Higher Order Constitutional Dynamic Networks: [2×3] and [3×3] Networks Displaying Multiple, Synergistic and Competitive Hierarchical Adaptation

Article abstract of DOI:10.1021/jacs.6b13072

The present study investigates the constitutional dynamic networks (CDNs) underlying dynamic covalent libraries (DCLs) that extend beyond the [2×2] case toward higher orders, namely [2×3] and [3×3] CDNs involving respectively six and nine constituents generated from the recombination of five and six components linked through reversible chemical reactions. It explores the behavior of such systems under the action of one or two effectors. More specifically and for the sake of proof of principle, it makes use of DCLs involving dynamic organic ligands and analyzes their single and double adaptive response under the action of one and two metal cation effectors. Thus, interconversions within [2×3] DCLs of six constituents (hydrazone, acylhydrazone, and imine ligands) give access to the generation of [2×3] CDNs of 3D trigonal prismatic type consisting of three [2×2] sub-networks and presenting specific responses to the application of Cu⁺ and Zn²⁺ metal cation effectors, in particular double agonistic amplification. More complex [3×3] CDNs based on nine ligand constituents of imine, hydrazone, and acylhydrazone types were also designed and subjected to the application of one or two effectors, e.g., Cu⁺ and Fe²⁺ metal cations, revealing novel types of adaptive behavior: (i) agonistic amplification between a single constituent and a full [2×2] sub-network, and (ii) agonistic amplification along a single diagonal connecting three constituents. Of special interest is also the dependence of the response of the system to hierarchical sequence of effector application, whereby initial interaction with Cu⁺ ions results in the destruction of the network, whereas the sequence Fe²⁺ followed by Cu⁺ yields a clean three-constituent DCL. Finally and strikingly, the present results also demonstrate that the increase in complexity of the system by introduction of an additional entity leads to a simpler output through dynamic competition between components.

Full text of DOI:10.1021/jacs.6b13072

Article
pubs.acs.org/JACS
Higher Order Constitutional Dynamic Networks: [2×3] and [3×3]
Networks Displaying Multiple, Synergistic and Competitive
Hierarchical Adaptation
Guangwen Men†,‡ and Jean-Marie Lehn*,†
†Laboratoire de Chimie Supramolec
allee Gaspard Monge, 67000 Strasbourg, France
́ ́ ́ ́
ulaire, Institut de Science et d’Ingenierie Supramoleculaires (ISIS), Universite de Strasbourg, 8
́
‡State Key Laboratory of Supramolecular Structure and Materials, Jilin University, 2699 Qianjin Avenue, Changchun 130012, P. R.
China
S
* Supporting Information
ABSTRACT: The present study investigates the constitutional dynamic
networks (CDNs) underlying dynamic covalent libraries (DCLs) that extend
beyond the [2×2] case toward higher orders, namely [2×3] and [3×3] CDNs
involving respectively six and nine constituents generated from the
recombination of five and six components linked through reversible chemical
reactions. It explores the behavior of such systems under the action of one or
two effectors. More specifically and for the sake of proof of principle, it makes
use of DCLs involving dynamic organic ligands and analyzes their single and
double adaptive response under the action of one and two metal cation
effectors. Thus, interconversions within [2×3] DCLs of six constituents
(hydrazone, acylhydrazone, and imine ligands) give access to the generation
of [2×3] CDNs of 3D trigonal prismatic type consisting of three [2×2] sub-
networks and presenting specific responses to the application of Cu+ and Zn2+
metal cation effectors, in particular double agonistic amplification. More complex [3×3] CDNs based on nine ligand constituents
of imine, hydrazone, and acylhydrazone types were also designed and subjected to the application of one or two effectors, e.g.,
Cu+ and Fe2+ metal cations, revealing novel types of adaptive behavior: (i) agonistic amplification between a single constituent
and a full [2×2] sub-network, and (ii) agonistic amplification along a single diagonal connecting three constituents. Of special
interest is also the dependence of the response of the system to hierarchical sequence of effector application, whereby initial
interaction with Cu+ ions results in the destruction of the network, whereas the sequence Fe2+ followed by Cu+ yields a clean
three-constituent DCL. Finally and strikingly, the present results also demonstrate that the increase in complexity of the system
by introduction of an additional entity leads to a simpler output through dynamic competition between components.
type, as is the case for the splitting of a single homogeneous
solution into two separate liquid phases.3e,f Taking steps toward
more complex systems7 requires the exploration of DCLs of
higher order containing a larger number of interconverting
constituents and defining higher order networks, including
networks of networks. Such connectomes lead to the emergence
of novel features under coupling to multiple agents in the
environment and responding to them in diverse fashions.
In the present study, we designed several higher order systems
and investigated their multiple responses to several chemical
effectors, displaying multiple sequential and hierarchical
adaptation features. On the first stage, DCLs of six constituents,
hydrazone, acylhydrazone, and imine ligands, were designed to
give access to the generation of [2×3] CDNs consisting of three
[2×2] sub-networks and presenting novel types of responses to
the application of different metal cation effectors. In a further
1. INTRODUCTION
Constitutional dynamic libraries (CDLs)1 of dynamic covalent2 or
non-covalent supramolecular type, formed respectively from the
reaction or interaction of n with m components, may be
represented in terms of [n×m] constitutional dynamic networks
(CDNs)1b−e,3,4 of order [n×m] that connect the interconverting
constituents through agonistic and antagonistic relation-
ships1b−e,3 and define the connectome of the system.1d,3e Such
CDLs respond to the application of chemical5 or physical6 agents
by a redistribution of their constituents, thus displaying adaptive
behavior. Sets of four constituents resulting from the
recombination of four components linked through reversible
chemical reactions form a dynamic covalent library (DCL) that is
represented by a square network of order [2×2].1b−e,3a−d,g The
response of such networks to a single agent (such as a metal
cation) or their double adaptive responses to two orthogonal
agents, i.e., photoselection and metalloselection, have been
investigated.3d Interconversions between [2×2] DCLs located in
two separate domains define 3D networks of square prismatic
Received: December 20, 2016
© XXXX American Chemical Society
A
DOI: 10.1021/jacs.6b13072
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
a
Scheme 1. Generation and Behavior of a 3D Trigonal Prismatic [2×3] CDN
a
(Top left) Any two of benzaldehyde components among A1, A2, and A3 (1 equiv each, 30 mM) and the two hydrazines B1 and B2 (1 equiv each, 30
mM) with 5% (in volume) aniline as a catalyst were used to establish the three 2D [2×2] square CDNs (N1−N3). These CDNs formed from the
four hydrazone constituents AaBb are indicated with statistical distribution of constituents. (Top right) Adaptation of the N1 and N3 CDNs in
response to the addition of zinc triflate with an amplification of A1B1 as its zinc complex and a simultaneous up-regulation of the corresponding
agonists A2B2 and A3B2; the network N3 is perturbed by the destruction of the constituent A3B1 by demethylation to A8B1. (Middle) Generation of
the 3D [2×3] trigonal prismatic CDN NA by combination of the three 2D [2×2] square sub-networks into a linear array N1-N2-N3 by
superposition of the vertices linking the same constituents followed by folding along these common vertices; treatment of NA with Zn2+ cations leads
to an adaptation with amplification of A1B1 as its zinc complex and a simultaneous up-regulation of the corresponding agonists A2B2 and A3B2 on
two faces of the prism. The larger size and bold letters as well as the bold diagonal lines indicate simultaneous up-regulations of these agonistic
constituents in response to the effector. (Bottom) Structures of the hydrazine constituents AaBb and of the imines generated from aniline D5. The
salicylaldehyde residue, produced by the demethylation side reaction, has been labeled A8. For the effect of Zn2+ binding to A3B1 and A8B1, see
Figure S5a−c. The quantitative values of the constituent distributions are given in Table S1.
step, more complex [3×3] CDNs based on nine constituents
were also designed and subjected to one effector or two effectors,
revealing novel adaptive behaviors. The present work is meant to
be a general demonstration of principle and an exploration of the
behavior of dynamic systems of increasing complexity. It thus
implements chemical entities derived from previous inves-
tigations in a logical development and paves the way for further
studies. [Note: The general description of the CDNs is given in
the main text, while the corresponding numerical data are
provided in the Supporting Information (SI). In order to
facilitate the perusal of the rather complicated sets of data, the
representations of the network patterns shown in Schemes 1−5
highlight in a graphical way the main features of the constituent
distributions and the changes they undergo in response to the
effectors. The disposition of the numerical values in the tables
given as SI follow the same organization of the network patterns,
again to facilitate reading.]
2. RESULTS AND DISCUSSION
2.1. [2×3] CDN with One Effector: Construction of a 3D
Network from Several 2D Networks. According to our
previous studies, hydrazones may be implemented for the design
B
DOI: 10.1021/jacs.6b13072
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Scheme 2. Multiple and Sequential Adaptation in the [2×3] Trigonal Prismatic CDN of the Six Constituents Generated from (top)
the Components (1 equiv each, 30 mM) A4, D1, C2, and D2 and A1 (2 equiv) in Response to (middle) a Single Effector or (bottom)
a
the Two Effectors Cu+ and Zn2+ (0.5 equiv each)
a
The longer and shorter left and right arrows indicate respectively the additions of the effector Cu+ (top left) or Zn2+ (top right) to the mixture of
components or to the preformed DCL of constituents. A1C2 exists in E and Z forms (see SI for details). The larger size and bold letters as well as the
bold diagonal lines indicate simultaneous up-regulation of these agonistic constituents in response to the effectors. See Table S2 for numerical values.
of adaptive libraries in view of their thermodynamic stability at
equilibrium state and their ability to undergo reversible
component exchange at the CN bond facilitated by catalysts
such as aniline.3c,d,g We here investigated the DCLs generated
from any two aldehydes among A1, A2, and A3 and the two
hydrazines B1 and B2 (Scheme 1) in order to establish the three
2D [2×2] square sub-networks that may then be combined to
construct the corresponding [2×3] full network. First, 1 equiv of
each component was mixed with 5% (in volume) aniline D5 used
as organo-catalyst3d,8 in CD3CN. After heating at 78 °C for 17 h,
the DCLs were cooled and subjected to 1H NMR measurement
for the analysis of the composition of the mixture of compounds
generated. As shown in Scheme 1 (top panel), three DCLs of four
hydrazone constituents, corresponding to the [2×2] square
CDNs A1B1/A1B2/A2B1/A2B2 (N1), A2B1/A2B2/A3B1/A3B2
(N2), and A3B1/A3B2/A1B1/A1B2 (N3), were produced from
condensation, transimination, hydrolysis, and recondensation of
the respective A and B components. The quantitative values
determined for all three DCLs were statistical within
experimental error (Table S1, top; Figures S1, S2, and S5a).
When 0.5 equiv of zinc triflate was added, N1 and N3, which
involve the only tridentate ligand A1B1, underwent reorganiza-
tion to amplify the hydrazone A1B1 in the form of its ZnII(A1B1)2
complex as well as simultaneously its diagonally linked agonist
A2B2 or A3B2 (agonist amplification;1b−e Scheme 1, top right; see
Figures S3−S5 for NMR spectra) giving strongly modified
distributions of 49%/1%/1%/49% for the A1B1/A1B2/A2B1/
A2B2 N1-Zn network and 1%/49%/49%/1% for the A3B1/A3B2/
A1B1/A1B2 N3-Zn network. Compared to the DCLs treated with
scandium triflate (20%) as exchange catalyst, these systems (N1
and N3) catalyzed by aniline display more pronounced
amplifications as a result of less hydrolysis.3d However, the
DCL with the constituents A2B1/A2B2/A3B1/A3B2 (N2),
without the pyridine carboxaldehyde moiety A1, exhibited a
C
DOI: 10.1021/jacs.6b13072
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
different behavior on addition of zinc cations compared to the
libraries of N1 and N3. In this case, 10% of a byproduct was
obtained, produced by an irreversible demethylation of the
methoxyl group in A3B1, giving A8B1 and leading to a partial
destruction of the network N3 (Scheme 1, top center). In absence
of the pyridine aldehyde component A1, required for the
formation of a tridentate NNN binding site, the added Zn2+
cation favored binding to the free NN bidentate component B1,
thus giving ZnII(B1)x complexes and increasing hydrolysis in the
DCL. In separate experiments, the notable demethylation
reaction of the O-methoxyl substituent of A3B1 in the presence
of Zn2+ and 5% aniline as catalyst was further investigated. As it is
not central for the present purposes, the results are given in the SI
(see Figure S5a−c for more details).
The derivatives of benzaldehyde A1 and A4 were selected as
components in the DCL, using 2 equiv of A1 to generate the two
constituents, acting respectively as specific receptors for the two
effectors. Compared to the unsubstituted A2 and the methoxyl-
substituted A3, the utilization of p-chloro substitution in A4 led to
a reduction of hydrolysis within the DCL on formation and
exchange of the imine and acylhydrazone constituents. On the
other hand, for the amine components of the network, the
hydrazine compounds were replaced by the aromatic amines D1
and D2 and the hydrazide C2 so as to adapt to the multiresponsive
transformations of the DCL. The p-dimethylamino substitution
of D1 might not only reduce the level of hydrolysis because of its
strong electron-donating effect, but also enhance the binding of
Cu+ ion compared to the methyl-substituted D2, considering the
known increase in stability of the copper complex.9
A [2×3] DCL was set up by mixing the separately equilibrated
libraries of N1, N2, and N3 in a NMR tube and re-equilibrating
the mixture at 78 °C for 17 h. It contained the full set of six
constituents A2B1/A2B1/A1B1/A1B2/A3B1/A3B2 with the dis-
tributions corresponding to a retention of the distributions
present in the starting [2×2] DCLs (see Table S1; see Figure S6
for 1H NMR spectrum). This [2×3] DCL may be represented by
a trigonal prism-shaped 3D [2×3] network NA, constructed by
the connection and folding of the previous three 2D sub-
networks along the vertices sharing the components B1 and B2
(Scheme 1, middle left). When 1 equiv Zn2+ was added, the
strong binding of these effector cations to the tridentate NNN
ligand A1B1 drives the amplification of this constituent as its
complex ZnII(A1B1)2 as well as the concomitant double up-
regulation of its two agonists A2B2 and A3B2 located all three in
the two [2×2] sub-networks sharing the A1B1−A1B2 edge of the
trigonal prismatic [2×3] network. Simultaneously, A1B2, which
presents a triply antagonistic relationship, is strongly down-
regulated, whereas this effect is weaker for A2B1 and A3B1, which
are connected to only two antagonistic partners. The trapping of
the Zn2+ cations by A1B1 suppresses the demethylation of the
A3B1 constituent, thus preventing the destruction of the fragile
A2A3B1B2 sub-network. Compared to the simple [2×2] CDNs,
the more complex integrated [2×3] network displays not only
the expected reorganization of the constituents through double
agonist amplification but also provides improved stability to the
system by protecting it from degradation of one of its
constituents, A3B1 (Scheme 1, middle right; see Figure S7 for
1H NMR spectrum).
The new DCL was then generated from a mixture of the five
components, A4, D1, C2, and D2 (30 mM for 1 equiv each) and 2
equiv of A1 in CD3CN, heated at 60 °C for 51 h to reach the
equilibrium. Six expected constituents A1D1, A1C2, A1D2, A4D1,
A4C2, and A4D2 were obtained and confirmed by NMR
determination (see Scheme 2, network NB, and Table S2,
middle center; Figure S8a,b), giving a biased distribution of 20%/
31%/14%/13%/3%/11% with 8% hydrolysis. The origin of this
bias may be attributed to the favored formation of acylhydrazone
A1C2 (in E and Z forms, E-A1C2 9%, Z-A1C2 22%, confirmed by
2D NMR shown in Figure S9a−c and both taken into account).10
As a consequence, the percentages of its dual diagonal agonists,
A4D1 and A4D2, were amplified on the side faces A1A4D1C2 and
A1A4C2D2 of the 3D trigonal prismatic network. Meanwhile,
A4C2, involved in a triply antagonistic relationship, was sharply
down-regulated. No additional catalyst was needed because the
aniline-type components D1 and D2 themselves acted as catalysts
during the exchange reactions. Starting from this initial DCL, its
constitutional variations induced by four sequential pathways of
cation additions were investigated: (I) Cu+ followed by Zn2+; (II)
Cu+ followed by another addition of the same amount of Cu+;
(III) Zn2+ followed by Cu+; and (IV) Zn2+ and Cu+ added
together.
Pathway I. When the starting mixture of six constituents
(NB) was subjected to CuI(OTf) (0.5 equiv, 15 mM), the DCL
underwent reorganization to amplify the imine constituent A1D1
as expected, in the form of its tetrahedral complex CuI(A1D1)2
acting as the second driving force besides the formation of the
stable acylhydrazone constituent A1C2 and favored by its
electron-donating dimethylamino substitution compared to the
other bidentate constituent A1D2. As a consequence, the
constituent A4D2, the common agonist of A1C2 and of the
complex CuI(A1D1)2, as well as A1C2 (as E-A1C2 8% and Z-A1C2
18%) maintained their original up-regulation, giving a new
expression of double diagonal agonistic amplification located on
the two side-faces A1A4D1D2 and A1A4C2D2 of the trigonal
prismatic network. Furthermore, the same distribution could also
be obtained directly by equilibration of the starting mixture of the
five components, A4, D1, C2, D2 and 2 equiv of A1 in the presence
of 0.5 equiv Cu+ ion, due to the thermodynamic control of the
DCL (network NB-Cu shown in Scheme 2 and Table S2, middle
left; see Figure S10a,b for spectral data). Subsequent addition of
ZnII(OTf)2 into this Cu+-pretreated network NB-Cu promoted
the generation of the octahedral complex ZnII(E-A1C2)2, due to
the preferential binding between Zn2+ ion and the only tridentate
ligand constituent A1C2, which displayed in addition a
configurational adaptation from the starting E/Z mixed isomers
of A1C2 to fully E in the complex. However, the original
Thus, remarkably, the larger DCL behaves in an improved
fashion, presenting an increased stability with less sensitivity
toward destructive perturbation by virtue of the relationships
between constituents established within the more complex
network.
2.2. [2×3] CDNs with Two Effectors: Multiple Adapta-
tion. 2.2.1. Multiple Adaptation in a DCL of Imines and
Acylhydrazones. A further step consists in devising a CDN that
would alternately respond to two different effectors, so that it
may be driven along different pathways to the expression of
multiple distributional patterns corresponding to different
constitutional information. The design of such a CDN requires
higher dimensionality than a 2D square network and involves
two specific receptors, one for each of the two effectors operating
in combination. To this end, two metal ion effectors Zn2+ and
Cu+ were utilized, in view of their different coordination features,
and the former CDN, solely based on hydrazone constituents,
was developed into a more complex imine/acylhydrazone mixed
system.
D
DOI: 10.1021/jacs.6b13072
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Scheme 3. Hierarchical Adaptation in the 3D Trigonal Prismatic CDN Representing the Response of the DCL Generated from
(top) the Components (1 equiv each, 30 mM) A4, B1, D1, and C1 and A1 (2 equiv) to (middle) a Single Effector or (bottom) the Two
a
Effectors Cu+ and Zn2+ (0.5 equiv each)
a
The larger size and bold letters as well as the bold diagonal line indicate simultaneous up-regulation of these agonistic constituents in response to
the effectors. The longer and shorter left and right arrows indicate respectively the additions of the effector Cu+ (top left) or Zn2+ (top right) to the
mixture of components or to the preformed DCL of constituents. The network in brackets NC-Cu (middle left) is partially destroyed by oxidation of
Cu+ to Cu2+ facilitated by coordination of the latter to B1. See Table S3 for numerical values.
distribution was not strongly affected, as A1C2 (E + Z) was
already strongly amplified in the parent DCL (see network NB-
ZnCu in Scheme 2 and Table S2, bottom right).
(network NB-2Cu in Scheme 2, bottom left; see Figure S12a.b for
1H NMR spectra). This case showed another expression of
conformational, configurational, and constitutional adaptation
besides network NB-ZnCu.
Pathway III. When the initial network NB was first treated
with 0.5 equiv of Zn2+, the DCL kept a distribution quite similar
to the initial one (network NB-Zn, 16%/33%/14%/11%/2%/
14% respectively for constituents A1D1/A1C2/A1D2/A4D1/
A4C2/A4D2 with about 10% hydrolysis and with A1C2 mainly
in its coordinated E form, 29%). Addition of Cu+ into this Zn2+-
pretreated CDN resulted in the same distribution as NB-ZnCu,
showing a fourth type of adaptation (Figures S11 and S13 for 1H
NMR data).
Pathway II. On the other hand, further addition of another 0.5
equiv of Cu+ into NB-Cu gives another change in the distribution
of the DCL by amplification the newly formed copper complex
CuI(A1D2)2. Consequently, the constituent A4C2, the common
double diagonal agonist of the two copper complexes
CuI(A1D1)2 and CuI(A1D2)2, underwent a significant up-
regulation from 7% to 19%. Engaged in triply antagonistic
relationships, all the E-A1C2 was consumed, and interestingly, Z-
A1C2 survived (9% left) the down-regulation of A1C2 owing to
the stabilization of the Z isomer by an internal hydrogen bond
E
DOI: 10.1021/jacs.6b13072
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Scheme 4. Graphical Representation of the Adaptation of a [3×3] CDN of the Nine Constituents Generated from the Six
Components (top) A1, A3, A5, B1, B2, and B3 (1 equiv each, 30 mM) with 5% (in Volume) Aniline as a Catalyst in CD3CN as
Weighted Graphs Corresponding to a DCL of Nine Hydrazones as Constituents under the Pressure of an Effector Zn2+ ion,
Leading to the Constitutional Variation of the System from (left) an Approximately Statistical Distribution to (right) an Enforced
Distribution Displaying (bottom) the Effector-Enhanced Fittest Constituent A1B1 in the Form of Its ZnII(A1B1)2 Complex and a
a
Square [2×2] Sub-network as Unfittest Agonist of Four Amplified Constituents A3B2/A3B3/A5B2/A5B3
a
See Table S4 for numerical values.
Pathway IV. As expected, the same distribution was also
obtained on simultaneous treatment with Zn2+ and Cu+ together
from the start.11
The six component DCL thus led to diverse constitutional
expressions on adaptation in response to the individual effectors
Cu+ and Zn2+ and their synergistic action.
Based on these considerations, A4, B1, D1, and C1 (30 mM
each) together with 2 equiv of A1 (Scheme 3, top) were mixed in a
NMR tube with CD3CN as solvent. These components may
generate a set of six constituents A1B1, A1D1, A1C1, A4B1, A4D1,
and A4C1. However, after equilibrating at 78 °C for 26 h, the
solution contained a mixture of mainly three constituents
identified as A1B1, A4D1, and A1C1, forming a strongly biased
initial DCL. This bias may be attributed to the favored formation
of the hydrazone A1B1 and of the acylhydrazone A1C1 together
with the corresponding amplification of their common agonist
A4D1 on side-faces A1A4B1D1 and A1A4D1C1 of the trigonal
prismatic network NC (Scheme 3, middle center; Figure S14a,b).
This DCL was subjected to treatment by Cu+ and Zn2+ effectors
along three pathways.
2.2.2. Competitive Hierarchical Adaptation in a DCL of
Imines, Acylhydrazones, and Hydrazones. For the purpose of
reaching higher complexity, a six constituent DCL that included
imines, acylhydrazones, and hydrazones was set up, and the
response of its [2×3] CDN to two effectors was investigated.
Besides keeping the aldehydes A1 and A4, hydrazine B1 and
aniline D1 were chosen as components aimed at forming ligand
constituents for binding of Zn2+ and Cu+, respectively. The
hydrazide C1 was expected to form NNO or NO coordination
sites that would be not only weaker Zn2+ binders than an NNN
site formed by hydrazone ligands but also potentially weaker Cu+
binders than the bidentate NN site provided by imine ligands
generated from anilines. In addition, by suppressing the risk of
ionization of the N−H site, such an N-methylated hydrazide has
a lower risk of interference with the binding of Zn2+ to
hydrazones compared to the non-methylated hydrazide C2.
Pathway I. First addition of Cu+ caused the destruction of this
DCL due to the oxidation of Cu+ to Cu2+ favored by the binding
of the latter to the component B1 during the component
exchange (network NC-Cu in Scheme 3 and Table S3, middle
left; see Figure S15 for the details).
Pathway II. On the other hand, addition of Zn2+ alone did not
greatly affect the original biased distribution, as it preferentially
bound the constituent ligand A1B1 that was already strongly
F
DOI: 10.1021/jacs.6b13072
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Scheme 5. Weighted Graph Representation of the Adaptation of the [3×3] CDN of a DCL of Nine Constituents Formed from a
Mixture of Six Components of (top) A6, A7, A4, D4, D3, and C1 (1 equiv each, 30 mM) in Response to (middle) a Single Effector or
a
(bottom) the Two Effectors Fe2+ and Cu+ (0.5 equiv each)
a
The size of the colored circles grossly represents the amounts of the corresponding constituents on the network. For clarity, only the dominant free
components D4 and A4 are represented in form of colored arches located out of the network NE. The longer and shorter left and right arrows
indicate respectively the addition of the effector Cu+ (top left) or Fe2+ (top right) to the mixture of components or to the preformed DCL of
constituents. The network in brackets NE-Cu (middle left) is partially destroyed by oxidation of Cu+ to Cu2+ facilitated by coordination of the latter
to D4. See Table S5 for numerical values.
amplified in the initial biased DCL (network NC-Zn in Scheme 3
and Table S3, middle right; see Figure S16a,b for H NMR
complexes ZnII(A1B1)2 and CuI(A1D1)2 as well as simulta-
neously their common agonist A4C1 (network NC-ZnCu in
Scheme 3 and Table S3, bottom; see Figure S17a,b for 1H NMR
1
spectra). However, subsequent addition of Cu+ into this mixture
led to a pronounced constitutional change, and what is even
more interesting, such Zn2+-pretreated DCL and its network
survived the addition of Cu+ ion because B1 was already trapped
and isolated from the dynamic exchange process by the
formation of the complex ZnII(A1B1)2. In this case, constituents
A1B1 and A1D1 were amplified via formation of their respective
spectra).
Pathway III. Addition of both Zn2+ and Cu+ together to the
initial network NC led to the same distribution, indicating also
that the interaction of the Zn2+ ions with the DCL supersedes
that of the Cu+ ions.
G
DOI: 10.1021/jacs.6b13072
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Thus, treating the original library with effectors Zn2+ and Cu+
either in fixed order (first zinc and second copper) or at the same
time avoids the destruction of the DCL caused by Cu+ oxidation
when it is added first, pointing to a hierarchical relationship
between the two effectors. Comparison of pathways I, II, and III
demonstrates that the two effectors operate in a synergistic way:
Cu+ required Zn2+ to trap the B1 constituent, while Zn2+ needed
the assistance of Cu+ to generate a novel state of the library, in a
sort of “co-evolution” of the DCL.12
components were selected on the basis of preliminary experi-
ments exploring binding affinities of metal cations and ligands.
We first observed the almost exclusive formation of FeII(A6D4)2
with respect to FeII(A7D4)2 through self-sorting from a mixture
of A6, A7 and D4 (30 mM each) treated with 0.5 equiv of
FeII(BF4)2, indicating that there was a strong metalloselection by
Fe2+ ions in favor of A6 with respect to A7 on condensation with
8-aminoquinoline D4, resulting from the steric hindrance
introduced by the phenyl ring in A7 on formation of the
octahedral complex (Figure S20a−c, including 7% mixed ligand
complex FeII(A6D4, A7D4), which was confirmed by the mass
spectrometric analysis shown in SI). On the other hand, the p-
methoxy-substituted aniline D3 was selected to generate the
second constituent for preferential coordination with Cu+
instead of D1 and D2, to take advantage of its clear methoxyl
proton signal in the 1H NMR spectra. Moreover, another
preliminary experiment involving the components A6, A7, and D3
showed that the bidentate ligand (A7D3) was a preferential
choice for Cu+ binding as compared to A6D3, in line with the
known stability of tetrahedral Cu+ complexes with ligands
bearing a phenyl ring in α position with respect to a coordinating
pyridyl nitrogen site13 (Figure S21a,b). Consequently, one might
expect Fe2+ and Cu+ to pick up the constituents A6D4 and A7D3
orthogonally, if the preconstructed [2×2] sub-network coming
from a mixture of A6, A7, D3 and D4 was treated simultaneously
with both Fe2+ and Cu+ effectors. In order to extent such a [2×2]
sub-network into a [3×3] network, a third pair of components,
aldehyde A4 and amine C1, was selected in consideration of the
reactivity of their condensation reaction and the low risk of
interference with the two cation effectors.
Heating an equimolar mixture of all six components A6, A7, A4,
D4, D3, and C1 (30 mM each) in CD3CN at 60 °C for 5 h,
produced initially a [2×2] square sub-network A6D3/A7D3/
A6C1/A7C1. Subsequently as a function of time, A4D3 formed
and increased gradually by condensation from the free A4 and D3
over a long time (at least 28 h) and higher temperature (78 °C)
for equilibration. After equilibration, large amounts of the free
components A4 and D4 (20% and 32%, respectively) were
obtained as a result of their comparatively low reactivity in the
imine formation reaction (network NE in Scheme 5 and Table
S5, middle center; and Figure S23a,b), in agreement with the
results from the competition experiments (Figure S22).
2.3. [3×3] CDN with One Effector: Amplification of the
Unfittest Sub-network. A further step toward complex
adaptive systems was to investigate the behavior of an even
larger [3×3] network formed by a DCL of nine hydrazone
constituents first in response to a single effector (see Scheme 4).
The components A1 and B1 that form the only NNN tridentate
constituent for effector Zn2+ were used as the driving force for the
adaptation, and the components A3, A5, B2, and B3, not
containing the pyridine group, were selected as the other
components. The six-component mixture (1 equiv, 30 mM each)
was reacted at 78 °C for 17 h in CD3CN containing 5% aniline
(in volume) as catalyst to reach thermal equilibrium. These
components underwent condensation, transamination, and
recondensation reactions, giving an almost statistical distribution
of 12%/8%/12%/9%/10%/12%/12%/9%/7% respectively for
A1B1/A1B2/A1B3/A3B1/A3B2/A3B3/A5B1/A5B2/A5B3 with a
total of 9% hydrolysis over all nine constituents (network ND,
in which A1B3 mainly exists in E form; see Scheme 4 and Table
S4, middle left; 1H NMR spectra shown in Figure S18a,b). When
0.5 equiv (15 mM) of Zn2+ was added into this DCL, it
reorganized to amplify the fittest ligand constituent A1B1 in the
form of its ZnII(A1B1)2 complex and to up-regulate simulta-
neously four constituents A3B2/A3B3/A5B2/A5B3, together with
a total of 3% hydrolysis (network ND-Zn, Scheme 4 and Table
1
S4, middle right; H NMR spectra shown in Figure S19a,b). In
this case, it was the 2D square network A3A5B2B3, showing an
approximate statistical distribution and representing an “un-
fittest” sub-network, that acted as the agonist to the fittest
constituent A1B1 and was amplified in concert with the
adaptation of this [3×3] CDN toward the single effector Zn2+.
The behavior of the present system provides the first example
of evolution of a CDN in which a full [2×2] sub-CDN occupies
the “unfittest” ecological niche (see refs 1c−e).
2.4. [3×3] CDN with Two Effectors: Hierarchical
Adaptation. In the studies described above, we chose different
amines (hydrazines and hydrazides providing NN and NO
binding sites as well as anilines providing a single N binding site)
for condensation with the same pyridine-2-carboxaldehyde A1.
The mixture produced ligand constituents, presenting tridentate
NNN and NNO or bidentate NN coordination subunits, that
selectively formed complexes with Zn2+ and Cu+, respectively,
based on their different coordination features. For the purpose of
construction of a [3×3] network that would respond to two
effectors, three aldehyde components, two pyridyl-aldehydes A6
and A7 with markedly different steric hindrance features together
with A4, were chosen to generate, in combination with the
amino-components C1, D3, and D4, a DCL of nine constituents
that would contain two ligand constituents presenting
preferential binding for two different metal cation effectors.
[One may note that the 8-aminoquinoline component provides a
tridentate NNN coordination site on reaction with a pyridine-2-
carboxaldehyde, while presenting exchange rates of imine-type
compounds, which are much faster than those involving
hydrazone NNN sites derived from hydrazines.] These
As expected, on addition of 15 mM FeII(BF4)2 to this
preformed DCL the difference of steric hindrance between A6
and A7 led to a strong metalloselection (see Figure S20a), with
generation of the octahedral coordination complex FeII(A6D4)2,
in preference to the complex containing the corresponding A7
component, with very pronounced amplification of this ligand at
the expenses of the antagonist constituents containing either A6
or D4 as well as by the almost full incorporation of the free D4,
which was unreacted in the starting library. As a consequence, the
square sub-network A7A4D3C1 underwent overall up-regulation
driven by the amplification of its A6D4 agonist, resulting also in
the concomitant marked increase of A4C1 due to the
condensation of free A4 with the C1 liberated in the process by
down-regulation of the A6C1 antagonist. This agonist amplifica-
tion between a single entity and a square sub-network is similar to
that discussed above (see Scheme 4, middle right, network ND-
Zn). Furthermore, the distribution in this A4A7C1D3 sub-
network of the Fe2+-treated DCL showed a bias in favor of the
A7C1-A4D3 diagonal (network NE-Fe in Scheme 5 and Table S5,
1
middle right; see Figure S24a,b for H NMR data). Thereafter,
when this NE-Fe DCL was further treated with Cu+, the square
H
DOI: 10.1021/jacs.6b13072
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
sub-network A7A4D3C1 underwent an amplification of the
opposite diagonal A7D3−A4C1 driven by the formation of the
complex CuI(A7D3)2. A very pronounced up-regulation (from
lower than 1% to 24%) of A4C1, resulting from its double
agonistic relationship to both complexes FeII(A6D4)2 and
CuI(A7D3)2, was achieved in a single operation when the initial
DCL was treated simultaneously with Fe2+ and Cu+ (Scheme 5
and Table S5, bottom center; see Figure S25a,b for NMR data).
On the other hand, first addition of Cu+ led to the destruction of
the starting network, which may be attributed to the fact that
strong binding of free D4 and Cu2+ promoted the oxidation of
Cu+ (see Figure S26 for more details; see also the similar case of
NC-Cu, Scheme 4). As an extension of the results obtained from
the [2×3] network above, the present [3×3] network displays
the behavior of a hierarchical system through competitive
effectors and constituents, where operation of the full
connectome results in a much higher degree of control of the
DCL via the underlying CDN.
ASSOCIATED CONTENT
■
S
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/jacs.6b13072.
Experimental details, NMR spectra, and synthetic
procedures, including Tables S1−S5 and Figures S1−
S26 (PDF)
AUTHOR INFORMATION
■
Corresponding Author
*lehn@unistra.fr
ORCID
Jean-Marie Lehn: 0000-0001-8981-4593
Notes
The authors declare no competing financial interest.
Thus, increased complexity allows for (1) implementation of
sequential addition of different effectors; (2) increased efficiency
(higher yield) of imine formation, from about of 32%
uncondensed free components (of both aldehyde and amine
type) in NE to 4% in NE-Fe and to only 3% in NE-FeCu (see
Table S5, middle and bottom); and (3) higher selectivity in
constituent formation (which can be expressed by the increase in
the sum of the % of constituents A6D4−A7D3−A4C1 on the
diagonal from only 10% for NE, to 55% for NE-Fe and 80% for
NE-FeCu) corresponding to a strong reduction of multiplicity of
the entities present in the dynamic system through competition
between components.12b,14
ACKNOWLEDGMENTS
■
This work is dedicated to the memory of J. D. (Jack) Roberts.
The authors thank the ERC (Advanced Research Grant
SUPRADAPT 290585) and the University of Strasbourg for
financial support. G.M., on leave from the State Key Laboratory
of Supramolecular Structure and Materials at Jilin University,
gratefully acknowledges Prof. Shimei Jiang (Jilin University,
Changchun), the National Natural Science Foundation of China
(51673082), and the 111 Project (B06009), as well as the
University of Strasbourg Institute of Advanced Study (USIAS)
for postdoctoral fellowship support. He also thanks Jan Holub
and Jean-Franco
procedures.
̧ is Ayme for suggestions on experimental
3. CONCLUSION
The present study has explored dynamic covalent libraries
comprising six and nine interconverting constituents in
thermodynamic equilibrium and forming constitutional dynamic
networks of respectively [2×3] and [3×3] order. It has revealed
novel adaptive behaviors of these DCLs in response to the
application of one or two effectors. Although they specifically
implement ligand molecules and metal cation effectors, the
features displayed are of broad significance, as they represent a
demonstration of principle for the emergence of higher order
adaptive properties, namely,
REFERENCES
■
(1) For a selection of papers on constitutional dynamic chemistry, see,
for instance: (a) Lehn, J.-M. Proc. Natl. Acad. Sci. U. S. A. 2002, 99,
4763−4768. (b) Lehn, J.-M. Chem. Soc. Rev. 2007, 36, 151−160.
(c) Constitutional Dynamic Chemistry; Barboiu, M., Ed.; Topics in
Current Chemistry 322; Springer: Berlin, 2012. (d) Lehn, J.-M. Angew.
Chem., Int. Ed. 2013, 52, 2836−2850. (e) Lehn, J.-M. Angew. Chem., Int.
Ed. 2015, 54, 3276−3289.
(2) For a selection of reviews specifically on dynamic covalent/
combinatorial chemistry, see, for instance: (a) Lehn, J.-M. Chem. - Eur. J.
1999, 5, 2455−2463. (b) Rowan, S. J.; Cantrill, S. J.; Cousins, G. R.;
Sanders, J. K. M.; Stoddart, J. F. Angew. Chem., Int. Ed. 2002, 41, 898−
952. (c) Corbett, P. T.; Leclaire, J.; Vial, L.; West, K. R.; Wietor, J.-L.;
Sanders, J. K. M.; Otto, S. Chem. Rev. 2006, 106, 3652−3711.
(d) Ladame, S. Org. Biomol. Chem. 2008, 6, 219−226. (e) Miller, B. L.
Dynamic Combinatorial Chemistry; Wiley: Chichester, 2010. (f) Reek, J.
N. H.; Otto, S. Dynamic Combinatorial Chemistry; Wiley-VCH:
Weinheim, 2010. (g) Hunt, R. A. R.; Otto, S. Chem. Commun. 2011,
47, 847−855. (h) Belowich, M. E.; Stoddart, J. F. Chem. Soc. Rev. 2012,
41, 2003−2024. (i) Herrmann, A. Chem. Soc. Rev. 2014, 43, 1899−1933.
(j) Li, J. W.; Nowak, P.; Otto, S. J. Am. Chem. Soc. 2013, 135, 9222−
9239.
• multiple agonistic and antagonistic regulation
• agonist amplification involving a single constituent and a
sub-network or a diagonal of three constituents
• synergistic operation of several agonists
• conditional response of the system on the hierarchical
sequence of effector application
Finally and strikingly, the present results also demonstrate that
an increase in complexity of the system by introduction of an
additional entity (constituent or effector) may result in a simpler
output, with higher efficiency and selectivity through dynamic
competition between entities.3g,14
(3) For recent examples of constitutional dynamic networks, see, for
instance: (a) Giuseppone, N.; Lehn, J.-M. Chem. - Eur. J. 2006, 12,
1715−1722. (b) Ulrich, S.; Lehn, J.-M. Chem. - Eur. J. 2009, 15, 5640−
5645. (c) Lao, L.; Schmitt, J.-L.; Lehn, J.-M. Chem. - Eur. J. 2010, 16,
4903−4910. (d) Vantomme, G.; Jiang, S. M.; Lehn, J.-M. J. Am. Chem.
Soc. 2014, 136, 9509−9518. (e) Hafezi, N.; Lehn, J.-M. J. Am. Chem. Soc.
2012, 134, 12861−12868. (f) Vantomme, G.; Hafezi, N.; Lehn, J.-M.
Chem. Sci. 2014, 5, 1475−1483. (g) Holub, J.; Vantomme, G.; Lehn, J.-
M. J. Am. Chem. Soc. 2016, 138, 11783−11791.
Steps toward systems of increasing complexity15 require the
exploration of DCLs of higher order containing a larger number
of interconverting constituents and defining higher order
networks, including networks of networks. Such connectomes
lead to the emergence of novel features under coupling to
multiple agents in the environment and responding to them in
diverse fashions, thus revealing novel behaviors toward the
implementation of adaptive chemistry.
(4) For a chemical interaction network, see: Ghosh, S.;
Mukhopadhyay, P.; Isaacs, L. J. Syst. Chem. 2010, 1, 6.
I
DOI: 10.1021/jacs.6b13072
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(5) For chemical effectors, see, for instance: (a) Giuseppone, N.;
Schmitt, J.-L.; Lehn, J.-M. J. Am. Chem. Soc. 2006, 128, 16748−16763.
(b) Ulrich, S.; Lehn, J.-M. J. Am. Chem. Soc. 2009, 131, 5546−5559.
(c) Klein, J. M.; Saggiomo, V.; Reck, L.; Luning, U.; Sanders, J. K. M.
̈
Org. Biomol. Chem. 2012, 10, 60−66. (d) Sreenivasachary, N.; Lehn, J.-
M. Proc. Natl. Acad. Sci. U. S. A. 2005, 102, 5938−5948.
(6) For physical stimuli, see, for instance: (a) Eliseev, A. V.; Nelen, M. I.
Chem. - Eur. J. 1998, 4, 825−834. (b) Giuseppone, N.; Lehn, J.-M.
Angew. Chem., Int. Ed. 2006, 45, 4619−4624. (c) Barboiu, M.; Dumitru,
F.; Legrand, Y.-M.; Petit, E.; van der Lee, A. Chem. Commun. 2009, 16,
2192−2194. (d) Ingerman, L. A.; Waters, M. L. J. Org. Chem. 2009, 74,
111−117. (e) Belenguer, A. M.; Friscic, T.; Day, G. M.; Sanders, J. K. M.
Chem. Sci. 2011, 2, 696−700.
(7) For recent reviews about systems chemistry, see, for instance:
(a) Kauffman, S. A. ACS Symp. Ser. 2008, 981, 310−324. (b) Ludlow, R.
F.; Otto, S. Chem. Soc. Rev. 2008, 37, 101−108. (c) Nitschke, J. R. Nature
2009, 462, 736−738. (d) Peyralans, J. J. P.; Otto, S. Curr. Opin. Chem.
Biol. 2009, 13, 705−713. (e) Hunt, R. A. R.; Otto, S. Chem. Commun.
2011, 47, 847−858. (f) Taran, O.; von Kiedrowski, G. In Chemical
Synthetic Biology; Luisi, P. L., Chiarabelli, C., Eds.; John Wiley & Sons:
Chichester, 2011; pp 289−319. (g) Huck, J.; Philp, D. In Supramolecular
Chemistry: From Molecules to Nanomaterials; Gale, P. A., Steed, J. W.,
Eds.; John Wiley & Sons: Chichester, 2012; pp 1415−1445.
(h) Giuseppone, N. Acc. Chem. Res. 2012, 45, 2178−2188. (i) Li, J.;
Nowak, P.; Otto, S. J. Am. Chem. Soc. 2013, 135, 9222−9239.
(j) Grzybowski, B.; Otto, S.; Philp, D. Chem. Commun. 2014, 50,
14924−14925. (k) Mattia, E.; Otto, S. Nat. Nanotechnol. 2015, 10, 111−
119.
(8) (a) Crisalli, P.; Kool, E. T. J. Org. Chem. 2013, 78, 1184−1189.
(b) Cordes, E. H.; Jencks, W. P. J. Am. Chem. Soc. 1962, 84, 826−831.
(c) Dirksen, A.; Dirksen, S.; Hackeng, T. M.; Dawson, P. E. J. Am. Chem.
Soc. 2006, 128, 15602−15603.
(9) Schultz, D.; Nitschke, J. R. J. Am. Chem. Soc. 2006, 128, 9887−
9892.
(10) A similar case concerning the dynamic E/Z isomerization of
pyridyl-acylhydrazone in a DCL is shown in ref 3g.
(11) For an example of sequential Zn/Cu and Cu/Zn coordination-
controlled assembly of a heterobimetallic [2×2] grid-type metallo-
supramolecular architecture, see: Petitjean, A.; Kyritsakas, N.; Lehn, J.-
M. Chem. - Eur. J. 2005, 11, 6818−6828.
(12) (a) Ulrich, S.; Lehn, J.-M. J. Am. Chem. Soc. 2009, 131, 5546−
5559. (b) Dhers, S.; Holub, J.; Lehn, J.-M. Chem. Sci. 2017,
DOI: 10.1039/C6SC04662B.
(13) Dietrich-Buchecker, C. O.; Marnot, P. A.; Sauvage, J. P.;
Kintzinger, J. P.; Maltes
(14) Kovarícek, P.; Meister, A. C.; Flídrova,
K.; Lehn, J.-M. Chem. Sci. 2016, 7, 3215−3226.
̀
e, P. Nouv. J. Chim. 1984, 8, 573−582.
̌
̌
́
K.; Cabot, R.; Kovaríckova,
̌
̌
́
(15) For complexity and complex systems, see, for instance: (a)
Reference 1d. (b) Mainzer, K. Thinking in Complexity, 5th ed.; Springer:
Berlin, 2007. (c) Cohen, R.; Havlin, S. Complex Networks; Cambridge
University Press: Cambridge, 2010. (d) Nicolis, G.; Nicolis, C.
Foundations of Complex Systems, 2nd ed.; World Scientific Publishing
Co.: Singapore, 2012. (e) Alfonso, I. Chem. Commun. 2016, 52, 239−
250.
J
DOI: 10.1021/jacs.6b13072
J. Am. Chem. Soc. XXXX, XXX, XXX−XXX