Binding of Hydrophobic Guests in a Coordination Cage Cavity is Driven by Liberation of “High‐Energy” Water

Abstract The cavity of an M8L12 cubic coordination cage can accommodate a cluster of ten water molecules in which the average number of hydrogen bonds per water molecule is 0.5 H‐bonds less than it would be in the bulk solution. The presence of these “hydrogen‐bond frustrated” or “high‐energy” water molecules in the cavity results in the hydrophobic effect associated with guest binding being predominantly enthalpy‐based, as these water molecules can improve their hydrogen‐bonding environment on release. This contrasts with the classical form of the hydrophobic effect in which the favourable entropy change associated with release of ordered molecules from hydrophobic surfaces dominates. For several guests Van't Hoff plots showed that the free energy of binding in water is primarily enthalpy driven. For five homologous pairs of guests related by the presence or absence of a CH2 group, the incremental changes to ΔH and TΔS for guest binding—that is, ΔΔH and TΔΔS, the difference in contributions arising from the CH2 group—are consistently 5(±1) kJ mol−1 for ΔΔH and 0(±1) kJ mol−1 for TΔΔS. This systematic dominance of ΔH in the binding of hydrophobic guests is consistent with the view that guest binding is dominated by release of “high energy” water molecules into a more favourable solvation environment, as has been demonstrated recently for some members of the cucurbituril family.


Electronic supporting information
Measurements of binding constants (experimental details) p1 Fig

Measurements of binding constants
Binding constants of guests were measured by NMR spectroscopy in D2O using a Bruker AV3-400 spectrometer. Concentration of host cage H w was always 0.2 mM; guest binding was always in slow exchange such that separate signals for free H w and the H w •G complex could be observed which could be separately integrated to allow calculation of the binding constant at a range of temperatures. In addition, signals for bound guests appeared in the region -5 to -10 ppm (shifted by the cage paramagnetism) and integration of these compared to those for unbound G provided additional measurements of binding constants. When signals for free H w and the H w •G complex overlapped they were separated by deconvolution in the Bruker Topspin â software. When this was required the region around overlapping signals was carefully baselined first and then the peaks deconvoluted using a mixed Lorentzian / Gaussian model, manually adjusting parameter such as threshold and peak overlap to ensure a good fit. This comparison of integral measurements was repeated multiple times across a spectrum whenever distinct signals for free H w and the H w •G complex (whether overlapping or completely separate) could be identified, and these were averaged. In addition each measurement was repeated at least twice. The averaging of multiple independent measurements of each cage/guest system at each temperature ensured that issues associated with integral measurements of weak, broad or overlapping signals are minimised, with uncertainties on individual K values in the range 1 -10%. Thus, for the data analyses shown below, we have allowed a pessimistic uncertainty of ±10% in all data points; this gives errors in ∆G of < 1 kJ/mol.
The following figures showing how the 1 H NMR spectra changed with temperature for each H w /G pair, followed by the Van't Hoff plot for each system from which the values of ∆H and T∆S for binding for each guest (used in Table 1) were derived. Error bars are based on an estimated uncertainty of ±10% in each K value as described above. S1. Parts of the 1H NMR spectrum of a mixture of H w (0.2 mM) and guest 1 (8.47 mM) in D2O at various temperatures. Left: changes in signals of the host when guest binds (blue = free H w ; red = H w •1 complex). Right: changes in signals for bound guest (highlighted in red).

Fig. S2.
Van't Hoff plot for the H w / 1 system based on the NMR data shown above, from which the data in Table 1 were derived. Error bars are based on an estimated uncertainty of ±10% in each K value.
H w + guest 3 (cyclononanone) Fig. S3. Parts of the 1H NMR spectrum of a mixture of H w (0.2 mM) and guest 3 (0.14 mM) in D2O at various temperatures. Left: changes in signals of the host when guest binds (blue = free H w ; red = H w •3 complex). Right: changes in signals for bound guest (highlighted in red).

Fig. S4.
Van't Hoff plot for the H w / 3 system based on the NMR data shown above, from which the data in Table 1 were derived. Error bars are based on an estimated uncertainty of ±10% in each K value.   Table 1 were derived. Error bars are based on an estimated uncertainty of ±10% in each K value.   S8. Van't Hoff plot for the H w / 5 system based on the NMR data shown above, from which the data in Table 1 were derived. Error bars are based on an estimated uncertainty of ±10% in each K value. Fig. S9. Parts of the 1H NMR spectrum of a mixture of H w (0.2 mM) and guest 6 (1.33 mM) in D2O at various temperatures. Left: changes in signals of the host when guest binds (blue = free H w ; red = H w •6 complex). Right: changes in signals for bound guest (highlighted in red).

Fig. S10.
Van't Hoff plot for the H w / 6 system based on the NMR data shown above, from which the data in Table 1 were derived. Error bars are based on an estimated uncertainty of ±10% in each K value.   S12. Van't Hoff plot for the H w / 7 system based on the NMR data shown above, from which the data in Table 1 were derived. Error bars are based on an estimated uncertainty of ±10% in each K value.   S14. Van't Hoff plot for the H w / 8 system based on the NMR data shown above, from which the data in Table 1 were derived. Error bars are based on an estimated uncertainty of ±10% in each K value.