Adjusting the binding thermodynamics, kinetics, and orientation of guests within large synthetic hydrophobic pockets

Molecular Cavity Synthesis.†

With an eye to adding functionality to the hydrophobic pocket of molecules such as 2, it can be noted that there are several points on the framework amenable to functionalization. However, the position taken up by H_c (Fig. 1) in parent basket 2a is attractive for several reasons. First, because it is situated at the entrance to the cavity, a group at this position should be able to affect guest complexation rates. Second, the site is remote from the crown of benzal hydrogens and therefore gives a juxtaposing handle by which it may be possible to alter guest orientation within the cavity. Third, incorporation of functional groups into this position is relatively straightforward. For example, basket 2c (Scheme S1) was formed in 88% yield by reacting benzal-bridged cavitand (33, 34) 1 (35) with 2-methyl resorcinol (4, R = 2-methyl). However, it should be noted that when methyl groups occupy all these rim positions, no binding of adamantanes has been observed (Z. Laughrey, C.L.D.G., T. Senechal, and B.C.G., unpublished data). This is unfortunate because adamantanes are ideal if guest orientation and movement within the pocket are of interest; both because adamantanes possess no conformation ambiguity, and because they are readily available with a number of different functional groups. The challenge therefore is how to synthesize molecular baskets that possess fewer groups pointing into the binding cavity that will influence binding without shutting it down. Two molecular baskets were chosen as targets: 5a, with one methyl group pointing into the cavity, and 5b, which possesses a hydroxyl group pointing into the cavity (Scheme S2).‡ Key to the synthesis of baskets 5a and 5b is the dibromide 3, itself formed by the combination of three molecules of resorcinol and octabromide 1, using a six-fold Ullmann aryl ether reaction (Scheme S3) (20). In wishing to optimize the formation of 3, we considered each reaction variable in turn. However, we were unable to increase the yield beyond 35%. Apparently, each bridged can be inserted with equal ease (or difficulty), and so statistical yields of the tris-bridged were generally obtained along with quantities of the bis-bridged cavitand and tetra-bridged species 2a (18 and 25%, respectively). Interestingly, of the two possible bis-bridged species, it was only the A/C derivative that was isolated in sizeable quantities. Only trace quantities of its bis-A/B compatriot were ever observed in these reactions. Thus, after insertion of the first resorcinol moiety, the two remaining bromine atoms of the bridged benzal groups are in some way made less accessible than the other four bromine atoms.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Position designations used for the baskets 2 and 5.

• Download figure
• Open in new tab
• Download powerpoint

Scheme 2

Synthesis of monosubstituted baskets 5a–c (R = CH₂CH₂Ph).

• Download figure
• Open in new tab
• Download powerpoint

Scheme 3

Synthesis of key dibromide 3 (R = CH₂CH₂Ph).

With quantities of 3 in hand, it was relatively straightforward to react it with 2-methyl resorcinol (4, R¹ = 2-methyl) and form the monomethyl derivative 5a in 90% yield. In contrast, the formation of the monophenol derivative 5b was a slightly more protracted process. Direct reaction between 3 and pyrogallol (4, R¹ = OH) gave a new compound in near quantitative yield. However, close scrutiny of the ¹H NMR of this compound indicated that it was not the required derivative 5b. With the possibility that the 1,2-diol moiety of the pyrogallol was acting as a reductant, we considered the option that the product was in fact the corresponding debrominated derivative. Indeed, this hypothesis was supported by the observation that a spectroscopically identical compound was formed when dibromide 3 was treated with n-BuLi, and the resulting lithiate was quenched with water. Mass spectrometry provided further conformation that under either of these sets of conditions, 3 underwent a two-fold debromination. A way around this problem was envisioned to be the mono-protected pyrogallol 4 (R¹ = OCH₂OEt), which does not possess the requisite H atom that must be lost during oxidation. Therefore, this derivative was prepared from pyrogallol in the usual manner (36). As anticipated, this protected derivative successfully underwent insertion into the cavitand framework to yield a product that was, with the help of nuclear Overhauser effect NMR, identified as basket 5c. The normal deprotection conditions of about 2% aqueous HCl in tetrahydrofuran at room temperature did not sufficed in removing the protecting group. Apparently, the protecting group is itself protected within the confines of the cavity. However, by using slightly more vigorous conditions the group was removed to give basket 5b in 31% yield for the two steps.

Mechanism of Guest Exchange.

Either a dissociative (Eq. 1) or an associative mechanism (Eq. 2) (37) can be proposed for the exchange of adamantanes into and out of molecular baskets such as 2a. An appreciation of which mechanism is in operation in these types of systems is essential if a firm understanding of the exchange kinetics and guest reorientation kinetics of these caviplexes is to be developed. We chose 1-bromoadamantane (G5, Scheme S4) binding to basket 2a, using CDCl₃ as solvent, to investigate the mechanism of complexation. At room temperature, binding is slow on the NMR time scale, thus kinetic determinations using one-dimensional nuclear Overhauser effect ¹H NMR (38) are an ideal approach (see supporting information) for ascertaining the observed rate constants for exchange, k_a and k_b (Eq. 3). 1 2 3 The precise experimental procedures necessary for determining k_a and k_b have been presented elsewhere by Sherman and coworkers (38) and Perrin and Dwyer (39). The mechanism can be determined by examining the changes in k_a and k_b when host concentration is varied and guest concentration is kept constant (37). If only a dissociative mechanism is present, then k_a will vary as it relates to a second-order process, but k_b, relating to a first-order process, will remain constant. The results from five different experiments in which the host concentration was varied from 0.91 to 8.72 mM are given in Table 1 and indicate that a dissociative mechanism is in operation. Knowing this fact, the measured rates k_a and k_b can be translated into the chemical rate constants k₁ and k₋₁ (see supporting information). The average of these five experiments gives k₁ as 1880 s⁻¹ M⁻¹ and k₋₁ as 10.5 s⁻¹, which corresponds to a free energy of activation (ΔG‡) for the rate-determining guest egression step of 16 kcal mol⁻¹. Thus, as suggested by models, the highly complementary guest must clear the cavity before a new guest can enter. However, a partial “vacuum” is probably created in the process because the host cannot completely collapse to fill the void. Consequently, there is a relatively high energy barrier to the process, and on average only 10 or so of these events occur every second.

• Download figure
• Open in new tab
• Download powerpoint

Scheme 4

Guests and numbering used in this study.

Caviplex Properties.

As alluded to above, we recently determined the thermodynamics of association for baskets 2a-c in three different solvents: CDCl₃, toluene-d₈, and DMSO-d₆. We wished to determine how the presence of a methyl group or a hydroxyl group affected both the strength of association between the host and adamantane guests and the orientation of the guest within the host. To begin our studies, we turned first to a series of monosubstituted adamantanes, 1-CO₂H, 1-NH₂, 1-OH, and 1-Br (G2–5, Scheme S4), and the determination of the association constants between these guests and hosts 2a, 5a, and 5b. We chose toluene-d₈ as a solvent for this study both because it engenders relatively strong association constants and because it would allow the monophenol basket 5b (R¹ = OH) to participate in hydrogen bonding to suitably functionalized guests. The association constants were recorded at room temperature for the stronger complexations, and −25°C for the weaker ones, and are presented in Table 2. For an accurate comparison we repeated the reported value of G5 binding to 2a (19). The placement of a methyl group at the rim of the cavity has a significant impact on guest binding. For example, the binding of G5 to 2a and 5a was 4.4 and 3.1 kcal mol⁻¹, respectively. This drop arises from only a 6% reduction in the volume of the estimated 280 ų cavity of parent 2a. The impact of the methyl group in 5a resulted in the poorer guests G2–4, demonstrating only marginal binding at best at room temperature. However, even at lower temperatures, G2 did not bind to 5a. For host 5b, the smaller OH group had less of a negative impact after binding. Thus, the approximate 3% reduction in cavity volume reduces the strength of binding for G5 by less than 0.3 kcal mol⁻¹. With respect to guests G2 and G4, the opportunity to form a hydrogen bond with the phenol group of the host was not sufficient to induce measurable binding at room temperature. However, at the lower temperature binding was apparent, especially with the alcohol guest G4 where the free energy of complexation was measured at 3.4 kcal mol⁻¹. Finally, even at room temperature, the amino derivative G3 bound strongly (5.5 kcal mol⁻¹) to host 5b, presumably through a hydrogen bond–salt bridge interaction.

When adamantanes of C_3v symmetry or less bind to these hosts, to a first approximation, two possible diastereomeric complexes can form. Remaining with the monosubstituted adamantanes such as G5 (Fig. 2, pastel red substituent = hydrogen) for the moment, it is possible that the guest can bind bromine atom down or bromine atom up. The binding of G5 to 2a has been shown to occur exclusively in a halogen-down manner (Z. Laughrey, C.L.D.G., T. Senechal, and B.C.G., unpublished data). In contrast, the sizable stabilization arising when G3 binds to 5b undoubtedly arises when the guest binds functional group up. Thus, the association constants quoted in Table 2 are actually summations of these two types of binding mode (e.g., K_{1-Br up} and K_{1-Br down}) (40). An examination of NMR shift data for selected protons on the host and the guest gives an insight into the distribution of these two isomers. Shift values, recorded at −25°C where the four aforementioned guests exchange slowly on the 500 MHz NMR time scale, are shown in Table 3. Two methods of monitoring binding were used. First, the benzal proton H_B (Fig. 1) is an ideal proton for monitoring binding from the point of view of the host. The shifts (Δδ) after binding of the respective guest are reported in Table 3. Alternatively, the most useful way to monitor binding from the point of view of a monosubstituted guest is to measure the difference in the shift of the signal from two sets of protons when the guest is bound and the guest is free. These ΔΔδ values, for the two sets of equivalent protons H_2/8/9 and H_4E/6E/10E at opposite “poles” of the guest (Scheme S4) are also reported. The binding of G5, which binds strongly halogen atom down, acts as a reference. For each host there is a large downfield shift in the benzal proton when guests bind into the cavity; a phenomenon that can be attributed to the formation of C-H⋅⋅⋅X-R hydrogen bonds between host and guest. However, the ΔΔδ values are more informative for the two sets of protons on the guest. For each host, these values are strongly negative, indicating that the protons nearer the bromine atom spend more time at the base of the cavity where the π-electron density is higher than the protons at the opposite pole. On the other hand, for guest G2 binding to 2a, there is little shift in H_b and virtually no difference between the shift in the signals originating from proton groups H_2/8/9 and H_4E/6E/10E. This latter result demonstrates that in the complex G2@2a, no one isomer is preferred over the other. G2 did not bind to 5a; however, it does bind to host 5b and engenders a positive ΔΔδ value. Thus, the switch from R¹ = H to R¹ = OH in the host structure allows the formation of a hydrogen bond between host and guest and swings the isomer ratio of about 1:1 in G2@2a to one in which the carboxylic acid group is predominantly up in G2@5b. A more dramatic shift is seen for guests G3 and G4. For example, in the former the amino group is noted to bind principally functional group down when binding to 2a or 5a, but the guest switches to a functional group up orientation when binding to 5b (ΔΔδ goes from negative to positive). These results demonstrate how subtle changes to the structure of the host can be used to change the thermodynamic binding profile of the host and adjust the preferred orientation of the guest.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Isomerism in adamantyl guests binding to the molecular baskets.

A question that has not yet been addressed concerns the possibility that there is an equilibrium between these diastereomeric complexes that does not involve guest egression and reentry (Fig. 2). Examining this possibility requires a disubstituted adamantane where similar functional groups, located at opposite poles of the adamantane cage, provide two significantly strong binding modes that are independent of each other. We chose the commercially available mixture of the Z and E isomers of 1,4-dibromoadamantane (G7 and G8 in Scheme S4) as a source for such a guest. The 3:2 ratio mixture of Z and E isomers was readily separated by gravity column chromatography, using hexanes as a mobile phase. As shown in Fig. 3, in the E isomer G8, the bromine atom on the tertiary carbon is topologically equivalent to 1-bromoadamantane G5, whereas the bromine atom on the secondary carbon is topologically equivalent to 2-bromoadamantane G6. Furthermore, as both brominated “ends” of the molecule point in opposite directions, each can independently bind down into the cavity. The binding energies for G5 and G6 binding to host 2a have been respectively recorded as 3.35 and 2.58 kcal mol⁻¹ in CDCl₃, 4.38 and 3.52 kcal mol⁻¹ in toluene-d₈, and 6.17 and 5.44 kcal mol⁻¹ in DMSO-d₆ (Z. Laughrey, C.L.D.G., T. Senechal, and B.C.G., unpublished data). Thus, to a first approximation, we can expect the difference in free energy of association for each end of G8 to be no more than an average of these difference, i.e., <0.79 kcal mol⁻¹. Guest G7 differs considerably. Here, the two bromine atoms are roughly 90° to each other. A halogen atom cannot bind down into the crown of benzal hydrogens without forcing the other to impact the walls of the host. Therefore, binding is expected to be much weaker.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Space-filling modes of (a) 1-bromoadamantane G5, (b) 2-bromoadamantane G6, (c) (Z) 1,4-dibromoadamantane G7, and (d) (E)1,4-dibromoadamantane G8. The bromine attached to the secondary position (either C-2 or C-4) is shown in pastel red.

To study the complexation of these bromide guests we opted for the mixed solvent system of 10% DMSO-d₆ in CD₂Cl₂, which offered strong slow (relative to the NMR time scale) binding over a wide temperature range. The association constants for guests G1, and G5–8 binding to the three hosts at −13°C§ are shown in Table 4. The parent adamantane G1 was included because molecular models indicated that it could bind under the methyl group of 5a, whereas the halogen-down mode of binding of G5 was expected to induce it to bind less deeply and hence interact more with the methyl group. However the results indicate that placing the methyl group at the top of the cavity causes about a 30% reduction in the association of both G1 and G5. Overall, the same trend for the three hosts was seen for this series of guests as was observed for the previous ones (Table 2). One methyl group at the rim of the cavity has a considerable detrimental effect on guest binding, whereas the smaller hydroxyl group has less of an impact. A comparison of the association constants for the two isomers G7 and G8, binding to hosts 2a and 5a indicated the expected different levels of independence of each (bromine) binding point in the two guests. For example, the latter host binds G8 1.1 kcal mol⁻¹ more weakly than does 2a, but there is nearly a 3.7 kcal mol⁻¹ difference between the two hosts binding G7. When one bromine atom of the Z isomer atom binds down, the other is forced to impact the methyl group at the rim of 5a and binding is impaired. Overall, for all of the hosts examined, a strong preference is shown for the E isomer G8. On the other hand, the binding points of this guest are not entirely independent. In all three hosts examined, the sum of the free energies of association for guests G5 and G6 is greater than the free energies of association of the “combined” guest G8.

Although slow guest exchange was in operation between room temperature and −13°C, the ¹H NMR spectra of caviplexes G7@2a and G7@5b did not show the complexity in the guest region that would be expected if two different isomers were present. On the other hand, the ¹H NMR spectra of caviplexes G8@2a, G8@5a, and G8@5b were quite different. For these caviplexes, especially the latter two, the peaks attributed to either the free or the bound guest were very broad over this temperature range. To begin to explain this discrepancy we examined the binding of guest G7 to each of the three hosts by using variable temperature ¹H NMR. From room temperature down to −85°C, only one set of sharp guest signals was observed for each of the caviplexes. This finding is consistent with one or two possibilities. Either caviplexes G7@2a and G7@5b each consist of two isomers but the energy barrier between them is very low, or only one isomer predominates. An examination of shift values suggests that the latter is true. Hence, when binding to 2a, the proton H₂ of guest G6 (Scheme S4) shifts −2.68 ppm. In contrast, the shift of the equivalent proton of G7 (H₄) shifts only −0.67 after binding. This proton spends little time at the base of the cavity. Additionally, the protons H₈ in G7, i.e., adjacent to C₁ but distal from C₄, are shifted −2.29 when the guest binds to 2a. The “equivalent” (α) protons in G5 shift −2.35 when it binds to 2a. Thus, although guest G7 has two bromine atoms, only the tertiary halide group is recognized by the crown of benzal hydrogens.

Subsequently, we turned our attention to caviplex G8@5b and monitored its NMR as a function of temperature. We monitored the signal from the single OH group from room temperature down to −58°C. Over this temperature range, two significant changes occurred (see supporting information). First, and not surprisingly given the negative entropy changes that occur when these types of complexes are formed (Z. Laughrey, C.L.D.G., T. Senechal, and B.C.G., unpublished data), the signal attributed to the OH of the empty host decreased in intensity. More interestingly, as the corresponding signal from the occupied host increased it also went through a coalescence point at −19°C, and at −58°C is fully resolved into two signals.¶ Integration of these signals gave a ratio of the possible isomers, G8_1-Br↓@5b to G8_1-Br↑@5b, of 52:48.‖ From these results, the isomer equilibrium constant K_{1-Br up to down} = 1.1 (Fig. 2), as well as the isomerization rates k_{1-Br down} = 320 s⁻¹ and k_{1-Br up} = 300 s⁻¹, could be calculated (41) from the populations of the two species (see supporting information). These values were independent of the host to guest ratio, i.e., the process is a first-order one, and corresponds to an average activation barrier ΔG of around 10 kcal mol⁻¹. By similar approaches it was also possible to determine the rates of flipping for G8 within the confines of the other two hosts 2a and 5a.** The full results, given in Table 5, demonstrate that our choice of guest was reasonable (the two modes of binding involve similar releases of energy), and that for all hosts the slow step is the reorientation of the guest into the 1-Br up position. They also reveal that the placement of a functional group at the rim of the cavity does have a subtle effect on the distribution of isomers. For example, relative to the other two hosts examined, the OH at the rim of the cavity of 5b promotes the stabilization of the isomer with the 1-Br up. The result is that the binding of G8 to 5b is comparable to the binding of G8 to 2a, whereas in all other cases examined where hydrogen bonds are not formed, the hydroxyl group of 5b decreases guest affinity relative to the parent host.

A direct comparison with the egress rate determined for G5@2a in CDCl₃ is not possible. However, as an approximate guide to the differences between the rate of egression and the rates of reorientation, we examined the egress rate of G8 from 2a in 10% DMSO-d₆ in CD₂Cl₂. This value was determined to be 0.9 s⁻¹ at 260 K. This rate compares to the isomerization rate of around 100 s⁻¹ measured at 188 K. Hence, the reorientation process is more than two orders of magnitude faster than the rate at which the guest can completely clear the cavity and reenter. Apparently, the limited flexibility of the host is sufficient to allow the cavity to elongate along one axis and allow the guest to invert without exiting the cavity. We are not aware of other examples of open-host diastereomeric complexes that interconvert without the guest necessarily exiting the cavity.‡‡

NMR shift data at 260 K also supports the overall orientation conclusions. For example, returning to caviplex G8@2a, the distal H₈ atoms in G8 (Scheme S4) shifts −1.79 ppm when it binds to the host, whereas the proton H₄ shifts −2.22 ppm. A comparison of these shifts to the corresponding shifts after the host binding, G5 (H_2/8/9 atoms shift −2.35 ppm), G6 (H₂ shifts −2.68 ppm), and G7 (H₈ shifts −2.29 and H₄ shifts 0.67), indicates that the two isomers of G8@2a are prominent species.