Resorcinarenes are hexameric capsules in solution

Abstract

The host–guest complexes of resorcin[4]arenes with small molecules in organic solutions are examined using modern NMR spectroscopic methods. The complexation of glutaric acid and β-methyl d-glucopyranoside in chloroform were investigated through 2D COSY, 2D NOESY, 1D NOE, and diffusion-ordered NMR spectroscopy (DOSY) techniques. These methods indicate that the complex is a self-assembled capsule composed of six resorcinarenes that surround six guest molecules of glutaric acid or three molecules of β-methyl d-glucopyranoside inside. The multiplicity of guest proton signals shows that the capsule provides an asymmetric magnetic environment that persists on the ¹H NMR time scale. The encapsulation of these guests and common solvents suggests that the phenomenon of reversible encapsulation in chemistry may be a century old.

The easy synthesis of resorcin[4]arenes (1, Fig. 1) developed by Högberg more than 25 years ago fostered their widespread use in studies of self-assembly and supramolecular chemistry (ref. 1; for an early review of resorcinarenes, see ref. 2). They provided the modules from which open-ended container molecules, the cavitands, were elaborated and led to the first covalent structures that completely surrounded other molecules, the carcerands (3). These shallow, bowl-shaped structures have intrinsic properties as receptors for molecular recognition as introduced by Aoyama and coworkers (4, 5). They described complexes of 1 in organic solvents with a number of guests, small diacids, alcohols, and even steroids, featuring a 1:1 stoichiometry. Intermolecular hydrogen bonding was proposed to hold the guest in the concavity of the resorcinarene host, as shown for glutaric acid in Fig. 2. The NMR spectra confirmed that the signals of the guest were upfield-shifted, as expected for their positions above the four aromatic units of the resorcinarene. However, the same spectra showed an unexpected feature: guest exchange took place slowly on the NMR time scale, with separate signals for free and bound guests. The broad range of molecules recognized and the slow on/off dynamics appeared at odds with the simple structures proposed for the complexes, and raised some questions as to the extent of the intermolecular forces involved. We have now examined the glutaric acid, β-methyl d-glucopyranoside, and related complexes of 1 in solution through modern NMR techniques and report on them here. We find that the unprecedented behavior can be reconciled through encapsulation complexes involving the self-assembly of six resorcinarenes in a capsular host. The host surrounds six glutaric acid or three β-methyl d-glucopyranoside guests.

Fig. 1.

Resorcin[4]arenes assemble to give hexameric capsules. Peripheral alkyl groups (R = C₁₁H₂₃) have been removed for viewing clarity.

Fig. 2.

The 1:1 host–guest complex of glutaric acid with a resorcin[4]arene proposed by Aoyama (21); the oval cartoon represents the macrocycle 1.

A blueprint to the solution structure came from the crystallographic studies of Atwood and MacGillivray (6). They found that, unlike the many previous solid-state structures of resorcinarenes and guests that involve layered arrays of the molecules (7–10), resorcin[4]arene 1′ crystallized from hot nitrobenzene as a rather large closed shell, a hexameric capsule surrounding a space of nearly 1,400 ų (6). The six resorcin[4]arene molecules appear at the sides of a notional cube, and one molecule of water is at each of the eight corners (Fig. 1). An unknown number of highly disordered solvent molecules are inside the cavity. Since that structure appeared, evidence has grown that the resorcinarene self-assembles as a capsule in solution as well (11–15). In scrupulously dry solvents such as benzene and chloroform, the molecules are rather insoluble or show slight solubility and broadened, featureless NMR spectra characteristic of multiple aggregated states. However, when small amounts of water are added, well defined host structures emerge; they show resolved signals in the NMR spectra that can be characterized by the methods described below.

Results and Discussion

Hexamer Assembly in Organic Solvents.

Resorcinarene 1 in wet, nondeuterated solvents encapsulates six and eight molecules of chloroform and benzene, respectively, as shown by ¹H NMR (12–15). The encapsulated solvent molecules show upfield-shifted resonances in their NMR spectra, separated from those of the solvents in bulk solution, indicating slow in–out exchange on the NMR time scale (up to 600 MHz). The diffusion NMR spectra in wet organic solvents show diffusion coefficients of 1 that are consistent with a hexameric assembly (12, 13). The solvents can be the only occupants of the capsule 1₆, or they can share it with additional guests (16–19), when the interaction between coguests is favorable. The recent solid-state studies by Atwood and colleagues (20) indicate that strong host–guest interactions relative to the interactions between coencapsulated guests are also possible.

The requirement of structural water molecules and the potential for these waters to occupy part of the cavity complicates studies with other solvents. Typical solvents that compete for hydrogen bonds (e.g., methanol or DMSO) disrupt the assemblies and reduce them to their constituent monomers (12, 14). Many other polar solvents also interfere with capsule formation. For example, no assembly is observed in the presence of diglyme (2-methoxyethylether) or other oligo(ethylene oxide)s.

Encapsulation of Polar Small Molecules.

Aoyama and coworkers (21) found that the odd-numbered carbon diacids, especially glutaric acid (2), complexed well with 1 in a 1:1 stoichiometric ratio shown by ¹H NMR (Fig. 2). Compounds lacking one or both acid groups (3 and 4) or with a longer spacer (5) showed only poor complexation (Fig. 3). Their IR measurements further indicated that the carboxylic acid was participating in hydrogen bonding. Some of the NMR methods described below were unavailable at the time of the original investigations, but these now show that the glutaric acid complex is also an assembly: a hexameric capsule surrounding six molecules of 2. Molecular modeling suggests that six molecules of glutaric acid can fit within the hexamer with a packing coefficient (PC) of 0.46, a value also seen with other capsules. The guest selectivity appears to arise from molecular packing considerations, rather than the geometry of intermolecular interactions between 1 and the guest.

Fig. 3.

Representative small molecules encapsulated within the cavity of 1₆ and related compounds.

Fig. 4 shows the NMR spectra of 1 with and without added glutaric acid (2), in which the upfield-shifted signals of the encapsulated guest are clearly seen. The 2D COSY, 2D NOESY, and 1D NOE spectra all indicate the following proton chemical shift assignments of the encapsulated glutaric acid: H1/H3 at 0.7 ppm, H1′/H3′ at 0.1 ppm, and H2/H2′ at −1.3 ppm; this requires that the symmetry of the guest glutaric acid is broken: enanatiotopic pairs of geminal hydrogens become diastereotopic upon encapsulation. This observation would be consistent with the formation of a chiral capsule, as is observed in the crystalline state (6). The ¹H NMR of methylated glutaric acid derivatives gave even more complicated spectra (see Fig. 10, which is published as supporting information on the PNAS web site).

Fig. 4.

¹H NMR spectra of 1 (30 mM) with glutaric acid (2) (90 mM) (a) and without in CDCl₃ (b). Peaks for encapsulated 2 can be seen far upfield. This concentrated sample was measured three months after preparation and according to this spectrum the stoichiometry of the formed complex, based on integration, is ≈1:1, that is, a hexamer with five to six encapsulated glutaric acid molecules.

The extent of encapsulation appears to depend on both concentration of glutaric acid (Fig. 5) and time before acquiring the spectra (Fig. 6). Furthermore, Fig. 6 also shows that the encapsulation of 2 occurs at the expense of the encapsulated chloroform molecules, as indicated by the disappearance of the two smaller peaks in the range of 4.9–5.1 ppm attributed to encapsulated chloroform molecules. However, these results do not necessarily indicate coencapsulation of chloroform and glutaric acid. Instead, the molecules of CHCl₃ and 2 may segregate into separate cavities. Of the aryl resorcinarene resonances, the peak at 6.1 ppm appears to be quite sensitive to the encapsulation process as shown in Fig. 5.

Fig. 5.

Sections of the ¹H NMR spectra (400 MHz, 298K) of 1 (30 mM) in CDCl₃ titrated with varying amounts of 2 and collected 1 h after brief heating of the sample. The concentration of 2 was 5 mM (a), 10 mM (b), 15 mM (c), 20 mM (d), and 25 mM (e).

Fig. 6.

Sections of the ¹H NMR spectra (400 MHz, 298K) of 1 (30 mM) in CHCl₃ titrated with varying amounts of 2 and collected at different time points after sample preparation. The sections of the spectra are of 1 (a); 1 and 15 mM of 2, 1 h (b); 1 and 25 mM of 2, 1 h (c); 1 and 25 mM of 2, 4 days (d); 1 and 30 mM of 2, 4 days (e); 1 and 60 mM of 2, 4 days (f); and 1 and 25 mM of 2, 3 months (g).

The pulsed field-gradient methods were developed in the 1960s and 1970s as a means to determine diffusion coefficients (D) by NMR spectroscopy (22, 23). Recently, the technique has found increasing application in both biological and supramolecular chemistry (for an early review, see ref. 24 and references therein; for a recent review on diffusion NMR in supramolecular chemistry, see ref. 25 and references therein). For the mixture of 1 and 2, the similarity of the signal decay of the peaks for resorcinarene and glutaric acid indicates that these species have the same diffusion coefficients (≈0.24 × 10⁻⁵ cm²·s⁻¹) within experimental errors (Table 1 and Fig. 7). These values are consistent with those reported for similar hexameric capsules with solvent or cationic guests inside (12–14). As anticipated, glutaric acid in bulk solvent diffuses very rapidly under these conditions (≈1.1 × 10⁻⁵ cm²·s⁻¹). The results clearly establish the proposed structures as discrete, 6:6 encapsulation complexes that diffuse as a single supramolecular entity.

Table 1.

Diffusion coefficients (D, × 10⁻⁵ [cm²·s⁻¹]) for 1 (30 mM) in the presence of 90 mM glutaric acid (2) in CDCl₃ measured by diffusion NMR (400 MHz, 298 K), using two different diffusion times

	ppm	Diffusion coefficient (D) (× 10−5 [cm2·s−1])*
		Short diffusion time†	Long diffusion time‡
Chloroform	7.26	2.176 ± 0.002	2.182 ± 0.033
Octol 1 (hexamer)	0.89	0.249 ± 0.006	0.241 ± 0.001
Octol 1 (hexamer)	4.31	0.242 ± 0.005	0.245 ± 0.001
Encapsulated glutaric acid 2	−1.33	0.235 ± 0.001	0.241 ± 0.001
Encapsulated glutaric acid 2	0.12	0.235 ± 0.002	0.244 ± 0.002
Unencapsulated glutaric acid 2	2.46	1.075 ± 0.032	1.054 ± 0.007

*All values are averages of three experiments.

^†Data acquired with relatively short diffusion time when Δ was set to 30 ms.

^‡Data acquired with relatively long diffusion time when Δ was set to 60 ms.

Fig. 7.

Diffusion data of the mixture of 1 (30 mM) and 2 (90 mM) in CDCI₃. (a) Stack-plots (400 MHz, 298 K, CDCl₃) of the signal decay as a function of the gradient strength (G) of two peaks representing the hexamer of 1 and two peaks assigned to encapsulated 2. (b) Natural log of the normalized signal decay (ln(I/I₀)) as a function of b value of representative peak of 1 (blue stars), encapsulated 2 (pink squares), unencapsulated 2 (green triangles). b value = γ²δ²G²(2/π)²(Δ − δ/4). For parameter definitions see Materials and Methods.

Aoyama’s research group (26–28) also found that many simple sugars could be extracted from aqueous solution into a CCl₄ solution containing 1. The binding showed discrimination, with selectivity for certain sugar stereoisomers. Although natural d-glucose is a relatively poor guest under these conditions, the β-anomer of methyl d-glucopyranoside (6, V = 136 ų) was selectively extracted into CCl₄ to give a different sort of host–guest complex with a 2:1 stoichiometric ratio (29).

We repeated these experiments in chloroform (CHCl₃ and CDCl₃) and found good encapsulation of 6 with a sharp singlet at −0.04 ppm and a particularly complex set of resonances for the methine proton of 1 (Fig. 8). The complexity is likely due to the formation of diastereomeric assemblies, but the stoichiometry observed can result from either a dimeric capsule with a single sugar inside (as proposed by Aoyama and colleagues, ref. 29) or from a hexameric capsule containing three sugar molecules.

Fig. 8.

Sections of the ¹H NMR spectra (400 MHz, 298 K) of the following CDCl₃ solutions: 12 mM 1 (a), 1:7 in a 1:1 ratio (b), 1:6 in a 2:1 ratio (c), 1:6 in a 1:1 ratio (d), and 1:6 in a 1:2 ratio (e).

As with the capsule containing glutaric acid, the observed spectra depend on both sugar concentration, time before acquiring the spectra, and heating. Mixing 1 and 6 in a 1:2 ratio in CHCl₃ and heating for several minutes gives 1/encapsulated-6/encapsulated-CHCl₃ in a 6:2:2 ratio (see Fig. 11 and Table 2, which are published as supporting information on the PNAS web site). The peaks for all three species showed similar diffusion coefficients. Further heating of the sample results in complete displacement of CHCl₃ by the sugar to give exclusively three encapsulated sugars per hexamer. There are no detectable NOEs between the encapsulated CHCl₃ and encapsulated sugar molecules, even for long mixing times (up to 1,000 ms). This finding does not entirely rule out coencapsulation, but suggests that the capsules contain either CHCl₃ or 6 exclusively.

Diffusion NMR shows that all of the peaks assigned to 1 and encapsulated 6 have the same diffusion coefficients: ≈0.24–0.25 × 10⁵ cm²·s⁻¹, consistent with the diffusion coefficient of a capsule containing tetrahexylammonium bromide (THABr), which is known to be well encapsulated by 1₆ (Fig. 9) (12, 30). To further verify this point, a CDCl₃ solution of 1:6:THABr in a ≈12:3:1 ratio, where the concentration of 1 was 24 mM, was subjected to diffusion NMR. Indeed, all species were found to have the same diffusion coefficient (see Table 3, which is published as supporting information on the PNAS web site). In addition, the diffusion coefficient of the hexameric capsule of 1 with 6 was measured in the presence of biscalix[5]arene, having a molecular weight of 2,398 g·mol⁻¹ (a molecular weight slightly larger that that of a resorcinarene dimer) (31). It was found that the biscalix[5]arene has a significantly higher diffusion coefficient than the hexameric capsule as shown in Fig. 9 (see also data in Fig. 12 and in Table 4, which are published as supporting information on the PNAS web site). We conclude that, with an excess of 6, three molecules of the sugar are encapsulated inside the hexamer with no additional chloroforms. The capsule displays high selectivity for 6 over other monosaccharides. For example, both ¹H NMR and diffusion measurements showed no encapsulation of methyl α-glucopyranoside (7) under these conditions (Fig. 8b). ¹H NMR similarly showed no encapsulation of methyl galactopyranoside, methyl ribofuranoside, and myo- and scyllo-inositol.

Fig. 9.

Natural log of the normalized signal decay (ln(I/I₀)) as a function of b value (400 MHz, 298 K, CDCl₃) of representative peak of 1 (blue stars), encapsulated 6 (filled squares), encapsulated tetrahexylammonium bromide (green triangles), and the biscalix[5]arene (red circles) (31).

Conclusions

The identification of glutaric acid in 1₆ as a hexameric capsule pushes back the history, if not the discovery, of reversible encapsulation as a form of molecular recognition to 1989. Some years before, in 1980, the Högberg synthesis produced 1 in the presence of wet organic solvents so encapsulation was surely there, but this synthesis was adapted from one that is more than a century old (32, 33). Arguably, reversible encapsulation (at least of solvents) is deeply rooted in the past. Multiple copies of small polar molecules are similarly encapsulated with or without additional molecules of solvent. Six copies of glutaric acid or three copies of β-methyl d-glucopyranoside fit inside without any additional solvent. The present cases also demonstrate the utility of diffusion NMR in identifying aggregates having the same symmetry as their monomer. However, questions about the role of water and the dynamics of guest motions inside the capsule remain for future investigations.

Materials and Methods

Materials.

Deuterated solvents (no tetramethylsilane, TMS) were purchased from Cambridge Isotope Laboratories (Andover, MA). All other reagents were purchased from Sigma-Aldrich (St. Louis, MO) or Acros Organics and used without further purification. Alkyl-footed resorcin[4]arene 1 was prepared according to published procedures (7). CHCl₃ was purchased with amylene stabilizer; however, no amylene encapsulation was observed by ¹H NMR.

NMR Experiments.

NMR spectra were acquired on a 400-MHz Brüker spectrometer, except as noted. All diffusion NMR experiments were performed with the LED sequence with sine-pulsed-field shaped pulsed-field gradient (34) and were carried out in triplicate. Only data where the correlation coefficients of ln(I/I₀) versus γ²δ²G²(2/π)²(Δ-δ/4) (in which γ is the gyromagnetic ratio, G is the pulsed gradient strength, and Δ and δ are the time separation between the pulsed gradients and their duration, respectively), generally termed the “diffusion weighting” and denoted as the b values, were >0.999 are reported. In some cases, diffusion experiments were preformed with two different diffusing times by changing Δ from 30 to 60 ms. In a few cases, samples were aged for 3 months before data acquisition.