two, or no metal ions is energeti-
cally favored. In total, four different base pairs were consid-
ered (Scheme 3).
The optimized geometries of the artificial base pairs bp1–
bp4 are shown in Figure 5. Each base pair displays an
almost coplanar orientation of the nucleobases, apart from
the gas-phase and solution structures of bp3. The increased
distortion of bp3 is most likely caused by the sp3 hybridiza-
(Dmꢂ1.8 kDa compared with a hypothetical nonmetalated
duplex). The mass shift Dm of about 1.8 kDa suggested that
the duplex contained about 17 metal ions, corresponding to
approximately one metal ion per base pair. For a duplex
containing 18 twofold-metalated base pairs, a mass shift Dm
of about 3.9 kDa would have been expected.
However, the metal ions coordinated by the exocyclic
functional groups are less strongly bound than those coordi-
nated by the endocyclic donor groups (see below for compu-
tational results). Hence, it is not necessarily surprising that
they are not detectable under the experimental conditions.
Surprisingly, a small peak E at m/zꢂ20.2 kDa appeared in
the presence of Ag+, which we tentatively assign to three
oligonucleotide strands and 22 Ag+ ions. The structure of
the aggregate giving rise to this peak is not known. It might
have formed as a result of the unusually high Ag+ concen-
tration used during sample preparation for the mass spectro-
metric studies. In fact, UV melting experiments on I in the
presence of a large excess of Ag+ also indicated the exis-
tence of an additional species (Figure S4 in the Supporting
Information). The mass spectra of II (Figure 4b) are similar
to those of I. A peak A for the single-stranded oligonucleo-
tide was detected in the absence of Ag+ (m/zꢂ5.5 kDa), ac-
Scheme 3. Possible Hoogsteen-type base pairs between thymine and 1,3-
dideazaadenine in the absence and presence of Ag+.
6538
www.chemeurj.org
ꢃ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 6533 – 6544

Base Pairs Mediated by Two AgI Ions
FULL PAPER
Table 3. Bond analysis for bp3 and bp3*: energy decomposition analysis
(in kcalmolꢀ1).[a]
bp3
bp3*
DEPauli
DVelst
DEoi
DEdisp
DEint
150.2
ꢀ150.3
ꢀ73.5
ꢀ10.1
ꢀ83.7
G
21.2
ꢀ12.9
ꢀ17.5
ꢀ7.0
ACHTUNGTRENNUNG
E
ACHTUNGTRENNUNG
E
ACHTUNGTRENNUNG
R
ACHTUNGTRENNUNG
A
ꢀ16.2
ACHTUNGTRENNUNG
[a] Computed at the ZORA-BLYP-D/TZ2P//ZORA-BLYP-D/TZ2P
level (in parentheses: non-relativistic values at BLYP-D/TZ2P//ZORA-
BLYP-D/TZ2P). See Experimental Section for
energy decomposition analysis (EDA).
a description of the
the interaction between the two frozen fragments is weak-
ened, from ꢀ83.7 to ꢀ16.2 kcalmolꢀ1 (Table 3). This weak-
ening reflects the interruption of the N-Ag-N bridges. This
interruption is reflected by concomitant increases in the par-
tial positive charges on the silver cations from +0.24 and
+0.26 a.u. to +0.36 and 0.41 a.u., respectively, according to
the VDD scheme (Table 4). Interestingly, however, the Ag–
Figure 5. Top and side views of the geometry-optimized structures of
bp1–bp4 in water.
Table 4. Bond analysis for bp3 and bp3*: VDD atomic charges (in a.u.)
of Ag in the different fragments.[a]
Separate fragments[b]
bp3
bp3*
tion of the exocyclic amino group. Accordingly, bp4 with an
sp2-hybridized amide as the metal-binding site is almost
planar. Interestingly, those metal-mediated base pairs con-
taining two metal ions display Ag···Ag distances of about
2.88 ꢀ, far below the sum of their van der Waals radii
(3.44 ꢀ),[28] suggesting “argentophilic d10–d10 interactions”
between the Ag+ ions.[29a] In fact, an even shorter Ag···Ag
distance of 2.805(4) ꢀ was previously observed by single-
crystal X-ray diffraction analysis.[29b]
Argentophilic interactions are often accompanied by the
emergence of fluorescent properties, although no definite
correlation has yet been found between the presence or ab-
sence of closed-shell interactions and optical properties.[29a]
In the presence of two equivalents of Ag+, oligonucleotides
I and II exhibit fluorescence at 475 and 454 nm, respectively,
upon excitation at 270 nm (Figure S7 in the Supporting In-
formation). In the absence of either Ag+ or the oligonucleo-
tides, no fluorescence is observed. Hence, the fluorescent
properties support the presence of an argentophilic interac-
tion as a result of the formation of metal-mediated base
pairs, as suggested by the computational analysis.
We have investigated the nature of the interaction be-
tween the two silver centers in our base pair in more detail,
by computing the interaction between the thyminate–Ag
and dideazaadenine–Ag+ fragments in the equilibrium
structure bp3 and in a deformed variant bp3*, in which (pro-
ceeding from an otherwise frozen bp3 structure) thymine
was rotated by 908 about the N3–Ag axis and dideazaade-
nine by 1808 about the Ag–N6 axis.[30] In the resulting bp3*
structure (Figure S5 in the Supporting Information), the two
N-Ag-N bridges no longer exist and the only contact be-
tween the thyminate–Ag and dideazaadenine–Ag+ frag-
ments is an Ag–Ag interaction. On going from bp3 to bp3*,
ꢀ
N3 Ag
0.407
0.484
0.240
0.263
0.357
0.411
+
ꢀ
N6 Ag
[a] Computed at the ZORA-BLYP-D/TZ2P//ZORA-BLYP-D/TZ2P
level. [b] Separate fragments thyminate–Ag and dideazaadenine–Ag+ as
they occur in both bp3 and bp3*.
Ag interaction contributes some 16 kcalmolꢀ1 to the overall
stability of the doubly-metal-mediated base pair. As can be
seen in Table 3, the Ag–Ag interaction in bp3* is little af-
fected by switching off relativistic effects in the computa-
tions, which causes a weakening of only 1 kcalmolꢀ1. In
comparison, in the base pair bp3, the N-Ag-N silver bridges
are destabilized by a larger amount, namely 9 kcalmolꢀ1, if
relativistic effects are not taken into account. We therefore
conclude that the Ag–Ag interaction is not caused by an
onset of relativistic effects.
Further analyses of the orbital electronic structure in con-
junction with energy decomposition analyses[31] showed that
the dominant contribution to the Ag–Ag bonding in bp3* is
a donor–acceptor interaction between occupied silver 4d-
type orbitals and low-lying, empty 4s orbitals. Note that the
orbital interaction component (ꢀ17.5 kcalmolꢀ1) is more
stabilizing in bp3* than the electrostatic term (ꢀ12.9 kcal
molꢀ1). However, note also that the latter is still stabilizing,
despite the partial positive charge on each silver center. This
illustrates that point-charge models cannot be used to repro-
duce the electrostatic interaction as soon as the charge dis-
tributions of two fragments begin to overlap. The Ag–Ag
donor–acceptor interactions do not substantially reduce the
partial positive charges on the silver centers, which decrease
from +0.41 and +0.48 a.u. in the separate fragments to
+0.36 and +0.41 a.u., respectively, in bp3*.
Chem. Eur. J. 2011, 17, 6533 – 6544
ꢃ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6539

J. Mꢁller, F. M. Bickelhaupt et al.
To determine which of these base pairs is formed prefer-
entially, the reaction energies DEreac for base-pair formation
were calculated from the total bonding energies of the prod-
ucts EP and the starting compounds of the base-pair forma-
tion ES according to Equation (1).
with reaction energies of ꢀ8.9 and ꢀ9.1 kcalmolꢀ1, respec-
tively. Similar to the gas-phase situation, the incorporation
of the second Ag+ ion results in an energy gain (ꢀ33.2 kcal
molꢀ1). A subsequent deprotonation of the exocyclic amino
group is unfavorable. Hence, in solution the formation of
the twofold-metalated base pair bp3 is favored.
Additional calculations were performed considering mi-
crosolvation (Figure 6b). Here, the total bonding energies of
DEreac ¼ SEPꢀSES
ð1Þ
geometry-optimized H2O, AgACHTNUGRTNEUNG
(H2O)+, and H3O+ were used
As can be seen from Figure 6a, the formation of the non-
metalated base pair bp1 is favored in the gas phase, because
DEreac amounts to ꢀ16.0 kcalmolꢀ1. A metalation that re-
during the calculations of DEreac instead of Ag+ and H+.
Such an inclusion of explicit water molecules that may act
as either proton acceptors or aqua ligands is expected to de-
scribe the energetic devolution of the base-pair formation
more precisely because, in this approach, the occurrence of
charge transfer (donor–acceptor orbital) interactions be-
tween the solute and one of the solvent molecules is includ-
ed explicitly. In fact, the experimentally observed trends are
even reproduced in the gas phase by this calculation, since
the formation of bp3 (DEreac =ꢀ63.1 kcalmolꢀ1) also be-
comes preferred. This stems from the fact that the substitu-
tion of the thymine imino proton by Ag+, as occurs during
the formation of bp2, is much less endothermic in solution
than in the gas phase when microsolvation is not considered.
Nonetheless, the deprotonation of the exocyclic amino
group of 1,3-dideazaadenine, which is necessary for the for-
mation of bp4, remains unfavorable. Similar trends were ob-
served in solution, with the exception of the reaction ener-
gies of the twofold-metalated base pair bp3, which differ sig-
nificantly. However, it is obvious that the inclusion of micro-
solvation is useful for the correct description of metal com-
plexes with anionic ligands that are formed by the
substitution of a proton by a metal ion.
To compare the overall stabilities of the artificial base
pairs, the energies DErel relative to separate bases and silver
cations were calculated according to Equation (2) from the
total bonding energies of the base pairs EBP and their re-
spective monomeric subunits EM. In the course of the calcu-
lations, the geometries of the monomeric subunits were also
fully optimized.
DErel ¼ EBPꢀSEM
ð2Þ
A comparison of DErel can be used to estimate the
strength of the Ag+–N interactions within the twofold-meta-
lated base pair. Accordingly, the metal ion coordinated by
N3 of thyminate and N7 of 1,3-dideazaadenine is bonded
more strongly than the one coordinated by the exocyclic
functional groups (Table 5). The increase in the overall
base-pair stability resulting from the coordination of a
Figure 6. Energy diagrams showing a) the total reaction energies DEreac
(kcalmolꢀ1) for the formation of bp1–bp4 in the gas phase (black) and in
water (grey), and b) the corresponding DEreac values (kcalmolꢀ1) ob-
tained under consideration of microsolvation.
quires the substitution of a proton by Ag+ seems to be un-
favorable, as shown by the highly positive DEreac values for
bp2 and bp4. In contrast, the formation of the coordinative
Table 5. Total interaction energies DErel (kcalmolꢀ1) of bp2 and bp3 in
the gas phase and in water.
ꢀ
ꢀ
bonds O4 Ag and N6 Ag in bp3 results in an energy gain
of ꢀ97.6 kcalmolꢀ1 compared with bp2. These data suggest
that bp1 will be preferentially formed in the gas phase. The
results obtained in solution are in better agreement with the
experimentally observed trends. Here, the non- and mono-
metalated base pairs bp1 and bp2 are almost isoenergetic,
DErel
bp2
bp3
gas phase
water
ꢀ215.0
ꢀ55.5
ꢀ312.5
ꢀ88.7
6540
www.chemeurj.org
ꢃ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 6533 – 6544

Base Pairs Mediated by Two AgI Ions
FULL PAPER
Scheme 4. Reaction paths of base-pair formation including the partial reaction energies DEreac(P) (kcalmolꢀ1) in water.
second Ag+ amounts to approximately one-half of that re-
sulting from the coordination by N7 and N3. The higher sta-
bilization in the case of the first Ag+ is probably due to the
fact that thyminate acts as an anionic ligand. These compu-
tational results are in good agreement with the mass spec-
trometric data, which suggest that one of the metal ions in
the base pair is weakly bonded and therefore not detectable
under the experimental conditions.
It should be noted that all of our calculations have been
performed on the basis of individual base pairs. In the con-
text of an entire DNA duplex, slightly different geometries
might arise due to the more hydrophobic environment
within a base pair stack. Moreover, the possibility that
neighboring nucleosides might serve as additional ligands
for the Ag+ ions cannot be ruled out completely. Such a co-
ordination pattern has previously been observed in the con-
text of pyridine-2,6-dicarboxylate···Cu2+···pyridine base
pairs[9d] as well as in calculated structures of DNA duplexes
comprising either hydroxypyridonate···Cu2+···hydroxypyrido-
nate base pairs or salen···Cu2+ base pairs.[32] In fact, the ob-
servations that oligonucleotide I forms significantly more
stable duplexes than oligonucleotide II and that it binds the
Ag+ ions more tightly might be the result of such an addi-
tional coordination in the case of the alternating purine-pyr-
imidine sequence in I.
A reaction scheme indicating DEreac(P) of each reaction
step is shown in Scheme 4. Three reaction paths are possible
for the formation of bp2. One of these involves the initial
formation of bp1 and a concomitant metalation. In the
other two cases, a metal complex of one of the nucleobases
is formed prior to formation of the base pair. The latter two
reaction paths seem to be unlikely, because the whole reac-
tion is only exothermic along all elementary steps when it
proceeds in water via bp1, with a DEreac(P) of ꢀ8.9 and
ꢀ0.2 kcalmolꢀ1 for base pairing and metalation, respective-
ly. In contrast, high-energy steps of +18.3 and +13.1 kcal
molꢀ1 are expected if a metal complex with a single nucleo-
base is formed first. Hence, prior to the formation of a
metalated base pair a nonmetalated one is probably formed,
even though stable Hoogsteen-type duplexes of the non-
metalated type have not been observed experimentally.[17]
However, the intermediate formation of a metalated nucleo-
base prior to base-pair formation cannot be ruled out com-
pletely, as both reaction pathways (i.e., via metalated thymi-
nate or via metalated 1,3-dideazaadenine) involve reaction
steps with a highly negative DEreac(P) (ꢀ27.4 and ꢀ22.2 kcal
molꢀ1, respectively).
Conclusion
Mechanism of base-pair formation: To gain initial insight
into the mechanistic details of the process of the formation
of a metal-mediated base pair, the reaction energies DEreac
were divided into partial reaction energies DEreac(P), which
are calculated as the energy of the product relative to the
energy of the preceding intermediate base pair (instead of
relative to the separate monomeric subunits plus silver ions)
as defined in Equation (3).
The oligonucleotides d(XT)9 (I) and dACTHNGUTERN(NUG X9T9) (II), containing
1,3-dideazaadenine, have been found to form stable double
helices in the presence of Ag+. Duplex formation has been
confirmed by LILBID mass spectrometry, UV and CD spec-
troscopies, as well as dynamic light scattering. Most interest-
ingly, the data indicate the formation of a base pair contain-
ing two metal ions. This is in contrast to previous experi-
ments with 1-deazaadenine, which only binds one Ag+ per
metal-mediated base pair with deprotonated thymine,[5d] and
is most likely attributable to the increased basicity of the
exocyclic amino group in 1,3-dideazaadenine. Dispersion-
DEreacðPÞ ¼ EBPꢀEprecedingBP
ð3Þ
Chem. Eur. J. 2011, 17, 6533 – 6544
ꢃ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6541

J. Mꢁller, F. M. Bickelhaupt et al.
biguously a 0.06 ratio from a 0.11 ratio, a different approach has been
used to obtain interpretable Job plots. Instead of defining one duplex as
one of the components, we have defined one nucleobase as this compo-
nent. Hence, a 1:1 stoichiometry in the Job plot (one Ag+ ion per nucleo-
base) corresponds to two Ag+ ions per base pair. Accordingly, the incor-
poration of only one Ag+ ion per base pair would be represented by a
2:1 stoichiometry in the Job plot (two nucleobases per Ag+ correspond-
ing to one Ag+ per base pair). For example, for the Job plot shown in
Figure 3d, the total concentration was 150 mm. Hence, at cnucleobase =1, the
nucleobase concentration was 150 mm (corresponding to 8.3 mm oligonu-
corrected DFT calculations have corroborated these experi-
mental findings, revealing the favored formation of a two-
fold-metalated base pair bearing one single positive charge.
Bond analyses have indicated a substantial “argentophilic”
Ag-d10–Ag-d10 interaction of about 16 kcalmolꢀ1 caused by
charge transfer from silver 4d to 4s orbitals, which contrib-
utes to the stability of doubly metal-mediated base pairs
such as bp3.
The observations presented herein allow important con-
clusions to be drawn regarding the generation and use of
metal-mediated base pairs in the context of nucleic acid
functionalization. 1) Subtle changes of a nucleobase in-
volved in metal-mediated base pairing may have far-reach-
ing consequences with respect to the metal-binding proper-
ties. 2) The oligonucleotide sequence can also play a signifi-
cant role in determining the stability of the metal-mediated
base pair. 3) The possibility of incorporating two metal ions
into one base pair represents a significant step forward to-
wards obtaining highly functionalized nucleic acids.
4) Mechanistic insights gained from calculations suggest that
the metal ions are inserted in one of the late steps of base-
pair formation. Interestingly, even though Hoogsteen du-
plexes comprising nonmetalated 1,3-dideazaadenine–thy-
mine base pairs are not stable, this base pair might still be
involved as an intermediate during the formation of its
Ag+-mediated analogue.
cleotide single strand or 4.2 mm double helix). Accordingly, at cnucleobase
=
0.1, the nucleobase concentration was 15 mm (corresponding to 0.83 mm
oligonucleotide single strand or 0.42 mm double helix), and the Ag+ con-
centration was 135 mm.
DLS experiments were performed on a DynaPro Titan instrument from
Wyatt Technology. The oligonucleotide samples (c=0.3 mm, 150 mm
NaClO4, 5 mm MOPS, pH 6.8) were incubated in the presence of the re-
spective amounts of AgNO3 by heating to 908C and then slowly cooled
to 108C (cooling rate: 0.5 Kminꢀ1). The samples were then centrifuged at
13500 rpm for at least 2 h. The supernatant was transferred to a 12 mL
cuvette and measured.
LILBID experiments were performed on an in-house-constructed time-
of-flight mass spectrometer (TOF-MS) of the Wiley–McLaren type. De-
tails of this method have been published previously.[33] Briefly, the ion
source contains a commercial droplet dispenser (Microdrop), which in-
jects on demand tiny micro droplets (Øꢂ50 mm; Vꢂ65 pL) from
300 Torr through
a pressure-reduction aperture into high vacuum
(10ꢀ6 Torr). There, they are individually irradiated by high-intensity mid-
IR laser pulses (lꢂ3 mm) generated by an in-house-constructed
Nd:YAG-pumped LiNbO3 optical parametric oscillator. Due to the ab-
sorption of the laser light by the solvent, beyond a certain intensity
threshold the droplets are disrupted, ejecting preformed analyte ions into
the vacuum. For the LILBID MS experiments, oligonucleotide samples
(20 mm oligonucleotide, 0 or 1 mm AgNO3 in 100 mm NH4HCO3/1 mm
Mg
ACHTUNGTRENNUNG
Experimental Section
buffer exchange to 20 mm NH4HCO3/1 mm MgACHTUNGTRENNNUG
All calculations were performed using dispersion-corrected density func-
tional theory (DFT-D) as implemented in the Amsterdam Density Func-
tional program (ADF 2009.1).[34] Geometry optimizations were per-
formed using analytical gradient techniques and without any symmetry
or geometrical constraints. All geometries were verified as true energy
minima, having no imaginary frequencies, by means of vibrational analy-
ses. The dispersion-corrected GGA functional BLYP-D,[35] which has
proved to be adequate for calculations of DNA base dimers,[36] was used
with the doubly polarized, triple-z-quality basis set TZ2P.[37]
In BLYP-D, the BLYP functional[35c,d] is augmented with an empirical cor-
rection for long-range dispersion effects, described by a sum of damped
interatomic potentials of the form C6Rꢀ6 added to the usual DFT ener-
gy.[35a,d] The TZ2P basis set used herein is a large uncontracted set of
Slater-type orbitals (STOs) containing diffuse functions (no Gaussian
functions are involved).[34] The basis set is of triple-z quality for all atoms
and is augmented with two sets of polarization functions, that is, 3d and
4f on C, N, O and 2p, 3d on H. The 1s core shells of C, N, and O were
treated by the frozen-core approximation. An auxiliary set of s, p, d, f,
and g STOs was used to fit the molecular density and to accurately repre-
sent the Coulomb and exchange potentials in each self-consistent field
cycle. The basis set superposition error (BSSE) in the bond energy has
previously been shown to be only 1 kcalmolꢀ1 or less in GGA calcula-
tions involving the TZ2P basis set. Moreover, the dispersion correction in
the present calculations has been developed such that these small BSSE
effects are absorbed into the empirical potential.[35a] Therefore, no explic-
it counterpoise corrections had to be carried out. In addition, scalar rela-
tivistic effects were taken into account by using the zeroth-order regula-
tor approximation (ZORA).[38]
1,3-Dideaza-2’-deoxyadenosine 5 and its FMOC-, DMT-, and CEDIP-
protected analogue 8 were synthesized according to (partially modified)
previously published procedures.[5b,17–19] DNA syntheses were performed
in the DMT-off mode on a K&A Laborgerꢂte H8 DNA/RNA synthesizer
following standard protocols (except for a threefold increase in coupling
time for the artificial phosphoramidite). The oligonucleotides were
cleaved from the solid support and deprotected by exposing them to tert-
butylamine/methanol/water (1:2:1) for 3 h at 658C. They were purified by
polyacrylamide gel electrophoresis under denaturating conditions (gel so-
lution: 7m urea, 1m TBE buffer; 8–18% polyacrylamide/bisacrylamide
(29:1); loading buffer: 11.8m urea; 42 mm Tris·HCl, pH 7.5; 0.83 mm
EDTA, pH 8.0; 8% sucrose). After purification, the oligonucleotides
were desalted by passage through NAP 10 columns. The desalted oligo-
nucleotides were identified by MALDI-TOF mass spectrometry (I: calcd
for [M+H]+: 5475 Da; found: 5476 Da; II: calcd for [M+H]+: 5475 Da;
found: 5474 Da). MALDI-TOF mass spectra were recorded on a Bruker
Reflex IV instrument using a 3-hydroxypicolinic acid/ammonium citrate
matrix. During quantification of the oligonucleotides, a molar extinction
coefficient e260 =7.5 cm2 mmolꢀ1 was used for 5.
UV/Vis spectra were recorded on a Varian CARY BIO 100 spectropho-
tometer. CD spectra were recorded at 108C on a Jasco J-815 spectrome-
ter. UV melting curves were recorded at 260 nm with a heating rate of
1 Kminꢀ1 and a data interval of 0.2 K. Melting temperatures were deter-
mined as the maxima of the first derivatives of the melting curves. Prior
to each measurement, the samples were equilibrated by heating to 908C
and then slowly cooling to 108C (cooling rate: 0.5 Kminꢀ1).
For the generation of a Job plot, the sum of the molar fractions of both
components needs to be constant during all measurements. When using a
double helix (with 18 base pairs) as one component and an Ag+ ion as
the other component, a 1:18 stoichiometry (one Ag+ ion per base pair)
would have to be distinguished from a 1:36 stoichiometry (two Ag+ ions
per base pair). As it would be almost impossible to differentiate unam-
For calculations in water, solvent effects were estimated using the con-
ductor-like screening model (COSMO).[39] In the COSMO calculations, a
corrected ion radius of Ag+ and a corrected solvation energy of H+ were
used. Parameterization was accomplished by performing
a series of
6542
www.chemeurj.org
ꢃ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 6533 – 6544

Base Pairs Mediated by Two AgI Ions
FULL PAPER
single-point calculations on Ag+ using COSMO with different radii. The
radius resulting in a solvation energy that matched the experimentally
d) G. H. Clever, K. Polborn, T. Carell, Angew. Chem. 2005, 117,
7370–7374; Angew. Chem. Int. Ed. 2005, 44, 7204–7208.
observed value was subsequently chosen for all COSMO calculations
[8] a) B. D. Heuberger, D. Shin, C. Switzer, Org. Lett. 2008, 10, 1091–
1094; b) D. Shin, C. Switzer, Chem. Commun. 2007, 4401–4403;
c) C. Switzer, D. Shin, Chem. Commun. 2005, 1342–1344; d) C.
Switzer, S. Sinha, P. H. Kim, B. D. Heuberger, Angew. Chem. 2005,
117, 1553–1556; Angew. Chem. Int. Ed. 2005, 44, 1529–1532.
[9] a) L. Zhang, E. Meggers, J. Am. Chem. Soc. 2005, 127, 74–75; b) N.
Zimmermann, E. Meggers, P. G. Schultz, Bioorg. Chem. 2004, 32,
13–25; c) N. Zimmermann, E. Meggers, P. G. Schultz, J. Am. Chem.
Soc. 2002, 124, 13684–13685; d) S. Atwell, E. Meggers, G. Spraggon,
P. G. Schultz, J. Am. Chem. Soc. 2001, 123, 12364–12367; e) E. Meg-
gers, P. L. Holland, W. B. Tolman, F. E. Romesberg, P. G. Schultz, J.
Am. Chem. Soc. 2000, 122, 10714–10715.
[10] a) H. Torigoe, A. Ono, T. Kozasa, Chem. Eur. J. 2010, 16, 13218–
13225; b) I. Okamoto, K. Iwamoto, Y. Watanabe, Y. Miyake, A.
Ono, Angew. Chem. 2009, 121, 1676–1679; Angew. Chem. Int. Ed.
2009, 48, 1648–1651; c) A. Ono, S. Cao, H. Togashi, M. Tashiro, T.
Fujimoto, T. Machinami, S. Oda, Y. Miyake, I. Okamoto, Y. Tanaka,
Chem. Commun. 2008, 4825–4827; d) Y. Tanaka, S. Oda, H. Yama-
guchi, Y. Kondo, C. Kojima, A. Ono, J. Am. Chem. Soc. 2007, 129,
244–245; e) Y. Miyake, H. Togashi, M. Tashiro, H. Yamaguchi, S.
Oda, M. Kudo, Y. Tanaka, Y. Kondo, R. Sawa, T. Fujimoto, T.
Machinami, A. Ono, J. Am. Chem. Soc. 2006, 128, 2172–2173; f) A.
Ono, H. Togashi, Angew. Chem. 2004, 116, 4400–4402; Angew.
Chem. Int. Ed. 2004, 43, 4300–4302.
[40]
(DG
G
;
DGsolvACHNUTGTRENNUNG
(calcd)=ꢀ102.6 kcalmolꢀ1 for
r=1.670 ꢀ). For isolated H+, the experimental solvation energy
(ꢀ250.8 kcalmolꢀ1
)
was used directly in the evaluation of the corrected
[40]
total bonding energy.
Analyses of the base-pairing interactions were carried out in the frame-
work of Kohn–Sham DFT using the Kohn–Sham molecular orbital (MO)
model in conjunction with a quantitative energy decomposition analysis
(EDA).[30,31a] The interaction energy is decomposed into four physically
meaningful terms: DEint =DVelst +DEPauli +DEoi +DEdisp. The term DVelst
corresponds to the classical electrostatic interaction between the unper-
turbed charge distributions of the deformed bases and is usually attrac-
tive. The Pauli repulsion DEPauli comprises the destabilizing interactions
between occupied orbitals and is associated with steric repulsion. The or-
bital interaction DEoi accounts for charge transfer and polarization. The
term DEdisp accounts for the long-range dispersion effects. For a more de-
tailed explanation of the energy terms, see references [30,31a].
Acknowledgements
Generous financial support by the Deutsche Forschungsgemeinschaft
(MU 1750/2–1, IRTG 1444), the National Research School Combination–
Catalysis (NRSC-C), and the Netherlands Organization for Scientific Re-
search (NWO-CW and NWO-NCF) is gratefully acknowledged. BB ac-
knowledges the support of the Cluster of Excellence Frankfurt (CEF)
“Macromolecular Complexes” and of the Centre for Membrane Proteo-
mics (CMP). The authors thank Eva Freisinger for access to the DLS in-
strument and Nicole Megger for performing the DLS measurements. We
thank Luisa De Cola for access to the fluorescence spectrophotometer
and Cristian Strassert for assisting in the fluorescence measurements.
[11] K. Tanaka, G. H. Clever, Y. Takezawa, Y. Yamada, C. Kaul, M.
Shionoya, T. Carell, Nat. Nanotechnol. 2006, 1, 190–194.
[12] S. Johannsen, S. Paulus, N. Dꢁpre, J. Mꢁller, R. K. O. Sigel, J. Inorg.
Biochem. 2008, 102, 1141–1151.
[13] a) R. M. Franzini, R. M. Watson, G. K. Patra, R. M. Breece, D. L.
Tierney, M. P. Hendrich, C. Achim, Inorg. Chem. 2006, 45, 9798–
9811; b) R. M. Watson, Y. A. Skorik, G. K. Patra, C. Achim, J. Am.
Chem. Soc. 2005, 127, 14628–14639; c) D.-L. Popescu, T. J. Parolin,
C. Achim, J. Am. Chem. Soc. 2003, 125, 6354–6355; d) A. Kꢁsel, J.
Zhang, M. AlvariÇo Gil, A. C. Stꢁckl, W. Meyer-Klaucke, F. Meyer,
U. Diedrichsen, Eur. J. Inorg. Chem. 2005, 4317–4324; e) K. Ohr,
R. L. McLaughlin, M. E. Williams, Inorg. Chem. 2007, 46, 965–974;
f) B. P. Gilmartin, K. Ohr, R. L. McLaughlin, R. Koerner, M. E. Wil-
liams, J. Am. Chem. Soc. 2005, 127, 9546–9555.
[14] a) M. K. Schlegel, L. Zhang, N. Pagano, E. Meggers, Org. Biomol.
Chem. 2009, 7, 476–482; b) M. K. Schlegel, L.-O. Essen, E. Meggers,
J. Am. Chem. Soc. 2008, 130, 8158–8159.
[15] a) T. Ihara, T. Ishii, N. Araki, A. W. Wilson, A. Jyo, J. Am. Chem.
Soc. 2009, 131, 3826–3827; b) M. G. Santangelo, P. M. Antoni, B.
Spingler, G. Jeschke, ChemPhysChem 2010, 11, 599–606.
[1] a) S. A. Benner, Acc. Chem. Res. 2004, 37, 784–797; b) A. T. Krueg-
er, H. Lu, A. H. F. Lee, E. T. Kool, Acc. Chem. Res. 2007, 40, 141–
150.
[2] a) E. T. Kool, H. O. Sintim, Chem. Commun. 2006, 3665–3675; b) Z.
Johar, A. Zahn, C. J. Leumann, B. Jaun, Chem. Eur. J. 2008, 14,
1080–1086.
[3] a) J. Mꢁller, Eur. J. Inorg. Chem. 2008, 3749–3763; b) G. H. Clever,
M. Shionoya, Coord. Chem. Rev. 2010, 254, 2391–2402; c) G. H.
Clever, C. Kaul, T. Carell, Angew. Chem. 2007, 119, 6340–6350;
Angew. Chem. Int. Ed. 2007, 46, 6226–6236; d) K. Tanaka, M. Shio-
noya, Coord. Chem. Rev. 2007, 251, 2732–2742; e) W. He, R. M.
Franzini, C. Achim, Prog. Inorg. Chem. 2007, 55, 545–611.
[4] a) J. Mꢁller, Nature 2006, 444, 698; b) T. Carell, C. Behrens, J. Gier-
lich, Org. Biomol. Chem. 2003, 1, 2221–2228.
[5] a) S. Johannsen, N. Megger, D. Bçhme, R. K. O. Sigel, J. Mꢁller,
Nat. Chem. 2010, 2, 229–234; b) D. A. Megger, J. Mꢁller, Nucleo-
sides Nucleotides Nucleic Acids 2010, 29, 27–38; c) D. Bçhme, N.
Dꢁpre, D. A. Megger, J. Mꢁller, Inorg. Chem. 2007, 46, 10114–
10119; d) F.-A. Polonius, J. Mꢁller, Angew. Chem. 2007, 119, 5698–
5701; Angew. Chem. Int. Ed. 2007, 46, 5602–5604; e) J. Mꢁller, D.
Bçhme, P. Lax, M. Morell Cerdꢅ, M. Roitzsch, Chem. Eur. J. 2005,
11, 6246–6253.
[6] a) Y. Takezawa, K. Tanaka, M. Yori, S. Tashiro, M. Shiro, M. Shio-
noya, J. Org. Chem. 2008, 73, 6092–6098; b) K. Tanaka, A. Tengeiji,
T. Kato, N. Toyama, M. Shionoya, Science 2003, 299, 1212–1213;
c) K. Tanaka, Y. Yamada, M. Shionoya, J. Am. Chem. Soc. 2002,
124, 8802–8803; d) K. Tanaka, A. Tengeiji, T. Kato, N. Toyama, M.
Shiro, M. Shionoya, J. Am. Chem. Soc. 2002, 124, 12494–12498.
[7] a) G. H. Clever, S. J. Reitmeier, T. Carell, O. Schiemann, Angew.
Chem. 2010, 122, 5047–5049; Angew. Chem. Int. Ed. 2010, 49, 4927–
4929; b) G. H. Clever, T. Carell, Angew. Chem. 2007, 119, 254–257;
Angew. Chem. Int. Ed. 2007, 46, 250–253; c) G. H. Clever, Y. Sçltl,
H. Burks, W. Spahl, T. Carell, Chem. Eur. J. 2006, 12, 8708–8718;
[16] F. Seela, T. Wenzel, Helv. Chim. Acta 1994, 77, 1485–1499.
[17] F. Seela, T. Wenzel, Helv. Chim. Acta 1995, 78, 833–846.
ˇ
[18] V. Milata, J. Salon, Org. Prep. Proced. Int. 1999, 31, 347–348.
[19] T. A. Devlin, D. J. Jebaratnam, Synth. Commun. 1995, 25, 711–718.
[20] E. Cubero, N. G. A. Abrescia, J. A. Subirana, F. J. Luque, M.
Orozco, J. Am. Chem. Soc. 2003, 125, 14603–14612.
[21] D. M. Gray, R. L. Ratliff, M. R. Vaughan, Methods Enzymol. 1992,
211, 389–406.
´
[22] L. E. Kapinos, A. Holy, J. Gꢁnter, H. Sigel, Inorg. Chem. 2001, 40,
2500–2508.
[23] a) N. G. A. Abrescia, C. Gonzꢆlez, C. Gouyette, J. A. Subirana, Bio-
chemistry 2004, 43, 4092–4100; b) G. Raghunathan, H. T. Miles, V.
Sasisekharan, Biopolymers 1994, 34, 1573–1581.
[24] a) P. Wu, B. S. Fujimoto, J. M. Schurr, Biopolymers 1987, 26, 1463–
1488; b) B. S. Fujimoto, J. M. Miller, N. S. Ribeiro, J. M. Schurr, Bio-
phys. J. 1994, 67, 304–308.
[25] H. M. Berman, B. Schneider, in Oxford Handbook of Nucleic Acid
Structure (Ed.: S. Neidle), Oxford University Press, Oxford, 1999,
pp. 295–312.
[26] N. Morgner, T. Kleinschroth, H.-D. Barth, B. Ludwig, B. Brutschy, J.
Am. Soc. Mass Spectrom. 2007, 18, 1429–1438.
[27] J. Hoffmann, T. L. Schmidt, A. Heckel, B. Brutschy, Rapid
Commun. Mass Spectrom. 2009, 23, 2176–2180.
Chem. Eur. J. 2011, 17, 6533 – 6544
ꢃ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6543

J. Mꢁller, F. M. Bickelhaupt et al.
[28] A. Bondi, J. Phys. Chem. 1964, 68, 441–451.
[35] a) S. Grimme, J. Comput. Chem. 2004, 25, 1463–1473; b) S.
Grimme, J. Comput. Chem. 2006, 27, 1787–1799; c) A. D. Becke,
Phys. Rev. A 1988, 38, 3098–3100; d) C. Lee, W. Yang, R. G. Parr,
Phys. Rev. B 1988, 37, 785–789.
[29] a) P. Pyykkç, Chem. Rev. 1997, 97, 597–636; b) P. D. Harvey, M.
Drouin, A. Michel, D. Perreault, J. Chem. Soc. Dalton Trans. 1993,
1365–1369.
[36] a) C. Fonseca Guerra, T. van der Wijst, J. Poater, M. Swart, F. M.
Bickelhaupt, Theor. Chem. Acc. 2010, 125, 245–252; b) C. Fonseca
Guerra, F. M. Bickelhaupt, J. G. Snijders, E. J. Baerends, J. Am.
Chem. Soc. 2000, 122, 4117–4128.
[37] J. G. Snijders, P. Vernooijs, E. J. Baerends, At. Data Nucl. Tables
1981, 26, 483–509.
[38] E. van Lenthe, E. J. Baerends, J. G. Snijders, J. Chem. Phys. 1994,
101, 9783–9792.
[39] a) A. Klamt, J. Phys. Chem. 1995, 99, 2224–2235; b) A. Klamt, G.
Schꢁꢁrmann, J. Chem. Soc. Perkin Trans. 2 1993, 799–805; c) C. C.
Pye, T. Ziegler, Theor. Chem. Acc. 1999, 101, 396–408.
[40] Y. Marcus, J. Chem. Soc. Faraday Trans. 1991, 87, 2995–2999.
[30] C. Fonseca Guerra, F. M. Bickelhaupt, J. G. Snijders, E. J. Baerends,
Chem. Eur. J. 1999, 5, 3581–3594.
[31] a) F. M. Bickelhaupt, E. J. Baerends, Rev. Comput. Chem. 2000, 15,
1–86; b) C. Fonseca Guerra, J.-W. Handgraaf, E. J. Baerends, F. M.
Bickelhaupt, J. Comput. Chem. 2004, 25, 189–210.
[32] S. S. Mallajosyula, S. K. Pati, Angew. Chem. 2009, 121, 5077–5081;
Angew. Chem. Int. Ed. 2009, 48, 4977–4981.
[33] N. Morgner, H.-D. Barth, B. Brutschy, Austr. J. Chem. 2006, 59, 109–
114.
[34] a) G. te Velde, F. M. Bickelhaupt, E. J. Baerends, C. Fonseca Guerra,
S. J. A. van Gisbergen, J. G. Snijders, T. Ziegler, J. Comput. Chem.
2001, 22, 931–967; b) C. Fonseca Guerra, J. G. Snijders, G. te Velde,
E. J. Baerends, Theor. Chem. Acc. 1998, 99, 391–403;
c) ADF2009.01, SCM, Theoretical Chemistry, Vrije Universiteit,
Amsterdam, The Netherlands, http://www.scm.com/.
Received: October 12, 2010
Revised: February 12, 2011
Published online: May 5, 2011
6544
www.chemeurj.org
ꢃ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 6533 – 6544