Self-assembly in systems of subcomponents: Simple rules, subtle consequences

Article abstract of DOI:10.1002/anie.200703877

(Chemical Equation Presented) All together now: Five subcomponents (see scheme) come together cleanly with Cu^I to form simultaneously all four product structures shown. Simply by changing the stoichiometry, any given subset of product structures is accessible. To explain the observed selectivity, it is necessary to consider the stability not only of individual products, but of the system as a whole.

Full text of DOI:10.1002/anie.200703877

Angewandte
Chemie
DOI: 10.1002/anie.200703877
Systems Chemistry
Self-Assembly in Systems of Subcomponents: Simple Rules, Subtle
Consequences**
Rupam J. Sarma and Jonathan R. Nitschke*
A defining feature of living organisms is the ability of their
biomolecular machinery to create a highly complex set of
functional structures from a relatively limited set of basic
building blocks.[1] Although the same subcomponents are
employed across a wide variety of structures, and although the
linkages that connect them are in many cases formed
reversibly, under thermodynamic control,[2] biological struc-
tures have evolved not to interfere with each othersꢀ
assembly: A high degree of compartmentalization occurs
during assembly processes in living systems. In many cases,
this compartmentalization is physical, involving the spatial
separation of two systems by a membrane, for example. In
other cases, particularly in prokaryotes, no such physical
separation exists, and the systems must rely upon self-sorting
of chemical species to avoid “crosstalk”,[3] the unwanted
sharing of subcomponents between systems.[4] In seeking to
understand and mimic natural self-sorting, chemists have
investigated a variety of model systems,[5] which have
contributed to the understanding of the principles that
underlie their behavior.
Herein we describe a self-sorting system,[6] shown in
Scheme 1, for which we have deciphered the basic rules that
govern which products will be observed under what circum-
stances. Although these rules are simple, the complexity of
the overall system and the degree of control obtained
represent advances in the state of the art. By controlling the
Scheme 1. The CuI-templated formation of structures 1–4 from amines A and B together with aldehydes X, Y, and Z.
stoichiometry of the five starting subcomponents, any arbi-
[*] Dr. R. J. Sarma, Dr. J. R. Nitschke
University of Cambridge
trary subset of the four product structures, in any relative
proportion, may be prepared. The self-assembly rules gov-
erning this system are thus deterministic and may be used as
programming instructions for the selection of “output”
structures based upon “input” chemical species.
Department of Chemistry
Lensfield Road Cambridge, CB2 1EW (UK)
Fax: (+44)223-793-215
E-mail: jn34@cam.ac.uk
As shown in Scheme 1, the reaction of amines A and B
with aldehydes X, Y, and Z in acetonitrile/DMSO produced a
dynamic library[7] of imines in equilibrium with the starting
materials and half-formed products (not shown) in which only
a single aldehyde group of Y or Z reacted with A or B. The
[**] This work was supported by the Walters-Kundert Charitable Trust,
the Swiss State Secretariat for Education and Research and the
Swiss National Science Foundation.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
Angew. Chem. Int. Ed. 2008, 47, 377 –380
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
377

Communications
addition of copper(I) tetrafluoroborate resulted in the clean
formation of a mixture of structures 1–4, which were
identified by NMR spectroscopy and ESI-MS. Complexes
1,[8] 2,[9] 3,[10] and 4[11] could also be prepared individually from
their constituent subcomponents and copper(I), as has been
previously reported.
Although it might be imagined that each of the inter-
mediate imine structures shown in Scheme 1 could interact in
a variety of ways with CuI ions, only four products were
observed after thermodynamic equilibration. Most of the
imines were eliminated from the initial dynamic library[12]
which indicated that the CuI ions exerted a strong template
effect[13] upon this library.
This selectivity may be encapsulated as a rule of valence
satisfaction:[11] The smallest possible structures will be formed
in which all CuI ions are tetracoordinate and all nitrogen atoms
are bound to a CuI ion. In structures incorporating dialde-
hydes Y and Z, such structures are di- or trinuclear helicates,
because imine ligands derived from these subcomponents are
poorly configured to chelate a single CuI ion pseudotetrahe-
drally.[11] This rule does not preclude the formation of a
mixed-ligand dicopper helicate incorporating one ligand from
each of 2 and 4. ESI-MS indicated the presence of such a
product in only very small quantities (ca. 1%), however, and
we were not able to identify its resonances in NMR spectra.
This rule allows us to predict the outcome of self-assembly
reactions involving subsets of the full array of subcomponents
of Scheme 1. In simple cases, this outcome is relatively
straightforward, as shown in Scheme 2. The presence of one
amine (A) and two aldehydes (X and Y) only allows two
Scheme 3. Although 1, 2, and 3 are all individually allowed, only 1 and
3 are observed in this system for the given stoichiometry.
template. The only other system reported to do this required
two separate metal ions, CuI and FeII.[14]
Both dialdehyde Y and aniline A were present in the
system shown in Scheme 3, ostensibly permitting the forma-
tion of “allowed” structure 2. The removal of Y and A leaves
X and B to form 5. This structure, however, is a “forbidden”,
as the CuI centers are not tetrahedrally coordinated. Avoid-
ance of forbidden 5 thus leads to the suppression of allowed 2,
and only 1 and 3 are observed as products.
This result demonstrates that the observed set of products
cannot be predicted only upon the basis of the stability of
individual products. The rule of valence satisfaction must be
applied to the system as a whole; all product structures must
obey it.
Scheme 2. Since A, X, and Y may only be used to construct the ligands
of 1 and 2, these are the only products observed.
Although 1 and 3 are the sole products observed in the
system of Scheme 3, a simple change in stoichiometry altered
the composition of this product mixture. As shown in
Scheme 4, product mixtures consisting of 1 + 2, 1 + 3, and
1 + 2 + 3 are all accessible by altering the relative proportions
of A and B and of X and Y. When X is present, A must react
preferentially with X to form 1, since X cannot form an
allowed product with B. Since Y may generate either 2 or 3,
the ratio of A and B after the formation of 1 determines the
ratio of 2 to 3. NMR spectra showing the conversion of a
imines to be formed; the path from these two imine ligands to
complexes 1 and 2 is thus clear-cut.
A more complex system is generated from two amines and
two aldehydes. For example, when the four subcomponents
shown in Scheme 3 were mixed with CuI in the proportions
shown, only products 1 and 3 were observed by NMR
spectroscopy and ESI-MS. Notably, this system allowed for
the clean sorting of four subcomponents into two independent
“baskets” (product structures) using a single metal-ion
378
www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 377 –380

Angewandte
Chemie
aꢀxꢀz ꢁ 0
ð1Þ
ð2Þ
ð3Þ
bꢀz ꢁ 0
y ¼ 2 a þ 2 bꢀ2 xꢀz
The condition imposed by Equation (1) ensures that a
sufficient quantity of A is present to react with all X and Z
present in the system to produce 1 and 4. Equation (2)
likewise ensures that enough B is available to react with all
dialdehyde Z. Equation (3) requires that the amount of
dialdehyde Y present in the system be sufficient to react with
any residual A and B, once the requirements of X and Z have
been met. Outside of these boundary conditions, determin-
istic control is lost. For example, when A, B, and Yare present
in quantities such that 2a + 2b < y, mixtures of different
proportions of 2 and 3 are observed. Higher concentrations of
CuI tend to favor the formation of 3, as we have observed
previously.[10]
Table 1 lists the correct stoichiometry of subcomponents
required to arrive at an arbitrary equimolar subset of the four
products. All of these combinations have been tested under
Table 1: The quantities of subcomponents and CuI required to generate
any arbitrary subset of the structures shown in Scheme 1.
Scheme 4. By changing the stoichiometry of A, B, X, and Y, any
combination of 1, 2, and 3 may be selected as products.
Amine(s)
Aldehyde(s)
Equiv CuI [15]
Observed product(s)
2A
4A
4B
2A+2B
6A
2A+4B
4A+2B
4A+4B
6A+2B
2A+6B
6A+4B
8A+2B
4A+6B
6A+6B
8A+6B
2X
2Y
2Y
2Z
1
2
3
2
3
4
3
5
4
5
6
5
6
7
8
1
2
3
4
mixture of subcomponents A, B, X, and Y into a mixture of 1,
2, and 3 after the addition of CuI are shown in Figure 1.
The system may thus be said to follow a program during
the self-assembly process. First, aldehydes X and Z react to
produce 1 and 4, removing the corresponding amounts of A
and B from the system. Second, aldehyde Y reacts to produce
2 and 3, the relative proportions of which depend upon the
amounts of A and B left in solution. This program may be
quantified; from a mixture containing a equivalents of A, b of
B, x of X, y of Y, and z of Z,[15] the following products would be
observed: x/2 equivalents of 1, (aꢀxꢀz)/4 of 2, (bꢀz)/4 of 3,
and z/2 of 4. For these rules to hold, it is necessary to impose
the boundary conditions of Equations (1)–(3):[15]
2X+2Y
2X+2Y
2X+2Z
4Y
2Y+2Z
2Y+2Z
2X+4Y
2X+2Y+2Z
2X+2Y+2Z
4Y+2Z
2X+4Y+2Z
1+2
1+3
1+4
2+3
2+4
3+4
1+2+3
1+2+4
1+3+4
2+3+4
1+2+3+4
conditions where each product is present at a concentration of
approximately 10 mm; NMR spectra are presented in the
Supporting Information. Products within mixtures were
identified by ESI-MS of the mixtures and by direct compar-
ison of 1H NMR spectra of pure products with the spectra of
mixtures. Attempts to use fingerprinting to identify the
components of these mixtures using 2D NMR spectroscopy
(COSY, ROESY) were unsuccessful; we attribute this to the
1
wide variation in H relaxation times between the various
products.
More complex mixtures, having arbitrary product ratios,
are predicted to be readily accessible within the boundaries
set by Equations (1)–(3), by simply mixing together subcom-
ponents in the ratio in which they are found in the desired
collection of product structures.[15] The system of Scheme 1 is
thus fully deterministic. The four degrees of freedom[16]
associated with the quantities of the inputs A, B, X, Y, and
Figure 1. 1H NMR spectra corresponding to the dynamic library
formed by subcomponents A, B, X, and Y (bottom) and the mixture of
products 1+2+3 (top) generated by addition of CuI to this library.
Angew. Chem. Int. Ed. 2008, 47, 377 –380
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
379

Communications
volume of solvent was reduced by half and further [D6]DMSO
(0.2 mL) was added. The progress of the reaction was monitored
using 1H NMR spectroscopy over 7 d, after which time the spectrum
corresponded in all cases to the mixture of products indicated in
Table 1.
Z are fully reflected in the quantities of outputs 1, 2, 3, and 4
observed after self-assembly.
Examination of ways in which the aldehydes and amines
of Scheme 1 might come together reveals how this system
might or might not be extended, in keeping with the rule of
valence satisfaction. The addition of an amine that contains
an additional pyridyl nitrogen atom, such as 2-(2-pyridyl)-8-
aminoquinoline C (Scheme 5), would lead to an undefined
Received: August 23, 2007
Published online: November 16, 2007
Keywords: combinatorial chemistry ·
.
dynamic covalent chemistry · self-assembly · systems chemistry ·
template synthesis
Scheme 5. Amine C, whose addition to the system of Scheme 1 would
result in a loss of deterministic control over the product mixture in the
presence of one other amine and more than one aldehyde, and
aldehydes of type W, the addition of any number of which to this
system would result in the conservation of control.
[1] L. M. Greig, D. Philp, Chem. Soc. Rev. 2001, 30, 287 – 302.
[2] J. M. Lehn, Science 2002, 295, 2400 – 2403.
[3] A. X. Wu, L. Isaacs, J. Am. Chem. Soc. 2003, 125, 4831 – 4835.
[4] F. M. Harold, Microbiol. Mol. Biol. Rev. 2005, 69, 544 – 564.
[5] a) R. Krämer, J. M. Lehn, A. Marquis-Rigault, Proc. Natl. Acad.
Sci. USA 1993, 90, 5394 – 5398; b) K. A. Jolliffe, P. Timmerman,
D. N. Reinhoudt, Angew. Chem. 1999, 111, 983 – 986; Angew.
Chem. Int. Ed. 1999, 38, 933 – 937; c) S. J. Rowan, D. G.
Hamilton, P. A. Brady, J. K. M. Sanders, J. Am. Chem. Soc.
1997, 119, 2578 – 2579; d) T. Kamada, N. Aratani, T. Ikeda, N.
Shibata, Y. Higuchi, A. Wakamiya, S. Yamaguchi, K. S. Kim,
Z. S. Yoon, D. Kim, A. Osuka, J. Am. Chem. Soc. 2006, 128,
7670 – 7678; e) H. B. Yang, K. Ghosh, B. H. Northrop, P. J. Stang,
Org. Lett. 2007, 9, 1561 – 1564; f) I. Saur, R. Scopelliti, K.
Severin, Chem. Eur. J. 2006, 12, 1058 – 1066; g) G. Ashkenasy, R.
Jagasia, M. Yadav, M. R. Ghadiri, Proc. Natl. Acad. Sci. USA
2004, 101, 10872 – 10877; h) P. Mukhopadhyay, P. Y. Zavalij, L.
Isaacs, J. Am. Chem. Soc. 2006, 128, 14093 – 14102.
[6] R. F. Ludlow, S. Otto, Chem. Soc. Rev. 2007, DOI: 10.1039/
B611921M.
[7] a) P. T. Corbett, J. Leclaire, L. Vial, K. R. West, J. L. Wietor,
J. K. M. Sanders, S. Otto, Chem. Rev. 2006, 106, 3652 – 3711;
b) J. M. Lehn, A. V. Eliseev, Science 2001, 291, 2331 – 2332.
[8] D. Schultz, J. R. Nitschke, J. Am. Chem. Soc. 2006, 128, 9887 –
9892.
[9] J. R. Nitschke, D. Schultz, G. Bernardinelli, D. Gꢁrard, J. Am.
Chem. Soc. 2004, 126, 16538 – 16543.
[10] M. Hutin, R. Franz, J. R. Nitschke, Chem. Eur. J. 2006, 12, 4077 –
4082.
outcome when present with amine A and two or more
aldehydes. Any collection of monoaldehydes of type W
(Scheme 5) could be added, however, and would lead to a
well-defined mixture of products. The key difference is that
amine C could add three different products to the dynamic
library[7] of products of Scheme 1, whereas any collection of
aldehydes of type W would add only one product per
aldehyde. A more complete discussion is presented in the
Supporting Information.
The present system thus demonstrates one simple mech-
anism by which the complex self-organizing systems of
biomolecules within prokaryotic cells[4] might avoid interfer-
ing with one another. The question of how molecules might
organize into mutually noninterfering complex systems is also
related to the question of how prebiotic chemical systems
became alive.[17]
Still more complex self-organizing systems might be
created by adding further layers of dynamic linkages[18] that
do not interfere chemically with the imines and the metal
coordination of the present system. These additional linkages
would impose their own selection rules upon the overall self-
assembly process, eventually allowing access by thermody-
namic self-assembly to structures of sufficient complexity to
generate complex function.[19]
[11] M. Hutin, G. Bernardenelli, J. R. Nitschke, Proc. Natl. Acad. Sci.
USA 2006, 103, 17655 – 17660.
[12] S. M. Voshell, S. J. Lee, M. R. Gagne, J. Am. Chem. Soc. 2006,
128, 12422 – 12423.
[13] a) D. H. Busch, Science 1971, 171, 241 – 248; b) S. Brooker, Y.
Lan, J. R. Price, Dalton Trans. 2007, 1807 – 1820.
[14] D. Schultz, J. R. Nitschke, Angew. Chem. 2006, 118, 2513 – 2516;
Angew. Chem. Int. Ed. 2006, 45, 2453 – 2456.
Experimental Section
Full experimental and characterization details, including copies of
NMR spectra, are presented in the Supporting Information. In a
typical experiment, calculated amounts of amines A and B and
aldehydes X, Y, and Z (as noted for each of the entries in Table 1) and
[D3]MeCN (0.4 mL) were added to a teflon-capped NMR tube, and
the corresponding quantity[15] of [Cu(NCMe)4]BF4 was added.
Quantities were chosen such that the concentration of the product
copper complexes would be approximately 10 mm. The dark brown
mixture thus obtained was degassed and purged three times with
argon. The reaction mixture was kept overnight at room temperature
and then at 508C for 24 h. During this period dark brown solids
precipitated. The volume of solvent was reduced to half the original
volume under vacuum, and subsequently [D6]DMSO (0.2 mL) was
added, which resulted in dissolution of the precipitated solids. The
reaction mixture was again kept at 508C for 24 h, after which the
[15] The number of equivalents of copper(I) ions required to create a
set of coordinatively saturated [L4CuI] structures will be
equivalent to the total number of nitrogen atoms present in all
subcomponents divided by four: a/4 + b/2 + x/4 + y/2 + z/4.
[16] In order to satisfy Equation (3), the quantity of one of the
subcomponents employed must be determined by the quantity
of the other four, which eliminates a degree of freedom. An
infinite number of solutions exist for Equations (1) and (2),
which thus do not eliminate any degrees of freedom from the
system.
[17] R. M. Lemmon, Chem. Rev. 1970, 70, 95 – 109.
[18] J. M. Lehn, Chem. Soc. Rev. 2007, 36, 151 – 160.
[19] See molecular machines special issue: Acc. Chem. Res. 2001, 34,
410 – 522.
380
www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 377 –380