Exploring the formation pathways of donor-acceptor catenanes in aqueous dynamic combinatorial libraries

Article abstract of DOI:10.1021/ja111407m

The discovery through dynamic combinatorial chemistry (DCC) of a new generation of donor-acceptor [2]catenanes highlights the power of DCC to access unprecedented structures. While conventional thinking has limited the scope of donor-acceptor catenanes to

Full text of DOI:10.1021/ja111407m

ARTICLE
pubs.acs.org/JACS
Exploring the Formation Pathways of Donor-Acceptor Catenanes in
Aqueous Dynamic Combinatorial Libraries
Fabien B. L. Cougnon,† Ho Yu Au-Yeung,† G. Dan Panto-s,*,†,‡ and Jeremy K. M. Sanders*,†
†University Chemical Laboratory, University of Cambridge, Lensfield Road, CB2 1EW, Cambridge, U.K.
‡Department of Chemistry, University of Bath, BA2 7AY, Bath, U.K.
S Supporting Information
b
ABSTRACT: The discovery through dynamic combinatorial
chemistry (DCC) of a new generation of donor-acceptor
[2]catenanes highlights the power of DCC to access unpre-
cedented structures. While conventional thinking has limited
the scope of donor-acceptor catenanes to strictly alternating
stacks of donor (D) and acceptor (A) aromatic units, DCC is
demonstrated in this paper to give access to unusual DAAD,
DADD, and ADAA stacks. Each of these catenanes has specific
structural requirements, allowing control of their formation.
On the basis of these results, and on the observation that the
catenanes represent kinetic bottlenecks in the reaction path-
way, we propose a mechanism that explains and predicts the structures formed. Furthermore, the spontaneous assembly of
catenanes in aqueous dynamic systems gives a fundamental insight into the role played by hydrophobic effect and donor-acceptor
interactions when building such complex architectures.
electron-deficient 1,4,5,8-naphthalenetetracarboxylic diimide
(NDI) vary with the polarity of the solvent, from ca. 2 M-1 in
chloroform to ca. 2000 M-1 in water. As expected, alternating
donor-acceptor interactions are clearly dominant over weaker
acceptor-acceptor (Ka < 1 M-1 in chloroform, ca. 200 M-1 in
water) or donor-donor interactions (Ka < 1 M-1 in chloroform,
ca. 20 M-1 in water).5 From these results it is generally believed that
the strong solvophobic component of aromatic interactions can be
used to dramatically increase the magnitude of electrostatic inter-
actions and that the alternating D-A stack is favored in any solvent.
In the light of these numbers, it is perhaps surprising that in
recent years A-A and D-D stacks have spontaneously emerged
from the self-assembly of donor and acceptor units in polar
media.6-8 In these cases, the solvophobic effect seems to over-
come, at least partially, pure electrostatic interactions between
donor and acceptor units, although it remains difficult to evaluate
to what extent the two phenomena cooperate and/or compete.
To our knowledge, only three examples of these unconventional
architectures have been reported in the literature. The first
contribution to this field was that of Li and co-workers, who
reported the synthesis of an AAAA [2]catenane based on
perylenediimide.6 Most recently, Fujita et al. showed that
discrete ADDA and AADA stacks could be obtained within a
box-shaped cavity.7 We reported the dynamic combinatorial
synthesis of DAAD and DADD catenanes8 at the expense of
1. INTRODUCTION
We present here the discovery of new donor-acceptor
[2]catenanes assembled from simple electron-donor (D) and
electron-acceptor (A) building blocks using disulfide chemistry.
Moving on from our previous dynamic combinatorial syntheses of
catenanes with unusual DAAD and DADD stacking arrangements,
we now describe the formation of catenanes with DADA and
unprecedented AADA stacks under the same conditions. These four
types of catenanes form a complete set and can be viewed as a whole
new generation of donor-acceptor [2]catenanes (Figure 1). More
importantly, perhaps, we outline a mechanism that explains and
predicts the formation of each of these catenanes, which turn out to
be kinetic bottlenecks on the reaction pathway.
Interactions between aromatic molecules are mainly described
in term of electrostatic interactions between electron-deficient
and electron-rich π-systems.1 The complementary alternating
stack of donor and acceptor aromatic moieties, leading to the
optimum electronic overlap, has always been thought to be the
most favorable arrangement. Alternating D-A stacks adopt a
parallel and compact face-to-face geometry and have been
extensively used as scaffolds for complex structures such as
foldamers,2 rotaxanes,3 and catenanes.4
The formation of stacked structures reduces the total solvent-
exposed surface area, so efficient desolvation of the aromatic
systems plays a crucial role in this process. Iverson et al.5 showed
that the effective strength of donor-acceptor interactions is
closely related to the solvent environment: association constants
between the electron-rich 1,5-dialkoxynaphthalene (DN) and the
Received: December 18, 2010
Published: February 15, 2011
r
2011 American Chemical Society
3198
dx.doi.org/10.1021/ja111407m J. Am. Chem. Soc. 2011, 133, 3198–3207
|

Journal of the American Chemical Society
ARTICLE
Figure 1. Possible arrangements of the π units of a donor-acceptor [2]catenane. Aromatic π-donor and π-acceptor are represented by red and blue
cartoons, respectively.
the expected DADA structure. According to the D-A stacking
model, the DADD stack would be unlikely to be formed in
significant quantities in competition with the more obvious D-A
stacks. By expanding the range of building blocks, we now open
the possibility of accessing a larger variety of donor-acceptor
catenanes, such as DDDD, DADD, DADA, DAAD, AADA, or
AAAA (Figure 1).
Each catenane presented in Figure 1 represents a major syn-
thetic challenge for the classical approach but can, in principle, be
achieved using dynamic combinatorial chemistry (DCC).9,10
We show here that we were able to obtain four out of the six
possible arrangements, with the AAAA and DDDD stacks
remaining elusive at present. The size and geometry of the
building blocks influences dramatically the composition of the
dynamic combinatorial libraries (DCLs) such that it allows a
precise control over the type of catenane formed on the basis of
its structural requirements.
The new generation of donor-acceptor [2]catenanes re-
ported in this work not only constitutes a success in the synthesis
of topologically complex targets but it may also open new
horizons for the use of catenanes as molecular machines, as
one of the DAAD catenanes was recently shown to exhibit two
switchable conformations, with a parallel arrangement of the
π-units or a Gemini-sign-like conformation.8b Hence, these cate-
nanes present a major advance in understanding, designing, and
using complex architectures based on donor and acceptor units.
These systems also represent a complex environment in which
both thermodynamic and kinetic factors play a role in determin-
ing the composition of the library.
We present first a proposed mechanistic pathway for the
synthesis of the donor-acceptor [2]catenanes which rationa-
lizes our earlier results and predicts the composition of new
libraries. The remainder of this paper explores and confirms these
predictions experimentally.
We propose that the formation of catenanes in aqueous
donor-acceptor DCLs occurs through three successive steps:
formation of a dimer, threading of a linear species into the dimer’s
cavity, and closing of the catenane. The key elements in the for-
mation of catenanes are the donor-acceptor interactions and the
polarity of the medium, manipulated by the addition of salt.
In pure water, DCLs composed of donor and acceptor
building blocks do not usually lead to the formation of catenanes,
although traces can be observed in some isolated cases (see the
Supporting Information). This indicates that in water, in the time
frame of the libraries’ oxidation, the sum of donor-acceptor and
hydrophobic interactions is insufficient to bring stacks of build-
ing blocks into the close proximity necessary for catenation.
In high salt8,12 (1 M NaNO3) libraries, the increased polarity
of the medium promotes the decrease of solvent-exposed
hydrophobic surfaces; this is achieved by either a building block
or a linear oligomer14 sliding into the hydrophobic cavity of a
preformed cyclic dimer. This threaded species is the first inter-
mediate in the formation of a catenane. Besides the threading
mechanism described above, it can also be formed by the
templated cyclization of a linear precursor in the presence of
a complementary linear template.13 If the donor-acceptor
interactions have a significant role, then the [-A-A-] cyclic
dimer should preferentially bind a donor over an acceptor unit;
similarly, the [-D-D-] and the [-A-D-] cyclic dimers are
more likely to interact with an acceptor rather than a donor
moiety. Each of the three favorable threaded species corresponds
to the initial step of a different pathway for formation of
catenanes, pathways I, II, and III, respectively (Figure 2).
Once a threaded complex is formed through pathway I, II, or
III, linking a fourth building block completes the catenane
formation. The factors involved in the closing step are more difficult
to predict, but electrostatic interactions are expected to play a lesser
role than in the threading step: in the closing step, there are only two
π surfaces interacting, while in the threading step there are four. The
catenanes can then be closed with either an acceptor or a donor
building block, depending on which event leads to the maximum
stabilization. Therefore, each pathway should lead to a characteristic
pair of catenanes: DADD and DADA catenanes from pathway I,
DAAD and AADA catenanes from pathway II, and DADA and
AADA catenanes from pathway III (Figure 2).
2. MECHANISM OF FORMATION OF DONOR-ACCEP-
TOR CATENANES IN AQUEOUS DCLS
Contrary to our initial thoughts,8 the dithiol building blocks used
in this work are fully oxidized into disulfide and stop exchanging
before the libraries reach thermodynamic equilibrium.11 The kinetics
of formation of the library members consequently play a major role,
allowing us to map out the pathways leading to the different types of
catenanes. For each of these pathways, we can trace the intermediates
involved and evaluate the role played by hydrophobic and donor-
acceptor interactions in each step of the catenation process.
This mechanism predicts that no homocatenane (AAAA or
DDDD) will be formed in this system: although the formation
of catenanes is dominated overall by the hydrophobic effect,
donor-acceptor interactions play a determining role in the first
step of threading. However, further increasing the polarity of our
3199
dx.doi.org/10.1021/ja111407m |J. Am. Chem. Soc. 2011, 133, 3198–3207

Journal of the American Chemical Society
ARTICLE
Figure 2. Proposed mechanism for formation of dynamic combinatorial donor-acceptor [2]catenanes. For clarity, the mechanistically equivalent
possibilities of threading a linear oligomer through each cyclic dimer or of forming complexes between linear dimers before cyclization are not shown.
medium or increasing the surface area of the hydrophobic core6
could potentially lead to the formation of homocatenanes.
On the basis of this set of intermediates and interactions
involved in the mechanism, it is possible to identify the para-
meters allowing efficient formation of the catenanes in the
libraries. First, the [-A-D-], [-D-D-] and [-A-A-]
dimers constitute the cornerstone for the formation of any
catenane, so their abundance in the DCL is paramount. To form
a catenane, a dimer needs to fulfill two requirements: it must be
formed efficiently in the library and be large enough to accom-
modate an aromatic thread. Catenane formation will be favored
when an optimum overlap of the aromatic moieties favors both
donor-acceptor and hydrophobic interactions in the two steps
of threading and closing. This happens when each ring forming
the catenane is tight; when the rings are too flexible, the enthalpic
gain is low and cannot compensate for the entropic costs derived
from the catenation process, so the catenane is not formed.
We can take advantage of these factors by designing carefully
the libraries, not only to form selectively one type of catenane but
also to select the pathway by which it will be formed.
acceptor building blocks NDI(Cys)2 (A1),13b,15 NDI(GlyCys)2
(A2),13a and NDI(GSH)2 (A5)13a have already been published,
while NDI(DMGlyCys)2 (A3) and the asymmetrically substi-
tuted NDI(Cys)(GSH) (A4) were prepared using similar methods
(see the Supporting Information). The acceptor building blocks
differ in the length, hydrophilicity, and bulk of the side chains.
Screening of libraries containing different combinations of these
donor and acceptor building blocks allows us to investigate
the extent to which all of these parameters play a role in the
formation of catenanes.
Aqueous disulfide DCLs were created by dissolving the
building blocks to a total concentration of 5 mM in water at
pH 8. The pH was adjusted with aqueous NaOH, and the libraries
were stirred under air in capped vials to allow oxidation of the thiol
building blocks. After 5 days, the libraries were fully oxidized and
analyzed by HPLC and LC-MS. Absorbance was recorded at the
optimum wavelength for each building block: acceptor units at 383
nm and donor units at 292 nm (1,5-substituted D1a and D1b) or
260 nm (2,6-substituted D2a and D2b).
3.1. Exploring Pathway I. The efficient formation of a [-
D-D-] dimer required in pathway I (Figure 4) is achieved by
using donor building blocks with short side chains.8a,8c,13a The
short side chains of the building block result in a [-D-D-]
dimer whose small cavity size enables a relatively good interac-
tion with an acceptor moiety, favoring the threading mechanism
described above. The geometry of the 1,5-isomer D1a favors the
formation of the [-D-] cyclic monomer over the [-D-D-]
dimer and does not lead to the formation of any catenane (see the
Supporting Information). On the other hand, the 2,6-isomer D2a
forms a stable [-D-D-] dimer, which leads to the formation of
catenanes through pathway I. Hence, all the following libraries
were prepared with the D2a donor building block and different
acceptor building blocks or mixtures of acceptor building blocks
(see the Supporting Information). To summarize the results
described in this section, three DADA catenanes were identified,
3. RESULTS AND DISCUSSION
All building blocks in this study (Figure 3) are constructed on
a similar design: a large hydrophobic π-surface, either NDI as
electron-acceptor (A) or DN as electron-donor (D), to which are
attached two cysteine-terminated hydrophilic side chains of
various lengths. The cysteine-derived side chains provide easy
access to building blocks of different sizes, along with the thiol
necessary for the initiation of disulfide exchange and carboxylic
acids for water solubility.
The donor building blocks D1a,13a D1b,8d D2a,8c and D2b8b
were synthesized following previously reported procedures.
They differ in the substitution pattern of the central core (1,5-
or 2,6-) and the length of the side chains. Syntheses of the
3200
dx.doi.org/10.1021/ja111407m |J. Am. Chem. Soc. 2011, 133, 3198–3207

Journal of the American Chemical Society
ARTICLE
Figure 3. Donor (D) and acceptor (A) building blocks used and their cartoon representations.
Figure 4. Formation of catenanes via pathway I.
Figure 5. HPLC analysis of an aqueous library composed of D2a and
A2 (1:1 molar ratio, 5 mM total), in the presence of 1 M of NaNO3.
Absorbance was recorded at 260 nm. The minor species present in the
library are described in the Supporting Information.
all containing a [-D2a-D2a-] dimer interlocked with either
[-A2-A2-], [-A1-A2-], or [-A1-A4-] acceptor dimers.
Their respective paired DADD catenanes were also identified.
The different yields of DADA catenanes suggest that the size of
the acceptor dimer influences their efficiency of formation, with
the tightest dimer [-A1-A4-] leading to a 70% yield of the
corresponding [2]catenane. Below we present a detailed discus-
sion of the libraries that follow pathway I.
Acceptor A1, with short side chains, forms an extremely tight
[-A1-A1-] dimer, which cannot possibly lead to the forma-
tion of alternating DADA catenanes.8,13 Hence, the first DADA
catenane (Cat-1) was identified, along with its predicted DADD
paired catenane (Cat-2), in a library containing the slightly
longer acceptor A2 and D2a (Figure 5).
doubly charged molecular ion (m/z of 1065.0), corresponding to
the mass of a tetramer composed of two donor and two acceptor
units. The largest fragments observed in MS/MS have an m/z of
1171.0 and 959.0, corresponding to the mass of the acceptor
homodimer and the donor homodimer, respectively. This frag-
mentation pattern is characteristic of a catenane and it follows the
expected fragmentation for a DADA catenane, i.e., a ring contain-
ing two acceptors interlocked with another ring consisting of two
donors.
The presence of the two paired catenanes (Cat-1 and Cat-2)
supports the proposed mechanism, whereby the choice of the
threaded building block is directed by donor-acceptor interac-
tions, while the closing step of catenation is driven by hydrophobic
Cat-1 was characterized by mass spectrometry (MS and MS/
MS), as shown in Figure 6. The ESI-MS (negative ion) shows a
3201
dx.doi.org/10.1021/ja111407m |J. Am. Chem. Soc. 2011, 133, 3198–3207

Journal of the American Chemical Society
ARTICLE
Figure 8. Partial 1H NMR spectrum (D2O, 298 K, 500 MHz) of DADA
catenane Cat-3 (the peaks labeled with * are associated with a [-A1-
A2-] dimer impurity). The outer NDI, corresponding to the singlet at
8.0 ppm, was assigned arbitrarily to A1.
Figure 6. Tandem (a) MS and (b) MS/MS fragmentation of DADA
catenane Cat-1.
Figure 7. HPLC analysis of an aqueous library composed of D2a, A1,
and A2 (2:1:1 molar ratio, 5 mM total), in the presence of 1 M of
NaNO3. Absorbance was recorded at 260 nm. The minor species
present in the library are described in the Supporting Information.
effects. The low yield of both catenanes (Cat-1, 2% and Cat-2,
6% of the library) is unsurprising, as their respective rings are
relatively large, reducing the likelihood of a strong interaction
between the aromatic units. Furthermore, even though the
building block ratio (1:1) should not favor the formation of a
structure containing three donor units, the DADD catenane Cat-2
is more efficiently formed, indicating that the overall catenation
process is not dominated by donor-acceptor interactions.
Considering the size of the [-A-A-] dimer, if [-A1-A1-]
is too tight, the larger [-A2-A2-] dimer is already too loose
to allow formation of a DADA catenane efficiently. Therefore, it
was not surprising to observe that replacing A2 by similarly sized
A3 did not bring much improvement. The behavior of the longer
A4 is relatively complex and will be discussed in the end of this
work, but as expected, in the case of the acceptor building block
with the longest side chains, A5, no catenane was observed: only
a limited number of small macrocycles, including the cyclic
acceptor monomer, were present at the stationary state (see
the Supporting Information).
Figure 9. Expanded region of the NOESY spectrum (D2O, 278 K, 500
MHz, evolution time: 800 ms) of DADA catenane Cat-3 (the peaks
labeled with * are associated with a [-A1-A2-] dimer impurity). The
outer NDI, corresponding to the broad singlet at 7.8 ppm, was assigned
arbitrarily to A1.
this catenane is higher than that of Cat-1. Once again, the DADD
paired catenane, Cat-5, is present in the library, supporting the
proposed formation pathway.
Naturally, increasing the complexity of the DCLs opens the
possibility for other pathways to compete with pathway I. This
explains the formation of other types of catenanes, such as the
DAAD Cat-4 (Figure 7), and will be discussed later.
Introducing two acceptor building blocks in the presence of
donor D2a allowed easy access to [-A-A-] dimers of inter-
mediate sizes to explore further the formation of catenanes via
pathway I. This is illustrated by a library composed of D2a, A1,
and A2 in a 2:1:1 molar ratio, shown in Figure 7. In this library,
the only DADA catenane formed is Cat-3, which contains the
tight mixed acceptor dimer [-A1-A2-]. The yield (15%) of
Cat-3 was isolated and characterized by 1H NMR (Figures 8
and 9). It displays two conformations in a molar ratio of 2:1, as
either A1 or A2 can be situated in the inner or outer positions of
the catenane. 1H NMR (500 MHz, 298 K, D2O) shows two sets
of signals for each conformation in the acceptor region (7.55-
8.15 ppm), assigned by correlation spectroscopy (COSY): one
upfield shifted (inner NDI) and one downfield shifted (outer
3202
dx.doi.org/10.1021/ja111407m |J. Am. Chem. Soc. 2011, 133, 3198–3207

Journal of the American Chemical Society
ARTICLE
NDI). A similar pattern, with two sets of signals for each con-
formation, was observed in the donor region (5.55-7.05 ppm).
At low temperature (278 K), cross-peaks between the donor and
acceptor units of the major conformation were observed in
NOESY spectra, confirming the alternating DADA structure of
Cat-3 (Figure 9).
the acceptor ring (Figure 11), as reflected by the yields that vary
from 2% to 70% of the library. Cat-3 and Cat-6 are deep red in
color in aqueous solution, which is indicative of the presence of
charge-transfer interactions between the complementary aromatic
units.
A library composed of D2a, A1, and A4 in a 2:1:1 molar ratio
(Figure 10) further illustrates the link between the size of the
acceptor ring and the efficiency of the formation of DADA
catenanes in the libraries. The [-A1-A4-] acceptor ring is very
tight, and the corresponding catenane Cat-6 is now formed in the
unusually high yield of 70%. The DADD paired catenane Cat-5
is still present, but is now a minor product. The formation of
Cat-6 in such high yield indicates a combination of optimal
donor-acceptor and hydrophobic interactions and their efficient
cooperation, leading to this particularly favorable catenane.
Because of its lack of symmetry, Cat-6 is produced as a pair of
diastereomers, each having two conformers, as either acceptor
can be situated in the inner or outer positions of the catenane.
Figure 12. Formation of catenanes via pathway II.
1
The H NMR of Cat-6 is consequently extremely complex.
However, analysis of the acceptor region by a combination of the
COSY and NOESY spectra (7.10-8.15 ppm) allowed correla-
tion of the signals expected from a mixture of four isomeric
catenanes (see the Supporting Information).
In conclusion, the D2a donor building block forms a stable
[-D-D-] dimer, which is thekey intermediate for theformation
of DADA and DADD catenanes via pathway I. The efficiency of
the DADA catenane formation is directly linked to the tightness of
3.2. Exploring Pathway II. Pathway II (Figure 12) occurs
under two circumstances. First, the donor building block must
have rather long side chains (D1b or D2b) so that it favors the
formation of the [-D-] cyclic monomer and restricts the forma-
tion of a stable [-D-D-] dimer,8a,8b,8d,13 thus preventing the
formation of catenanes through pathway I. Second, the formation of
the threaded complex into the cavity of the [-A-A-] dimer must
be prevented, blocking the formation of catenanes via pathway III.
This is the case when using A1: as the [-A1-A1-] dimer is too
tight to allow threading through its cavity,8,13 the [-D-A-]
heterodimer becomes the only species that can potentially form a
catenane. Composed of a long donor building block and a short
acceptor, the [-D-A-] dimer has a relatively small cavity, suitable
for strong interactions with an acceptor building block. A pair of
ADAA and DAAD catenanes should be formed through pathway II.
However, the necessity of having a tight [-A-A-] dimer implies
that the ADAA catenane cannot actually be formed; hence, only one
type of catenane, DAAD, was observed.
Many of those DAAD catenanes have previously been published
by our group8 and now find their place in the generic mechanism
described here, such as Cat-7, formed in a library of D1b and A1
(1:1 molar ratio),8d (Figure 13a). A similar catenane was observed
when replacing D1b by its 2,6-isomer D2b.8b
The short donor D2a is also suitable for the formation of
a DAAD catenane via pathway II when mixed with A1
(Figure 13b).8c However, we have previously shown that the
use of D2a also favors pathway I, and the two pathways now
compete. According to our mechanism, pathways I and II should
Figure 10. HPLC analysis of an aqueous library composed of D2a, A1,
and A4 (2:1:1 molar ratio, 5 mM total), in the presence of 1 M of
NaNO3. Absorbance was recorded at 260 nm. The minor species
present in the library are described in the Supporting Information.
Figure 11. Relationship between the size of the acceptor ring and the efficiency of DADA catenane formation. The donor ring is [-D2a-D2a-] in
each case.
3203
dx.doi.org/10.1021/ja111407m |J. Am. Chem. Soc. 2011, 133, 3198–3207

Journal of the American Chemical Society
ARTICLE
Figure 14. Formation of catenanes via pathway III.
Figure 13. HPLC analysis of aqueous libraries composed of (a) D1b
and A1 (1:1 molar ratio, 5 mM total) and (b) D2a and A1 (1:1 molar
ratio, 5 mM total). The libraries were prepared with 1 M of NaNO3.
Absorbance was recorded at (a) 292 nm and (b) 260 nm.8c,8d
lead to two pairs of two catenanes, respectively DADD/DADA,
and DAAD/ADAA. Once again, because the [-A1-A1-] does
not allow threading of an aromatic group through its cavity, only
one type of catenane can actually be observed for each pathway:
DAAD Cat-4 (pathway II) and DADD Cat-5 (pathway I). Both
catenanes, composed of short donor and acceptor building
blocks, are the tightest catenanes obtained to date for these
two types of structures and are consequently formed in the best
yields (55% for DAAD Cat-4 in a D:A = 1:1 molar ratio and 50%
for DADD Cat-5 in a D:A = 3:1 molar ratio). This agrees with
our previous observation that tight rings are necessary for the
efficient formation of catenanes.
The two catenanes were isolated and characterized. Interest-
ingly, the green DAAD catenane Cat-4 and the orange DADD
catenane Cat-5 exhibit different UV-vis spectra8c,8d from the
DADA catenanes described above, due to the particular arrange-
ment of their donor and acceptor units.
Figure 15. HPLC analysis of aqueous libraries composed of D1b, A1,
and A2 (5 mM total): (a) 2:1:1 molar ratio and (b) 1:2:1 molar ratio.
DCLs were prepared with 1 M of NaNO3. Absorbance was recorded at
383 nm. The minor species present in the libraries are described in the
Supporting Information.
limitation for this mechanism. In the previous cases described
(pathways I and II), the smallest and most rigid donor and/or
acceptor building blocks were necessary to lead efficiently to the
most compact catenanes. The libraries were consequently quite
simple, composed of catenanes and small macrocycles. To access
catenanes through pathway III, more flexible building blocks are
necessary, increasing the complexity of the libraries: large
macrocycles, such as trimers or tetramers, which can fold and
thus bury some of their large hydrophobic surfaces, become
serious competitors for the formation of the catenanes.
The first library that follows pathway III was composed of
donor D1b with the acceptor building blocks A1 and A2 in a
2:1:1 molar ratio (Figure 15a). The DADA Cat-9 is formed from
the tightest possible acceptor ring in this system [-A1-A2-],
interlocked with the donor ring that is much larger than in
previous catenanes of this type, leading to a very low yield. This
shows that both the size of the acceptor and the donor rings play
a role in the efficiency of formation of the DADA catenane. Its
paired AADA catenane Cat-8 is observed for the first time and
identified by tandem MS and MS/MS (Figure 16). The library
also contained the DAAD Cat-7 formed via pathway II.
3.3. Exploring Pathway III. Pathway III (Figure 14), which
leads to the formation of the new pair of catenanes DADA and
AADA, occurs under particular conditions. The donor building
block must have long side chains (D1b or D2b), to limit the
formation of the [-D-D-] dimer. The acceptor building block
must also have long enough side chains so that the [-D-A-]
dimer is too loose to form any stable threaded complex. How-
ever, it must be able to form a [-A-A-] dimer tight enough to
allow a favorable interaction with a donor moiety. Therefore, the
size of the acceptor building block is critical to the formation of
catenanes through pathway III. Unfortunately, it is not the only
3204
dx.doi.org/10.1021/ja111407m |J. Am. Chem. Soc. 2011, 133, 3198–3207

Journal of the American Chemical Society
ARTICLE
Figure 16. Tandem (a) MS and (b) MS/MS fragmentation of AADA catenane Cat-8.
1
Figure 17. Partial H NMR spectrum (D2O, 298 K, 500 MHz) of
AADA catenane Cat-8. The inner NDI, corresponding to the doublets at
8.25 and 8.10 ppm, was assigned arbitrarily to A2.
ESI-MS (negative ion) of Cat-8 shows a doubly charged
molecular ion (m/z of 1031.7), corresponding to the mass of a
tetramer containing only one donor and three acceptor units.
The largest fragments observed in MS/MS have an m/z of
1056.5 and 1006.6, corresponding to the mass of the [-A1-
A2-] homodimer and the [-D2b-A1-] heterodimer,
respectively.
The AADA Cat-9 could be amplified up to 30% yield, along
with its isomeric macrocycle [-A2-D1b-A1-A1-], by using
a biased ratio of building blocks (Figure 15b). The amplification
of this tetramer clearly shows that large macrocycles can fold to
minimize hydrophobic exposure and be in direct competition
with the formation of the corresponding catenanes.
Both Cat-8 and its corresponding tetramer were isolated and
analyzed by 1H NMR spectroscopy (Figure 17 and Supporting
Information). The two compounds exhibit significantly different
1H NMR spectra (500 MHz, D2O), with the spectrum of the
catenane displaying the expected upfield shifts for the aromatic
protons due to the stacked structure. Only one conformation was
observed for the AADA catenane Cat-8. Three sets of signals
were correlated by COSY in the acceptor region (8.05-8.50
ppm), corresponding to the two outer and one inner NDIs. Only
one set of signals corresponding to one DN unit was found in the
Figure 18. Expanded region of the NOESY spectrum (D2O, 278 K, 500
MHz, evolution time: 800 ms) of AADA catenane Cat-8 (the peaks
labeled with * are associated with a [-A1-A2-] dimer impurity).
donor region (5.95-6.90 ppm). At 278 K, cross-peaks between
the aromatic units were observed by NOESY, confirming the
unusual AADA structure of Cat-8 (Figure 18).
As expected, a library composed of D2b, A1, and A2 (2:1:1
molar ratio, Figure 19) contains the DADA catenane Cat-11 and
its paired AADA catenane Cat-10. The presence of Cat-9 again
indicates that pathway II is competing with pathway III.
In a biased library with a ratio of D2b:A1:A2 of 1:2:1, the yield
of this catenane increased up to 10% of the library, allowing for its
isolation (see the Supporting Information). Other AADA cate-
nanes were identified in libraries based on either long donors
D1b or D2b. Only the smallest acceptor rings lead to the
3205
dx.doi.org/10.1021/ja111407m |J. Am. Chem. Soc. 2011, 133, 3198–3207

Journal of the American Chemical Society
ARTICLE
3.4. Competing Pathways. We have described above design
protocols that access catenanes selectively via pathway I, II, or III,
but we also observed in the most complex systems that catenanes
can be formed via two simultaneous pathways. It is therefore
conceivable that a library fulfilling the conditions of all three
Pathways will generate all the types of catenanes. This is beautifully
illustrated in Figure 20 by a library containing the short donor
building block D2a and the asymmetric acceptor building block A4
(1:1 molar ratio, 5 mM). The behavior of A4 is driven by its shorter
side chain, making the formation of a DAAD catenane possible
(pathway II). However, the [-A-A-] homodimer is now large
enough to allow the formation of DADA and AADA catenanes
(pathway III), and the use of D2a allows the formation of a DDAD
catenane (pathway I). Indeed, all the four possible catenanes DAAD
(Cat-15), DADD (Cat-16), DADA (Cat-14), and AADA (Cat-
13) were observed in the same library.
The presence of two peaks on the HPLC trace for each of Cat-
14 and Cat-15 can be explained by the existence of different
isomers or of slowly interconverting conformations. The synth-
esis of all four types of catenanes in the same library clearly shows
that there is little energetic difference between them. This allows
us to conclude beyond a doubt that the formation of the unusual
AADA, DAAD, and AADA stacked catenanes is not significantly
disfavored when compared to the typical DADA catenane.
Figure 19. HPLC analysis of an aqueous library composed of D2b, A1,
and A2 (2:1:1 molar ratio, 5 mM total), in the presence of 1 M of
NaNO3. Absorbance was recorded at 383 nm. The minor species
present in the library are described in the Supporting Information.
4. CONCLUSIONS
A total of 22 new donor-acceptor [2]catenanes with different
arrangements of the aromatic units (DAAD, DDAD, DADA, and
AADA) have been identified by dynamic combinatorial chem-
istry (see the Supporting Information). The observation that
complete oxidation freezes exchange in such a way that the
catenanes become kinetic bottlenecks (or traps) takes this work
away from pure dynamic combinatorial chemistry into a hybrid
realm where there is a complex interplay of kinetics and
thermodynamics. While this is in some sense disappointing, it
is not entirely surprising and it also provides an opportunity to
probe reaction pathways that are necessarily invisible in a truly
thermodynamic equilibrium.
Donor-acceptor interactions, which conventionally lead to
DADA catenanes only, were partially overcome by using a highly
polar medium, allowing efficient access to catenanes with unusual
DADD, DAAD, and ADAA stacking sequences. The formation
of these previously exotic catenanes in the DCLs can be convinc-
ingly explained by taking into account the relative contributions
of donor-acceptor and hydrophobic effects in their assembly.
A close analysis of the DCLs studied allowed us to propose a
mechanistic pathway for their synthesis and also to predict the
outcome of libraries a priori. Donor-acceptor interactions are
crucial in the mechanism, but the hydrophobic effect determines
the overall outcome of catenation process. Catenanes containing
different stacking sequences of the donor and acceptor units can
be selectively formed following one of the three synthetic path-
ways proposed simply by choosing the correct building blocks in
the DCL setup.
These results challenge conventional thinking that has limited
the scope of donor-acceptor [2]catenanes to strictly alternating
stacks of donor and acceptor aromatic units. They indicate that
an approach to catenane synthesis that takes into account only
donor-acceptor interactions is unnecessarily limiting. To date,
aromatic interactions have been satisfactorily described in terms
of electrostatic interactions, which can only be enhanced by a
solvophobic effect. We have shown here that the solvophobic
Figure 20. HPLC analysis of an aqueous library composed of D2a and
A4 (1:1 molar ratio, 5 mM total), in the presence of 1 M of NaNO3.
Absorbance was recorded at 260 nm.
formation of AADA catenanes, but most of the libraries are
relatively complex, and the formation of large macrocycles seems
to be favored over the formation of the catenanes, which are
only present, in most cases, as traces (see the Supporting
Information).
To conclude, using flexible donors D1b or D2b leads to the
formation of DADA and AADA catenanes through pathway III.
The formation of DADA catenane is not favored through this
mechanism, because the donor ring is too large to allow an
optimum overlap of the complementary aromatic units. There is
a delicate balance in the formation of AADA catenanes, as they
are in competition with the formation of large macrocycles such
as trimers and tetramers. However, AADA catenane Cat-8 can be
amplified up to 30% by optimizing the choice of the building
blocks used in order to access the tightest possible acceptor and
donor rings. Cat-8 and Cat-10 were isolated as yellow-brown
products, showing that AADA catenanes exhibit different optical
properties than the other types of catenanes described above.
3206
dx.doi.org/10.1021/ja111407m |J. Am. Chem. Soc. 2011, 133, 3198–3207

Journal of the American Chemical Society
ARTICLE
effect can overcome and direct syntheses based on supramole-
cular interactions between aromatic moieties. This observation
will allow chemists to design and indeed synthesize complex
structures that were previously inaccessible.
N.; Zhao, X.-L.; Huang, F. Org. Lett. 2009, 3350. (h) Liu, M.; Li, S.;
Zhang, M.; Zhou, Q.; Wang, F.; Hu, M.; Fronczek, F. R.; Li, N.; Huang,
F. Org. Biomol. Chem. 2009, 7, 1288. (i) Blanco, V.; Abella, D.; Pía, E.;
Platas-Iglesias, C.; Peinador, C.; Quintela, J. M. Inorg. Chem. 2009, 48,
4098. (j) Koshkakaryan, G.; Parimal, K.; He, J.; Zhang, X.; Abliz, Z.;
Flood, A. H.; Liu, Y. Chem.—Eur. J. 2008, 14, 10211. (k) Liu, Y.;
Bruneau, A.; He, J.; Abliz, Z. Org. Lett. 2008, 10, 765. (l) Raehm, L.;
Hamilton, D. G.; Sanders, J. K. M. Synlett 2002, 11, 1743. (m) Hansen,
J. G.; Feeder, N.; Hamilton, D. G.; Gunter, M. J.; Becher, J.; Sanders,
J. K. M. Org. Lett. 2000, 2, 449. (n) Zheng, Q.; Hamilton, D. G.; Feeder,
N.; Teat, S. J.; Goodman, J. M.; Sanders, J. K. M. New J. Chem. 1999, 23,
897. (o) Hamilton, D. G.; Prodi, L.; Feeder, N.; Sanders, J. K. M. J. Chem.
Soc., Perkin Trans. 1 1999, 1057.
’ ASSOCIATED CONTENT
S
Supporting Information. Detailed procedure for build-
b
ing block synthesis, library preparations, HPLC/LC-MS meth-
ods and data, and UV-vis and NMR spectra of isolated
catenanes and macrocycles. This material is available free of
charge via the Internet at http://pubs.acs.org.
(5) Cubberley, M. S.; Iverson, B. L. J. Am. Chem. Soc. 2001, 123,
7560.
(6) Wang, W.; Wang, L.; Palmer, B. J.; Exarhos, G. J.; Li, A. D. Q. J.
’ AUTHOR INFORMATION
Corresponding Author
g.d.pantos@bath.ac.uk; jkms@cam.ac.uk
Am. Chem. Soc. 2006, 128, 11150.
(7) Murase, T.; Otsuka, K.; Fujita, M. J. Am. Chem. Soc. 2010, 132,
7864.
(8) (a) Au-Yeung, H. Y.; Panto-s, G. D.; Sanders, J. K. M. J. Org.
Chem. DOI: 10.1021/jo201001981p. (b) Au-Yeung, H. Y.; Panto-s,
G. D.; Sanders, J. K. M. Angew. Chem., Int. Ed. 2010, 49, 5331. (c) Au-
Yeung, H. Y.; Panto-s, G. D.; Sanders, J. K. M. J. Am. Chem. Soc. 2009,
131, 16030. (d) Au-Yeung, H. Y.; Panto-s, G. D.; Sanders, J. K. M. Proc.
Natl. Acad. Sci. U.S.A. 2009, 106, 10466.
’ ACKNOWLEDGMENT
We thank EPSRC, the Croucher Foundation, and Pembroke
College (Cambridge) for financial support and Dr Ana Belenguer
for maintaining the HPLC facility.
(9) (a) Herrmann, A. Org. Biomol. Chem. 2009, 7, 3195. (b) Ladame,
S. Org. Biomol. Chem. 2008, 6, 219. (c) Lehn, J.-M. Chem. Soc. Rev. 2007,
36, 151. (d) Rozenman, M. M.; McNaughton, B. R.; Liu, D. R. Curr.
Opin. Chem. Biol. 2007, 11, 259. (e) Corbett, P. T.; Leclaire, J.; Vial, L.;
West, K. R.; Wietor, J.-L.; Sanders, J. K. M.; Otto, S. Chem. Rev. 2006,
106, 3652. (f) de Bruin, B.; Hauwert, P.; Reek, J. N. H. Angew. Chem., Int.
Ed. 2006, 45, 2660.
(10) Other examples of catenane discovery from DCLs: (a) Chung,
M.-K.; White, P. S.; Lee, S. J.; Gagnꢀe, M. R. Angew. Chem., Int. Ed. 2009,
48, 8683. (b) West, K. R.; Ludlow, R. F.; Corbett, P. T.; Besenius, P.;
Mansfeld, F. M.; Cormack, P. A.G.; Sherrington, D. G.; Goodman, J. M.;
Stuart, M. C. A.; Otto, S. J. Am. Chem. Soc. 2008, 130, 12218. (c) Lam,
R. T. S.; Belenguer, A.; Roberts, S. L.; Naumann, C.; Jarrosson, T.; Otto,
S.; Sanders, J. K. M. Science 2005, 308, 667.
(11) Disulfide exchange was shown to stop after 1 day of oxidation:
the catenanes are progressively formed in the libraries until the libraries
reach a steady state, corresponding to the full oxidation of the library
members. The quantity of catenane formed can be further increased by
adding 20% dithiothreitol to reinitiate the exchange process, showing
that thermodynamic equilibrium has not been reached yet (see the
Supporting Information).
(12) (a) Fujita, M.; Ibukuro, F.; Hagihara, H.; Ogura, K. Nature
1994, 367, 720. (b) Fujita, M.; Ibukuro, F.; Ogura, K. J. Am. Chem. Soc.
1995, 117, 4175.
’ REFERENCES
(1) (a) Hunter, C. A.; Lawson, K. R.; Perkins, J.; Urch, C. J. J. Chem.
Soc., Perkin Trans. 2 2001, 651. (b) Hunter, C. A. Chem. Soc. Rev. 1994,
23, 101. (c) Hunter, C. A.; Sanders, J. K. M. J. Am. Chem. Soc. 1990, 112,
5525.
(2) (a) Ramkumar, S. G.; Ramakrishnan, S. Macromolecules 2010,
43, 2307. (b) Zhang, W.; Dichtel, W. R.; Stieg, A. Z.; Benítez, D.;
Gimzewski, J. K.; Heath, J. R.; Stoddart, J. F. Proc. Natl. Acad. Sci. U.S.A.
2008, 105, 6514. (c) Bradford, V. J.; Iverson, B. L. J. Am. Chem. Soc.
2008, 130, 1517. (d) Zhou, Q.-Z.; Jia, M.-X.; Shao, X.-B.; Wu, L.-Z.;
Jiang, X.-K.; Li, Z.-T.; Chen, G.-J. Tetrahedron 2005, 61, 7117. (e) Zhao,
X.; Jia, M.-X.; Jiang, X.-K.; Wu, L.-Z.; Li, Z.-T.; Chen, G.-J. J. Org. Chem.
2004, 69, 270. (f) Ghosh, S.; Ramakrishnan, S. Angew. Chem., Int. Ed.
2004, 43, 3264. (g) Zhou, Q.-Z.; Jiang, X.-K.; Shao, X.-B.; Chen, G.-J.;
Jia, M.-X.; Li, Z.-T. Org. Lett. 2003, 5, 1955. (h) Gabriel, G. J.; Iverson,
B. L. J. Am. Chem. Soc. 2002, 124, 15174. (i) Ngyuen, J. Q.; Iverson, B. L.
J. Am. Chem. Soc. 1999, 121, 2639. (j) Lokey, R. S.; Iverson, B. L. Nature
1995, 375, 303.
(3) For some recent examples:(a) Mullen, K. M.; Davis, J. J.; Beer,
P. D. New J. Chem. 2009, 33, 769. (b) Lin, T.-C.; Lai, C.-C.; Chiu, S.-H.
Org. Lett. 2009, 11, 613. (c) Ikeda, T.; Higuchi, M.; Kurth, D. G. Chem.
—Eur. J. 2009, 15, 4906. (d) Yoon, I.; Benítez, D.; Zhao, Y.-L.; Miljaniꢀc,
ꢁ
O. S.; Kim, S,-Y.; Tkatchouk, E.; Leung, K. C.-F.; Khan, S. I.; Goddard,
W. A., III; Stoddart, J. F. Chem.—Eur. J. 2009, 15, 1115. (e) Ikeda, T.;
Higushi, M.; Kurth, D. G. J. Am. Chem. Soc. 2009, 131, 9158. (f)
Cagulada, A. M.; Hamilton, D. G. J. Am. Chem. Soc. 2009, 131, 902. (h)
Pascu, S. I.; Naumann, C.; Kaiser, G.; Bond, A. D.; Sanders, J. K. M.;
Jarrosson, T. Dalton Trans. 2007, 3874. (g) Pascu, S. I.; Jarrosson, T.;
Naumann, C.; Otto, S.; Kaiser, G.; Sanders, J. K. M. New J. Chem. 2005,
29, 80.
(4) For some recent examples:(a) Wang, C.; Olson, M. A.; Fang, L.;
Benítez, D.; Tkatchouk, E.; Basu, S.; Basuray, A. N.; Zhang, D.; Zhu, D.;
Goddard, W. A.; Stoddart, J. F. Proc. Natl. Acad. Sci. U.S.A. 2010, 107,
13991. (b) Stoddart, J. F. Chem. Soc. Rev. 2009, 38, 1802. (c) Zhao, Y.-L.;
Trabolsi, A.; Stoddart, J. F. Chem. Commun. 2009, 4844. (d) Spruell,
J. M.; Paxton, W. F.; Olsen, J.-C.; Benítez, D.; Tkatchouk, E.; Stern, C. L.;
Trabolsi, A.; Friedman, D. C.; Goddard, W. A., III; Stoddart, J. F. J. Am.
Chem. Soc. 2009, 131, 11571. (e) Ramos, S.; Alcalde, E.; Stoddart, J. F.;
White, A. J. P.; Williams, D. J.; Pꢀerez-García, L. New J. Chem. 2009, 33,
300. (f) Cao, D.; Amelia, M.; Klivansky, L. M.; Koshkakaryan, G.; Khan,
S. I.; Semeraro, M.; Silvi, S.; Venturi, M.; Credi, A.; Liu, Y. J. Am. Chem.
Soc. 2010, 132, 1110. (g) Li, S.; Liu, M.; Zheng, B.; Zhu, K.; Wang, F.; Li,
(13) (a) Au-Yeung, H. Y.; Cougnon, F. B. L.; Otto, S.; Panto-s, G. D.;
Sanders, J. K. M. Chem. Sci. 2010, 567. (b) Au-Yeung, H. Y.; Pengo, P.;
Panto-s, G. D.; Otto, S.; Sanders, J. K. M. Chem. Commun. 2009, 419.
(14) Equivalently, the complexation of linear oligomers and
single building blocks can lead to their cyclization to give the same
intermediates.
(15) Pengo, P.; Panto-s, G. D.; Otto, S.; Sanders, J. K. M. J. Org. Chem.
2006, 71, 7063.
3207
dx.doi.org/10.1021/ja111407m |J. Am. Chem. Soc. 2011, 133, 3198–3207