Anion recognition as a method for templating pseudorotaxane formation

Abstract

Two thread-shaped cations, pyridinium nicotinamide and imidazolium, as their chloride and hexafluorophosphate salts, were studied with regards to complexation with hydrogen-bond-donating acyclic and macrocyclic ligands. In the latter case, the cations form pseudorotaxanes templated by the chloride anion but not hexafluorophosphate. This formation is a function of the coupling of ion-pairing between the cation and chloride anion and subsequent recognition of the anion by the macrocyclic diamide, which provides the driving force for interpenetration. We propose that this anion template principle is a general method for the construction of pseudorotaxanes and could be applied to other cationic threads, anions, and macrocyclic species.

The synthesis and study of interlocked or mechanically bonded supramolecular species, known as rotaxanes, catenanes, and knots, has become an area of increasing interest in the last decade (1–4). This interest is largely because of the continuing emergence of effective template methods for their construction and the unique properties of the resulting ensembles—properties which, in many cases, are tied to the manipulation of the residual template effects used to construct them (5–10). These methods generally depend on noncovalent interactions between neutral or cationic components to generate the interlocked products. Anions, when present, rarely play an active role in these synthetic protocols and are generally chosen so as to interfere as little as possible in the template process (e.g., the use of relatively noncoordinating PF counterions) (11–13). We have recently shown that pseudorotaxane formation can be templated selectively by chloride anions in acetone solution (Fig. 1) (14). In this system, macrocycle 2 binds the pyridinium cation–chloride anion ion-pair thread 1a through a combination of first and second sphere coordination of the anion (i.e., complexation of the specific ion pair as a whole entity). Replacement of the chloride ion by less complementary anions, such as bromide or iodide, destabilizes the recognition at the macrocyclic amide-based binding site and, as a result, the ability of the cationic thread to form the desired pseudorotaxane superstructure. This pseudorotaxane motif depends, in part, on the incomplete saturation of the coordination sphere of the chloride anion by the cation, which leaves an empty meridian for complexation by the macrocyclic-amide anion recognition site (15). Although the initial goal of this research was to demonstrate both anion templation and selectivity in pseudorotaxane formation, the results suggested a wider scope for the template role of anions in the formation of interpenetrated and interlocked systems.† Here we report pseudorotaxane formation with two different ion-pair threads and propose a general template procedure based solely on anion recognition.

Figure 1.

The pseudorotaxane equilibrium between ion-pair 1a and macrocycle 2 (Hx = n-hexyl).

Materials and Methods

The syntheses of 1, 2, 3 (14), and 5b (16) have been reported. Compound 5a was obtained by extensive washing of a dichloromethane solution of 5b with aqueous NH₄Cl (1 M). 4⋅Br was obtained by alkylation of n-hexyl nicotinamide with 1- bromohexane in refluxing chloroform over 3 days. The syntheses of 4a and 4b were accomplished by anion exchange methods reported previously (14). ¹H NMR spectra were recorded on Varian 300-MHz or 500-MHz spectrometers. All compounds studied were characterized by ¹H and ¹³C NMR, elemental analysis, and MS. All ¹H NMR titrations were performed in acetone-d₆ dried over molecular sieves at a concentration of 2.5 × 10⁻³ M for 2 or 3 and monitoring the movement of the appropriate protons on addition of tetrabutylammonium (TBA) chloride, 1, 4, and 5 (Fig. 2). In the case of 3, data were obtained by observation of both the amide and mutually ortho aryl protons, which gave the same K_a (association constant) values within error. In the case of 2, data were obtained by observation of the amide, mutually ortho aryl, and hydroquinone protons, which gave the same K_a values within error. Association constants were determined from the data by using the program eqnmr (17) and fit 1:1 binding models. Estimated errors in all cases were <10%.

Figure 2.

Structures of 1b, 3, 4, and 5.

Results and Discussion

The design of macrocycle 2 was originally conceived as a wheel component for the complexation of thread 1a. Thus, the hydroquinone and polyether elements of the structure exist for the express purpose of binding the cationic pyridinium portion of the ion-pair thread. The question arose, however, as to whether second-sphere coordination of the anion was necessary for pseudorotaxane formation to occur in the presence of ion-pairing effects.

In acetone solution, ion-pairing between the pyridinium cation and chloride anion in 1a is very strong, and this is observed in the large K_a (>10⁵ M⁻¹) and in NMR shifts of protons a and b on titration of 1b with TBA chloride. Complexation of the chloride anion occurs in the hydrogen-bond donor cleft formed by the amide functions and their mutually ortho heteroaryl proton. In a similar manner, cationic threads 4 and 5 strongly bind a chloride anion in acetone as well. In the case of pyridinium nicotinamide 4⁺, an anion-binding cleft is formed with the protons at the 2 position on the pyridinium ring and the amide, whereas with imidazolium 5⁺ a strong hydrogen bond is formed with the proton at the 2 position of the imidazolium ring. Again these conclusions are reached as a result of the markedly larger NMR shifts of these protons and large association constants (K_a > 10⁵) on titration of both 4b and 5b with TBA chloride. All three of these ion-pair threads are designed to leave a meridian of the coordination sphere of the chloride anion partially unsaturated for complexation by hydrogen-bond-donating ligands 2 or 3 in a noncompetitive solvent such as acetone.

Acyclic compound 3 was chosen as a control, which would allow us to assess the contribution that anion recognition provides in the formation of the pseudorotaxanes. To determine any possible difference in the chloride-binding ability of 2 and 3 with minimal interference from the cation, both compounds were titrated with TBA chloride in acetone (Table 1). The macrocycle 2 complexes chloride anions marginally more strongly than the acyclic control compound 3 (ΔΔG = −2.0 kJ⋅mol⁻¹). This difference may be attributed to a greater preference of the amide groups in 2 for the syn–syn conformation (18), which is involved in anion binding, and preorganized by macrocyclization. Regardless, the difference in the ability of the two receptors to bind chloride anions is small.

Table 1.

Association constants [K_a (M⁻¹)] and free energies of complexation (−ΔG in kJ⋅mol⁻¹) in acetone-d₆ at 298 K of TBA Cl, 1, 4, and 5 with macrocycle 2 and acyclic ligand 3

Ligand	TBA Cl		1a		1b		4a		4b		5a		5b
	Ka	−ΔG	Ka	−ΔG	Ka	−ΔG	Ka	−ΔG	Ka	−ΔG	Ka	−ΔG	Ka	−ΔG
2	6,600	21.8	2,400	19.3	35	8.8	2,100	19.0	NA	NA	3,300	20.1	NA	NA
3	2,900	19.8	100	1.4	NA	NA	1,100	17.4	NA	NA	1,900	18.7	NA	NA

NA, no interaction observed. 

The results of the titration of either 2 or 3 with compounds 1, 4, and 5 as their chloride and PF₆ salts are summarized in Table 1. Throughout the titrations of 2 with 1a, 4a, and 5a we observed notable shifts in the hydroquinone proton resonances, which are attributed to π–π interactions between the cation and the hydroquinone rings (see, for example, Fig. 3). These shifts are indicative of threading of the cations through the cavity of the macrocycle. If these changes were caused by a noninterpenetrative geometry, it would be expected that they would be observed in the case of the PF₆ salts, which they are not. Only 1b shows any evidence of interpenetration without templation by a chloride anion, and the effect is weak at best. Neither of the other two PF₆ salts tested showed any interaction with either 2 or 3 up to the addition of 10 equivalents.

Figure 3.

¹H NMR spectra of 4a, 2, and a 1:1 mixture of the two demonstrating the shifts observed on pseudorotaxane formation.

Acyclic receptor 3 complexes 1a an order of magnitude less well than it does 4a and 5a. This difference is presumably a consequence of this particular cation's greater complementarity for, and therefore better encapsulation of, the chloride anion. Fortunately, weaker complexation in this case is compensated for by the additional interactions of the cationic thread with the macrocyclic cavity on pseudorotaxane formation with 2. This is not the case with pseudorotaxane formation involving either 4a or 5a and 2. In both of these cases, threading of the cation appears to be driven almost entirely by recognition of the anion, considering the small differences in the free energy of complexation when compared with the acyclic receptor 3 (ΔΔG ≤ −1.6 kJ⋅mol⁻¹). The fact that the cationic threads are all strongly ion-paired to the chloride counterion serves to entrain them through the cavity during the recognition process. Thus, recognition of the same chloride anion in all three cases provides the impetus for pseudorotaxane formation, despite the varied stereoelectronic nature of the different threads and the total lack of affinity 4⁺ and 5⁺ display for the macrocyclic cavity without this anion-templated driving force.

These results lead us to propose general template procedure for the fabrication of interpenetrated structures based on the coupling of anion recognition with ion-pairing pictured in Fig. 4. Such a strategy should operate effectively in noncompetitive solvents provided the cationic thread does not saturate the coordination sphere of the anion. In addition, the subsequent anion recognition step must orient the macrocyclic ligand orthogonally to the cation to provide an interpenetrated geometry. It is a simple matter to visualize this template in conjunction with other anions, cationic threads, and macrocyclic anion receptors, which is an ongoing topic of research in our laboratories. We anticipate that conversion of these types of pseudorotaxanes to their related rotaxane and catenane derivatives will provide unique mechanically bonded assemblies, which incorporate various anion specific recognition domains, as a consequence of the anionic templates used to construct them. Such supramolecular architectures should be of interest not only in their anion binding and recognition abilities, but also in the anion-controlled intramolecular movement of their interlocked parts.

Figure 4.

Template procedure for pseudorotaxane formation relying on ion-pairing and anion recognition.

Abbreviation

TBA

tetrabutylammonium