Monomeric Cryptophane with Record-High Xe Affinity Gives Insights into Aggregation-Dependent Sensing

Abstract

Cryptophane host molecules provide ultrasensitive contrast agents for ¹²⁹Xe NMR/MRI. To investigate key features of cryptophane–Xe sensing behavior, we designed a novel water-soluble cryptophane with a pendant hydrophobic adamantyl moiety, which has good affinity for a model receptor, beta-cyclodextrin (β-CD). Adamantyl-functionalized cryptophane-A (AFCA) was synthesized and characterized for Xe affinity, ¹²⁹Xe NMR signal, and aggregation state at varying AFCA and β-CD concentrations. The Xe–AFCA association constant was determined by fluorescence quenching, K_A = 114,000 ± 5000 M⁻¹ at 293 K, which is the highest reported affinity for a cryptophane host in phosphate-buffered saline (pH 7.2). No hyperpolarized (hp) ¹²⁹Xe NMR peak corresponding to AFCA-bound Xe was directly observed at high (100 μM) AFCA concentration, where small cryptophane aggregates were observed, and was only detected at low (15 μM) AFCA concentration, where the sensor remained fully monomeric in solution. Additionally, we observed no change in the chemical shift of AFCA-encapsulated ¹²⁹Xe after β-CD binding to the adamantyl moiety and a concomitant lack of change in the size distribution of the complex, suggesting that a change in the aggregation state is necessary to elicit a ¹²⁹Xe NMR chemical shift in cryptophane-based sensing. These results aid in further elucidating the recently discovered aggregation phenomenon, highlight limitations of cryptophane-based Xe sensing, and offer insights into the design of monomeric, high-affinity Xe sensors.

Graphical Abstract

Proton-based magnetic resonance imaging (¹H MRI) is a widely used technique for the detection and diagnosis of various human disease states. However, ¹H MRI is poorly suited for imaging void spaces and areas with low-proton spin densities, such as the lungs.¹ Furthermore, the low sensitivity of ¹H MRI at clinical magnetic field strengths limits its application for molecular imaging. To overcome these issues, hyperpolarized (hp) ¹²⁹Xe NMR and MRI have been investigated as complements to ¹H-based techniques. Readily obtained via spin-exchange optical pumping (SEOP),² hp ¹²⁹Xe offers a 10⁴–10⁸ signal enhancement over thermal polarization.^{3,4 129}Xe possesses additional properties that make it attractive for imaging applications, including good solubility in water and organic solvents.⁵ A highly polarizable electron cloud promotes binding to molecular capsules or proteins and produces a ~300 ppm NMR chemical shift window for host-bound ¹²⁹Xe in aqueous solution.⁶

This molecular sensitivity has motivated the use of Xe in sensing applications, where a specific-analyte affinity tag is conjugated to a Xe-binding host molecule. The best-studied Xe host is the 1 nm spherical, cagelike molecule known as cryptophane-A, formed by linking two cyclotriveratrylene (CTV) units by three ethoxy bridges.^7,8 In 2001, a biotin-modified cryptophane for detecting avidin via hp ¹²⁹Xe NMR provided the first example of a ¹²⁹Xe biosensor.⁹ This led our laboratory and others to synthesize cryptophane–¹²⁹Xe sensors that can detect proteins,^10–18 biothiols,¹⁹ pH,^20,21 metal ions,^22–26 and DNA²⁷ via hp ¹²⁹Xe NMR. However, only a few biomolecular interactions with cryptophanes have been shown to elicit ¹²⁹Xe NMR chemical shift changes larger than 5 ppm, which poses limitations for biosensing at clinical magnetic field strengths. Furthermore, we showed that cryptophanes can aggregate to form water-soluble nano-particles in solution, which further complicates efforts to control the ¹²⁹Xe NMR chemical shift in the absence and presence of analytes.²⁸ Here, we use a novel cryptophane–¹²⁹Xe NMR sensor as a model system to study aggregation-modulated Xe-cryptophane binding.

The cryptophane aggregation phenomenon has been discovered only recently;²⁸ consequently, its details are not yet well known. However, it is likely that aggregation, or lack thereof, of cryptophane units in solution has a considerable effect on Xe accessibility to the cage interior. We previously observed that the aggregation state can modulate the ¹²⁹Xe NMR chemical shift and the observed hyper-chemical exchange saturation transfer (CEST) contrast.²⁸ To obtain a better understanding of these effects, we set out to design a cryptophane-based sensor for use as a model system to further study the solution behavior of cryptophanes with respect to Xe binding. In our previous studies of cryptophane solution behavior, we reported on a benzenesulfonamide-functionalized cryptophane biosensor (C8B) that transitioned from heavily aggregated (average diameter = 345 ± 13 nm) to monomeric (average diameter = 4.4 ± 0.1 nm) upon binding to its carbonic anhydrase II (CAII) protein target.²⁸ We observed that as CAII was added to solutions of C8B, the chemical shift of ¹²⁹Xe–C8B/CAII increased to 5.9 ppm downfield of ¹²⁹Xe–C8B at a 1:20 C8B/CAII ratio. The C8B studies indicated that changes in the biosensor aggregation state contributed significantly to the observed ¹²⁹Xe NMR chemical shift. However, we were not able to study ¹²⁹Xe NMR chemical shift changes upon target binding with nonaggregated cryptophane systems nor changes in Xe affinity upon biosensor-target binding because of strong background fluorescence from aromatic amino acids in CA. Here, we hypothesized that water-soluble β-cyclodextrin (β-CD), which exhibits a minimal fluorescence signal when excited at 280 nm and thus enables a fluorescence quenching technique for determining cryptophane-Xe affinity in the absence and presence of the target, could be used as a model target receptor for a functionalized cryptophane. Additionally, β-CD has not been shown to interact with Xe in aqueous solution, which is thought to be due to either its poor Xe affinity or, if transient binding occurs, to fast exchange making such binding undetectable by NMR and fluorescence-based techniques.²⁹ Adamantane derivatives have good association constants (K_A = 10³ to 10⁵ M⁻¹ at rt) for β-CD,³⁰ which motivated the use of an adamantyl recognition moiety in our sensor design. As in our prior trifunctionalized cryptophane biosensors, we added two carboxylates to ensure water solubility at physiologic pH. We report herein the synthesis and characterization of adamantyl-functionalized cryptophane-A (AFCA, Scheme 1), a novel cryptophane sensor for ¹²⁹Xe NMR. Using this model system, it was possible to further study the effects of cryptophane aggregation on Xe affinity and ¹²⁹Xe NMR signal.

Scheme 1.

Synthesis of AFCA 5^a

^a(a) 1 (1 equiv), 2 (1.1 equiv), CuSO₄ (1.2 equiv), TBTA (1 equiv), 2,6-lutidine (1 equiv), NaAsc (20 equiv), 12 h. (b) 3 (1 equiv), 4 (10 equiv), CuSO₄ (1.2 equiv), 2,6-lutidine (1 equiv), NaAsc (45 equiv), 12 h.

EXPERIMENTAL SECTION

Synthetic Methods.

Scheme 1 shows the synthesis of AFCA from tripropargyl cryptophane 1, which was synthesized with slight modifications to previously published methods,³¹ as given in the Supporting Information. 1 (60.2 mg, 0.0622 mmol, 1 equiv) was dissolved in 4 mL of dry dimethyl sulfoxide (DMSO) and added to a 15 mL round-bottom flask. 2-Azido-N-adamantylacetamide (synthesized from 1-adamantylamine using the literature procedure,³² 16.0 mg, 0.0684 mmol, 1.1 equiv) was added to the flask, which was stirred under N₂ gas at room temperature (rt). Separate solutions of 100 mg/mL CuSO₄ and 860 mg/mL sodium ascorbate (NaAsc) were prepared. The reaction flask was degassed, and 119.2 μL of 100 mg/mL CuSO₄ solution (0.0746 mmol, 1.2 equiv) was added.

The flask was degassed again, and the tris-(benzyltriazolylmethyl)amine (TBTA) ligand (33.0 mg, 0.0622 mmol, 1 equiv) was added. The flask was sonicated and degassed, and 2,6-lutidine (7.2 μL, 0.0622 mmol, 1 equiv) was added. The flask was degassed once more, and 286.8 μL of 860 mg/mL NaAsc solution (1.24 mmol, 20 equiv) was added. The flask was covered in foil and degassed, and the reaction was allowed to stir overnight under N₂. The following day, ddH₂O (40 mL) was added to the reaction and was carried out three times with EtOAc. The combined organics were washed with ddH₂O and brine, and the aqueous layers were extracted with EtOAc. The combined organics were dried over Na₂SO₄ and concentrated in vacuo. The dry product was purified by using column chromatography [1:20 to 3:20 acetone/CH₂Cl₂, gradient method and thin-layer chromatography 3:20 acetone/CH₂Cl₂ R_f(3) = 0.65] to yield 29.6 mg (0.0246 mmol, 40% yield) of 3. ¹H NMR (CDCl₃, 500 MHz): δ 7.84 (s, 1H), 6.78 (m, 12H), 5.62 (s, 1H), 5.16 (ABq, 2H, J_AB = 12.4 Hz), 5.16 (ABq, 2H, J_AB = 12.4 Hz), 5.00 (ABq, 2H, J_AB = 15.8 Hz) 4.63 (m, 10H), 4.18 (m, 12H), 3.83 (d, 6H), 3.58 (s, 3H), 3.41 (m, 6H), 2.70 (q, 2H), 2.08 (m, 3H), 1.97 (m, 6H), 1.66 (m, 6H).

3 (29.6 mg, 0.0246 mmol, 1 equiv) was dissolved in 4 mL of dry DMSO and added to a 15 mL round-bottom flask. 3-Azidopropionic acid (synthesized from 3-bromopropionic acid using the literature procedure,³³ 28.4 μL, 0.246 mmol, 10 equiv) was added to the flask, which was stirred under N₂ gas at rt. Separate solutions of 100 mg/mL CuSO₄ and 860 mg/mL NaAsc were prepared. The reaction flask was degassed, and 100 mg/mL CuSO₄ solution (47.2 μL, 0.0295 mmol, 1.2 equiv) was added. The flask was degassed, and 2,6-lutidine (2.9 μL, 0.0246 mmol, 1 equiv) was added. The flask was degassed once more, and 860 mg/mL NaAsc solution (255.4 μL, 1.11 mmol, 45 equiv) was added. The flask was covered in foil and degassed, and the reaction was allowed to stir overnight under N₂. In the work-up, 30 mL of ddH₂O and 10 mL of brine were added to the reaction and extracted three times with EtOAc. The combined organics were washed with brine, dried over Na₂SO₄, and concentrated in vacuo. The dry solid was then dissolved in dilute NaOH, transferred to a conical centrifuge tube, and precipitated by drop-wise addition of 1 M HCl. The mixture was centrifuged (277 K, 4000 rpm, 30 min), and the supernatant was removed. The solids were resuspended in ddH₂O, dissolved by drop-wise addition of 1 M NaOH, precipitated by drop-wise addition of 1 M HCl, and centrifuged again, and the supernatant was removed. Finally, the solids were resuspended in ddH₂O, sonicated for 10 min, centrifuged to remove the supernatant, and lyophilized to yield 24.6 mg (0.0172 mmol, 70% yield) of AFCA as a white powder. ¹H NMR (d₆-DMSO, 500 MHz): δ 8.20 (s, 2H), 8.17 (s, 1H), 7.93 (s, 1H), 7.09 (d, 3H), 6.85 (m, 9H), 5.08 (m, 8H), 4.64 (t, 4H, J = 6.7 Hz), 4.52 (m, 6H), 4.14 (m, 12H), 3.47 (s, 6H), 3.42 (s, 3H), 2.95 (t, 4H, J = 6.7 Hz), 1.98 (m, 9H), 1.62 (m, 6H). H_eq protons with a total integration of 6H were hidden under the residual water peak. ¹³C NMR (d₆-DMSO, 500 MHz): δ 171.9, 164.2, 148.9 (d), 148.8, 147.8 (d), 146.2, 146.1, 146.0, 145.7 (d), 145.5, 143.3, 143.2, 133.8 (d), 133.7, 133.5, 132.7, 132.5, 131.7 (d), 125.6, 124.2, 120.6, 119.7, 119.1, 117.2, 115.2, 115.0, 68.8, 68.5, 68.2, 62.9 (d), 62.8, 56.0, 55.9, 52.1, 51.3, 45.7, 40.1, 40.0, 39.9 (d), 39.8, 39.7, 39.6, 39.5, 39.4 (d), 39.2, 39.0, 35.9, 34.9, 34.7, 30.7, 28.8. HRMS m/z: calcd for C₇₈H₈₂N₁₀O₁₇ [M + 2H]²⁺, 716.2997; found, 716.3008. MALDI-MS m/z: calcd for C₇₈H₈₂N₁₀O₁₇ [M + Na]⁺, 1453.576; found, 1453.617 (Figure S1).

Dynamic Light Scattering.

Samples were prepared as cryptophane solution at the indicated concentrations in pH 7.2 phosphate-buffered saline (PBS) and filtered via 0.22 μm centrifugal filter units. Dynamic light scattering (DLS) data were collected at 298 K with the Zetasizer Nano Z system using software from Malvern Instruments Ltd. v. 2.0. The number distribution was analyzed as an average of three trials.

Direct Detection hp ¹²⁹Xe NMR Spectroscopy.

Hp ¹²⁹Xe was cryogenically separated, accumulated, thawed, and collected in controlled atmosphere valve 10 mm NMR tubes (New Era) containing the sample and 0.1% (v/v) Pluronic L81 (Aldrich). Prior to data collection, tubes were vigorously shaken to mix sample solutions with Xe. An EBURP-1 selective pulse centered at 63 ppm with an excitation bandwidth of 680 Hz, pulse length τ_pulse = 2.2 ms, and B_1,max = 288 μT was used, and the signal was averaged over 64 scans, unless otherwise noted. Fourier-transformed spectra were processed with zero-filling and Lorentzian line broadening of 40 Hz. Chemical shifts were referenced to aqueous ¹²⁹Xe in its respective solvent.

Xenon Quenching of AFCA Fluorescence.

This method was adapted from the one published previously by our laboratory.³⁴ To prepare the saturated Xe solution, PBS buffer (ca. 4 mL) was added to an acid-washed screw top vial with a total volume of ca. 7 mL. The vial was sealed with a PTFE-faced silicone septum (Ace Glass), and the solution was sonicated for 10 s. The vial was evacuated by high vacuum via a 22-gauge needle inserted through the septum into the headspace. After degassing was complete, the needle was pushed to the bottom of the vial, and Xe gas (Airgas) was bubbled into the vial for 5 min, with a 21-gauge needle inserted into the septum to vent excess pressure. Cryptophane solution was added to an acid-washed screw top cuvette (1 cm path length, 4 mL total volume, Starna Cells). The concentration and volume of cryptophane solution were adjusted, so that after the addition of saturated Xe solution, the concentration of cryptophane was either 15 μM for low concentration measurements or 100 μM for high concentration measurements, with 1 mL of the headspace in the cuvette. The cuvette was sealed with a PTFE-faced silicone septum (Ace Glass) and evacuated by high vacuum via the 22-gauge needle inserted through the septum into the headspace. Aliquots of saturated Xe solution (5.05 mM) could then be transferred from the vial to the cuvette via syringe. The cuvette was allowed to equilibrate at 293 K in the fluorimeter for 30 min before collection of fluorescence spectra. The excitation and emission wavelengths were 280 and 317 nm, respectively. The PMT voltage was 900 V for low concentration measurements and 850 V for high concentration measurements. For the saturated Xe measurement, Xe gas was directly bubbled into the degassed cuvette containing 3 mL of 15 μM cryptophane for 5 min, with a 21-gauge needle inserted into the septum to vent excess pressure. The curve fitting was done with OriginPro 8 using the following single-site binding model:

$$\frac{[\text{Xe}@\text{crypt}]}{[\text{ crypt }]+[\text{Xe}@\text{crypt}]}=\frac{[\text{Xe}]}{[\text{Xe}]+1/K_{\text{A}}}$$

where K_A is the Xe association constant. Fluorescence measurements were obtained using a Varian Cary Eclipse fluorescence spectrophotometer.

pH Titration of AFCA.

This method was adapted from the one previously published by our laboratory.³⁵ AFCA (2.7 mg) was suspended in a degassed 100 mM NaCl solution, and 1 μL aliquots of degassed 100 mM NaOH were added until all solid dissolved. pH measurements were made using a Thermo Scientific Orion Star A111 pH meter with a micro glass-body combination pH electrode. With stirring, 1 μL portions of degassed 50 mM HCl were titrated into the AFCA solution, and the pH values recorded after 30 s. The pK_a value was determined from the x-intercept of the second derivative of the titration curve.

ITC, AFCA, and β-CD.

All isothermal titration calorimetry (ITC) experiments were performed at 298 K on a GE Healthcare MicroCal iTC200 instrument. Solutions of 1.00 mM β-CD and 0.11 mM AFCA were prepared in pH 7.2 PBS. Prior to performing ITC, both the cryptophane and β-CD solutions were degassed by brief bath sonication, followed by centrifugation. The sample cell was filled with AFCA solution (325 μL), and the reference cell was filled with water. Calorimetric data were analyzed by performing nonlinear regression fitting to the binding isotherm using Origin software.

Characterization of AFCA Aggregates via Fluorescence Spectroscopy.

This method was adapted from our previous fluorescence-based studies of cryptophane aggregates.²⁸ Briefly, aliquots of 3.4 μL of 145.5 μM acridine orange (AO) solution were added to 100 μL samples of AFCA in pH 7.2 PBS buffer at varying concentrations. The solutions were mixed thoroughly before fluorescence measurements using an excitation wavelength of 436 nm and monitoring emission at 525 nm. The AO concentration was 4.8 μM for all samples. The PMT voltage was 950 V. Measurements were averaged over three trials. Fluorescence measurements were obtained using a Varian Cary Eclipse fluorescence spectrophotometer.

¹²⁹Xe NMR Hyper-CEST Frequency Scans.

Hp ¹²⁹Xe was generated using the SEOP method with a home-built ¹²⁹Xe polarizer based on the IGI.Xe.2000 commercial model by GE. A Shark 65 W tunable ultranarrow band diode laser (OptiGrate) set to 795 nm was used for optical pumping of Rb vapor. A gas mixture of 88% He, 10% N₂, and 2% natural abundance Xe (Linde Group, NJ) was used as the hyperpolarizer input. The output ¹²⁹Xe hyperpolarization level was roughly 10–15%. To determine the magnitude and frequency of the CEST effect for a given sample, shaped saturation pulses were scanned across a specific chemical shift range, and the normalized integral of the resulting ¹²⁹Xe(aq) signal was plotted as a function of saturation frequency, generating what is known as a z-spectrum. For each data point, hp ¹²⁹Xe was bubbled into a 10 mm NMR tube containing 2.5 mL of the sample through capillaries for 20 s, followed by a 3 s delay to allow bubbles to collapse. The pressure of the gas downstream of the inlet valve to the NMR tube was ca. 63 psi, and the flow rate of gas was ca. 0.70 standard liters per minute. Dsnob saturation pulses with total saturation time τ_sat = 2.29 s and the indicated presaturation field strength were used. NMR experiments were performed using a Bruker BioDRX 500 MHz NMR spectrometer and a 10 mm PABBO probe at 300K. A 90° hard pulse of this probe has a pulse length of 40.6 μs. Measurements were averaged over three trials.

RESULTS AND DISCUSSION

Xe Binding to AFCA.

In order to determine the Xe association constant for AFCA, we used a fluorescence quenching method previously published by our laboratory.³⁴ By adding aliquots of Xe-saturated buffer to a sealed cuvette containing 15 μM AFCA solution and monitoring the decrease in fluorescence, we obtained a Xe association constant of 114,000 ± 5000 M⁻¹ at 293 K in pH 7.2 PBS (Figure 1), which is currently the highest measured K_A for a single cryptophane host in physiologic buffer solution. This value is slightly higher than that measured under the same conditions for the structurally similar tris-(triazole propionic acid)-cryptophane (TTPC, K_A = 99,000 ± 4000 M⁻¹, Figure S2), which features a third propionate moiety in place of the adamantyl group.¹ We hypothesize that this elevated Xe affinity results from a low average number of water molecules binding inside the cryptophane core. It was previously reported that the average number of water molecules residing in the cryptophane cavity and the experimental free energy of Xe binding are negatively correlated³⁶ because of Xe needing to displace any cryptophane-bound water upon binding. It was noted that cryptophanes with more hydrophobic side chains contained fewer water molecules on average, resulting in greater Xe affinity. Thus, we believe that replacing propionic acid with the hydrophobic adamantyl moiety directly contributed to the elevated K_A value measured by fluorescence quenching. We note that this could be a promising strategy to develop high-affinity Xe hosts, with the only limitation being a decrease in cryptophane solubility in aqueous solution.^a

Figure 1.

Fluorescence quenching of 15 μM AFCA in pH 7.2 PBS, 293 K. The legend shows the concentrations of Xe used. K_A was determined to be 114,000 ± 5000 M⁻¹. Data shown are an average of three trials. Inset: Curve fit for a single-site binding model.

Additional confirmation of Xe binding to AFCA came from ¹H NMR studies performed in D₂O in the absence and presence of Xe. The ¹H NMR spectra (Figure S3) show remarkably sharper resonances upon Xe addition, particularly in the region corresponding to the ethoxy linkers (δ = 4.5−4.3 ppm) and the axial protons on the two CTV caps (δ = 4.2−3.9 ppm). This is due to the change in conformation of the linkers from gauche to trans, as the cryptophane adopts a more open conformation to accommodate the Xe atom.³⁷ The sharpening of these resonances has previously been reported in cryptophanes with Xe and CHCl₃ as guests.^35,38

However, when we studied Xe binding to AFCA via direct detection ¹²⁹Xe NMR spectroscopy, no hp ¹²⁹Xe NMR signal corresponding to Xe bound inside AFCA was observed with a solution of 100 μM AFCA in pH 7.2 PBS (Figure 2a). At similar concentrations in the same buffer, we were previously able to detect ¹²⁹Xe@cryptophane NMR signals with a cryptophane–benzenesulfonamide biosensor¹² and TTPC.¹⁷

Figure 2.

hp ¹²⁹Xe NMR spectra of AFCA in pH 7.2 PBS, 300 K. (a) 100 μM AFCA and (b) 15 μM AFCA; the Xe@AFCA peak was observed at 63.5 ppm. (c) 15 μM AFCA and 75 μM β-CD; the Xe@AFCA-β-CD peak was observed at 63.6 ppm. The encapsulated ¹²⁹Xe peaks were referenced to the ¹²⁹Xe@aq peak at 193.0 ppm.

DLS with AFCA.

To confirm that the absence of a bound ¹²⁹Xe NMR peak was not due to substantial aggregation of AFCA in solution, we measured the size distribution of solutions of AFCA by DLS. Figure 3 shows the size distribution obtained by DLS for a solution of 100 μM AFCA in pH 7.2 PBS in comparison to TTPC. It was observed that AFCA exhibited only minor aggregation in solution, with an average diameter of 5.7 ± 0.2 nm, ~10-fold smaller than TTPC and similarly trifunctionalized water-soluble cryptophanes that our laboratory previously analyzed by DLS.²⁸ Figure 3 also shows the size distribution for 100 μM solutions of AFCA and TTPC in 10 mM NaOH. We observed that in 10 mM NaOH, the average diameter of 100 μM AFCA decreased to 1.6 ± 0.3 nm, which is in good agreement with the predicted molecular structure, as the cryptophane itself is nearly 1 nm in diameter, and the fully extended and solvated propionate and adamantyl side chains extend the molecular diameter to approximately 2 nm. This represents the smallest reported size for a functionalized cryptophane, likely resulting from the increased solvation of AFCA at elevated pH. Titration of AFCA with HCl (Figure S4) determined a pK_a of 5.8 for the less acidic carboxylate, confirming that both carboxylates are completely ionized in 10 mM NaOH. This decrease in the aggregation state in basic solution was also observed with TTPC: in 10 mM NaOH, the average diameter of 100 μM TTPC decreased to 17 ± 2 nm from 55 ± 5 nm in pH 7.2 PBS (Figure 3).

Figure 3.

DLS data showing size distribution by number at 298 K of 100 μM AFCA; average diameter = 5.7 ± 0.2 nm in pH 7.2 PBS, polydispersity index (PdI) = 0.55 and 1.6 ± 0.3 nm in 10 mM NaOH, PdI = 0.70 and of TTPC; average diameter = 55 ± 5 nm in pH 7.2 PBS, PdI = 0.49 and 17 ± 2 nm in 10 mM NaOH, PdI = 0.54. Data shown are an average of three trials.

¹²⁹Xe NMR, DLS, and Fluorescence Quenching Studies with AFCA.

Interestingly, when the AFCA concentration was lowered to 15 μM, the concentration at which the Xe affinity measurements were performed (Figure 1), a distinct peak at 63.5 ppm corresponding to AFCA-encapsulated ¹²⁹Xe was able to be directly observed by ¹²⁹Xe NMR (Figure 2b). DLS experiments revealed that at this concentration in pH 7.2 PBS, AFCA becomes monomeric, with an average diameter of 1.4 ± 0.6 nm (Figure S5). These results suggest that even small AFCA aggregates can preclude observation of the ¹²⁹Xe NMR signal. This hypothesis is further supported by fluorescence studies of AFCA using a method previously developed by our laboratory to estimate critical concentration of cryptophane.²⁸ Briefly, the fluorescence emission of solutions of varying concentrations of cryptophane and constant concentrations of AO dye is monitored, and an increase in AO fluorescence is observed as the dye transitions from an aqueous to a micellar environment. Using this technique, we observed a distinct difference in fluorescence behavior as AFCA solutions exceeded ca. 50 μM, indicating the inception of aggregate formation (Figure S6).

To gain additional understanding of the mechanism behind this phenomenon, we performed fluorescence quenching experiments with 100 μM AFCA to determine if the lack of observable ¹²⁹Xe NMR signal is due to decreased Xe affinity at elevated cryptophane concentration (Figure S7). The Xe association constant was determined to be 96,000 ± 10,000 under these conditions, which is only a 16% decrease relative to 15 μM AFCA. This reveals that Xe is still able to bind with high affinity to cryptophanes within small aggregates, as was also indicated by the sharpening of key ¹H resonances for low-mM AFCA in the presence of Xe (Figure S3). Additionally, our previous studies with aggregated cryptophane suggest that elevated Xe exchange is more favored in monomeric cryptophane constructs;²⁸ thus, it is unlikely that a lack of ¹²⁹Xe NMR signal at high AFCA concentrations is due to fast exchange. We believe that the most probable explanation for this phenomenon is that AFCA aggregates promote relaxation of the encapsulated ¹²⁹Xe, which precludes observation of the signal. We note that the intensity of the ¹²⁹Xe(aq) signal is not affected by the AFCA concentration (Figure 2), suggesting that it is only the encapsulated ¹²⁹Xe pool that is prone to relaxation effects within these small cryptophane aggregates.

A secondary explanation of this phenomenon is that the Xe–AFCA exchange is significantly slow and so does not efficiently deplete the pool of bulk hp ¹²⁹Xe, resulting in the intensity of the ¹²⁹Xe(aq) signal to remain unchanged. To semiquantitatively estimate this exchange rate, we used an approximation³⁹ to a simplified solution derived from the Bloch–McConnell equations for CEST between hp nuclei in solvent and bound to host.⁴⁰ Briefly, the CEST effect can be described as a function of the saturation time, t, and the depolarization rate, λ, using the following equation

$$\text{CEST}=1-{\text{e}}^{-t\lambda }$$

The on-resonance depolarization rate can then be described as a function of the saturation pulse strength, B₁, using the following equation

$$\lambda =\frac{f{k}_{\text{off}}{\left(\gamma B_1\right)}^2}{{\left(\gamma B_1\right)}^2+\left(k_{\text{off}}{}^2\right)}$$

where f is the ratio of Xe encapsulated by the host to Xe in bulk solvent, k_off is the Xe dissociation rate, which approximates the Xe exchange rate, and γ is the gyromagnetic ratio of ¹²⁹Xe. Thus, it becomes possible to express the CEST effect as a function of B₁ using the equation below

$$\frac{1}{\text{ln}(1-\text{CEST})}=-t^{-1}{f}^{-1}\left[\frac{k_{\text{off}}}{{\left(\gamma {\text{B}}_1\right)}^2}+\frac{1}{k_{\text{off}}}\right]$$

Plotting 1/ln(1 − CEST) as a function of (γB₁)⁻² yields a linear relationship with a slope equal to (−t⁻¹f⁻¹)(k_off) and y-intercept equal to (−t⁻¹f⁻¹)/(k_off), from which one can extract k_off as (slope/y-intercept)^0.5. This method was recently used by our laboratory to estimate the Xe exchange rate of an organocobalt tetrahedron capable of encapsulating Xe.⁴¹

Figure S8 shows the z-spectra obtained with 15 μM AFCA at saturation pulse strengths ranging from 15 to 30 μT. The plot of 1/ln(1 − CEST) as a function of (γB₁)⁻² yielded a linear relationship (R² = 0.959, Figure S9, from which it was possible to extract the Xe–AFCA dissociation rate. This value was determined to be 53 ± 3 s⁻¹ at 300 K, which is similar to the Xe exchange rate for a triacetic acid-functionalized cryptophane-A derivative (45 s⁻¹ at 297 K)⁴² previously synthesized by our lab. These data suggest that the Xe–AFCA exchange is not significantly reduced relative to other functionalized cryptophanes. We note that these studies were performed with a solution of 15 μM AFCA, a concentration at which the compound was observed to be monomeric (Figure S5). Performing these experiments with AFCA aggregates at concentrations at or above 100 μM was not feasible because of substantial foaming of cryptophane solution during Xe bubbling. Therefore, it may be possible that slow Xe–aggregate exchange contributes to a lack of encapsulated ¹²⁹Xe signal, especially when in tandem with an elevated relaxation time.

The hypothesis that AFCA aggregation can affect observation of the ¹²⁹Xe NMR signal is supported by the restoration of signal in 10 mM NaOH, a solvent in which AFCA was observed to be monomeric (Figure 3); a strong peak at 63.7 ppm corresponding to AFCA-encapsulated ¹²⁹Xe was observed with a 200 μM solution of AFCA (Figure S10). Having shown that we could completely disaggregate high concentrations of AFCA using 10 mM NaOH, we performed fluorescence quenching experiments in this solution to study the Xe affinity of the completely monomeric cryptophane. Using this method with 15 μM AFCA, we determined a Xe affinity for AFCA of 136,000 ± 5000 M⁻¹ at rt in 10 mM NaOH (Figure S11), a 19% increase relative to pH 7.2 PBS. These data suggest that Xe affinity for cryptophane is modulated by its aggregation state, although additional studies are required to elucidate the molecular–structural details behind this phenomenon. Additionally, future studies will aim to clarify why observation of “bound” ¹²⁹Xe NMR signal is possible in more aggregated cryptophane species (TTPC,¹⁷ C8B¹²), but not in less aggregated ones (AFCA).

¹²⁹Xe NMR, DLS, and Fluorescence Quenching Studies with the AFCA:β-CD Complex.

ITC confirmed that AFCA shows decent affinity for β-CD, with a measured association constant of 38,000 ± 5000 M⁻¹ at rt (Figure S12). A direct detection ¹²⁹Xe NMR spectrum of 15 μM AFCA and 75 μM β-CD revealed a peak at 63.6 ppm corresponding to ¹²⁹Xe encapsulated by the AFCA:β-CD complex (Figure 2c). No significant chemical shift change of this peak relative to ¹²⁹Xe bound by free AFCA (Figure 2b) was observed. Importantly, we also observed no significant change in the size distribution of the AFCA:β-CD complex relative to AFCA only; the average diameter of the sensor-target complex was found by DLS to be 1.5 ± 0.3 nm (Figure S13), which is almost identical to unbound AFCA at the same concentration (1.4 ± 0.6 nm, Figure S5). We conclude that without inducing a large change in the aggregation state, β-CD binding to the pendant adamantane is too small of a structural perturbation to result in a ¹²⁹Xe NMR chemical shift change. This finding supports our previous hypothesis²⁸ that many analytes will not elicit a ¹²⁹Xe NMR chemical shift change unless they induce a significant change in the cryptophane aggregation state.

Finally, we performed fluorescence quenching experiments to compare Xe affinity of the 1:5 AFCA:β-CD complex to that of AFCA alone. The Xe affinity for this complex was determined to be 66,000 ± 3000 M⁻¹ in pH 7.2 PBS at rt (Figure S14), 42% lower than that observed with AFCA alone in PBS. This finding is expected, as the presence of the bound target is likely to limit Xe accessibility to the cryptophane interior. We note that this decrease in Xe affinity is not due to a change in the aggregation state, as the system remains monomeric at these concentrations of AFCA and β-CD (Figure S13). Additionally, it is likely that hydrophilic targets such as β-CD may serve to attract water molecules to the periphery of the cage and thus facilitate access of water to the cryptophane interior, thereby lowering Xe affinity.

Overall, this study sheds additional light on the complexities of cryptophane aggregation phenomena. By using AFCA as a model system, we determined that even small aggregates of several cryptophane units can significantly limit observation of the ¹²⁹Xe NMR signal, as evidenced by the necessity to decrease AFCA concentration below the normal working regime in order to be able to observe the ¹²⁹Xe NMR signal. This is an important finding that may help to explain the lack of a signal for cryptophane-based sensors with experimentally determined high Xe affinity. For example, in our earlier studies with C8B, we synthesized additional CA-targeted biosensors with varying numbers of carbons between the triazole and the benzenesulfonamide moieties.¹² Some of these biosensors gave much stronger ¹²⁹Xe–cryptophane NMR signals than others for reasons that were unknown at the time, but now are suspected to be because of varying cryptophane aggregation states. Additionally, this study illustrates the possibility of designing turn-on Xe sensors that report on changes in their chemical environment, such as an increase in pH, via a Xe-bound signal. Such turn-on biosensors would be advantageous in ¹²⁹Xe NMR/MRI applications, as the absence of signal corresponding to “free” biosensor would facilitate spectral analysis. Finally, this system provides greater understanding of protein biosensing, as previously it was challenging to study protein binding effects because of high background fluorescence and to separate aggregation effects from NMR studies. We note that having a minimally aggregated sensor such as AFCA allows for more in-depth studies of cryptophane aggregation phenomena. These new insights about cryptophane solvation and aggregation, Xe accessibility, and sensor-analyte binding could thus drive the development of new classes of “smart” Xe sensors.

CONCLUSIONS

In summary, we have synthesized and characterized a novel AFCA to provide new insights into the nature of cryptophane aggregation and Xe–cryptophane binding. Using DLS, fluorescence quenching, and hp ¹²⁹Xe NMR spectroscopy, we determined that AFCA possesses several notable features, including being more monomeric in aqueous solution than previously studied trifunctionalized cryptophanes and exhibiting the highest affinity for Xe in water of any currently known cryptophane host. We observed a concentration-dependent effect on the direct detection ¹²⁹Xe NMR signal, which was only visible at low (15 μM) AFCA concentration, where the sensor was monomeric. We note that sensors with high Xe affinity such as AFCA are highly advantageous in this regard, as ¹²⁹Xe NMR signals can be observed at lower sensor concentrations, without the impedance of aggregation and relaxation-based mechanisms. Dissolving AFCA in 10 mM NaOH gave a strong Xe@AFCA peak via direct detection ¹²⁹Xe NMR at elevated (200 μM) cryptophane concentration. AFCA:β-CD studies revealed that β-CD binding to the adamantyl moiety does not result in a ¹²⁹Xe NMR chemical shift relative to free AFCA, as expected for a Xe sensor bound to a small target, and for which the aggregation state does not change upon target binding. These studies provide insights into the design of monomeric, high-affinity Xe–cryptophane sensors and give valuable cautionary notes with respect to effects of cryptophane aggregation on Xe-based sensing.