pH-Controlled selection between one of three guests from a mixture using a coordination cage host

We demonstrate the use of a simple pH swing to control the selection of one of three different guests from aqueous solution by a coordination cage host.


Introduction
The ability of self-assembled molecular containers to accommodate guest molecules in the central cavity 1 is underpinning the development of a range of interesting functions from catalysis to drug delivery. 2 These cavities provide a shape-and size-constricted environment that is quite different from that in bulk solution. This environment can be controlled to some extent synthetically, providing a degree of control over which types of guest can bind and under what conditions: for example, incorporation of uorinated groups, 3 aromatic panels 4 or an array of H-bonding units 5 on the internal surface of molecular containers have all been used to provide selective binding of different types of guest. Given the importance of understanding guest binding quantitatively, we 6,7 and others 8 have performed systematic studies of the specic thermodynamic contributions to guest binding in particular container families. Very recently we have shown how we could use our knowledge of guest binding properties in a specic coordination cage to develop a scoring function which allows protein/ligand docking soware to predict new guest types for the cage, with high reliability, from a virtual screen: 9 this is the rst such application of the methodology of drug discovery to identifying new synthetic supramolecular host/guest systems.
Beyond the ability to design molecular containers as hosts, and to put guest binding on a quantitative and predictable footing, the next level of control is to be able to switch guest uptake/release by an external stimulus. Several examples are known of the release of guests from containers following disassembly or irreversible decomposition of the host. 10 Reversible stimulus-responsive uptake and release of guests is much rarer, with a handful of examples including the use of a redox swing, 11 light-induced isomerisation, 12 or a pH swing 13,14 to control guest binding. We demonstrated recently how acidic or basic guest molecules with pK a values in the range 3-11, including some drug molecules, could undergo fullyreversible changes in binding constant of up to three orders of magnitude in a coordination cage host as the pH changed. 14 This occurred irrespective of the sign of the charge on the guest. Thus, neutral amine guests were expelled from the cavity on protonation to give a cation, and neutral carboxylic acid guests were expelled from the cavity on deprotonation to give an anion, with the driving force in each case being the improved solvation (i.e. loss of hydrophobic character) in water.
Aer the stimulus-responsive uptake/release of individual guests, the next stage of control in host/guest complex formation would be to use the external stimulus (here, the pH swing) to switch an assembly not just between bound and unbound states, but between several different bound states. We describe here the rst demonstration of this behavior, showing how a simple change in pH can result in one of three different guests binding in a coordination cage host, with each one being bound and then released in turn as the pH is varied. This represents a signicant advance in the control that can be achieved with host/guest systems, which therefore opens the door to more sophisticated forms of functional behavior in which one of several different guests can be selected at will from a mixture.

Results and discussion
The host cage used for these studies is the Co 8 L 12 assembly ( Fig. 1) whose host/guest chemistry we have described in previous reports. 6,7,9,14 It contains a high-spin Co(II) ion at each vertex of an approximate cube, and a bis-bidentate ligand containing two pyrazolyl-pyridine chelating termini 15 spanning each edge of the cube. The pendant hydroxyl groups on the external surface make the cage water-soluble, 7 and its hydrophobic interiorlined with CH groups from the ligandresults in strong binding of suitably-sized guests in water with binding constants of up to 10 8 M À1 . 7, 9,14 It is stable over a wide pH range, and the paramagnetism of the Co(II) ions acts as a shi reagent dispersing the 1 H NMR signals over the range ca. +100 to À100 ppm, greatly facilitating NMR-based analysis of guest binding. 6,7,14 For these experiments we have selected three guests: acidic adamantane-1,3-dicarboxylic acid (H 2 A) which binds with K ¼ 2.3 Â 10 5 M À1 at low pH when it is neutral, but very weakly above pH 5 when it is deprotonated to A 2À ; †, 14 basic 1-aminoadamantane (B) which binds with K ¼ 1.0 Â 10 4 M À1 at high pH when it is neutral, but very weakly below pH 11 when it is protonated to HB + ; 14 and cyclononanone (C) whose binding constant of 1.1 Â 10 4 M À1 is pH independent. 7b These are summarized in Scheme 1.
In all cases, guest binding is signaled by a shi of the 1 H NMR signals of the bound guest to the region À6 to À10 ppm (Fig. 2) as a consequence of the array of paramagnetic ions surrounding the bound guest in the cavity. Each guest gives a quite distinct pattern of signals in this region which, fortuitously, is clear of signals from the host cage. This provides a convenient way to monitor replacement of one guest by another as the pH changes.
Initially we performed two separate pH-based switching experiments, involving competition between guests H 2 A and C, and then between guests B and C. The protocol in every case was the same: a solution of the host cage (0.2 mM) and the two guestsat concentrations determined by their binding constant in the cagewas prepared in D 2 O and the pH was adjusted by addition of NaOD or DCl, and the 1 H NMR spectrum and pH were recorded aer each addition. ‡ The results of the rst experiment (switching between H 2 A and C) are in Fig. 3. At low pH, neutral H 2 A binds much more strongly than C, and at the concentrations used we can only detect the cage$H 2 A complex in the 1 H NMR spectrum with no competing bound state cage$C. As the pH is raised, the characteristic signals of bound H 2 A decrease in intensity and are replaced by a new set of signals from bound C in the complex cage$C (Fig. 3). The physical interpretation of this is that as the pH rises and H 2 A is deprotonated to A 2À , it becomes hydrophilic and therefore more weakly binding than C which is not affected by pH. Thus, H 2 A is replaced completely (within the limits of sensitivity of the NMR experimentsee spectrum at pH 8.8 in Fig. 3) by C as guest, with A 2À being ejected from the host due to  its hydrophilicity. 14 The effect is fully reversible, with binding switching between the cage$H 2 A and the H$C states as the pH is changed. § A similar effect is seen in the experiment with guests B and C (Fig. 4). In this case the two guests have similar binding constants, so excess of B was used to allow binding of B to dominate over C when B is in its neutral form. At neutral pH, the only complex present is cage$C, because B is fully protonated as hydrophilic HB + whose binding is very weak. 14 As the pH rises and HB + is deprotonated to neutral B, the signals for bound C are reduced in intensity, and a new set of signals characteristic of bound B grows in as cage$C is replaced by cage$B. By pH 12.2, cage$B is clearly the dominant complex as we would expect given the presence of excess B over C.
Finally, we performed a combined experiment to demonstrate switching between all three bound guest states as a function of pH. This is a simple combination of the previous two experiments: a D 2 O solution containing 0.2 mM cage, H 2 A (0.75 mM), B (7.1 mM) and C (0.2 mM) was prepared, and 1 H NMR spectra were measured over the pH range 3-12. The results are summarized in Fig. 5 and 6. The evolution of 1 H NMR spectra in the À6 to À11 ppm range (Fig. 5) shows very clearly how, as pH increases, cage$H 2 A (dominant complex at low pH) is successively replaced by cage$C (dominant complex at neutral pH) and then by cage$B (dominant complex at high  4.1, 4.3, 4.6, 5.0, 5.6, 7.3, 8.8. Replacement of bound H 2 A (red signals) by C (green signals) as the pH rises is clear.  7.4, 8.9, 9.6, 9.8, 10.6, 11.3, 12.2. Replacement of bound C (green signals) by B (blue signals) as the pH rises is clear.  2.8, 4.3, 4.9, 5.7, 6.0, 6.5, 7.1, 8.1, 9.3, 9.7, 10.0, 10.6, 11.0, 11.5, 12.2. The change in occupancy of the cavity by the three different guests in succession is clear as the pH rises; the red, green and blue signals arise from the bound guests in the complexes cage$H 2 A, cage$C; and cage$B (see Fig. 2). Fig. 6 Graphical representation of the data obtained from the NMR spectra in Fig. 5, showing the proportion of each type of complex (red, cage$H 2 A; green, cage$C; blue, cage$B), as a percentage of total complexed cage present, across the pH range. Dots represent measured data. The blue and red lines are calculated fits for pHdependent binding of monobasic (B) and dibasic (H 2 A) guests, respectively (see ref. 14); the green line represents the calculated residual fraction of bound guest whose binding is not pH-dependent, i.e. cage$C. pH), associated with (i) deprotonation of H 2 A to A 2À at pH z 5 and then (ii) deprotonation of HB + to B at pH z 11. The proportions of each complex throughout the pH range, expressed as a fraction of total complex concentration, are summarized in Fig. 6 and illustrate very clearly the switching between the three different bound states as a function of pH. The pK a values for H 2 A and HB + are far enough apart to allow for near-quantitative conversion between the three bound states: at the extremes, the complexes cage$H 2 A (low pH) and cage$B (high pH) constitute close to 100% of the total complex present, and at pH 7.5, the population of bound cage is >97% cage$C with <2% of each of the other two complexes.

Conclusions
In conclusion, we have demonstrated how a host cage can select one of three possible guests from a mixture using a single external stimulus (a pH change)an unprecedented degree of control over guest binding. For any potential applications of molecular containers in which stimulus-responsive guest binding is an important factor, this ability to switch reversibly between any one of multiple bound states using a single stimulus represents a new level of sophistication and control in host guest chemistry which will expand the range of functions that can be developed.