Anion-induced reconstitution of a self-assembling system to express a chloride-binding Co10L15 pentagonal prism

ARTICLES PUBLISHED ONLINE: 5 AUGUST 2012 | DOI: 10.1038/NCHEM.1407

Anion-induced reconstitution of a self-assembling system to express a chloride-binding Co10L15 pentagonal prism Imogen A. Riddell1, Maarten M. J. Smulders1, Jack K. Clegg1,2, Yana R. Hristova1, Boris Breiner1, John D. Thoburn3 and Jonathan R. Nitschke1 * Biochemical systems are adaptable, capable of reconstitution at all levels to achieve the functions associated with life. Synthetic chemical systems are more limited in their ability to reorganize to achieve new functions; they can reconfigure to bind an added substrate (template effect) or one binding event may modulate a receptor’s affinity for a second substrate (allosteric effect). Here we describe a synthetic chemical system that is capable of structural reconstitution on receipt of one anionic signal (perchlorate) to create a tight binding pocket for another anion (chloride). The complex, barrel-like structure of the chloride receptor is templated by five perchlorate anions. This second-order templation phenomenon allows chemical networks to be envisaged that express more complex responses to chemical signals than is currently feasible.

I

ncreasingly complex synthetic supramolecules are finding use in diverse applications, including systems capable of converting light energy into chemical potential1, electrical current2 or torque3, activity-switchable polymerization catalysts4 and biocompatible membranous sacs5. Many of these synthetic supramolecular devices take inspiration from the functional complexity of biomolecular systems, the supramolecular architectures of which create function from form. Biological systems are protean, that is, adaptable: an amino acid residue may be incorporated into a protein, used in the biosynthesis of a nucleobase or oxidized to meet an organism’s energetic needs. Synthetic systems have so far demonstrated allostery6, whereby a binding event at one site influences molecular function elsewhere4,7, but the goal of signal-induced reconstitution to achieve new function remains elusive8. The supramolecular construction technique of subcomponent self-assembly9, in which metal–ligand coordinative interactions10–17 and metal-templated dynamic covalent imine bonds18–20 are formed during the same process, has been employed to create architectures of increasing complexity21–27. Here we describe the use of this approach to create a synthetic molecular network that exists in three distinct states. The first disordered state may be transformed into a second more-ordered state, which consists of tetrahedral capsules, by the addition of a chemical signal (either triflate or hexafluorophosphate anions, which act as templates for the capsules). A third state may be accessed from the first or second state through the addition of perchlorate, which induces the formation of a highly ordered, structurally unprecedented pentagonal prism assembled from 60 molecular components. This prism encapsulates five peripheral anions, which play a structural role in stabilizing its pentagonal framework. A sixth anion, such as chloride, can be bound with great affinity (affinity constant Ka . 6 × 105 M21) within a central pocket. This abiological system is thus capable of signal transduction28, wherein one anion triggers a structural reorganization29 that allows the newly

formed structure to function as a highly efficient binder of another anion. This ability to reconstitute allows for more complex abiological signalling cascades to be conceived, within which individual building blocks may play multiple roles.

Results and discussion

The reaction of p-toluidine, 6,6′ -diformyl-3,3′ -bipyridine and cobalt(II) triflimide hydrate (Co[N(SO2CF3)2]2.H2O) in acetonitrile resulted in a dynamic mixture of coordination complexes 1 (Fig. 1). Dynamic library 1 displayed a broad 1H NMR spectrum that contained many signals, indicative of a complicated mixture of species (Supplementary Fig. S1). Electrospray ionization mass spectrometry (ESI-MS) suggested that the components of this mixture had the general formula ConLm , where L is 6-formyl-6′ -tolyliminomethyl-3,3′ -bipyridine, 1 , n , 3 and 1 , m , 3n. The use of cobalt(II) triflate hexahydrate (Co(SO3CF3)2.6H2O) instead of the triflimide salt within the system shown in Fig. 1 produced a tetrahedral Co4L68þ cage, 2, as the unique product observed by paramagnetic 1H NMR spectroscopy30 and ESI-MS. The appearance of a new signal in the 19F NMR spectrum of the solution confirmed the encapsulation of a triflate ion (OTf 2) within 2, which suggests that this anion acted as a template that provided a driving force for the reorganization of the dynamic library31 1 into a single structure. This templation role was confirmed by the addition of sodium triflate to 1, which generated 2.OTf 2, as confirmed by NMR spectroscopy, ESI-MS and single crystal X-ray diffraction. Addition of potassium hexafluorophosphate to dynamic library 1 similarly resulted in the formation of the hexafluorophosphate adduct of tetrahedron 2; 19F NMR spectroscopy again confirmed the binding of the hexafluorophosphate anion within the capsule. Using the organic subcomponents and the conditions employed for the system depicted in Fig. 1, an analogous Fe4L68þ cage was prepared from iron(II) triflimide32. We hypothesize that a template is

1

Department of Chemistry, University of Cambridge, Lensfield Road, Cambridge CB2 1EW, UK, 2 School of Chemistry and Molecular Biosciences, The University of Queensland, Brisbane St Lucia, Queensland 4072, Australia, 3 Department of Chemistry, Randolph-Macon College, Ashland, Virginia 23005, USA. * e-mail: [email protected]

NATURE CHEMISTRY | VOL 4 | SEPTEMBER 2012 | www.nature.com/naturechemistry

751

ARTICLES

NATURE CHEMISTRY ii

DOI: 10.1038/NCHEM.1407

2Co(OTf)2

N N

N N

N

N N

N

N

N N

N

N

N N

N N

6 NH2 O N

N N N

i

N N

N

N N

N

N iii

N

2

iv

3

2Co(NTf2)2

3 N

N

N

N

N

N

N

O

N

N

N

N N N

plus others 1 Dynamic library

Co4L6

Co10L15

Figure 1 | States of a chemical system. p-Toluidine, 6,6′ -diformyl-3,3′ -bipyridine and cobalt(II) triflimide react to form a complex dynamic library of interconverting coordination complexes 1 (i). A Co4L6 tetrahedron 2 (shown in diagrammatic form above the crystal structure) may be prepared either directly via subcomponent self-assembly (by starting with the Co(II) triflate salt) (ii) or through templation on the addition of triflate or hexafluorophosphate anions to dynamic library 1 (iii). On the addition of LiClO4 to 1 or 2 (iv), quantitative conversion was observed to a Co10L15 pentagonal prism 3 (shown in diagrammatic form above the crystal structure). The schematic representations of 2 and 3 show each unique ligand once, additional ligands being represented as lines between the Co2þ ions, which are shown as orange spheres. In 3 the individual ligands are assigned separate colours for clarity. X-ray crystal structures: nitrogen atoms, blue; carbon atoms, grey; cobalt ions, orange.

necessary for the formation of cobalt cage 2 to overcome the effects of the Jahn–Teller distortion at the metal ion. This distortion leads to a disruption of the threefold symmetry axis of the metal coordination sphere. This threefold axis is a symmetry element of tetrahedrally symmetric 2; the breakdown of symmetry in turn disrupts the edge-to-face CH–p interactions that stabilize the analogous Fe4L68þ cage without the need for a template. The addition of lithium perchlorate to either library 1 or cage 2 resulted in the transformation into a Co10L1520þ pentagonal prism 3 (Fig. 1), which was the unique product observed by 1H NMR spectroscopy and ESI-MS. Complex 3 is composed of two parallel Co5L5 pentagonal rings21,33,34. Five ‘equatorial’ ligands (Fig. 2a, depicted in red and blue) comprise each ring, and at each cobalt(II) ion the two rings are linked by an ‘axial’ ligand (Fig. 2a, depicted in green). The overall architecture is barrel-like and has idealized D5 point symmetry. In total, 60 chemical species are brought together to generate each equivalent of 3. All of the metal centres have the same stereochemical L or D handedness; both enantiomeric forms of 3 are present in the crystal. In contrast to metallosupramolecular capsules reported previously, which generally incorporate facial coordination25,26,35,36, the cobalt(II) centres of 3 display meridional coordination of lower symmetry15,37,38 of the ligands around the metal centres. The interweaving of the ligands creates six distinct anion-binding pockets, five of which lie along the twofold symmetry axes (Fig. 2c) between the upper and lower rings. In the crystal, each of these pockets is filled by a perchlorate anion. A channel surrounded by ten inward-pointing pyridyl hydrogen atoms is located in the centre of the molecule (Fig. 2d). 752

At the midpoint of this channel is a sixth binding site, occupied by a chloride anion (Fig. 2d). The formation of this unprecedented structure is favoured by three factors. First, the perchlorate anions are well-suited to the pockets that they occupy within 3: 58% of the free void volume of each pocket is occupied by its perchlorate guest39, which corresponds well to the 55% optimum suggested by Mecozzi and Rebek40. By comparison, perchlorate would occupy only 32% of the free void volume of 2. Each perchlorate anion is surrounded by four cobalt(II) cations, which leads to electrostatic attraction and stabilization. Second, model compound studies (described in the Supplementary Information) revealed that a meridional coordination environment is favoured to a greater degree than a facial environment for a high-spin tris(pyridylimine)cobalt(II) model compound than for its low-spin iron(II) analogue. This observation suggests that 3, with all metal centres meridional, would be more favoured for cobalt(II) than for iron(II), for which the observed structure (2) has all facial metal centres. Finally, the compact structure of 3 allows its electron-rich toluidine residues to stack favourably with the electron-poor pyridine groups (Fig. 2)41,42. Confirmation of the persistence of structure 3 in solution was obtained through 1H NMR spectroscopy30 and high-resolution ESI-MS. Mass spectrometry also confirmed the crystallographic observation that 3 associates with a single chloride anion. As no chloride was added, we inferred that 3 had scavenged chloride from the glassware and laboratory implements used in its preparation. Treatment of 3 with a silver perchlorate solution did not result in the removal of the internally bound chloride ion43. This observation allowed us to conclude that the chloride affinity of 3 NATURE CHEMISTRY | VOL 4 | SEPTEMBER 2012 | www.nature.com/naturechemistry

NATURE CHEMISTRY

ARTICLES

DOI: 10.1038/NCHEM.1407

a

b

c

d

Figure 2 | Crystal structure of 3.(ClO4)5.Cl. a, Side view of the cationic part of the crystal structure of 3.(ClO4)5.Cl, which shows the parallel rings of equatorial ligands (red and blue) and the axial bridging ligands (green); cobalt, orange. b, Side view of a ball-and-stick representation of the crystal structure overlaid on a space-filling representation, emphasizing the compact nature of the structure. Carbon, grey; nitrogen, blue. c, View down the centre of the pentagonal prism, which shows the empty central channel and five distinct anion-binding pockets (shaded mesh in distinct colours). d, Cutaway view of the central binding channel, which emphasizes the orientations of the central pyridyl hydrogens of the axial ligands towards the central chloride anion (green).

is greater than 6 × 105 M21 (see the Supplementary Information). Later attempts to synthesize chloride-free 3 using glassware treated by silver perchlorate and metal salts recrystallized from silver-containing solutions also resulted in the isolation of the chloride-bound complex. We infer that the high chloride affinity of 3 results from the structure of its preorganized channel, optimally sized for chloride and surrounded by an almost spherical array of stabilizing hydrogen-bond donors44. The presence of these interactions is confirmed in the crystal structure through the observation of ten CH...Cl2 contacts of 2.60–2.77 Å (Fig. 2d). Under conditions in which the presence of adventitious chloride was minimized, other anions, including F2, Br2, N32, OCN2 and SCN2, were observed to bind within the central cavity of 3. Guest uptake was confirmed by the appearance of a new set of 1H NMR signals attributed to the host–guest complex. In each case a new host–guest peak was observed alongside the parent host peak; however, no evidence of anion-free 3 could be obtained by ESIMS, which led to the observation that 3 binds a central anion (adventitious chloride45 if no other anion is added) either during or immediately after its formation. No evidence could be found for the binding of the anions HF22, CN2, SeCN2 or the neutral molecules acetylene and carbon dioxide. NATURE CHEMISTRY | VOL 4 | SEPTEMBER 2012 | www.nature.com/naturechemistry

To create a host with an empty central pocket, an analogue of 3 was prepared by employing 4-hydroxyaniline in place of the toluidine residues of 3; ESI-MS and NMR spectra were consistent with the formation of a new product isostructural to 3. When this new pentagonal prism, 3′ , was prepared using silver-treated glassware and reagents (as noted above), new peaks were observed in the ESI-MS and NMR spectra that could be assigned to chloride-free 3′ , which was observed alongside the chloride adduct. The more electron-rich 4-hydroxyaniline residues of 3′ appear to have led to a lower overall positive charge surrounding the central cavity of 3′ and thus to a lower chloride affinity, in comparison to the more electron-poor 3. The observation of 3′ free of a central anion allows us to infer that only the peripheral template ions are necessary to template the pentagonal prismatic structure. Although the central site has great affinity for the anions noted above, no central anionic template is required. As discussed above, different anions play distinct templation roles in the interconversion between 1, 2 and 3. Although triflate served as a classical or first-order template for the formation of the triflate-binding tetrahedral cage 2, perchlorate induced a more complex templation process. It served not only as a first-order template for the five peripheral cavities of pentagonal prism 3, but also 753

ARTICLES

NATURE CHEMISTRY OTf −

ClO4−

ClO4−

3

363 K

DOI: 10.1038/NCHEM.1407

3

11 days ClO4−

OTf −

PF6−

1 2

2

Dynamic library

OTf −

313 K 35 days

PF6–

3 2

2

Mixture of isomers

Figure 3 | Chemical network showing the effects of the sequential addition of anions. The perchlorate anion is shown in red, triflate in green and hexafluorophosphate in blue. Co4L6 tetrahedra 2 shown in brackets are kinetic products observed to evolve to the corresponding thermodynamic Co10L15 products, 3, on heating. The tetrahedra not shown in brackets are the thermodynamic products.

as a second-order template, which generated the central cavity that serves to bind chloride. This second-order templation phenomenon, whereby the presence of one anion leads to the generation of a receptor for another, is conceptually related to the allosteric effect4. In this case it serves to allow the function of chloride binding to be turned on by the receipt of a signal (perchlorate) that does not otherwise interact with chloride. The addition of potassium hexafluorophosphate to 1 initially yielded the hexafluorophosphate adduct of cage 2 (Fig. 1). However, after heating this sample at 363 K for 11 days, the 1 H NMR signals that correspond to 2 had disappeared and a new set of peaks was observed, identical in number and relative intensities to those observed in the spectrum of 3, but at slightly different chemical shift values. Single-crystal X-ray diffraction of this new product revealed that it was isomorphous to 3, but with hexafluorophosphate anions instead of perchlorate in its five peripheral binding pockets, which led us to formulate this product as 3.(PF6)5. The ligands in this structure are less planar than those in the perchlorate analogue, and cause the binding pockets in 3.(PF6)5 to be expanded slightly in comparison with those of 3.(ClO4)5 , which allows the larger PF62 guests to fit snugly in the available void spaces. Adventitious chloride binding was also observed in 3.(PF6)5; the accommodation of the hexafluorophosphate anions did not interfere with chloride binding. The rate of conversion of the hexafluorophosphate adduct of 2 into 3 was found to depend both on the temperature and on the concentration of chloride. In the absence of externally added chloride, we observed the transformation from 2 into 3 to be slow, proceeding with a half-life of two days at 363 K. Faster kinetics were observed in the presence of added chloride; one equivalent of Cl2 per 3.(PF6)5 at 363 K resulted in the conversion of 2 into 3 with a half-life of 12 hours. When 2.PF6 was heated with one equivalent of Cl2 at 343 K, the consumption of 2 proceeded at a greater rate than the formation of 3, which suggests the presence of intermediate species. Such intermediates could not be identified during the course of this transformation because of their broad 1H NMR signals (Supplementary Fig. S17). 754

The observation that the addition of different template anions to dynamic library 1 could provoke a transformation of this library into different self-assembled host structures motivated us to explore the scope and limitations of such anion-mediated structural transformations in the context of a broader chemical network (Fig. 3). The addition of OTf 2 to the hexafluorophosphate adduct of 2 resulted in the formation of a mixture of the triflate and hexafluorophosphate adducts of 2. The relative binding constants of 2 2 these two anions could be calculated as KOTf /KPF6 ¼ 1.5+0.11. The minor increase in binding strength of OTf 2 relative to PF62 stands in contrast to the 50-fold preference for PF62 over OTf 2 for the analogous iron(II) cage32. The inversion of the preference of 2 between these two anions could be a consequence of the increase in cavity volume for 2 in comparison to that of the iron(II) analogue, as reflected in the increase in metal-to-metal distance (9.8 Å and 9.5 Å, respectively). The mixture of triflate and hexafluorophosphate adducts of cage 2 was observed to rearrange into host 3 after heating at 313 K over 35 days. Complex 3 was observed, by ESI-MS, to encapsulate different combinations of the two anions. The addition of OTf 2 to 3.(PF6)5 resulted in partial displacement of the PF62 anions to generate a similar mixture of hexafluorophosphate and triflate adducts of 3. This mixture displayed identical spectroscopic features to that obtained on equilibration of a mixture of 2.PF6 and 2.OTf. The addition of perchlorate to the hexafluorophosphate adduct of either 2 or 3 resulted in the displacement of all the encapsulated PF62 anions and generation of the perchlorate adduct of 3, which was thus identified as the system’s thermodynamic product.

Conclusion Within the protean system mapped out in Fig. 3, different anionic signals can either induce the expression of a simple receptor for the added anion, or initiate a more complex self-assembly process that results in the generation of a receptor architecture that encapsulates both the added template anion and a second, smaller anion (that is, chloride) with great affinity. Such signal-processing steps, whereby the application of a signal (perchlorate) turns on a function NATURE CHEMISTRY | VOL 4 | SEPTEMBER 2012 | www.nature.com/naturechemistry

NATURE CHEMISTRY

DOI: 10.1038/NCHEM.1407

(chloride binding), mirror how biological systems are able to manifest complex responses to environmental stimuli. The cylindrical form of host 3 evokes the barrel-like structures of natural46 and synthetic47,48 ion channels, which suggests that suitably functionalized derivatives of 3 might serve to gate the passage of biologically relevant anions, such as chloride, through membranes49, although 3 itself appears too small to span a biological membrane. This gating function could be turned on through the templatedirected transformation of 2 into 3, and chloride flux might be modulated through the incorporation of varying proportions of perchlorate and hexafluorophosphate into the peripheral binding pockets of 3. In the crystal, the presence of the larger PF62 anions is observed to result in a channel with a minimum diameter of 2.28 Å (calculations are presented in the Supplementary Information), whereas the smaller ClO42 guests provide a central channel 2.37 Å wide at its narrowest point. Investigations are underway into the ion-transport abilities of longer derivatives of 3.

Methods

Synthesis of [2](OTf )8. 6,6′ -diformyl-3,3′ -bipyridine (3.0 mg, 13.7 mmol), toluidine (3.03 mg, 28.3 mmol) and cobalt(II) triflate hexahydrate (4.26 mg, 9.1 mmol) were added to a J-Young NMR tube containing CD3CN (0.7 ml). The tube was sealed and subjected to three evacuation/nitrogen fill cycles. The solution was then sonicated and heated at 363 K for 24 hours. Red–orange crystals of [2](OTf )8 suitable for X-ray diffraction studies were grown by slow diffusion of diethyl ether into an acetonitrile solution over several days and were used directly in the analysis. X-ray analyses are detailed in the Supplementary Information. 1 H NMR (400 MHz, 298 K, CD3CN): d ¼ 237.2 (s, 12H, imine), 83.5 (s, 12H, 2,2′ -bipyridine), 67.7 (s, 12H, 5,5′ -bipyridine), 15.8 (s, 12H, 4,4′ -bipyridine), 10.3 (s, 36H, methyl), 10.0 (s, 24H, 3-aniline), 224.3 (s, 24H, 2-aniline). 19F NMR (400 MHz, 298 K, CD3CN): d ¼ 276.2 (s, OTf2), 2176.4 (s, OTf2 , 2). ESI-MS: m/z ¼ 1,108.0 (([2](OTf2)5)3þ), 793.9 (([2](OTf2)4)4þ), 605.2 (([2](OTf2)3)5þ), 479.56 (([2](OTf2)2)6þ). Synthesis of [3](ClO4)20. All glassware and implements were washed thoroughly with a dilute solution of AgClO4 in CD3CN (30 mM) before use. 6,6′ -diformyl3,3′ -bipyridine (24.0 mg, 113 mmol), toluidine (24.24 mg, 226 mmol) and freshly recrystallized cobalt(II) perchlorate hexahydrate (27.6 mg, 75.4 mmol) were added to a 10 ml Schlenk flask that contained degassed CD3CN (4.8 ml). The flask was sealed and subjected to three evacuation/nitrogen fill cycles. The reaction was stirred for 24 hours at 363 K, allowed to cool and the resultant orange solution was used without further purification. Square red–orange crystals of [Cl2 , 3](ClO4)19.C4H10O suitable for X-ray diffraction studies were grown from the slow diffusion of diethyl ether into an acetonitrile solution of [3](ClO4)20 over several days and were used directly in analysis. X-ray analyses are detailed in the Supplementary Information. 1 H NMR (400 MHz, 298 K, CD3CN): d ¼ 236.9 (s, 10H, imine), 225.2 (s, 10H, imine), 202.4 (s, 10H, imine), 132.1 (br s, 10H, 2,2′ -bipyridine), 130.0 (br s, 10H, 2,2′ -bipyridine), 96.9 (s, 10H, 5,5′ -bipyridine), 91.8 (br s, 10H, 2,2′ -bipyridine), 70.5 (s, 10H, 5,5′ -bipyridine), 64.2 (s, 10H, 5,5′ -bipyridine), 38.9 (s, 10H, 4,4′ bipyridine), 12.2 (s, 30H, methyl), 10.6 (s, 10H, 4,4′ -bipyridine), 10.5 (s, 30H, 3-aniline), 9.6 (s, 30H, methyl), 8.0 (s, 30H, 3-aniline), 7.8 (s, 10H, 4,4′ -bipyridine), 7.5 (s, 30H, methyl), –22.9 (s, 30H, 2-aniline), –50.4 (s, 30H, 2-aniline). ESI-MS: m/z ¼ 1,993.3 (([Cl2 , 3](ClO4)15)þ), 1,575.0 (([Cl2 , 3](ClO4)14)5þ), 1,295.9 (([Cl2 , 3](ClO4)13)6þ, 1,096.5 (([Cl2 , 3](ClO4)12)7þ, 946.9 (([Cl2 , 3](ClO4)11)8þ. Matrix-assisted laser desorption/ionization: 8,274 ([Cl2 , 3](ClO4)18)þ, 8,472 ([Cl2 , 3](ClO4)20)2. Elemental analysis (%): calculated for C390H330Cl20Co10N60O76.16H2O.C4H10O, C 54.18, H 4.29, N 9.62; found, C 53.95, H 4.03, N 9.41. The X-ray crystallographic coordinates for structures reported in this article are deposited at the Cambridge Crystallographic Data Centre (CCDC) under deposition numbers CCDC 853787, Co(II)L3; CCDC 853788, compound 3 with perchlorate; CCDC 853789, compound 3 with hexafluorophosphate (laboratory source); CCDC 853790, compound 3 with hexafluorophosphate (synchrotron source); CCDC 879992, tetrahedron 2. These data can be obtained free of charge (http://www.ccdc. cam.ac.uk/data_request/cif ). Full experimental details and crystallographic analysis are given in the Supplementary Information.

Received 24 February 2012; accepted 19 June 2012; published online 5 August 2012; corrected online 14 August 2012

References 1. Serreli, V., Lee, C. F., Kay, E. R. & Leigh, D. A. A molecular information ratchet. Nature 445, 523–527 (2007). NATURE CHEMISTRY | VOL 4 | SEPTEMBER 2012 | www.nature.com/naturechemistry

ARTICLES 2. Bhosale, S. et al. Photoproduction of proton gradients with p-stacked fluorophore scaffolds in lipid bilayers. Science 313, 84–86 (2006). 3. Koumura, N., Zijlstra, R. W. J., van Delden, R. A., Harada, N. & Feringa, B. L. Light-driven monodirectional molecular rotor. Nature 401, 152–155 (1999). 4. Yoon, H. J., Kuwabara, J., Kim, J-H. & Mirkin, C. A. Allosteric supramolecular triple-layer catalysts. Science 330, 66–69 (2010). 5. Capito, R. M., Azevedo, H. S., Velichko, Y. S., Mata, A. & Stupp, S. I. Selfassembly of large and small molecules into hierarchically ordered sacs and membranes. Science 319, 1812–1816 (2008). 6. Rebek, J. Jr & Wattley, R. V. Allosteric effects. Remote control of ion transport selectivity. J. Am. Chem. Soc. 102, 4853–4854 (1980). 7. Willans, C. E., Anderson, K. M., Potts, L. C. & Steed, J. W. Allosteric effects in a tetrapodal imidazolium-derived calix[4]arene anion receptor. Org. Biomol. Chem. 7, 2756–2760 (2009). 8. Lehn, J-M. From supramolecular chemistry towards constitutional dynamic chemistry and adaptive chemistry. Chem. Soc. Rev. 36, 151–160 (2007). 9. Nitschke, J. R. Construction, substitution, and sorting of metallo-organic structures via subcomponent self-assembly. Acc. Chem. Res. 40, 103–112 (2007). 10. Chakrabarty, R., Mukherjee, P. S. & Stang, P. J. Supramolecular coordination: self-assembly of finite two- and three-dimensional ensembles. Chem. Rev. 111, 6810–6918 (2011). 11. Sauvage, J-P. Transition metal-containing rotaxanes and catenanes in motion: toward molecular machines and motors. Acc. Chem. Res. 31, 611–619 (1998). 12. Gianneschi, N. C., Masar, M. S. & Mirkin, C. A. Development of a coordination chemistry-based approach for functional supramolecular structures. Acc. Chem. Res. 38, 825–837 (2005). 13. Whittell, G. R., Hager, M. D., Schubert, U. S. & Manners, I. Functional soft materials from metallopolymers and metallosupramolecular polymers. Nature Mater. 10, 176–188 (2011). 14. Sun, Q-F. et al. Self-assembled M24L48 polyhedra and their sharp structural switch upon subtle ligand variation. Science 328, 1144–1147 (2010). 15. Ward, M. D. Polynuclear coordination cages. Chem. Commun. 4487–4499 (2009). 16. Li, F., Clegg, J. K., Lindoy, L. F., Macquart, R. B. & Meehan, G. V. Metallosupramolecular self-assembly of a universal 3-ravel. Nature Commun. 2, 205, http://dx.doi.org.210.1038/ncomms1208 (2011). 17. Ronson, T. K. et al. Stellated polyhedral assembly of a topologically complicated Pd4L4 ‘Solomon cube’. Nature Chem. 1, 212–216 (2009). 18. Rowan, S. J., Cantrill, S. J., Cousins, G. R. L., Sanders, J. K. M. & Stoddart, J. F. Dynamic covalent chemistry. Angew. Chem. Int. Ed. 41, 899–952 (2002). 19. Belowich, M. E. & Stoddart, J. F. Dynamic imine chemistry. Chem. Soc. Rev. 41, 2003–2024 (2012). 20. Osowska, K. & Miljanic´, O. Sˇ. Self-sorting of dynamic imine libraries during distillation. Angew. Chem. Int. Ed. 50, 8345–8349 (2011). 21. Ayme, J-F. et al. A synthetic molecular pentafoil knot. Nature Chem. 4, 15–20 (2012). 22. Beves, J. E., Blight, B. A., Campbell, C. J., Leigh, D. A. & McBurney, R. T. Strategies and tactics for the metal-directed synthesis of rotaxanes, knots, catenanes, and higher order links. Angew. Chem. Int. Ed. 50, 9260–9327 (2011). 23. Forgan, R. S., Sauvage, J-P. & Stoddart, J. F. Chemical topology: complex molecular knots, links, and entanglements. Chem. Rev. 111, 5434–5464 (2011). 24. Chichak, K. S. et al. Molecular borromean rings. Science 304, 1308–1312 (2004). 25. Meng, W. et al. A self-assembled M8L6 cubic cage that selectively encapsulates large aromatic guests. Angew. Chem. Int. Ed. 50, 3479–3483 (2011). 26. Caulder, D. L. & Raymond, K. N. Supermolecules by design. Acc. Chem. Res. 32, 975–982 (1999). 27. Domer, J., Slootweg, J. C., Hupka, F., Lammertsma, K. & Hahn, F. E. Subcomponent assembly and transmetalation of dinuclear helicates. Angew. Chem. Int. Ed. 49, 6430–6433 (2010). 28. Park, J. S. et al. Ion-mediated electron transfer in a supramolecular donor– acceptor ensemble. Science 329, 1324–1327 (2010). 29. Meudtner, R. M. & Hecht, S. Helicity inversion in responsive foldamers induced by achiral halide ion guests. Angew. Chem. Int. Ed. 47, 4926–4930 (2008). 30. Turega, S. et al. Selective guest recognition by a self-assembled paramagnetic cage complex. Chem. Commun. 48, 2752–2754 (2012). 31. Otto, S., Furlan, R. L. E. & Sanders, J. K. M. Selection and amplification of hosts from dynamic combinatorial libraries of macrocyclic disulfides. Science 297, 590–593 (2002). 32. Hristova, Y. R., Smulders, M. M. J., Clegg, J. K., Breiner, B. & Nitschke, J. R. Selective anion binding by a ‘Chameleon’ capsule with a dynamically reconfigurable exterior. Chem. Sci. 2, 638–641 (2011). 33. Hasenknopf, B. et al. Self-assembly of tetra- and hexanuclear circular helicates. J. Am. Chem. Soc. 119, 10956–10962 (1997). 34. Giles, I. D., Chifotides, H. T., Shatruk, M. & Dunbar, K. R. Anion-templated self-assembly of highly stable Fe(II) pentagonal metallacycles with short anion-p contacts. Chem. Commun. 47, 12604–12606 (2011). 35. Glasson, C. R. K. et al. Unprecedented encapsulation of a [FeIIICl4]2 anion in a cationic [FeII4L6]8þ tetrahedral cage derived from 5,5′′′ -dimethyl-2,2′ :5′ ,5′′ :2′′ ,2′′ quaterpyridine. Chem. Sci. 2, 540–543 (2011). 755

ARTICLES 36. Clegg, J. K., Li, F., Jolliffe, K. A., Meehan, G. V. & Lindoy, L. F. An expanded neutral M4L6 cage that encapsulates four tetrahydrofuran molecules. Chem. Commun. 47, 6042–6044 (2011). 37. Paul, R. L. et al. Structures and anion-binding properties of M4L6 tetrahedral cage complexes with large central cavities. Dalton Trans. 3453–3458 (2004). 38. Hall, B. R. et al. Structures, host–guest chemistry and mechanism of stepwise self-assembly of M4L6 tetrahedral cage complexes. Dalton Trans. 40, 12132–12145 (2011). 39. Mingos, P. M. D. & Rohl, A. L. Size and shape characteristics of inorganic molecular ions and their relevance to crystallization problems. Inorg. Chem. 30, 3769–3771 (1991). 40. Mecozzi, S. & Rebek, J. Jr The 55% solution: a formula for molecular recognition in the liquid state. Chem. Eur. J. 4, 1016–1022 (1998). 41. Stephenson, A. & Ward, M. D. Molecular squares, cubes and chains from selfassembly of bis-bidentate bridging ligands with transition metal dications. Dalton Trans. 40, 10360–10369 (2011). 42. Howson, S. E. et al. Origins of stereoselectivity in optically pure phenylethaniminopyridine tris-chelates M(NN′ )3 nþ (M¼Mn, Fe, Co, Ni and Zn). Dalton Trans. 40, 10416–10433 (2011). 43. Alexander, R., Ko, E. C. F., Mac, Y. C. & Parker, A. J. Solvation of ions. XI. Solubility products and instability constants in water, methanol, formamide, dimethylformamide, dimethylacetamide, dimethyl sulfoxide, acetonitrile, and hexamethylphosphorotriamide. J. Am. Chem. Soc. 89, 3703–3712 (1967). 44. Li, Y. & Flood, A. H. Pure C–H hydrogen bonding to chloride ions: a preorganized and rigid macrocyclic receptor. Angew. Chem. Int. Ed. 47, 2649–2652 (2008). 45. Camiolo, S., Gale, P. A., Hursthouse, M. B. & Light, M. E. Nitrophenyl derivatives of pyrrole 2,5-diamides: structural behaviour, anion binding and colour change signalled deprotonation. Org. Biomol. Chem. 1, 741–744 (2003).

756

NATURE CHEMISTRY

DOI: 10.1038/NCHEM.1407

46. Jentsch, T. J., Stein, V., Weinreich, F. & Zdebik, A. A. Molecular structure and physiological function of chloride channels. Physiol. Rev. 82, 503–568 (2002). 47. Matile, S., Som, A. & Sorde´, N. Recent synthetic ion channels and pores. Tetrahedron 60, 6405–6435 (2004). 48. Meng, W., Clegg, J. K. & Nitschke, J. R. Transformative binding and release of gold guests from a self-assembled Cu8L4 tube. Angew. Chem. Int. Ed. 51, 1881–1884 (2012). 49. Haynes, C. J. E. & Gale, P. A. Transmembrane anion transport by synthetic systems. Chem. Commun. 47 (2011).

Acknowledgements This work was supported by the Engineering and Physical Sciences Research Council (EPSRC), the Netherlands Organization for Scientific Research (M.M.J.S.) and the Marie Curie International Incoming Fellowship Scheme of the Seventh European Union Framework Program (J.K.C.). We thank the EPSRC Mass Spectrometry Service at Swansea for MALDI/time-of-flight experiments, D. Howe for help in running the NMR experiments and C. Sporikou for the synthesis of 6,6′ -diformyl-3,3′ -bipyridine. The authors thank Diamond Light Source (UK) for synchrotron beam time on I19 (MT7114).

Author contributions Synthetic and spectroscopic work was carried out by I.A.R. Experiments were conceived by I.A.R., J.R.N., M.M.J.S., Y.R.H. and J.K.C. X-ray data were collected, solved and refined by J.K.C. Mass spectrometry was performed by B.B. and I.A.R. Data analysis was performed by I.A.R., M.M.J.S, J.D.T. and J.R.N. All authors contributed to the writing of the paper.

Additional information The authors declare no competing financial interests. Supplementary information accompanies this paper at www.nature.com/naturechemistry. Reprints and permission information is available online at http://www.nature.com/reprints. Correspondence and requests for materials should be addressed to J.R.N.

NATURE CHEMISTRY | VOL 4 | SEPTEMBER 2012 | www.nature.com/naturechemistry

ERRATUM

Anion-induced reconstitution of a self-assembling system to express a chloride-binding Co10L15 pentagonal prism Imogen A. Riddell, Maarten M. J. Smulders, Jack K. Clegg, Yana R. Hristova, Boris Breiner, John D. Thoburn and Jonathan R. Nitschke Nature Chemistry 4, 751–756 (2012); published online 5 August 2012; corrected after print 14 August 2012. In the version of this Article previously published, in the final paragraph of the Methods section the accession number CCDC 878882 should have read CCDC 879992. This has been corrected in the online Article.