Blurring the Lines between Host and Guest: A Chimeric Receptor Derived from Cucurbituril and Triptycene

The preparation of triptycene host 1 follows our building block approach (Figure 1). [5a, 8] To prepare a triptycene based aromatic wall we performed the Diels-Alder reaction between anthracene (2) and benzoquinone (3) followed by aromatization under acidic conditions to yield 4 in 66% yield over both steps. [9] Next, we allowed 4 to react with 1,3-propanesultone (5) under basic conditions (NaOH, dioxane / H2O) to give aromatic wall 6 in 75% yield.

Finally, the double electrophilic aromatic substitution reaction between aromatic wall 6 and glycoluril tetramer 7 was performed (TFA, Ac2O (1:1)) which gave triptycene host 1 in 30% yield after purification by silica gel chromatography and gel permeation chromatography (Sephadex G25). Host 1 was characterized by electrospray ionization mass spectrometry as well as 1 H and 13 C NMR spectroscopy which provide strong evidence for the molecular formula and time averaged C2v-symmetry of 1 (Supporting Information). The solubility of host 1 in water (≈ 3 mM) is sufficient to allow an investigation of its molecular recognition properties (vide infra). We were fortunate to obtain x-ray crystal structures of both aromatic wall 6 (CCDC 1829725) and triptycene host 1 (CCDC 1829724) as shown in Figure 2. [10] Figure 2a shows the packing of 6 in the crystal along the a-axis. Interestingly, two molecules of 6 associate with one another by π-π interactions (mean interplanar separation of 3.57 Å) between the dialkoxy substituted aromatic walls. In addition, one of the unsubstituted

aromatic blades of triptycene 6 nestles into the cleft of an adjacent molecule in a geometry reminiscent of the stack of shuttlecocks. Quite interestingly, triptycene host 1 assumes a lower symmetry conformation whereby one unsubstituted aromatic ring folds into its cavity and reduces the overall cavity volume (Figure 2b). These two aromatic rings are not coplanar and the Ar C-atom to mean plane distances range from 3.55 -3.86 Å (Figure 2b) which is longer than the optimal π-π stacking distance of ≈ 3.4 Å. The cavity of 1 is filled with CF3CO2H. The packing of 1 in the crystal is even more interesting. Three molecules of 1 form a trimer in the ab-plane that again features a shuttlecock type arrangement between the tips of the unsubstituted aromatic rings of the triptycene walls (Figure 2c). Finally, these trimers are linked together by Na + ions that bridge between the ureidyl C=O groups to form a hexagonal honeycomb-type network (Figure 2d). The solvent filled holes in the hexagonal honeycomb-type network form continuous solvent filled channels that extend along the c-axis in the crystal. Overall, the packing of 1 in the crystal can be seen as a direct consequence of the preferences of its acyclic CB[n] and triptycene components. Given the propensity of 1 for self-assembly in the crystal we first sought to determine if 1 is also trimeric in solution. [11] Accordingly, we performed a 1 H NMR dilution experiment (3 mM to 0.03 mM, Supporting Information Figure S40) but did not observe any significant changes in chemical shift. In addition, we measured the diffusion coefficients for 1 alone (D = 2.45 x 10 -10 m 2 s -1 ) and for monomeric 1•12 complex (D = 2.57 x 10 -10 m 2 s -1 ). Having established that 1 remains monomeric in water below its solubility limit of ≈ 3 mM, we next turned our attention to exploring the recognition properties of triptycene host 1.

Based on the structure of 1 whose cavity is shaped by four glycoluril rings and four aromatic rings we envisioned that 1 might be able to accomodate voluminous guests. Conversely, the x-ray crystal structure of 1 (Figure 2b) shows a smaller cavity with a self-folded conformation based on π-π interactions that would need to be interrupted to accommodate larger guests. Accordingly, we decided to investigate the interaction of 1 with guests 8 -19 (Figure 3) which differ greatly in size. We started with narrow guest 13 with a hexylene linker which is appropriate for a CB[6] sized cavity. Figure 4a-c shows the 1 H NMR spectra recorded for 1 alone and for 1:1 and 1:2 mixtures of 1 and 13. In Figure 4a, the resonances for Hc and Hd of the triptycene wall are upfield of those for Ha and Hb which likely reflects the shielding effect of the opposing triptycene wall in the π-stacked geometry shown in Figure 2b. Interestingly, when the 1 H NMR spectrum of 1 was recorded in DMSO-d6, Hc and Hd resonate at 7.44 and 7.12 ppm, respectively, which suggests that the πstacking between triptycene walls is disrupted in this solvent. The large upfield shifts observed for Hu and Hv (Figure 4b; δΔ, Hs: -0.25, Ht: -0.17, Hu: -1.25, Hv: -2.03 ppm) within the 1•13 complex reflect the anisotropic shielding effect of the four aromatic rings that help define the cavity of 1 whereas the observation of separate resonances for both free and bound 13 (Figure 4c) establishes the slow nature of guest exchange on the chemical shift time scale. Interestingly, the host resonances Ha -Hd, Hn, and Ho undergo downfield shifts upon complexation which may reflect the disruption of the self-folded geometry (Figure 4a). and -1.53 ppm) when complexed to 1 (Supporting Information) which establishes binding inside the cavity of 1. Even larger guests, which are typical of CB[7] or CB[8] sized cavities (viologens 10 -12, dye 15, 4,4'-dipiperidinium 16, adamantane derivatives 17 and 19) also bind inside the cavity of 1 and show significant upfield shifts of the guest protons upon binding (Supporting Information) although the larger adamantane derived guests 17 -19 display fast exchange kinetics on the chemical shift timescale. To gain quantitative information on the strength of host•guest binding we performed isothermal titration microcalorimetry experiments between 1 and guests 10, 12, 13, and 17 (Supporting Information and Table 1) which show enthalpically dominated binding events with Ka values in the 10 6 -10 7 M -1 range. The dominant enthalpic driving force observed complexes of 1 likely reflects the presence of high energy waters similar to the situation recently delineated for macrocyclic CB[n] receptors. [12] Interestingly, adamantane ammonium guest 17 which is an ultratight guest for CB [7] binds more than 20-fold weaker to 1 than aromatic guests 10, 12, and 13 which likely reflects the competing recognition preferences of the chimeric host 1 (e.g. CB[n]-type regions prefer alicyclic guests versus triptycene regions which prefer aromatics).

Table 1. Binding constants and thermodynamic parameters for host 1 and selected guests.

Guest Ka (M -1 ) ΔG (kcal mol -1 ) ΔH (kcal mol -1 ) -TΔS (kcal mol -1 ) 10 a 3.5 ± 0.4×10 7 -10.3 -10.1 ± 0.08 -0.19 12 a 2.8 ± 0.4×10 7 -9.0 -18.1 ± 0.3 9.09 13 a 2.6 ± 0.4×10 7 -10.1 -8.89 ± 0.08 -1.23 17 a 1.4 ± 0.1×10 6 -8.4 -4.86 ± 0.4 -3.54 20 a 3.7 ± 0.2×10 7 -10.5 -10.1 ± 0.1 -0.46 22 b 1.2 ± 0.2×10 6 -8.3 -11.3 ± 0.14 3.05

Note.

[a] H2O, 25 ˚C. [b] 20 mM sodium phosphate buffered H2O containing 1M

NaNO3, pH 7.4, 25 ˚C.

Subsequently, we sought to test the limits of the capacity of host 1 by offering it guests that are too large to fit inside macrocyclic CB [8] which is also composed of eight building blocks. An obvious choice was blue box 20 which is known to bind CB [10] but not CB [8]. [13] Accordingly, we titrated an aqueous solution of 1 with 20 and observed the formation of a precipitate that contained a 1:1 mixture of the components. We believe the poor solubility of the 1•20 complex reflects the formation of a zwitterion (e.g. tetraanion 1 complexed with tetracation 20). Although the poor solubility of 1•20 in water prevented use of 1 H NMR as an analytical tool, we were able to measure the Ka value for 1•20 by ITC (3.7 ± 0.2 × 10 7 M -1 ) by using low concentrations of 1 (8.75 µM) in the ITC cell. Fortunately, we found that the 1•20 complex is stable in 30% DMSO-d6 in D2O which allowed us to record the 1 H NMR spectra shown in Figure 5 which could be assigned with the help of the COSY and NOESY spectra (Supporting Information). Most interesting is that the symmetry of both the host and the guest are reduced (and the number of resonances increased) upon formation of the 1•20 complex. For example, host 1 exhibits four resonances for CH3-groups n and o, five pairs of resonances for the diastereotopic CH2-groups of the glycoluril tetramer

backbone, four resonances for Hl and Hm, and four sets of resonances for the aromatic triptycene walls which indicates that the two ends of host 1 are different in complex 1•20. Similarly, two sets of upfield shifted resonances are observed for each of the viologen protons Hs and Ht of 20 (Δδ: s = -0.27 and -0.38 ppm; t = -0.73 and -2.29 ppm) in the 1•20 complex. One aromatic wall (Ar4, Figure 5) and one viologen proton (Hs) are strongly upfield shifted by cavity inclusion. These observations are not consistent with a geometry where guest 20 is fully engulfed by host 1 but rather point to a geometry whereby one of the aromatic walls of 1 (Ar4) inserts into the cavity of blue box 20 and one of the viologen units of 20 inserts into the cavity of 1 as depicted in Figure 6a. Cross peaks observed in the 2D NOESY spectrum are consistent with this geometrical assignment (Supporting Information). We find this geometry to be intriguing because it blurs the typical lines of division between host and guest in that regions of 1 and 20 perform both host and guest roles within a single structure. In analogy with the clipped rotaxane previously reported by Klärner, [14] the geometry of 1•20 can be viewed as a clipped catenane. Encouraged by the intriguing geometry of the 1•20 complex, we decide to investigate the binding of 1 with Fujita's bipyridine squares (21: M = Pd, and 22: M = Pt). [15] After much experimentation, on the basis of 1 H NMR titration and ITC experiments (Supporting Information) and considering the kinetic lability of the Pd-N coordination bonds [15] we determined that addition of four equivalents of 1 to a solution of 21 causes its disassembly into four equivalents of bipyridine•Pd(en) which bind to 1 in the usual way. The situation was completely different for the platinum analogue 22 because of its kinetically inert Pt-N coordination bonds. [15] Beyond the intriguing geometrical features of the complex between 1 and 20 or 22 lies interesting optical properties. Initially, we found that wall 2 absorbs at 213 nm (ε = 4.7 × 10 4 M -1 cm -1 ). Compound 2 is fluorescent with an emission band at 337 nm (λex = 213 nm). The fluorescence of 2 is quenched by the addition of Fe 3+ , partially quenched by Cu 2+ , but not Ag + and NH4 + (Supporting Information). In contrast, host 1 displays a UV/Vis absorbance band at 214 nm (ε = 6.6× 10 4 M -1 cm -1 ) and a longer wavelength fluorescence emission maximum at 377 nm in water which is probably due to interaction between the two triptycene walls. Accordingly, we hypothesized that guest binding might lead to significant fluorescence changes. Figure 7 shows the fluorescence spectra recorded for 1 (12 µM) in the presence of different guests and Table 2 presents key parameters. Interestingly, in all cases guest binding results in a hypsochromic shift in emission maximum probably due to disruption in the π-π interactions between the tryptycene walls. Most interesting is the divergent behavior of adamantane derivatives 17, 18, and 19. Whereas the 1•17 and 1•18 complexes display enhanced fluorescence, complex 1•19 is completely quenched presumably due to photoinduced electron transfer from the excited state of 1 to the pyridinium guest 19. In accord with this interpretation is the observation that methyl viologen complex 1•12 is also heavily quenched. Related sensing materials have been pioneered by Swager for detection of nitroaromatics. [16] Host 1 and related compounds -with their high affinity and selectivity toward hydrophobic cations -may offer unique opportunities as aqueous sensors for pyridinium and related quaternary ammoniums that are present in drugs and other natural products (e.g. NADP). In summary, we have reported the synthesis of a chimeric receptor 1 combining the recognition preferences of the cucurbituril and triptycene. The x-ray crystal structure of 1 reveals a self-folded geometry based on π-π interactions between the triptycene walls of 1; 1 further organizes itself into a honeycomb arrangement that features infinite solvent filled channels along the c-axis. Host 1 binds to hydrophobic (di)cations as is typical of CB[n] derived receptors but also recognized larger guests like blue box 20 and Fujita square 22. Intriguingly, the geometries of 1•20 and 1•22 blur the lines between what constitutes host and guest in that each component contains binding epitopes that fulfil both roles in each complex.

Finally, 1 displays guest responsive fluorescence and is particularly sensitive to pyridinium derived guests 12 and 19 which quench host fluorescence by photoinduced electron transfer.

Overall, the work further establishes acyclic CB[n]-type receptors as versatile and readily functionalized systems that display both intriguing recognition behaviour and function. Φ(%) 6.9 2.6 4.3 7.6 3.5 5.5 8.8 0.33