Dissipative Assembly of Macrocycles Comprising Multiple Transient Bonds

Article abstract of DOI:10.1002/ange.202001523

Dissipative assembly has great potential for the creation of new adaptive chemical systems. However, while molecular assembly at equilibrium is routinely used to prepare complex architectures from polyfunctional monomers, species formed out of equilibrium

Full text of DOI:10.1002/ange.202001523

Angewandte
Forschungsartikel
Chemie
Zitierweise:
Internationale Ausgabe: doi.org/10.1002/anie.202001523
Deutsche Ausgabe: doi.org/10.1002/ange.202001523
Supramolecular Chemistry Hot Paper
Dissipative Assembly of Macrocycles Comprising Multiple Transient
Bonds**
Mohammad Mosharraf Hossain, Joshua L. Atkinson und C. Scott Hartley*
Abstract: Dissipative assembly has great potential for the
creation of new adaptive chemical systems. However, while
molecular assembly at equilibrium is routinely used to prepare
complex architectures from polyfunctional monomers, species
formed out of equilibrium have, to this point, been structurally
very simple. In most examples the fuel simply effects the for-
mation of a single short-lived covalent bond. Herein, we show
that chemical fuels can assemble bifunctional components into
macrocycles containing multiple transient bonds. Specifically,
dicarboxylic acids give aqueous dianhydride macrocycles on
treatment with a carbodiimide. The macrocycles are assembled
efficiently as a consequence of both fuel-dependent and fuel-
independent mechanisms; they undergo slower decomposition,
building up as the fuel recycles the components, and are a fa-
vored product of the dynamic exchange of the anhydride
bonds. These results create new possibilities for generating
structurally sophisticated out-of-equilibrium species.
synthesis, dynamic chemical libraries at equilibrium can
themselves exhibit complex responsive behavior.
[39,46–50]
Merging the structural sophistication of equilibrium as-
sembly with the functional possibilities of dissipative assem-
bly would enable new behavior. For example, transient hosts
could be used for active transport, and dissipative dynamic
libraries should exhibit new mechanisms of responding to
their environments. However, it is not yet possible to as-
semble polyfunctional monomers into closed architectures
using the energy of chemical fuels. The challenge is overco-
ming competing polymerization. For assembly at equilibrium,
target structures can be produced in high yields because un-
stable kinetic products decompose, providing a form of error
[
27]
correction.
In general, closed architectures are favored
entropically and thus can be produced efficiently so long as
[
51]
the precursors are suitably preorganized. It is inherently
more complex to achieve similar results away from equilib-
rium. By definition, transiently formed products are not
thermodynamically favored and the time scales of the
chemistry are limited. However, new mechanisms for error
correction are also possible.
Introduction
The use of chemical fuels to drive abiotic systems out of
equilibrium is challenging but offers great potential for novel
Herein, we demonstrate the efficient assembly of poly-
functional monomers into macrocycles using the energy of
a chemical fuel. As shown in Scheme 1, we focus on diacid
substrate DA1 dissolved in water. On treatment with a car-
bodiimide, it assembles into macrocycle M1 in competition
[
1–8]
adaptive behaviors.
Using a variety of fuel reactions, no-
[
9–12]
[13]
[14,15]
nequilibrium gels;
molecular switches, shuttles,
and
molecular
[
16]
[17–20]
motors;
hosts;
supramolecular assemblies;
and other species
[
21,22]
[23,24]
have been reported. The
behaviors of these dissipative systems can be complex, typi-
cally resulting from changes in noncovalent interactions fol-
lowing the generation of a new transient species by the fuel.
The underlying (covalent) chemical transformations, ho-
wever, have been very simple; the role of the fuel is generally
the formation of a single unstable covalent bond. This sim-
plicity contrasts sharply with the sophistication of dynamic
systems under thermodynamic control, in which the simulta-
with polymerization to oligomers O1 . Compound M1, com-
n
prising two unstable bonds (the anhydrides), is formed very
efficiently. While it is a simple two-component macrocycle,
analogous structures formed under thermodynamic control
exhibit interesting host–guest and self-assembling be-
[
52–56]
havior.
We find that the principles underlying the ef-
ficiency of assembly are distinctly different from those di-
recting equilibrium assembly.
The chemical fuel for this system is the well-known car-
bodiimide reagent EDC (N-(3-dimethylaminopropyl)-N’-
ethylcarbodiimide hydrochloride). Carbodiimides are simply
dehydrating agents that are kinetically stable in water; their
[
25,26]
neous formation of many covalent bonds
can be used to
ladders,
This structural
[
27–29]
[30–34]
[35,36]
self-assemble macrocycles,
cages,
[
37–39]
[40–42]
knots,
and other closed architectures.
complexity leads to new function. For example, the products
[
30,43]
have interior spaces that can be used to bind guests,
they can be incorporated into porous materials;
and
beyond
hydration is catalyzed by carboxylic acids, yielding the un-
[
44,45]
[57,58]
stable anhydrides as byproducts.
has been used by Boekhoven
Recently, this chemistry
[
23,24,59]
[11,22]
and us
to generate
transient anhydrides with functional behavior. It is well-suited
to polyfunctional assembly. Unlike other fuel reactions (e.g.,
methylation ), only the energy of the carbodiimide fuel is
used; it does not transfer structural components to the
transient state.
[
*] Dr. M. M. Hossain, J. L. Atkinson, Dr. C. S. Hartley
Department of Chemistry & Biochemistry, Miami University
Oxford, OH 45056 (USA)
[10]
E-Mail: scott.hartley@miamioh.edu
[
**] A previous version of this manuscript has been deposited on
a preprint server (https://doi.org/10.26434/chemrxiv.11426007).
Supporting information and the ORCID identification number(s) for
the author(s) of this article can be found under:
https://doi.org/10.1002/anie.202001523.
&
&&&
ꢀ 2020 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. 2020, 132, 2 – 9
These are not the final page numbers!

Angewandte
Forschungsartikel
Chemie
Scheme 1. Assembly of diacid DA1 fueled by EDC.
À1
Results and Discussion
appears at 1794 cm , as shown in Figure 2. This peak is
characteristic of the symmetric stretching mode of the an-
hydride functional group, confirming that the underlying
chemistry occurs effectively. The intensity of the band plate-
aus within 30 min then decays slowly. The overall time scale of
this experiment is 15–20 h, consistent with the NMR moni-
toring results (and HPLC monitoring, described later).
The NMR and IR spectroscopy data indicate clean con-
version of DA1 to anhydrides and then its regeneration over
the course of hours, but do not indicate whether the system
Compound DA1 was designed to have limited confor-
mational mobility so that it would be well (but not too well)
preorganized for assembly into M1. The hexaglyme groups
were included to ensure good solubility in water. The syn-
thesis of DA1 is straightforward and is described in the Sup-
porting Information. Ab initio geometry optimization of
a simplified version of macrocycle M1 shows that the core
should be relatively free of strain, but distorted from planarity
into a saddle conformation by twisting of the tolane moieties
(
Supporting Information, Figure S1).
The reaction of DA1 (Scheme 1) with EDC was in-
[
60]
vestigated in D O at room temperature, maintaining the
2
pD at 5.5 using a [D ]pyridine buffer and the ionic strength (I)
5
at 1.0m using NaCl. These conditions closely parallel those we
[
61]
have used for other benzoic acids, although the pD was
increased from 4.5 to 5.5 here. The higher pD was chosen for
experimental convenience; in preliminary experiments at pD
4
.5, the DA1 was consumed very quickly, making the reac-
tions more difficult to monitor.
1
The system was first studied by H NMR spectroscopy.
Immediately after addition of the EDC, the reaction mixture
was inserted into the spectrometer and monitored for 2 h.
Spectra for a representative reaction, where 25 mm DA1 was
treated with 50 mm EDC, are shown in Figure 1. The aliphatic
region of the spectrum shows clean conversion of the EDC to
the corresponding urea EDU (Scheme 1). Inspection of the
aromatic region shows that all of the DA1 is consumed within
approximately 10 min, with no concurrent appearance of
sharp peaks that could be assigned to new species. Closer
inspection reveals a very broad feature in the aromatic region,
[62]
suggesting aggregation of the products of assembly. The
reaction mixture remains homogeneous throughout the ex-
periment (as do all others reported here). The signals for DA1
are absent over the course of hours but return once the
reaction has sat overnight, indicating that it is regenerated
once the fuel is consumed.
Figure 1. Reaction of DA1 (25 mm) with EDC (50 mm) monitored by
H NMR spectroscopy at various reaction times (pD=5.5, 250 mm
1
[
D ]pyridine buffer, I=1.0m (NaCl), 298 K, 500 MHz). The blank solu-
5
tion (bottom) represents a control experiment with just the addition of
The reaction was also followed by in situ IR spectroscopy.
Again 25 mm DA1 was treated with 50 mm EDC in buffered
D O instead of the EDC solution. The spectra are normalized to the
2
intensity of an internal standard (N,N-dimethylacetamide).
D O. Immediately after addition of the EDC, a new band
2
Angew. Chem. 2020, 132, 2 – 9
ꢀ 2020 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.de
&&&&
These are not the final page numbers!

Angewandte
Forschungsartikel
Chemie
Figure 2. Treatment of DA1 (25 mm) with EDC (50 mm) monitored by
À1
IR spectroscopy at 1794 cm (pD=5.5, 250 mm [D ]pyridine buffer,
5
À1
I=1.0m (NaCl), rt). Inset: Absorption peak at 1794 cm at various
time points. The blank consists of all components of the reaction mix-
ture except EDC.
favors macrocyclization to M1, polymerization to O1 , or
n
neither. We then monitored the system by HPLC. As a refe-
rence, it was possible to synthesize an authentic sample of M1
by treating DA1 with N,N’-dicyclohexylcarbodiimide follo-
wing a previously reported method (see Supporting In-
[
22]
formation).
While somewhat unstable, the macrocycle
could be unambiguously characterized by ESI mass spec-
trometry and NMR spectroscopy (Figures S22–S24, S53, and
S54).
The reaction of DA1 with EDC was carried out as before.
Representative chromatograms are shown in Figure 3. The
components of the mixtures elute in three distinct regions for
Figure 3. a) Assembly of DA1 (25 mm) on treatment with EDC
(50 mm) monitored by reversed-phase HPLC at various reaction times,
detected by absorbance at 280 nm (pD=5.5, 250 mm [D ]pyridine buf-
fer, I=1.0m (NaCl), rt). b) Dependence of the M1 peak area on total
reaction time.
5
DA1, O1 , and M1, with each peak characterized by LC/MS
n
(
Figures S25–S32). The identities of DA1 and M1 were also
confirmed by comparison with the authentic samples. As it is
the least-polar possible component of the mixture, the peak
for macrocycle M1 is well-separated from the carboxylic-acid-
containing species. The extinction coefficients of the various
species at the detection wavelength (280 nm) should be si-
milar (Figure S11), allowing qualitative comparisons of their
relative populations (EDC and EDU are not observed be-
cause they do not absorb at this wavelength; detection at
to M1 (Figures S33 and S34). We assign this broad peak to
stacked aggregates formed at higher concentrations, consi-
1
stent with the signal broadening observed by H NMR spec-
[63]
troscopy. The broad HPLC peak grows in more slowly than
the one for unaggregated M1, reaching its maximum area
after 2.5–3 h, and persists longer as the macrocycle de-
composes, suggesting that aggregation and disaggregation are
relatively slow and that stacking impedes hydrolysis. The as-
signment of the broad peak to aggregates was further con-
firmed by HPLC analysis of the independently synthesized
M1. A comparable feature was observed in its chromatogram,
but only if the sample was prepared as a solution in the
aqueous buffer used for the EDC-fueling experiments (Fi-
gures S14–S16). When the independently synthesized M1 was
injected directly from acetonitrile solution, in which ag-
2
20 nm shows that they elute at roughly 1 min).
Immediately after the addition of EDC, the HPLC ana-
lysis shows that assembly of DA1 gives predominantly olig-
omers O1 , with only a relatively small amount of macrocycle
n
M1. Within less than an hour, however, the chromatogram is
dominated by M1, indicating that macrocyclization is highly
efficient given time. The consumption of EDC maintains the
system for approximately 5 h at a nearly steady state. Com-
pound M1 then hydrolyzes slowly back to the starting DA1
over the course of roughly 10 h. Notably, the macrocycle ap-
pears to hydrolyze directly back to DA1 without any si-
gnificant buildup of acyclic oligomers. That is, the relative
[
64]
gregation is less significant,
only a sharp peak was ob-
served. Aggregation was also observed by dynamic light
scattering (Figures S17 and S18). A solution of 25 mm DA1 in
water yields an apparent average particle size of 2–3 nm,
[
65]
consistent with its structure.
On treatment with 50 mm
populations of M1 and O1 are not solely a function of the net
anhydride concentration.
As the concentration of M1 builds up, a broad feature
appears in the chromatograms immediately following its
sharp peak. LC/MS analysis indicates that it also corresponds
EDC, the average particle size increases sharply, peaking at
12–13 nm, before returning slowly to the original value for the
DA1 over the course of roughly 20 h.
n
The assembly of shape-persistent macrocycles into stak-
[66–68]
ked solution-phase aggregates is well-precedented,
par-
&
&&&
www.angewandte.de
ꢀ 2020 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. 2020, 132, 2 – 9
These are not the final page numbers!

Angewandte
Forschungsartikel
Chemie
ticularly for amphiphiles, like M1, dissolved in solvents that
products in the short term can self-correct within its limited
[
69,70]
are matched to the peripheral groups.
Our expectation of
experimental time scale. Several distinct processes could ac-
count for the high proportion of macrocycles in the products:
1) They could simply be the kinetic products of the reaction,
formed preferentially on reaction of the diacid with the fuel.
2) They could be kinetically stable compared to polymeric
byproducts; that is, the (aggregated) anhydrides of M1 could
undergo slower hydrolysis. Under these circumstances, even if
this system is that it is an isodesmic polymerization, at least
initially, although we cannot exclude subsequent higher-order
[
71]
aggregation.
Comparable aggregation would not be ex-
pected for DA1 as it has a smaller available surface area for
arene–arene stacking and its core is much more hydrophilic
(
e.g., charged at the experimental pH). Further evidence for
the promotion of macrocycle aggregation by water was ob-
tained by extracting reaction mixtures containing M1 with
chloroform-d. In this solvent, the aromatic regions of the
H NMR spectra show clearly defined signals that could be
assigned to DA1 and M1 alone (Figure S19). There is no peak
the assembly is completely unselective, any O1 produced
would decompose and the regenerated DA1 redistributed
n
between M1 and O1 on reaction with additional EDC, gra-
n
1
dually building up the concentration of M1. 3) All of the an-
hydride-containing species could exchange, allowing the sy-
stem to favor local minima on its free energy surface, pro-
vided that exchange is fast compared to decomposition. Pa-
ralleling thermodynamically controlled assembly, this process
would favor unstrained closed architectures because they are
entropically favored relative to polymer and because of added
broadening or significant concentration dependence for the
[
72]
signals assigned to M1.
While the HPLC analysis quite
clearly showed that M1 was the major component of the
mixtures before extraction, substantial amounts of DA1 were
always observed in the NMR spectra of extracted mixtures.
This suggests that significant hydrolysis occurs once the ag-
gregates are broken up in (water-saturated) chloroform.
The lifetime of M1 is, of course, dependent on the initial
concentrations of diacid DA1 and EDC. The effect of varying
EDC concentration (12.5–50 mm) added to constant DA1
[
27,51,73]
stability from aggregation.
These three processes are not mutually exclusive. Ho-
wever, it is clear that macrocycle M1 is not the kinetic product
of this system, since substantial quantities of oligomers O1n
are always observed in HPLC traces obtained at short reac-
tion times. Thus, process 1 is not significant.
(
25 mm) is shown in Figure 4 (top) (results for other concen-
trations of DA1 are given in Figures S2–S10). In all three
experiments, the maximum concentration of assembled M1 is
achieved within an hour. The lifetimes of M1 are roughly
proportional to the amount of fuel added. Similarly, for
a constant 50 mm EDC, shown in Figure 4 (bottom), the li-
fetime is increased with decreasing concentration of DA1
over the range of 12.5–50 mm, consistent with expectations.
With 50 mm DA1, the equimolar EDC should be consumed
almost immediately, leaving macrocycle M1 to simply und-
ergo hydrolysis. For lower concentrations of DA1 the remai-
ning excess EDC will regenerate the anhydride as it de-
composes, prolonging the lifetime of the transient state.
The results described above show that architectures lin-
ked by multiple transient bonds can be generated efficiently,
and that a nonequilibrium system that gives misassembled
Assembly-induced slowing of hydrolysis has been ob-
served by Boekhoven for transient anhydrides that undergo
[
23,24]
phase separation once formed.
To test whether this is
occurring here, we examined the behavior of monoacid Ac1,
as shown in Scheme 2, which mimics the structure of the an-
hydrides within acyclic O1 . We have previously used Ac1 to
n
[61]
study the kinetics of EDC-fueled anhydride formation.
While the missing alkynyl group would certainly exert a sub-
stituent effect, it would be expected to accelerate anhydride
hydrolysis, and thus the measurements on Ac1 provide a lo-
wer limit for the rate. Under conditions identical to those for
DA1 assembly, the reaction is complete within minutes, with
almost no buildup of anhydride An1 (Figure S20). Hydrolysis
must therefore be faster than was observed for M1 by orders
of magnitude. This result strongly suggests that hydrolysis of
acyclic oligomers O1 is faster than that of (aggregated) ma-
n
crocycle M1, and thus process 2 must be operative. The results
from the extraction experiments (see above) also support this
assertion.
It has long been known that anhydride exchange by
transacylation is faster than hydrolysis, particularly in the
[
61,74,75]
presence of pyridine.
To probe its significance to as-
sembly, we designed the system shown in Figure 5a. A pen-
taglyme-functionalized diacid, DA2, was prepared. Der-
ivatives of DA2 can be distinguished from those of DA1 by
1
HPLC and mass spectrometry (see below). H NMR spec-
Figure 4. Effect of varying the concentration of DA1 and EDC on the
assembly of M1, monitored by HPLC (data are normalized to the ma-
ximum peak area for a given run, unnormalized data are in Fig-
ure S10). Top: Variable concentration of EDC added to 25 mm DA1.
Bottom: Variable concentration of DA1 treated with 50 mm of EDC.
Scheme 2. EDC-fueled assembly of monoacid Ac1.
Angew. Chem. 2020, 132, 2 – 9
ꢀ 2020 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.de
&&&&
These are not the final page numbers!

Angewandte
Forschungsartikel
Chemie
conversion. The two solutions were then mixed and the sy-
stem was monitored by HPLC.
A chromatogram acquired immediately after mixing
shows two separate peaks corresponding to macrocycles M1
and M2, as shown in Figure 5b. A new peak appears between
those for M1 and M2 within an hour. This peak was assigned
to the mixed macrocycle M3 (Figure 5a). The assignments
were confirmed by LC/MS (Figures S35–S38). As shown in
Figure 5c, the fraction of the total macrocycle population that
is M3 increases over the course of hours. It reaches a plateau
of roughly 45% after 10–12 h, consistent with the 50% ex-
pected for the macrocycles at their equilibrium populations
(1:1:2 ratio of M1/M2/M3 on the basis of symmetry).
Under the conditions of the mixing experiment, NMR
measurements show that the EDC should be completely
consumed within 5–6 h (Figure S12), at which point no new
anhydrides should be generated (i.e., process 2 should no
longer occur). As shown in Figure 5c, the concentration of M3
continues to rise for roughly 2 h after this point, even as the
total concentration of macrocycles falls. Further, while the
concentration of EDC should fall continuously over the first
6
h, the rate of buildup of M3 remains approximately con-
stant.
These results confirm that anhydride exchange is si-
gnificant in these systems, indicating that process 3, self-cor-
rection akin to that observed in thermodynamically con-
trolled self-assembly, should also play a role in directing the
efficient formation of macrocycle. The time scale of exchange
(10 h to reach a statistical distribution of macrocycles), is
much longer than the time scale required for nearly complete
assembly into the macrocycles (< 1 h). However, since the
aggregates undergo slower hydrolysis, it is reasonable to as-
sume that exchange would be as well; transacylation involving
acyclic oligomers should be quite rapid. At this point, it is not
possible for us to gauge the relative importance of the in-
herent stability of the free macrocycle compared to the added
stability from aggregation in driving the system toward M1.
A proposed mechanism for out-of-equilibrium assembly
in this system is shown in Figure 6. The generation of an-
hydrides is largely unselective, producing a mixture of ma-
crocycles and acyclic oligomers. Self-correction is both fuel-
mediated and fuel-independent; the fuel regenerates mixtu-
res of macrocycles and oligomers, but since the oligomers
preferentially decompose this should lead to a buildup of
macrocycle over time. Simultaneously, transacylation allows
the pool of anhydrides to rearrange to macrocycles, local
minima on the free energy surface, in much the same way that
equilibrium self-assembly eventually finds the global mini-
mum. The experimental data suggests that both mechanisms
play a significant role, but quantifying the relative contribu-
tions is challenging, particularly as they should both be time-
dependent.
Figure 5. a) Dynamic exchange of anhydrides (Hxg=(CH CH O) CH ,
2
2
6
3
Png=(CH CH O) CH ). The two macrocycles M1 and M2 (25 mm
2
2
5
3
each) were generated separately by treatment with EDC (50 mm), then
mixed once the acids were consumed (14 min) (pD=5.5, 250 mm
[
D ]pyridine buffer, I=1.0m (NaCl), rt). b) Monitoring of the mixture
5
by HPLC. c) HPLC peak heights vs. reaction time for the experiment in
b); the total peak intensity is shown at the top. NMR monitoring ex-
(
periments show that the EDC should be completely consumed within
the first 5–6 h.
troscopy monitoring experiments confirmed that the reac-
tivity of DA2 is effectively identical to that of DA1 and that
the pentaglyme groups provide sufficient solubility for it and
its associated macrocycle M2 (Figure S21).
We carried out a simple mixing experiment to test for
exchange. Separate solutions of DA1 and DA2 were prepared Conclusion
25 mm each), treated with EDC (50 mm), and allowed to
(
react for 14 min under conditions identical to those for ma-
crocycle assembly alone. At this point, the NMR experiments
had shown that the diacids should have reached maximum
Bifunctional diacid DA1 assembles effectively into ma-
crocycle M1, with two transient anhydride bonds, on treat-
ment with EDC, and is regenerated over the course of hours.
&
&&&
www.angewandte.de
ꢀ 2020 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. 2020, 132, 2 – 9
These are not the final page numbers!

Angewandte
Forschungsartikel
Chemie
[
1] M. Fialkowski, K. J. M. Bishop, R. Klajn, S. K. Smoukov, C. J.
Campbell, B. A. Grzybowski, J. Phys. Chem. B 2006, 110, 2482 –
496.
2
[
[
2] E. Mattia, S. Otto, Nat. Nanotechnol. 2015, 10, 111 – 119.
3] S. A. P. van Rossum, M. Tena-Solsona, J. H. van Esch, R. Eel-
kema, J. Boekhoven, Chem. Soc. Rev. 2017, 46, 5519 – 5535.
4] R. D. Astumian, Chem. Commun. 2018, 54, 427 – 444.
5] G. Ragazzon, L. J. Prins, Nat. Nanotechnol. 2018, 13, 882 – 889.
6] S. De, R. Klajn, Adv. Mater. 2018, 30, 1706750.
7] B. Rieß, J. Boekhoven, ChemNanoMat 2018, 4, 710 – 719.
8] M. Tena-Solsona, J. Boekhoven, Isr. J. Chem. 2019, 59, 898 – 905.
9] J. Boekhoven, A. M. Brizard, K. N. K. Kowlgi, G. J. M. Koper, R.
Eelkema, J. H. van Esch, Angew. Chem. Int. Ed. 2010, 49, 4825 –
[
[
[
[
[
[
Figure 6. Proposed mechanism for the chemically fueled self-assembly
of macrocycles in this system. Both the interconversion of the an-
hydrides and the difference in hydrolysis rates contribute to the ef-
ficiency.
4
828; Angew. Chem. 2010, 122, 4935 – 4938.
[
[
[
[
[
10] J. Boekhoven, W. E. Hendriksen, G. J. M. Koper, R. Eelkema,
J. H. van Esch, Science 2015, 349, 1075 – 1079.
11] B. Zhang, I. M. Jayalath, J. Ke, J. L. Sparks, C. S. Hartley, D.
Konkolewicz, Chem. Commun. 2019, 55, 2086 – 2089.
12] S. Bal, K. Das, S. Ahmed, D. Das, Angew. Chem. Int. Ed. 2019,
5
8, 244 – 247; Angew. Chem. 2019, 131, 250 – 253.
Compound M1 further undergoes aggregation in water si-
milar to that known for many other shape-persistent ma-
crocycles. The generation of anhydrides is unselective in these
systems, but two other mechanisms contribute to the ef-
ficiency of self-assembly. First, the acyclic oligomers undergo
rapid hydrolysis compared to the aggregated macrocycle;
subsequent regeneration of anhydrides by the fuel eventually
favors the more kinetically stable aggregated M1. Second,
transacylation allows the anhydride-containing species to
exchange; the macrocycle should be a local minimum on the
systemꢀs free energy surface, allowing it to be populated
preferentially. The latter mechanism is analogous to thermo-
dynamically controlled self-assembly, but the former is unique
to nonequilibrium systems.
This work demonstrates that well-defined closed ar-
chitectures connected by multiple transient bonds can be as-
sembled effectively using chemical fuels. In principle, it
should allow the functional properties of macrocycles and
cages, already well-established in thermodynamically con-
trolled systems, to be applied in nonequilibrium systems,
yielding new adaptive behavior.
13] C. Biagini, S. Albano, R. Caruso, L. Mandolini, J. A. Berrocal, S.
Di Stefano, Chem. Sci. 2018, 9, 181 – 188.
14] S. Erbas-Cakmak, S. D. P. Fielden, U. Karaca, D. A. Leigh, C. T.
McTernan, D. J. Tetlow, M. R. Wilson, Science 2017, 358, 340 –
343.
[15] Q. Shi, C.-F. Chen, Chem. Sci. 2019, 10, 2529 – 2533.
[16] M. R. Wilson, J. Solꢁ, A. Carlone, S. M. Goldup, N. Lebrasseur,
D. A. Leigh, Nature 2016, 534, 235 – 240.
[
[
[
17] A. K. Dambenieks, P. H. Q. Vu, T. M. Fyles, Chem. Sci. 2014, 5,
3396 – 3403.
18] B. G. P. van Ravensteijn, W. E. Hendriksen, R. Eelkema, J. H.
van Esch, W. K. Kegel, J. Am. Chem. Soc. 2017, 139, 9763 – 9766.
19] R. K. Grçtsch, A. Angı, Y. G. Mideksa, C. Wanzke, M. Tena-
Solsona, M. J. Feige, B. Rieger, J. Boekhoven, Angew. Chem. Int.
Ed. 2018, 57, 14608 – 14612; Angew. Chem. 2018, 130, 14817 –
14822.
[20] R. K. Grçtsch, C. Wanzke, M. Speckbacher, A. Angı, B. Rieger,
J. Boekhoven, J. Am. Chem. Soc. 2019, 141, 9872 – 9878.
21] C. S. Wood, C. Browne, D. M. Wood, J. R. Nitschke, ACS Cent.
[
[
[
[
[
Sci. 2015, 1, 504 – 509.
22] L. S. Kariyawasam, C. S. Hartley, J. Am. Chem. Soc. 2017, 139,
1
1949 – 11955.
23] B. Rieß, C. Wanzke, M. Tena-Solsona, R. K. Grçtsch, C. Maity, J.
Boekhoven, Soft Matter 2018, 14, 4852 – 4859.
24] M. Tena-Solsona, C. Wanzke, B. Riess, A. R. Bausch, J. Boek-
hoven, Nat. Commun. 2018, 9, 2044.
25] S. J. Rowan, S. J. Cantrill, G. R. L. Cousins, J. K. M. Sanders, J. F.
Stoddart, Angew. Chem. Int. Ed. 2002, 41, 898 – 952; Angew.
Chem. 2002, 114, 938 – 993.
Acknowledgements
This work was supported by the U.S. Department of Energy,
Office of Science, Basic Energy Sciences, under Award # DE-
SC0018645. We thank Lasith Kariyawasam for synthesizing
Ac1 and experimental assistance and Dr. Theresa Ramelot
for help with the LC/MS experiments.
[
[
[
[
[
[
[
[
26] Dynamic Covalent Chemistry, (Eds.: W. Zhang, Y. Jin), Wiley,
Hoboken, 2017.
27] W. Zhang, J. S. Moore, Angew. Chem. Int. Ed. 2006, 45, 4416 –
4439; Angew. Chem. 2006, 118, 4524 – 4548.
28] V. H. Gessner, J. F. Tannaci, A. D. Miller, T. D. Tilley, Acc.
Chem. Res. 2011, 44, 435 – 446.
29] B. Liu, C. G. Pappas, E. Zangrando, N. Demitri, P. J. Chmie-
lewski, S. Otto, J. Am. Chem. Soc. 2019, 141, 1685 – 1689.
30] M. L. C. Quan, D. J. Cram, J. Am. Chem. Soc. 1991, 113, 2754 –
Conflict of interest
The authors declare no conflict of interest.
2
755.
31] X. Liu, Y. Liu, G. Li, R. Warmuth, Angew. Chem. Int. Ed. 2006,
5, 901 – 904; Angew. Chem. 2006, 118, 915 – 918.
Stichwçrter: aggregation · anhydrides · dissipative assembly ·
macrocycles · self-assembly
4
32] M. Mastalerz, Angew. Chem. Int. Ed. 2010, 49, 5042 – 5053;
Angew. Chem. 2010, 122, 5164 – 5175.
33] T. Mitra, K. E. Jelfs, M. Schmidtmann, A. Ahmed, S. Y. Chong,
D. J. Adams, A. I. Cooper, Nat. Chem. 2013, 5, 276 – 281.
Angew. Chem. 2020, 132, 2 – 9
ꢀ 2020 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.de
&&&&
These are not the final page numbers!

Angewandte
Forschungsartikel
Chemie
[
[
[
[
34] Y. Jin, Q. Wang, P. Taynton, W. Zhang, Acc. Chem. Res. 2014, 47,
575 – 1586.
35] C. S. Hartley, E. L. Elliott, J. S. Moore, J. Am. Chem. Soc. 2007,
29, 4512 – 4513.
36] T. Wei, J. H. Jung, T. F. Scott, J. Am. Chem. Soc. 2015, 137,
6196 – 16202.
37] C. D. Pentecost, K. S. Chichak, A. J. Peters, G. W. V. Cave, S. J.
Cantrill, J. F. Stoddart, Angew. Chem. Int. Ed. 2007, 46, 218 – 222;
Angew. Chem. 2007, 119, 222 – 226.
[60] D O (and not H O) was used throughout this work for consi-
2 2
1
stency. It is necessary for the NMR and IR spectroscopy expe-
riments (for IR, the bending vibration of H O overlaps with the
2
1
key C=O stretching region).
[61] L. S. Kariyawasam, J. C. Kron, R. Jiang, A. J. Sommer, C. S.
Hartley, J. Org. Chem. 2020, 85, 682 – 690.
1
[62] We can exclude the aggregation of DA1 under these conditions.
The reaction mixtures are buffered, the concentration of DA1
drops during the experiments, and we know that the aggregation
of DA1 is not induced by EDU (otherwise the aggregates would
persist at the ends of the experiments). Thus, the aggregation of
DA1 would have to be induced by EDC independent of an-
hydride formation. However, the lifetimes of the aggregates
(DLS, NMR) are always significantly longer than the lifetimes of
the EDC (NMR) during the reactions, whereas they match the
lifetimes of the anhydrides (IR, HPLC).
[
[
38] N. Ponnuswamy, F. B. L. Cougnon, J. M. Clough, G. D. Panto s¸ ,
J. K. M. Sanders, Science 2012, 338, 783 – 785.
39] J.-F. Ayme, J. E. Beves, C. J. Campbell, D. A. Leigh, Angew.
Chem. Int. Ed. 2014, 53, 7823 – 7827; Angew. Chem. 2014, 126,
7957 – 7961.
[
40] K. R. West, R. F. Ludlow, P. T. Corbett, P. Besenius, F. M.
Mansfeld, P. A. G. Cormack, D. C. Sherrington, J. M. Goodman,
M. C. A. Stuart, S. Otto, J. Am. Chem. Soc. 2008, 130, 10834 –
[63] Quantitative comparisons of the amount of aggregated M1 are
of course impossible on the basis of HPLC because disag-
10835.
[
41] G. Zhang, O. Presly, F. White, I. M. Oppel, M. Mastalerz, Angew.
Chem. Int. Ed. 2014, 53, 5126 – 5130; Angew. Chem. 2014, 126,
gregation will occur during the HPLC run.
1
[64] H NMR spectroscopy run in CD CN gives sharp aromatic
3
5
226 – 5230.
42] F. Ren, K. J. Day, C. S. Hartley, Angew. Chem. Int. Ed. 2016, 55,
620 – 8623; Angew. Chem. 2016, 128, 8762 – 8765.
43] C. Zhang, Q. Wang, H. Long, W. Zhang, J. Am. Chem. Soc. 2011,
33, 20995 – 21001.
peaks, see Supporting Information.
[65] The fully extended DA1 with the core in the anti conformation is
a bent rod approximately 5.5 nm long.
[66] C. T. L. Ma, M. J. MacLachlan, Angew. Chem. Int. Ed. 2005, 44,
4178 – 4182; Angew. Chem. 2005, 117, 4250 – 4254.
[67] A. S. Shetty, J. Zhang, J. S. Moore, J. Am. Chem. Soc. 1996, 118,
1019 – 1027.
[68] Y. Tobe, N. Utsumi, K. Kawabata, A. Nagano, K. Adachi, S.
Araki, M. Sonoda, K. Hirose, K. Naemura, J. Am. Chem. Soc.
2002, 124, 5350 – 5364.
[
[
8
1
[
[
44] J. T. A. Jones, et al., Nature 2011, 474, 367 – 371.
45] N. Giri, M. G. Del Pꢂpolo, G. Melaugh, R. L. Greenaway, K.
Rꢃtzke, T. Koschine, L. Pison, M. F. C. Gomes, A. I. Cooper,
S. L. James, Nature 2015, 527, 216 – 220.
[
46] S. Otto, R. L. E. Furlan, J. K. M. Sanders, Science 2002, 297,
5
90 – 593.
[69] S. Lahiri, J. L. Thompson, J. S. Moore, J. Am. Chem. Soc. 2000,
122, 11315 – 11319.
[
[
47] S. Otto, Acc. Chem. Res. 2012, 45, 2200 – 2210.
48] J. M. Klein, V. Saggiomo, L. Reck, U. Lꢄning, J. K. M. Sanders,
Org. Biomol. Chem. 2012, 10, 60 – 66.
[70] S. Hçger, K. Bonrad, A. Mourran, U. Beginn, M. Mçller, J. Am.
Chem. Soc. 2001, 123, 5651 – 5659.
[
[
49] K. Ono, N. Iwasawa, Chem. Eur. J. 2018, 24, 17856 – 17868.
50] H. Lçw, E. Mena-Osteritz, M. von Delius, Chem. Sci. 2018, 9,
[71] We note that our DLS and NMR experiments do not provide
evidence for a well-defined critical aggregation concentration.
By DLS, for example, we see a smooth change in aggregate size
that parallels the expected concentration of M1.
[72] In fact, the signals for DA1 vary as a function of concentration in
4
785 – 4793.
51] G. M. Whitesides, J. P. Mathias, C. T. Seto, Science 1991, 254,
312 – 1319.
[
1
[
[
52] D. Zhao, J. S. Moore, J. Org. Chem. 2002, 67, 3548 – 3554.
53] F. B. L. Cougnon, H. Y. Au-Yeung, G. D. Panto s¸ , J. K. M. San-
ders, J. Am. Chem. Soc. 2011, 133, 3198 – 3207.
CDCl , suggesting that it may aggregate through hydrogen
bonding in this solvent.
[73] W. Zhang, J. S. Moore, J. Am. Chem. Soc. 2005, 127, 11863 –
3
[
54] C. Zhang, H. Long, W. Zhang, Chem. Commun. 2012, 48, 6172 –
11870.
6174.
[74] C. A. Bunton, N. A. Fuller, S. G. Perry, V. J. Shiner, Tetrahedron
[
[
55] S. W. Sisco, J. S. Moore, Chem. Sci. 2014, 5, 81 – 85.
Lett. 1961, 2, 458 – 460.
[75] A. R. Fersht, W. P. Jencks, J. Am. Chem. Soc. 1970, 92, 5432 –
56] C. Tsiamantas, X. de Hatten, C. Douat, B. Kauffmann, V.
Maurizot, H. Ihara, M. Takafuji, N. Metzler-Nolte, I. Huc, An-
gew. Chem. Int. Ed. 2016, 55, 6848 – 6852; Angew. Chem. 2016,
5442.
128, 6962 – 6966.
[
[
57] N. Nakajima, Y. Ikada, Bioconjugate Chem. 1995, 6, 123 – 130.
58] V. M. Dꢅaz Pꢆrez, C. Ortiz Mellet, J. Fuentes, J. M. Garcꢅa
Fernꢇndez, Carbohydr. Res. 2000, 326, 161 – 175.
Manuskript erhalten: 29. Januar 2020
Verꢃnderte Fassung erhalten: 18. April 2020
Akzeptierte Fassung online: 8. Mai 2020
Endgꢄltige Fassung online: &&. && &&&&
[
59] M. Tena-Solsona, B. Rieß, R. K. Grçtsch, F. C. Lçhrer, C.
Wanzke, B. Kꢃsdorf, A. R. Bausch, P. Mꢄller-Buschbaum, O.
Lieleg, J. Boekhoven, Nat. Commun. 2017, 8, 15895.
&
&&&
www.angewandte.de
ꢀ 2020 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. 2020, 132, 2 – 9
These are not the final page numbers!

Angewandte
Forschungsartikel
Chemie
Forschungsartikel
Supramolecular Chemistry
M. M. Hossain, J. L. Atkinson,
C. S. Hartley*
&&&& — &&&&
Dissipative Assembly of Macrocycles
Comprising Multiple Transient Bonds
Out-of-equilibrium assembly: Diacid pre-
cursors assemble into unstable aqueous
anhydride macrocycles on treatment with
carbodiimide chemical fuels, even though
initial anhydride generation is unselective
for cyclization versus polymerization. The
efficiency of assembly results from both
fuel-independent (anhydride exchange)
and fuel-dependent (selective hydrolysis)
mechanisms.
Angew. Chem. 2020, 132, 2 – 9
ꢀ 2020 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.de
&&&&
These are not the final page numbers!