Just Add Water: Modulating the Structure-Derived Acidity of Catalytic Hexameric Resorcinarene Capsules.

The hexameric undecyl-resorcin[4]arene capsule (C11R6) features eight discrete structural water molecules located at the vertices of its cubic suprastructure. Combining NMR spectroscopy with classical molecular dynamics (MD), we identified and characterized two distinct species of this capsule, C11R6-A and C11R6-B, respectively featuring 8 and 15 water molecules incorporated into their respective hydrogen-bonded networks. Furthermore, we found that the ratio of the C11R6-A and C11R6-B found in solution can be modulated by controlling the water content of the sample. The importance of this supramolecular modulation in C11R6 capsules is highlighted by its ability to perform acid-catalyzed transformations, which is an emergent property arising from the hydrogen bonding within the suprastructure. We show that the conversion of C11R6-A to C11R6-B enhances the catalytic rate of a model Diels-Alder cyclization by 10-fold, demonstrating the cofactor-derived control of a supramolecular catalytic process that emulates natural enzymatic systems.

Introduction

Supramolecular catalysis derives inspiration from enzymes, translating natural features into synthetic systems to attain higher levels of control in chemical processes. Approaches toward bioinspired supramolecular catalysis include the biomimicry,[1−4] second coordination sphere design,[5−7] and confinement of the catalytic site.[8−14]

Along these lines, the positioning of catalytic active sites within well-defined capsules has been demonstrated to enable the control of catalyst properties to promote selective catalytic transformations.[6,7] In natural systems, enzymatic activity that enables the self-steering of catalytic processes necessary for metabolism can be modulated via allosteric modifications by physiochemical inputs. Although it is an intrinsic feature of natural systems, analogous modulation of catalyst properties in synthetic mimics are rare.[15−18]

It is now more than 30 years ago that the Aoyama group described the host–guest chemistry of resorcin[4]arenes in nonpolar organic solvents.[19−22] As further characterization developed, the hexameric nature of these capsules was realized and its capacity for host–guest interactions were extensively characterized.[23−39] Analogous to an enzyme, R exhibits catalytic function from the elevated Brønsted acidity emerging from its supramolecular structure.[40] Illustrated in Figure , this capsule is formed in nonpolar solvents (e.g., chloroform) through the self-assembly of six facial monomers in a cubic arrangement, featuring eight water molecules (one per vertex).[23] The edges of R are held together by hydrogen-bond network edges between adjacent facial monomers, with each end point capped at the vertex with a water molecule, completing the cubic structure.[23]

Figure 1

Ball-and-stick rendering of a model R showing a cubic structure with 6 resorcin[4]arenes (CPK rendering) forming the faces of the cube (yellow) held together by an hydrogen-bond network continuous along each edge (black line), and capped by eight water molecules at the vertex positions (van der Waals volume renderings). To improve clarity, pendant alkyl groups and nonhydroxy hydrogen atoms were omitted from this figure.

The hydrogen-bond network of R results in the enhanced Brønsted acidity beyond that of the individual monomer units.[40] This feature has driven the application of R as a supramolecular, organic Brønsted acid catalyst for chemical transformations under mild conditions.[41−44] In addition, the hydrogen-bond rich environment of the internal cavity within R has been utilized as a supramolecular organocatalyst,[45−52] demonstrating a host-selective reactivity based on substrate size, and substrate–bond activation via supramolecular interactions. The use of a supplemental protic acid cocatalyst (typically HCl) extends the scope of R activity,[53−64] notably for application toward facile synthesis of high-value terpene derivatives.[60−64] Further reactivity has been demonstrated in host-catalyzed Diels–Alder cyclization.[65]

Beyond the intrinsic Brønsted acidity of R, this supramolecule possesses an internal cavity (ca. 1400 Å3),[23] permitting the encapsulation of transition metal catalysts[66−69] or organic catalysts[70−72] within its cavity. In these instances, the internal surface of the capsule serves as a second coordination-sphere to modulate or enhance catalytic function.[5−7]

Both the acidity and host-capacity of R are derived from its structure.[23,40] Recent work by Payne and Oliver have demonstrated structural modification of R by the incorporation of alcoholic solvent molecules into the hydrogen bond network,[73] complementing previous studies by Cohen[74−76] and Schnatwinkel,[77] which featured similar inclusion of long chain alcohols into the hydrogen-bond network. Interestingly, Katiyar has reported the association of free water to the capsule’s hydrogen-bond network,[78,79] beyond the 8 molecules needed for capsule assembly.[23,31,32] Studies by Merget suggest that the presence of additional water may impact the catalytic activity of the R capsule in acid-promoted cyclization of terpenes.[61] Together these findings suggest that polar molecules such as water may act as cofactors able to modulate the structure and acidity of R, analogous to the allosteric control of enzymes (e.g., cytochrome p450 oxidases, nitric oxide synthases, etc.). Understanding that these structural changes would provide insights into the previously observed water-dependent catalytic behavior and fine control the capsule’s catalytic activity.

In this work, we investigate the structural changes of R capsules through classical molecular dynamics (MD) simulations, which is further supported by 1H NMR spectroscopy. Using MD, we find that R interconverts between two assemblies as summarized by Figure . The R-A assembly features 8 water molecules at the vertex positions—in line with previous reports of R structure[23]—while R-B has 14–15 water molecules, 6–7 of which spontaneously incorporate into a single edge of the cubic suprastructures, referred to here as “incorporated water” (Scheme ). This computational finding is supported by NMR studies of water association, revealing a water-dependent equilibrium between the two capsule species differing significantly in their hydrogen-bond network. Differences between the R-A and R-B assemblies are substantiated by 31P NMR chemical shifts of an encapsulated phosphine oxide, revealing different internal acidities quantified by their Guttman–Beckett acceptor number (AN). This difference in internal acidity allows the rate modulation of R catalyzed Diels–Alder cycloaddition of maleimide and sorbic alcohol, demonstrating novel control of an abiotic homogeneous catalytic process.

Figure 2

(a) Plot of the relative Helmholtz free energies (ΔA), internal energy (ΔU) of the water–water or water–R interactions, and entropy (ΔS) for the incorporation of water molecules beyond 8 determined by the GIST method.[80] (b) Renderings depicting the structures of R-A and R-B containing 8 and 14 water molecules, respectively. Renderings feature highlights indicating structural (red) and incorporated (blue) water, which differentiate R6-A and R-B, respectively. Note that structural water highlights for R-B are omitted for clarity. Similarly, alkyl pendant groups, solvent, and nonhydroxy hydrogen atoms are omitted from both renderings for clarity. Both thermodynamic calculations (a) and model visualizations (b) were generated from 28 800 ns MD trajectories.

Scheme 1

Simplified Representation of the Water-Dependent Conversion between the Two Forms of R, Highlighting the 8 Water Molecules Necessary for Capsule Formation (Red) and the 7 Additional Water Molecules (Cyan) That Effect the Transition by Association to a Capsule Edge (Black Line)

Results and Discussion

MD Simulations Reveal Distinct Species

Simulations containing explicitly solvated R with a total of 8–24 explicit water molecules were propagated as molecular dynamics trajectories for a total of 10 μs using optimized force field parameters (Figure S1). Unfortunately, simulations featuring randomly placed water molecules and undecyl-resorcin[4]arene monomers (R) failed to self-assemble over several μs of MD propagation (results not reported). Therefore, we found it necessary to include the 8 structural water molecules, placed at the vertex positions of the capsule, while the remaining water molecules were positioned randomly in the periphery of the capsule.

In simulations containing 8–12 water molecules, we observe the external attachment of free water to the R in line with previous reports.[78,79] Simulations containing ≥14 water molecules reveal 6 additional incorporated water molecules along a single edge of the hydrogen-bond network of the R capsule (Scheme ), as depicted in Figure b. Although these incorporated water molecules are highly organized and an integral part of the hydrogen bond network (Figure S16), single water molecules still exchange rapidly with water molecules from the bulk solvent and the 8 structural waters needed to form the capsule. The mobility of the incorporated water is highlighted by the concerted migration between the hydrogen bond edges of the capsule. This migration phenomenon was qualitatively observed as a rare event in our MD simulations (Figure S15), but occur at a sub-microsecond time scale.

The incorporation of additional water into the edge of the hydrogen bond network results in a breakage of the hydrogen bond between adjacent R faces, altering the connectivity of the supramolecular system. This change in connectivity and composition distinguishes R-B from the typical R-A assembly. Analysis of hydrogen-bonding in our MD trajectories (Figure S2) reveal a minimum of 6 extra incorporated water molecules are required to form R-B.

Energetic analysis of the MD data using GIST (Figure a) distinguishes between both attached water[78,79] and the incorporated water we observe in R-B. While GIST does not provide complete free energy differences between R-A and R-B, it is useful in the analysis of favorable water structures found in our MD simulations. In simulations containing 8–12 water molecules the attached water is observed. Interestingly, the GIST-determined ΔA is similar to previously reported values (ca. −2.0 kcal mol–1),[78,79] and from our analysis this is driven entirely by a favorable water–water interaction (Figure a, ΔUwater–water). The inclusion of water along the hydrogen bond edge is optimal in the presence of 14 water molecules, where an additional favorable water-capsule interaction (Figure a, ΔUwater-), resulting in a very favorable association (ΔA = −6.3 kcal mol–1). While the incorporation of further water molecules within the suprastructure is possible, it incurs an increasing penalty from internal energy (Figure a, ΔU) and system entropy (Figure a, −TΔS). The specificity of R-B to incorporate 6 water molecules is a “goldilocks” number, originating from the required size of the hydrogen-bond network needed to fill a capsule edge (Figure b), resulting in favorable internal energy (Figure a). These “incorporated water” molecules are more mobile than their “structural water” counterparts, and are not as strongly localized. These simulations suggest that R is found in only two forms—R-A containing 8 water molecules and R-B containing 14 water molecules—and the ratio between the two may depend on water content.

1H NMR Identification of R-A and R-B

The formation of R-A and R-B was investigated by 1H NMR, by measuring spectra of R solution at various concentrations of water (44.12–103.01 mM; for details see Supporting Information). Contrasting previous reports, which broadly attribute all phenolic peaks (δ = 8.5–10.0 ppm) to a singular species of R,[13−16] our spectra, shown in Figure a, reveal a changing pattern in the phenolic peaks, concomitant with the changing water content. The separation of these phenolic peaks indicates slowly exchanging environments,[81] inconsistent with the 5 ns lifetimes of previously described water dynamics.[79] As these peaks increase (or decrease) in a correlated fashion, we attribute these spectral features to distinct assemblies: R-A (δ = 9.58 and 9.35 ppm) and R-B (δ = 9.65 and 9.46 ppm). This peak assignment is further supported by inversion-relaxation measurements (Figure S21), from which identical T1 relaxation times were obtained for the phenolic peaks of either capsule indicative of a shared environment. The increased sensitivity of T1 relaxation times of the peaks belonging R-B to changing water content is in line with the larger number of water molecules associate to its structure.

Figure 3

Quantitative 1H NMR spectra (a) and DOSY diffusograms (b) of R (5.38 mM) observed over a range of water concentrations (44.12–103.01 mM) corresponding to 8–19 water molecules per R capsule. Water peaks and diffusion traces are highlighted in blue or accompanied by a blue circle. Peaks and diffusion traces attributed to R-A (red highlight and diamond) and R-B (green highlight and triangle) assemblies are annotated. Overlapping diffusion traces are shown in purple. Quantitation of water concentration was determined by integration of the water peak (blue circle) compared to a peak corresponding to both R assemblies (δ = 6.2 ppm, 24 H, 5.38 mM). Complete diffusograms are provided in the Supporting Information (Figure S4). Long recycle delays (25 s) were necessary to obtain quantitative spectra for both water (T1 = 0.7–0.9 s, data not shown) and R (T1 = 1.39 s, data not shown).

Interestingly, the relative concentrations of these species vary with water content from 44.12 mM (ca. 8 water molecules per capsule) to 103.01 mM (ca. 19 water molecules per capsule). As these differences are only apparent in the phenolic region of the NMR spectrum, we surmise that these assemblies are distinguished by the structure of their respective hydrogen-bond networks. Therefore, we putatively assigned these peaks to R-A (δOH = 9.58, 9.35 ppm) and R-B (δOH = 9.65, 9.46 ppm) based on the increasing concentration of water and consistent with the structures observed in MD simulations (Figure ). The presence of incorporated water in R-B is further evidenced by stronger NOE correlations between its phenolic peaks and free water (Figure S18). Deuterium exchange of the OH-groups with D2O (Figure S23) is different for the two capsules, and evidence the discontinuous hydrogen bond network in line with our MD simulations (Figure S16).

Interestingly, only two peaks of equal area are observed for the phenolic protons of either assembly, despite the asymmetry derived by incorporated water molecules in R-B (Figure ). Our MD simulations show the specific arrangement of incorporated water shift between edges of the capsule on a sub-microsecond time scale (Figure S15). The environments of the phenolic protons of R-B, exchange at this rate, and as such are observed as a time-averaging signal. Exchange of water between R-B and R-A is relatively slow leading to distinct phenolic peaks that can be distinguished in the NMR spectra (Figure S14).[81] On the basis of the relative strength of NOE correlations between the phenolic peaks and water, we assign the upfield peaks of either assembly (δ = 9.35 and 9.46 ppm) to the 24 phenolic protons adjacent to the structural water sites (Figure ). Similarly the downfield peaks of either assembly (δ = 9.58 and 9.65 ppm), are assigned to the remaining 24 phenolic protons which participate in hydrogen bonding between and within the resorcin[4]arene monomer faces. Fortunately, the separation of the pairs (33 Hz) of the resolved R-A and R-B phenolic peaks constraints the rate constant for chemical exchange (kex) between the two assemblies to <155 s–1 (for a detailed discussion see, Figure S14).[81]

The apparent diffusion of these phenolic peaks appears faster than the other peaks (Figure b) due to proton exchange with water occurring within the diffusion time in the measurement (Δ = 100 ms). Fortunately, the pairing of these diffusion traces further supports the speciation of the two assemblies observed by the correlation of the peak areas (Figure a).

Further characterization of the capsule using 1H NMR (Figure S3), DOSY (Figures b and S4),[31−33,78] and solution state FTIR (Figure S5)[36] indicate that both assemblies are hexameric assemblies with a similar Stokes radius (16.6 Å) at [H2O] = 44 and 103 mM consistent with previous reports of R capsule structure.[23−39]

The single observed peak of water (Figure ) indicates that it is in a state of fast exchange between a free state in the bulk solution and a bound state, incorporated into the R capsule (Figure S14).[83] As previous reports detail, the available water is completely incorporated into the cage at low (i.e., 44 mM) water concentrations;[31−33,78] therefore, the measured chemical shift (δ = 5.1 ppm) can be attributed to the structural water (Figure ), as opposed to the free H2O water-saturated chloroform.[82] As the observed chemical shift is time-averaged,[81] the proportion and quantity of water associated with R (Bwat) was determined directly from 1H NMR spectra (Figure a).

Figure shows the total number of water molecules associated with R increases linearly with the proportion of R-B (θB) in the sample, with the slope showing an additional 7.27 ± 0.26 water molecules are incorporated per R-B formed. Thus, combined with the 8 structural waters native toR, a total of 15 water molecules are associated with R-B. From our MD simulations (Figure ) we surmise that these additional water molecules are incorporated into the hydrogen bonding network of the capsule. This number is in agreement with MD models (Figure ) that predict a minimum of 14 water molecules for the formation of R-B (Figure ). The water-dependent conversion between R-A and R-B was fit using an empirical model (Figure S13) to enable estimation of the proportion of R-B capsules (θB) via water content.

Figure 4

Plot of the total number of associated waters (Bwat) and proportion of R-B capsules (θB) determined from 1H NMR measurements (Figure a). The association of an additional 7.27 ± 0.26 water molecules concomitant to conversion is determined from the slope of the linear fit (red).

31P NMR Investigation of Structure-Dependent Acidity

Many catalytic applications of R rely on the intrinsic acidity derived from its supramolecular structure.[40] The 33 Hz downfield shift of the R-B phenolic protons (Figure a) suggest an increased acidity (compared to R-A),[83] a feature which is further supported by their apparent diffusivities observed by DOSY (Figure b).

Previously the Brønsted acidity of R assemblies were measured using nitrogen bases to estimate aqueous-equivalent pKa values.[40] Unfortunately, this protocol impairs the accurate determination of water content by either Karl–Fischer titration or 1H NMR integration, and could not be used to differentiate the acidity of R-A and R-B.

Therefore, we investigate the ability of structure-dependent acidity to modulate the interaction strength with tri-n-butyl phosphine oxide (Bu3PO) as guest through 31P NMR (Figure ).[84,85] The encapsulation of Bu3PO was readily confirmed by 1H NMR, showing the development of broad upfield peaks (δ = −2.0–0.5 ppm), typically observed for encapsulated guests.[24−39] The binding of Bu3PO within the capsule was further evidenced by 1H DOSY measurements (Figure S12), with similar diffusion for the R host and upfield peaks (log D = −9.0, see Figure b).

Figure 5

Chemical shift difference between free and encapsulated Bu3PO observed by 31P NMR at two concentrations, 3.50 mM (black) and 24.00 mM (red) in the presence of R (5.38 mM). Spectra were obtained at water contents spanning 43.76–110.19 mM (3.50 mM Bu3PO) and 43.05–86.53 mM (24.00 mM Bu3PO), which were subsequently converted to the proportion of R-B (θB) by an empirical model (Figure S13). Inset, a 31P NMR spectrum showing peaks corresponding to encapsulated (▼, green) and free (■, blue) Bu3PO.

A downfield chemical shift in 31P NMR is expected when a Bu3PO forms a hydrogen-bond adduct with another species, such as when encapsulated within R and the degree of this shift is proportionate to the acidity of the hydrogen-bond donor.[84,85] Three peaks (31Pδ ≈ 55.0–65.0 ppm) were consistently observed in the 31P NMR spectra of the encapsulated Bu3PO (Figures S9 and S10). The upfield peak (31Pδ ≈ 55.0–64.0 ppm) was assigned to the free Bu3PO by observed correlations to the protons of the free species by 1H–31P HMBC (Figure S11). A low intensity peak (31Pδ ≈ 64.0–65.0 ppm) was observed in all spectra, with a low intensity that waned with increasing water content. This spectral feature is particularly evident at a minimal water concentration (44.18 mM water, Figure S8), where the majority of the Bu3PO (3.50 mM) was observed to be encapsulated. Unfortunately, two-dimensional techniques (e.g., HMBC) could not provide sufficient cross-peaks with which to identify the originating species by other means. With additional water this minor peak broadens and diverges compared to the major peaks, and we infer that exchange between this minor species and the observed major peak is unlikely based on the diverging chemical shift. On the basis of the low intensity of the 31P signal, we surmise that this spectral feature does not correspond to the free or encapsulated Bu3PO, and its identity is unlikely to interfere with measurements of the R capsule’s internal acidity. The remaining peak was attributed to the R-associated Bu3PO (31Pδ ≈ 60.0–64 ppm) based on its apparent intensity (Figures S9 and S10). All three peaks were observed to move in a concerted fashion with changes in water content, which we ascribe to changes in bulk dielectric of the solvent medium.[86]

The free and encapsulated Bu3PO afford distinct peaks in slow exchange (Figure , inset). Similar to observations made with 1H NMR (Figure ), differentiation between phosphine oxide encapsulated within R-A and R-B was not observed by 31P due to the similarities of the magnetic environments experienced by the phosphorus nuclei. Due to this similarity, the shift of the observable peak corresponds to the time weighted average of the Bu3PO encapsulated within R-A and R-B (see Figure S14a for an example of the exchange of indistinguishable nuclei).[81] Further complications arise as a phosphine oxide guest within R-A or R-B may exchange hydrogen bonding partners within the capsule at a time scale faster than NMR measurement,[81] resulting in a single observable peak with a shift that is the time weighted average of the hydrogen bonding states (see Figure S14b for a detailed example of the exchange of a rapid process). The result of these exchange processes is a single observable peak corresponding to Bu3PO encapsulated by R-A or R-B, in all states of hydrogen bonding (see Figure S14 for a detailed explanation).[81]

Despite these limits in observation, the strength of the interaction between R and Bu3PO can be correlated to the downfield chemical shift of the single observable peak (31Pδ = 64.0–60.0 ppm). The strength of the interaction between Bu3PO and R can be determined by modulating the Brønsted acidity through changing the content of the sample (i.e., varying the water content of the sample) as shown in Figure . Two sets of experiments were performed where the R-A/R-B ratio was modulated through controlling water content (44.18–110.19 mM and 43.05–86.53 mM, respectively) in the presence of either a low (3.50 mM) or high (24.00 mM) concentration of Bu3PO. While the high concentration is analogous to catalytic conditions, at lower concentrations the Bu3PO probe selectively associates to the stronger interacting (i.e., more acidic) assembly. From these contrasting measurements we determine that the environment of R-B is more acidic than R-A, which may enhance its catalytic activity. We rationalize the increased acidity of R-B by the increased availability of protons within the capsule from the weakly bound incorporated water molecules (Scheme ).

Similar to 4 of the structural water molecules of R-A,[40] the 7 incorporated water molecules found in R-B are capable hydrogen-bond donors, and may also act as acids stabilized by the edge hydrogen-bond network (Figure S16).

The Guttman–Becket acceptor number (AN) is a measure of Lewis acidity that quantifies the differences in acidity between the two capsules, and allows comparison of acid catalysts in solvent media.[85] On the basis of 31P NMR spectra obtained at a minimal water concentration ([H2O] = 44.18 mM, Figure S8), we have estimated the Lewis acidity of R-A (AN = 51), similar to B(OMe)3 (AN = 51).[85] By extrapolating the chemical shift difference observed with Bu3PO (3.5 mM, Figure ), we estimate the Lewis acidity of R-B assemblies (AN = 68 ± 1), similar to TiCl4 (AN = 70).[87]

Structural Modulation of the R-Catalyzed Diels–Alder Cycloaddition

We investigated the catalytic activity of the two R assemblies in the Diels–Alder cycloaddition of maleimide and sorbic alcohol to produce 4-(hydroxymethyl)-7-methyl-3a,4,7,7a-tetrahydro-1H-isoindole-1,3(2H)-dione (Figure , inset). The Diels–Alder reaction was explicitly chosen as a probe for the structure-dependent catalytic activity of R as it proceeds without the generation of water or acid as a byproduct. Specifically, catalysis was performed at different water concentrations ([H2O] = 8.76–25.95 mM) enabling modulation of the R-B proportion (θB = 0.12–0.44) within the mixture. The dependency of catalytic activity on the proportion of R-B was revealed, with the result depicted in Figure .

Figure 6

Initial reaction velocities, Vobs, for the Diels–Alder cycloaddition of sorbic alcohol (6.00 mM), and maleimide (6.00 mM), catalyzed by R (1.34 mM, 22 mol %) in CDCl3 under ambient conditions (25 °C). Reaction velocity was measured by the depletion of maleimide (δ = 6.72 ppm) referenced to a TMS standard by 1H NMR. Reaction water content spanned 11.77–34.87 mM (8.76–25.95 molecules per capsule) as determined by Karl–Fischer titration. The water content was used to estimate R-B conversion by an empirical model (eq S1). Error bars correspond to the standard error in the linear fitting of the initial reaction velocity (Figure S6). A linear fit (red) reveals the turnover frequency of R-A (0.24 ± 0.06 h–1) and R-B (2.16 ± 0.29 h–1).

The initial reaction rates reveal that increases in water content afforded a doubling of the observed reaction rate (0.65–1.15 h–1), an effect not observed in the absence of R (Figure S7). As the ratio of R-A and R-B could not be directly observed by NMR, they were computed from the measured water content in conjunction with our empirical model (eq S1). The observed reaction velocity increases linearly (θB = 0.1–0.3) with the formation of R-B until it plateaus (θB = 0.3–0.5), where another process becomes rate limiting. We propose that this rate limitation is due to the slow isomerization of sorbyl alcohol from its inactive s-trans isomer to the active s-cis isomer (Figure S17). From this limitation we surmise that R6 acts primarily as an acid-catalyst for the activation of maleimide. A linear fit of the reaction rate to the proportion of R-B (θB) between 0.1–0.3 decomposes the overall reaction rate to the activity of either R-A or R-B assemblies. From this linear fit we find the more acidic R-B (2.16 ± 0.29 h–1) is significantly more active than R-A (0.24 ± 0.06 h–1). As the computed rate of R-A catalyzed cycloadditions is close to the uncatalyzed reaction (0.21 ± 0.01 h–1, Figure S7) we surmise that R-B is the sole active catalytic species. This result highlights the similarities between biological and supramolecular catalytic systems, where subtle changes in the arrangement of (supra)molecular features yield significant changes in catalytic output under mild conditions.

Conclusion

On the basis of NMR spectroscopy and computational data we demonstrate that the self-assembled hexameric undecyl-resorcin[4]arene capsule R can be switched between two distinct species—R-A and R-B—respectively featuring 8 and 15 water molecules within their hydrogen-bond networks. The internal environments of the two assemblies were probed by the binding of Bu3PO, revealing substantial shifts in the 31P NMR peak of this guest through changing the R-A/R-B ratio by the addition of water to the sample. These NMR experiments suggest a stronger acidity of R-B assemblies that translate into differences in catalytic activity. The catalytic activity of these two assemblies were investigated in a Diels–Alder cycloaddition reaction, revealing that R-B exhibits greater catalytic output by an order of magnitude. This study demonstrates the ability of water to effect structural changes in R capsules by modulating the structure-derived catalytic properties of the supramolecular assembly. We envisage that the present work will enable subsequent study of other small-molecules as structural effectors of R (and related supramolecular structures) with the goal of gated and self-steering catalytic applications.