Electrospray-ionization mass spectrometry for screening the specificity and stability of single-stranded-DNA templated self-assemblies

Article abstract of DOI:10.1002/chem.200801506

Supramolecular complexes consisting of a single-stranded oligothymine (dTn) as the host template and an array of guest molecules equipped with a complementary diaminotriazine hydrogen-bonding unit have been studied with electrospray-ionization mass spectr

Full text of DOI:10.1002/chem.200801506

DOI: 10.1002/chem.200801506
Electrospray-Ionization Mass Spectrometry for Screening the Specificity and
Stability of Single-Stranded-DNA Templated Self-Assemblies
Pim G. A. Janssen, Joost L. J. van Dongen, E. W. Meijer,* and
Albertus P. H. J. Schenning*[a]
Abstract: Supramolecular complexes
consisting of a single-stranded oligothy-
mine (dTn) as the host template and
an array of guest molecules equipped
with a complementary diaminotriazine
hydrogen-bonding unit have been stud-
ied with electrospray-ionization mass
spectrometry (ESI-MS). In this hybrid
construct, a supramolecular stack of
guest molecules is hydrogen bonded to
dTn. By changing the hydrogen-bond-
ing motif of the DNA host template or
the guest molecules, selective hydrogen
bonding was proven. We were able to
detect single-stranded-DNA (ssDNA)–
guest complexes for strands with
lengths of up to 20 bases, in which the
highest complex mass detected was
15 kDa; these complexes constitute 20-
component self-assembled objects.
Gas-phase breakdown experiments on
single- and multiple-guest–DNA as-
semblies gave qualitative information
on the fragmentation pathways and the
relative complex stabilities. We found
that the guest molecules are removed
from the template one by one in a
highly controlled way. The stabilities of
the complexes depend mainly on the
molecular weight of the guest mole-
cules, a fact suggesting that the com-
plexes collapse in the gas phase. By
mixing two different guests with the
ssDNA template,
a multicomponent
dynamic library can be created. Our re-
sults demonstrate that ESI-MS is a
powerful tool to analyze supramolec-
ular ssDNA complexes in great detail.
Keywords: mass spectrometry · nu-
cleic acids · self-assembly · supra-
molecular chemistry · templates
Introduction
structs and will help the exploitation of this approach. DNA
constructs are usually analyzed by spectroscopic or micro-
scopic techniques. A technique that is less explored is ESI-
MS, which is a soft technique that transfers solutes into the
gas phase and allows the study of noncovalent interactions
and large nonvolatile chargeable biomolecules such as pro-
teins[28–30] and nucleic acids.[31–38] Furthermore, this technique
has recently been used to characterize a variety of supra-
molecular complexes.[39–45] Qualitative information on the
binding strength of the supramolecular complexes can be
obtained by collision-induced-dissociation (CID) experi-
ments.[46]
DNA is a promising scaffold[1–11] for organizing small mole-
cules through a noncovalent bottom-up approach to create
monodisperse nanosized objects.[12–16] Noncovalent ap-
proaches such as groove binding,[1–3] intercalation,[1–3] hydro-
gen bonding,[7–11] and metal–ligand binding[17] have been re-
ported. Furthermore, DNA can, for example, be used as a
building block to construct complex nanosized objects
through sticky-end cohesion.[18–27] With this approach, DNA
polyhedra[25] and gridded structures[26,27] have been con-
structed. Insight into the strength and dynamics of binding is
essential for a deeper understanding of these DNA con-
Recently, we have reported the ssDNA-templated self-as-
sembly of naphthalene and oligo(p-phenylene)vinylene di-
AHCTUNGTREGaNNNU minotriazine derivatives (G1 and G4, respectively) that
[a] P. G. A. Janssen, J. L. J. van Dongen, Prof. Dr. E. W. Meijer,
Dr. A. P. H. J. Schenning
bind to a complementary strand of oligothymine (dT40, in
which 40 is the number of thymines) to yield uniform ob-
jects (Figure 1 and Scheme 1).[11,47] With ESI-MS, we have
been able to detect distinct complexes of G1 or G4 with
dT10.[11] The ESI-MS technique has also been used to study
noncovalent double-stranded-DNA (dsDNA)–guest com-
plexes[9,48] to enable drug screening.[49–51] In addition to the
probing of selective hydrogen bonds[9] or the binding stoichi-
Laboratory of Macromolecular and Organic Chemistry
Eindhoven University of Technology
P.O. Box 513, 5600 MB Eindhoven (The Netherlands)
Fax : (+31)40-245-1036
E-mail: e.w.meijer@tue.nl
a.p.h.j.schenning@tue.nl
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.200801506.
352
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 352 – 360

FULL PAPER
Scheme 1. Molecular structures of dTn and the guests (Gx).
(in G2),[52] to a naphthalene (in G1),[11] a stilbene (in G3),
and finally an oligo(p-phenylene)vinylene (in G4)[11] core
(Figure 1 and Scheme 1). The guests themselves were hardly
detectable in the negative mode because the basic diamino-
triazine-substituted guests are barely deprotonated under
our ESI-MS conditions in ultrapure water.[53] The ESI mass
spectra of solutions of dT10 show peaks for negatively
charged single strands [dT10]zꢀ, in which z is the number of
charges (Figure 2a). The ESI mass spectra of the solutions
containing ten equivalents of G1 relative to dT10 show
[dT10+mG1] complex peaks (in which m is the number of
G1 molecules) and the [dT10]zꢀ peaks; the guests are now
also visible as deprotonated guest molecules, [G1ꢀH]1ꢀ (Fig-
ure 2b). Deconvolution of this ESI mass spectrum yields
masses of distinct [dT10+mG1] complexes with up to 11
guests bound; [dT10+6G1] has the most intense complex
peak (Figure 2c).[54] Remarkably, the number of guests
bound is larger than the number of binding sites available
on the template. The detection of [dT10+11G1] complex
peaks could be caused by further aggregation of the guest
molecules to the complex during the formation of nanodrop-
lets.[55] The concentration inside such a droplet quickly in-
creases during the evaporation process[28,29] and can easily
reach 1.2 mm, at which point G1 starts to form aggregates at
room temperature.[11] The presence of the [G1ꢀH]1ꢀ peak
suggests that, during the evaporation process and flight of
the ions to the detector, the complex can dissociate, which
leads to a distribution of distinct complexes that is not nec-
essarily present in solution.
Figure 1. Schematic representation of the formation of the dTn–guest
complexes. Blue bars: guests; red bars: hydrogen-bonding motif; black
strand: DNA backbone.
ometry, the affinities and selectivities of minor-groove bind-
ers[49,50] have been proven with ESI-MS and CID experi-
ments. In the present study, ESI-MS is used to systematically
study the effects of template lengths and guest-molecule size
on complex detectabilities to explore the scope and limita-
tions of this technique for these supramolecular self-assem-
blies. With CID experiments, the relative strength of guest
binding to the ssDNA template in the gas phase was stud-
ied. Finally, ESI-MS was used for combinatorial screening of
these hybrid systems, in mixtures of two guests together
with the ssDNA template.
Results and Discussion
Detection of ssDNA–guest complexes: First, the supra-
molecular complexes in ultrapure water with the template
dT10 and different guests, Gx, were studied under mild con-
ditions with ESI-MS in the negative mode. We systematical-
ly increased the strength of possible p–p interactions in the
complexes by using guest derivatives equipped with differ-
ent diaminotriazine hydrogen-bonding units, from a benzene
Titration experiments in which up to 20 equivalents of G1
were added to dT10 show that, already at 6 equivalents of
G1, fully covered dT10–G1 complexes are observed
(Figure 3). For lower amounts of guest molecules
AHCTUNGTREG(NNUN <6 equiv), the concentration of G1 is too low for significant
Chem. Eur. J. 2009, 15, 352 – 360
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
353

E. W. Meijer, A. P. H. J. Schenning et al.
Figure 3. Deconvoluted ESI mass spectra for titration experiments in
which [G1] is increased from 0 to 0.45 mm and [dT10] is kept constant at
0.0225 mm in ultrapure water.
Figure 2. ESI mass spectra for a) [dT10]=0.05 mm and b) [G1]=0.5 mm
and [dT10]=0.05 mm in ultrapure water. c) Deconvoluted spectrum for
[G1]=0.5 mm and [dT10]=0.05 mm in ultrapure water.
binding, whereas the distribution and intensity of the peaks
hardly changes with 10 equivalents or more.[56] With 20
equivalents of guest molecule, the same distribution is pres-
ent as with 10 and 15 equivalents, which suggests that, at
these concentrations of G1, most dT10 templates are fully
covered with G1 in solution.
Figure 4. Deconvoluted ESI mass spectra for a) [G2]=1 mm and
[dT10]=0.1 mm, b) [G3]=1 mm and [dT10]=0.1 mm, and c) [G4]=
Other diaminotriazine-equipped guest molecules were
also tested for binding to dT10. Mixtures containing ten
equivalents of the gallic acid derivative G2, the stilbene de-
rivative G3, or the oligo(p-phenylene)vinylene derivative
G4 all showed complex peaks when mixed with dT10
(Figure 4).[57] For G2 and G3, similar spectral features to
those of G1 were observed, whereas only up to six guests
bound to dT10 have been detected for G4 complexes.
We also changed the size of the template for dTn/G1 mix-
tures with n=5, 20, or 40, in which we kept the concentra-
0.5 mm and [dT10]=0.05 mm in ultrapure water. indicates the mass of
*
the molecular ion for dT10.
tion of G1 the same and added a base equivalent of the
ssDNA template. The solution experiments with these mix-
tures all showed binding of G1 to the dTn template.[58] For
dT5–G1 and dT20–G1 complexes, the mass spectra could be
deconvoluted (Figure 5a and b) and up to 5 and 19 G1 mol-
354
www.chemeurj.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 352 – 360

Mass Spectrometry of Supramolecular Complexes
FULL PAPER
tration of complex, the high number of possible peaks, the
high molecular weights of the dT40–G1 complexes (from 12
to 30 kDa) that considerably broaden the peaks, or the
purity or fragmentation of dT40. The presence of a very
small amount of Na+ and K+ ions will further increase the
number of peaks for each [dT40+mG1] complex.
Selectivity study: To see whether the observed binding of
the guest to the template is selective, a variety of control ex-
periments were performed. When G1 is added to the homo-
DNA strands dA5, dA10, dA20, dG10, or dC10 (Scheme 2)
under the same conditions as those used for dTn, the most
intense peaks are from the free ssDNA, and only a few G1
molecules are bound to the ssDNA templates, which is in
line with the solution experiment.[11,58] In titration experi-
ments in which up to 20 equivalents of G1 were added to
dT10 or dA10, the detected amount of free dA10 is higher
than that of free dT10.[47] Also, a higher number of G1 guest
molecules bind to dT10 rather than to dA10 during the
whole titration. The control experiments with other homo-
ssDNA strands clearly show that there is a clear preference
for binding to dT10, which is logical when it is considered
that threefold hydrogen bonding is possible for a thymine–
G1 complex.
The methyl ester derivative G5 (Scheme 2) and adenine
(G6) did not show binding to dT10,[47] whereas up to six
adenosines (G7), which is more water soluble than adenine,
bound to dT10 have been detected.[47] These experiments
reveal that the binding of the diaminotriazine to the tem-
plate occurs through hydrogen bonding. This conclusion is
supported by a competition experiment in which an excess
of complementary dA20 strands is added to the dT20–G1
Figure 5. Deconvoluted ESI mass spectra for a) [G1]=1 mm and [dT5]=
0.2 mm and b) [G1]=1 mm and [dT20]=0.05 mm in ultrapure water. in-
*
dicates the mass of the molecular ions for dT5 or dT20, respectively.
ecules bound to the template, respectively, have been de-
tected.[59] The mass spectrum of the longer dT40 with G1
only shows the [G1ꢀH]1ꢀ peak and some dT40 degradation-
product peaks.[47] The [dT40]zꢀ peaks are not visible, where-
as they are visible without guest molecules.[47]
In all of the experiments above, distinct dTn–G com-
plexes have been detected; these can reach up to a maxi-
mum mass of 15 kDa in the case of a ssDNA length of 20
bases. For dT40/G1 mixtures, no distinct complexes have
been detected. Still, the presence of the [G1ꢀH]1ꢀ peak and
the lack of [dT40]zꢀ peaks is evidence that G1 binds to
dT40.[47,58] The binding of the guest molecules observed with
ESI-MS is in agreement with the solution experiments.[11]
The reason that no distinct complex peaks are observed for
the dT40–G1 complex could be the relatively lower concen-
Scheme 2. Molecular structures of the control ssDNA and control guests.
Chem. Eur. J. 2009, 15, 352 – 360
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
355

E. W. Meijer, A. P. H. J. Schenning et al.
complexes in solution to yield the more stable dsDNA.[47,60]
This proves that the diaminotriazine unit binds to the dT20
template through hydrogen bonding[61] and that the dT20–
dA20 double strand is more stable than the dT20–G1 com-
plex. These results agree with the previously reported solu-
tion experiments with CD spectroscopy analysis, in which
the complementary strand dA40 (Scheme 2) was added to a
mixture containing the dT40–G1 complex.[11]
Mg
Mg
ECM ¼ E
¼ zV
ð2Þ
lab MiþMg
acc MiþMg
For the CID experiment, we initially studied
[dT5+mG1]zꢀ complexes because these low-molecular-
weight ions had a sufficiently high intensity of the parent
ions. The CID ion-appearance plots show that, for up to
triply charged complexes, the breakdown occurs through se-
quential loss of neutral G1 molecules, as shown, for exam-
ple, with the [dT5+2G1]3ꢀ complex (Figure 6a). In contrast,
the dissociation of the quadruply charged complexes shows
Collision-induced-dissociation experiments: The relative sta-
bility of the ssDNA–guest complexes was studied by CID
experiments or tandem mass spectrometry on selected com-
plexes in the gas phase. After sample ionization, a single
complex of interest is isolated with the quadrupole mass an-
alyzer and only this isolated complex enters the collision
cell, in which the selected ion subsequently collides with the
collision gas. The degree of dissociation depends on the
mass of the collision gas used, the gas pressure in the colli-
sion cell, and the acceleration voltage, Vacc, which deter-
mines the kinetic energy of the ion. A higher acceleration
voltage yields a higher kinetic energy for the complex and,
thus, a higher collision energy between the complex and the
collision gas, argon in our case. With a higher Vacc value or a
higher argon pressure in the collision cell, more guest mole-
cules are dissociated, either as neutral or charged species,
and this eventually results in virgin ssDNA. The argon pres-
sure in the collision cell was set to 1.2ꢁ10ꢀ3 mbar, at which
value each ion faces multiple collisions with the collision
gas. We monitored the dissociative pathway and the degree
of dissociation of the isolated ion as a function of the volt-
age in the collision cell.[29,62] Quantitative values for guest-
binding energies cannot be obtained with these CID experi-
ments, because the threshold energy of dissociation cannot
be determined with our setup. The breakdown curve of a
single complex is measured by selecting a narrow m/z range
around the m/z value of the complex with the quadrupole
and by increasing the Vacc value in the collision cell. The
mass spectrum then shows the breakdown of the selected
complex and the appearance of the breakdown products of
the selected complex. The ion-survival yield (%) of the
parent ion and the appearance of collision products at each
Vacc value were calculated relative to the total intensities of
all ions detected according to Equation (1), in which
I
ACHTUNGTRENNUNG
[dTn+mG]zꢀ is the intensity of the corresponding ion.
zꢀ
I½dTnþm Gꢁ
zꢀ
Appearance ½dTnþm Gꢁ
¼
ꢂ 100 %
zꢀ
n
P
I½dTnþm Gꢁ
m¼0
ð1Þ
For a fair comparison between the different complexes,
the collision energy, ECM, was corrected for the center of
mass according to Equation (2), in which Elab is the kinetic
energy of the ion obtained in the accelerator, z is the
number of charges, Vacc is the acceleration voltage, Mg is the
mass of the collision gas (argon), and Mi is the mass of the
parent ion.[29]
Figure 6. Ion-appearance plots for the CID experiments of
a) [dT5+2G1]3ꢀ
(m/z
585.45 gmolꢀ1), and c) [dT5+4G1]3ꢀ (m/z 807.06 gmolꢀ1) as a function of
the collision energy, ECM
780.92 gmolꢀ1),
b) [dT5+2G1]4ꢀ
(m/z
.
356
www.chemeurj.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 352 – 360

Mass Spectrometry of Supramolecular Complexes
FULL PAPER
two fragmentation pathways. For instance, for the
[dT5+2G1]4ꢀ complex (Figure 6b), neutral G1 molecules
and charged [G1ꢀH]1ꢀ ions can be dissociated. It is remark-
able that, due to the increased charge repulsion,[50] the basic
amines of the diaminotriazine are deprotonated. The break-
down of the triply charged complex with four guests,
[dT5+4G1]3ꢀ, also shows a breakdown mechanism whereby
neutral G1 molecules are sequentially dissociated (Fig-
ure 6c). At ECM =0.2 eV, which corresponds to a Vacc value
of 5 V, a very low voltage for ESI-MS experiments, only
90% of the parent ion survives. This implies that, even
when all DNA templates are still fully covered after the ion-
ization process, a distribution will always be present in the
ESI mass spectrum due to fragmentation in the collision cell
at this pressure. The gas-phase stabilities of the parent ions
of the [dT5+mG1]zꢀ complexes as a function of the number
of guests, m, and charges, z, were compared through the
parent-ion survival yield as a function of the ECM and I50
(ECM value at which 50% of the parent ion survives) values
(Figure 7). In a comparison of a single, only hydrogen-
values do not change the I50 value to a large extent (Figure 7
and Table 1). We found that I50 =0.27 eV for [dT5+1G1]3ꢀ
and I50 =0.25 eV for [dT5+5G1]3ꢀ, that is, a difference of
Table 1. I50 values of [dT5+mGx]zꢀ ions.
Ion
I50 value [eV]
m=1
m=2
m=3
m=4
m=5
A
n.d.[a]
0.27
0.60
0.55
0.70
0.30
0.30
0.25
0.45
0.42
0.62
0.28
0.28
N
0.27
0.25
0.25
R
n.d.[a]
n.d.[a]
n.d.[a]
n.d.[a]
n.d.[a]
n.d.[a]
n.d.[a]
n.d.[a]
n.d.[a]
AHCTUNGTRENNUNG
AHCTUNGTRENNUNG
[a] n.d.=not determined.
less than 10%. Hence, the dissociation is neither coopera-
tive nor anti-cooperative. Apparently, these complexes do
not necessarily retain their solution structure when brought
into the gas phase, but collapse into a structure with more
host–guest interaction (see below). When compared, the
quadruply charged complexes appear to be slightly more
stable than the triply charged complexes (Figure 7b); for in-
stance, with five G1 molecules bound, I50 =0.28 and 0.25 eV
for the quadruply and triply charged complexes, respective-
ly.
To study the relative gas-phase stabilities of the com-
plexes with the different guests, CID experiments were per-
formed on [dT5+1Gx]3ꢀ and [dT5+3Gx]3ꢀ complexes
(Figure 8) under standard conditions. The corresponding ion
peaks were selected and the voltage of the collision cell was
increased until the parent ions were no longer detected. It
should be noted that the ECM value has been corrected for
the mass of the parent ion for a fair comparison. When
three guests are bound, it is clear that the larger guests bind
more strongly, as indicated by the higher I50 value (Figure 8a
and Table 1). This agrees well with an increased guest–guest
interaction for bigger guest molecules. When [dT5+3Gx]3ꢀ
and [dT5+1Gx]3ꢀ are compared, the I50 value of
[dT5+1Gx]3ꢀ is lower in all cases, results that are similar to
those already found for G1 (Figure 8 and Table 1). For
[dT5+1Gx]3ꢀ individually, no or hardly any difference in the
I50 value is expected if the guests are only hydrogen bonded
to the DNA template. By contrast, for collapsed structures
in which the sum of all the van der Waals interactions will
count, the I50 value should increase with the molecular
weight of the guests. Here, we notice an apparent linear in-
crease of the I50 value of the [dT5+1Gx]3ꢀ parent ions with
the molecular weight of the guest molecules (Figure 8 and
Table 1). This clearly shows that the templated structures
that are present in solution collapse in the gas phase as ob-
served earlier.[30,63,64] This structural collapse can easily
occur during the evaporation process or during the several
tens of microseconds when the ions travel through the colli-
sion cell.[64,65]
Figure 7. Parent-ion survival yields for the CID experiments of
a) [dT5+mG1]3ꢀ and b) [dT5+5G1]zꢀ. The lines are included to guide
the eyes.
bonded guest with multiple-bound guests, the I50 value when
multiple guests are bound should be significantly higher
than that when one guest is bound because of additional
guest–guest interactions However, clearly, both the m and z
Mixed guest systems: We have also performed competition
experiments in which two guests are mixed in solution with
Chem. Eur. J. 2009, 15, 352 – 360
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
357

E. W. Meijer, A. P. H. J. Schenning et al.
Figure 9. Deconvoluted ESI mass spectra at different Vacc values for
[G1]=[G2]=1 mm and [dT10]=0.1 mm.
the dT10 template and only ions in which G2 is bound are
detected; this shows that G2 binds more strongly to dT10
than G1. This is in good agreement with the I50 values ob-
tained from the CID experiments (Table 1). Mixtures of dT5
or dT10 with G1 and G3 or G1 and G4 showed comparable
results, in that G1 again dissociates prior to G3 or G4.[47]
To determine the relative stability of two different guests,
we performed CID experiments on [dT5+1G1+1Gx]3ꢀ
complexes. The results for the [dT5+1G1+1G2]3ꢀ complex
are shown in Figure 10, which is also representative for the
Figure 8. Parent-ion survival yields for the CID experiments of
a) [dT5+3G1]3ꢀ and b) [dT5+1Gx]3ꢀ. c) Plot of the I50 value of the
parent ions versus the mass of the guest molecule. The lines are included
to guide the eyes.
Figure 10. ESI mass spectra of the [dT5+1G1+1G2]3ꢀ ion at different
Vacc values.
G1 and G3 or G1 and G4 combinations.[47] At low Vacc
values, the parent ion is still intact. Upon an increase in the
voltage, the smaller G1 dissociates first and, upon a further
increase in the Vacc value to 30 V, G2 also dissociates. At this
high Vacc value, some fragmentation products of dT5 can
also be seen. These CID experiments again show that G2
binds more strongly to dT5 than G1. With G3 and G4, simi-
lar results were obtained.
the ssDNA template. We added one base equivalent of each
guest for a fair competition experiment. As an illustrative
example, G1 was mixed with G2 in the presence dT10
(Figure 9).[66] Under mild conditions, with Vacc =15 V, distinct
complexes of dT10 with both G1 and G2 bound in various
ratios can be detected with masses of up to 12000 Da. When
the Vacc value is increased, the ions are broken down to
lower mass ions and the mass spectrum becomes simpler.
Eventually, at Vacc =30 V, most of the G1 is removed from
358
www.chemeurj.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 352 – 360

Mass Spectrometry of Supramolecular Complexes
FULL PAPER
spectively. Before injection into the mass spectrometer, the solutions
were heated to above 708C to erase all memory effects and they were
subsequently annealed at room temperature for at least 2 h.
Conclusion
We have explored the scope and limitations of ESI-MS for
the detection of hydrogen-bonded supramolecular com-
plexes of ssDNA and guest molecules. Individual ssDNA–
guest complexes could be detected up to a mass of 15 kDa,
which is noteworthily high since the complex consists of 20
components. CID experiments show that the hydrogen-
bonded guest molecules can be dissociated from the DNA
template one by one in a highly controlled way. Remarka-
bly, we have found a linear relationship of the molecular
weight of the guest molecules with the I50 value of the
[dT5+1Gx]3ꢀ complexes, which shows that the stability is
not only determined by the hydrogen-bond interaction and,
therefore, a complex with a small number of guests does not
necessarily retain its solution hydrogen-bonded structure in
the gas phase. Interestingly, upon an increase in the collision
energy of two different guests bound to dTn in solution, the
weakest bound guest can be dissociated first. This promising
tool will allow for combinatorial screening of molecules
binding to DNA.
Acknowledgements
The authors wish to acknowledge L. Brunsveld and M. J. Pouderoijen for
supplying compounds G2 and G3, respectively, K. Pieterse for the art-
work, H. Eding for the elemental analysis, X. Lou for the MALDI-TOF
spectra, and the EURYI scheme for financial support.
[1] H. Ihmels, D. Otto, Top. Curr. Chem. 2005, 258, 161–204.
[2] U. Pindur, M. Jansen, T. Lemster, Curr. Med. Chem. 2005, 12, 2805–
2847.
[3] B. A. Armitage, Mol. Supramol. Photochem. 2006, 14, 255–287.
[4] E. J. Fechter, B. Olenyuk, P. B. Dervan, J. Am. Chem. Soc. 2005,
127, 16685–16691.
[5] K. C. Hannah, R. R. Gil, B. A. Armitage, Biochemistry 2005, 44,
15924–15929.
[6] K. C. Hannah, B. A. Armitage, Acc. Chem. Res. 2004, 37, 845–853.
[7] R. Iwaura, F. J. M. Hoeben, M. Masuda, A. P. H. J. Schenning, E. W.
Meijer, T. Shimizu, J. Am. Chem. Soc. 2006, 128, 13298–13304.
[8] R. Iwaura, K. Yoshida, M. Masuda, M. Ohnishi-Kameyama, M.
Yoshida, T. Shimizu, Angew. Chem. 2003, 115, 1039–1042; Angew.
Chem. Int. Ed. 2003, 42, 1009–1012.
[9] R. Iwaura, M. Ohnishi-Kameyama, M. Yoshida, T. Shimizu, Chem.
Commun. 2002, 2658–2659.
[10] R. Iwaura, Y. Kikkawa, M. Ohnishi-Kameyama, T. Shimizu, Org.
Biomol. Chem. 2007, 5, 3450–3455.
In general, we have shown that ESI-MS and CID experi-
ments are useful techniques for identifying and studying se-
lective and nonselective secondary interactions in templated
supramolecular assemblies. These methods provide impor-
tant information concerning secondary interactions in these
supramolecular assemblies in the gas phase.
[11] P. G. A. Janssen, J. Vandenbergh, J. L. J. van Dongen, E. W. Meijer,
A. P. H. J. Schenning, J. Am. Chem. Soc. 2007, 129, 6078–6079.
[12] R. F. Service, Science 2005, 309, 95.
[13] A. P. H. J. Schenning, P. Jonkheijm, F. J. M. Hoeben, J. Van Herri-
khuyzen, S. C. J. Meskers, E. W. Meijer, L. M. Herz, C. Daniel, C.
Silva, R. T. Phillips, R. H. Friend, D. Beljonne, A. Miura, S. De Feyt-
er, M. Zdanowska, H. Uji-I, F. C. De Schryver, Z. Chen, F. Wuerth-
ner, M. Mas-Torrent, D. Den Boer, M. Durkut, P. Hadley, Synth.
Met. 2004, 147, 43–48.
[14] F. J. M. Hoeben, P. Jonkheijm, E. W. Meijer, A. P. H. J. Schenning,
Chem. Rev. 2005, 105, 1491–1546.
[15] O. Ikkala, G. ten Brinke, Chem. Commun. 2004, 2131–2137.
[16] A. Ajayaghosh, V. K. Praveen, Acc. Chem. Res. 2007, 40, 644–656.
[17] K. Tanaka, A. Tengeiji, T. Kato, N. Toyama, M. Shionoya, Science
2003, 299, 1212–1213.
[18] B. Ding, N. C. Seeman, Science 2006, 314, 1583–1585.
[19] N. C. Seeman, Methods Mol. Biol. 2005, 303, 143–166.
[20] P. W. K. Rothemund, Nature 2006, 440, 297–302.
[21] M. Brucale, G. Zuccheri, B. Samori, Trends Biotechnol. 2006, 24,
235–243.
[22] K. Lund, B. Williams, Y. Ke, Y. Liu, H. Yan, Curr. Nanoscience
2006, 2, 113–122.
[23] J. Chen, N. C. Seeman, Nature 1991, 350, 631–633.
[24] J. D. Cohen, J. P. Sadowski, P. B. Dervan, J. Am. Chem. Soc. 2008,
130, 402–403.
[25] Y. He, T. Ye, M. Su, C. Zhang, A. E. Ribbe, W. Jiang, C. Mao,
Nature 2008, 452, 198–201.
[26] K. V. Gothelf, T. H. LaBean, Org. Biomol. Chem. 2005, 3, 4023–
4037.
[27] J. Zhang, Y. Liu, Y. Ke, H. Yan, Nano Lett. 2006, 6, 248–251.
[28] J. H. Gross, Mass Spectrometry: A Textbook, Springer, Heidelberg,
2004, p. 518.
[29] C. G. Herbert, A. W. Johnstone, Mass Spectrometry Basics, CRC,
Boca Raton, 2002, p. 496.
Experimental Section
Materials: The synthesis, purification, and characterization of G3 and G5
can be found in the Supporting Information. ssDNA was supplied freeze
dried and HPLC purified by MWG Biotech AG. All solvents purchased
from Acros Chemica, Aldrich, or Fluka were of p.a. quality. All other
chemicals were commercially available and were used without further pu-
rification.
General methods: UV/Vis and fluorescence spectra were performed on a
Perkin–Elmer Lambda40 and Perkin–Elmer LS-50B spectrophotometers.
CD, UV/Vis, and fluorescence spectra were recorded on a JASCO 815 in-
strument equipped with
a Peltier PFD-425S temperature controller.
High-resolution ESI-MS was performed on Q-Tof Ultima Global mass
spectrometer (Micromass, Manchester, UK) equipped with a Z-spray
source. The analysis was performed with MassLynx 4.1 software and the
spectra were deconvoluted with the MaxEnt-1 program. Electrospray
ionization was achieved in the negative mode by direct and continuous
injection of 0.5–4 mm solutions in MilliQ water at room temperature with
a rate of 5 mLminꢀ1 and by applying 5 kV on the needle. The collision
cell was filled with argon (1.2ꢁ10ꢀ3 mbar) as the collision gas and the
cone voltage was set to 150 V. For a typical single-run CID experiment,
the acceleration voltage was increased up to 50 V in small steps, with in-
jection for 30 s for each Vacc value. From the area underneath the intensi-
ty chromatogram of the specific mass region of each ion, the relative in-
tensities of each ion peak at every Vacc value could then easily be calcu-
lated.
Sample preparation: All ESI-MS samples were prepared by addition of
ssDNA in MilliQ water to the solid guest and subsequent sonification of
the mixture at 708C for at least 5 min. In the case of G4, a concentrated
solution of G4 in HPLC-grade tetrahydrofuran (THF) was injected into
the dTn solution in MilliQ water and the THF was removed by heating
the sample to 708C. For the titration experiments, dT10 or dA10 solu-
tions were mixed with a G1–dT10 (20:1) or G1–dA10 (20:1) solution, re-
[30] I. A. Kaltashov, S. J. Eyles, Applications of Mass Spectrometry in
Biophysics: Conformation and Dynamics of Biomolecules, Wiley,
New York, 2005, p. 368.
Chem. Eur. J. 2009, 15, 352 – 360
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
359

E. W. Meijer, A. P. H. J. Schenning et al.
[31] C. A. Schalley, Mass Spectrom. Rev. 2002, 21, 253–309.
[32] C. G. Huber, H. Oberacher, Mass Spectrom. Rev. 2001, 20, 310–343.
[33] A. Apffel, J. A. Chakel, S. Fischer, K. Lichtenwalter, W. S. Hancock,
J. Chromatogr. A 1997, 777, 3–21.
[34] A. Apffel, J. A. Chakel, S. Fischer, K. Lichtenwalter, W. S. Hancock,
Anal. Chem. 1997, 69, 1320–1325.
[53] In the positive mode, the guest molecules can easily be detected.
[54] Note that quantitative values for guest-binding equilibrium con-
stants cannot be obtained from these ESI-MS experiments because
the free-guest concentration is not linear with the intensity of
[GꢀH]1ꢀ and most of the complexes are partially dissociated, even
at low acceleration voltages.
[35] F. D. Lewis, L. Zhang, X. Zuo, J. Am. Chem. Soc. 2005, 127, 10002–
10003.
[55] R. D. Smith, K. J. Light-Wahl, Biol. Mass Spectrom. 1993, 22, 493–
501.
[36] A. Cuppoletti, Y. Cho, J.-S. Park, C. Straessler, E. T. Kool, Bioconju-
gate Chem. 2005, 16, 528–534.
[56] Note that the intensities of the peaks of the spectra in Figures 2c
and 3 differ, which is presumably due to the slightly different argon
pressure in the collision cell.
[57] For the original mass spectra, see the Supporting Information.
[58] With CD spectroscopy, all of the dTn–G1 mixtures showed a Cotton
effect in the region in which only G1 absorbs. None of the other
homo-ssDNA strands showed a Cotton effect in the G1 absorption
region. See the Supporting Information.
[37] M. Balaz, A. E. Holmes, M. Benedetti, P. C. Rodriguez, N. Berova,
K. Nakanishi, G. Proni, J. Am. Chem. Soc. 2005, 127, 4172–4173.
[38] H. Kashida, H. Asanuma, M. Komiyama, Angew. Chem. 2004, 116,
6684–6687; Angew. Chem. Int. Ed. 2004, 43, 6522–6525.
[39] M. Rueda, F. J. Luque, M. Orozco, J. Am. Chem. Soc. 2006, 128,
3608–3619.
[40] P. G. A. Janssen, P. Jonkheijm, P. Thordarson, J. C. Gielen, P. C. M.
Christianen, J. L. J. van Dongen, E. W. Meijer, A. P. H. J. Schenning,
J. Mater. Chem. 2007, 17, 2654–2660.
[41] M. Elhabiri, A. Trabolsi, F. Cardinali, U. Hahn, A.-M. Albrecht-
Gary, J.-F. Nierengarten, Chem. Eur. J. 2005, 11, 4793–4798.
[42] U. Hahn, M. Elhabiri, A. Trabolsi, H. Herschbach, E. Leize, A. Van
Dorsselaer, A.-M. Albrecht-Gary, J.-F. Nierengarten, Angew. Chem.
2005, 117, 5472–5475; Angew. Chem. Int. Ed. 2005, 44, 5338–5341.
[43] B. J. Holliday, C. A. Mirkin, Angew. Chem. 2001, 113, 2076–2097;
Angew. Chem. Int. Ed. 2001, 40, 2022–2043.
[44] F. Hof, S. L. Craig, C. Nuckolls, J. Rebek, Jr., Angew. Chem. 2002,
114, 1556–1578; Angew. Chem. Int. Ed. 2002, 41, 1488–1508.
[45] E. S. Baker, S. L. Bernstein, M. T. Bowers, J. Am. Soc. Mass Spec-
trom. 2005, 16, 989–997.
[46] M. A. C. Broeren, J. L. J. van Dongen, M. Pittelkow, J. B. Christen-
sen, M. H. P. van Genderen, E. W. Meijer, Angew. Chem. 2004, 116,
3641–3646; Angew. Chem. Int. Ed. 2004, 43, 3557–3562.
[47] See the Supporting Information.
[48] J. L. Beck, M. L. Colgrave, S. F. Ralph, M. M. Sheil, Mass Spectrom.
Rev. 2001, 20, 61–87.
[49] F. Rosu, V. Gabelica, C. Houssier, E. de Pauw, Nucleic Acids Res.
2002, 30, e82.
[50] F. Rosu, S. Pirotte, E. de Pauw, V. Gabelica, Int. J. Mass Spectrom.
2006, 253, 156–171.
[51] S. A. Hofstadler, R. H. Griffey, Chem. Rev. 2001, 101, 377–390.
[52] L. Brunsveld, J. A. J. M. Vekemans, J. H. K. K. Hirschberg, R. P. Sij-
besma, E. W. Meijer, Proc. Natl. Acad. Sci. USA 2002, 99, 4977–
4982.
[59] Similar results have been obtained for G2, G3, and G4 in combina-
tion with dT5 and dT20; see the Supporting Information.
[60] The dT20–G1 complex peaks that appeared between m/z values of
750 and 1750 Da are no longer present, which indicates that the
supramolecular strand of G1 is replaced by the dA20 strand. Also,
no complex peaks are present in the deconvoluted spectrum and it
now mainly contains a dA20 peak and a very small dT20 peak. Al-
though some peaks are the double-stranded dT20–dA20 peaks, they
could not be deconvoluted. It should be noted that dT20–dA20 can
polymerize into high-molecular-weight dsDNA; this makes the de-
tection of dT20–dA20 a lot harder.
[61] Another proof for the involvement of the amine protons in binding
to the template is an experiment in which the amine protons of the
diaminotriazine of G1 were replaced with deuterium atoms by using
D2O as a solvent instead of H2O. The mass found for the breakdown
product of the G1(D4)–dT10 complexes corresponds to the
[G1(D4)ꢀD]1ꢀ peak. See the Supporting Information.
[62] F. Rosu, C.-H. Nguyen, E. De Pauw, V. Gabelica, J. Am. Soc. Mass
Spectrom. 2007, 18, 1052–1062.
[63] C. A. Schalley, Analytical Methods in Supramolecular Chemistry,
2007, Wiley-VCH, Weinheim, p. 484.
[64] J. Gidden, A. Ferzoco, E. S. Baker, M. T. Bowers, J. Am. Chem. Soc.
2004, 126, 15132–15140.
[65] S. Sen, L. Nilsson, J. Am. Chem. Soc. 2001, 123, 7414–7422.
[66] For the same experiment with dT5, see the Supporting Information.
Received: July 24, 2008
Published online: November 28, 2008
360
www.chemeurj.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 352 – 360