Anion detection has proven to be of significant importance due to the critical roles these negatively charged species play in nature. Anions are involved in mediating essential processes such as metabolic regulation [
1–
3], aquatic eutrophication [
4–
6], and the maintenance of physiological homeostasis [
7–
10]. A widely adopted strategy for anion recognition involves designing macrocyclic hosts with intrinsic cavities that can accommodate anionic guests through noncovalent interactions, particularly hydrogen bonding, in a synergistic manner [
11–
15]. Recently, inspired by the enhanced binding affinities of cryptands toward cationic guests compared to their two-dimensional counterparts (
e.g., crown ethers) [
16,
17], chemists have discovered that cage-shaped molecules often exhibit stronger anion-binding capabilities than macrocyclic analogues [
18–
24]. A notable example of this was reported by Flood and co-workers, who synthesized a cage,
via a click reaction, that was capable of extracting chloride anions from aqueous solution [
25]. However, a frequent drawback of these irreversible synthetic approaches is the generation of oligomeric or polymeric byproducts during the formation of the target cage molecules.
Reversible covalent reactions provide a promising alternative by enabling error correction during self-assembly [
26–
30]. Nevertheless, this dynamic nature can compromise the robustness of the self-assembled molecules. For instance, acceptors containing imine bonds, a commonly used dynamic covalent bond, are prone to hydrolysis in aqueous media [
31–
33]. Therefore, the development of synthetic strategies that balance the error-correcting benefits of reversible reactions with high product stability remains a key challenge. As demonstrated in prior research [
34], a cage-shaped molecule was synthesized in near-quantitative yield by condensing a triscationic hexaformyl precursor with three equivalents of bisamine linker. Despite the presence of six imine bonds in its framework, the cage exhibits remarkable kinetic stability against hydrolysis in aqueous media, compared with those compounds each contains only one single imine bond. This stability arises from the fact that disrupting one of the three bisamino building blocks requires breaking two imine bonds simultaneously, a process that is significantly more energy-demanding than cleaving a single bond. Leveraging the hydrophobic effect, the cage can encapsulate alkane derivatives within its intrinsic cavity, where all the aromatic building blocks in the cage framework are oriented in a face-in configuration. Building on this, we hypothesize that by reorienting the cationic pyridinium units into an edge-in arrangement, the cage could potentially recognize anionic guests through hydrogen bonding interactions.
Herein, we report the self-assembly of two structurally analogous chiral cationic cages via the condensation of 3 equiv. of a bisamine linker with 1 equiv. of a hexaformyl compound, affording the desired products in near-quantitative yields. Within the framework of each cage, three acidic CH protons are oriented toward the interior cavity or the clefts, facilitating the accommodation of anionic guests through the formation of synergistic hydrogen bonds exhibiting modest binding affinities in both organic and aqueous media.
Two triscationic hexaaldehydes,
2a3+·3PF
6‒ and
2b3+·3PF
6‒, were synthesized
via S
N2 reactions between tris(bromomethyl)benzene and the corresponding pyridine or imidazole precursors, followed by counterion exchange from bromide (Br
‒) to hexafluorophosphate (PF
6‒). To construct the cage
R-
3a3+,
2a3+·3PF
6‒ and (1
R,2
R)-(-)-1,2-diaminocyclohexane ((
R,R)-CHDA) were combined in a 1:3 molar ratio in CD
3CN (
Fig. 1).
The mixture was heated at 40 ℃ for at least 12 h, after which the 1H NMR spectrum revealed a set of well-defined resonances corresponding to the formation of the cage molecule R-3a3+. No significant byproducts were observed in the spectrum. Upon addition of tetrabutylammonium chloride (TBA+·Cl‒) to the solution, R-3a3+·3Cl‒ precipitated and was collected by filtration in 90% yield relative to the aldehyde precursor 2a3+·3PF6‒. The isolated 3a3+·3Cl‒ was soluble in D2O, and its ¹H NMR spectrum confirmed its stability in aqueous media, with no observable degradation during either counterion exchange or dissolution (Fig. S22 in Supporting information). Addition of ammonium hexafluorophosphate (NH4+·PF6‒) to the aqueous solution resulted in the precipitation of R-3a3+·3PF6‒ in 95% yield relative to R-3a3+·3Cl‒. Dissolving R-3a3+·3PF6‒ in CD3CN yielded a ¹H NMR spectrum identical to that of the initial self-assembly solution, confirming the integrity of the cage structure (Fig. S17 in Supporting information). The enantiomer S-3a3+·3Cl‒ and S-3a3+·3PF6‒ were synthesized following a similar procedure, using (S,S)-CHDA as the amino precursor. Similarly, analogous reactions using imidazolium-based precursor 2b3+ afforded R-3b3+·3PF6‒ and R-3b3+·3Cl‒, as well as their enantiomers S-3b3+·3PF6‒ and S-3b3+·3Cl‒. All cages were obtained in high yields, a result attributed to the dynamic nature of the imine bonds, which facilitates error correction during the self-assembly process.
The planar chirality of the cages is induced by the point chirality of the bisamino building blocks, specifically (
R,
R)-CHDA or (
S,
S)-CHDA. This chirality was confirmed through circular dichroism (CD) spectroscopy. In the CD spectra recorded in water, the cages
S-
3a3+·3Cl
‒ and
S-
3b3+·3Cl
‒ displayed (
Fig. 2) mirror-image signals compared to their enantiomers,
R-
3a3+·3Cl
‒ and
R-
3b3+·3Cl
‒, respectively.
Diffraction-quality single crystals of
R-
3a3+·3PF
6‒,
S-
3a3+·3PF
6‒ and
R-
3b3+·3PF
6‒ were successfully obtained by vapor diffusion of isopropyl ether into their respective acetonitrile solutions (CCDC 2363838, 2363601 and 2308786, respectively). The resulting solid-state structures unequivocally confirmed the formation of the cages (
Fig. 3 and Fig. S118 in Supporting information). As anticipated, the frameworks of
R-
3a3+ and
S-
3a3+ are mirror images of each other, reflecting their enantiomeric relationship. In both the
R/
S-
3a3+ and
R-
3b3+ frameworks, each pyridinium or imidazolium unit adopts an edge-in orientation. In the case of
R-
3b3+, two of the three imidazolium units are oriented in a near perpendicular manner relative to the corresponding phenyl unit. As a consequence, the acidic imidazolium protons are directed toward the interior cavity. This conformation favors the formation of inclusion complexes, in which the anionic guest is located within the cage cavity. In the case of
R-
3a3+, as a comparison, dihedral angle between each of the pyridinium unit and the corresponding phenyl unit is around 40°, indicating a
pseudo face-in orientation. This conformation allows
R-
3a3+ to accommodate guests in each of the three windows. This hypothesis is supported by the solid-state structure of
R/
S-
3a3+·3PF
6‒, in which the PF
6‒ counterions are located in the windows between the two pyridinium units. The cavity sizes of
3a3+·3PF
6‒ and
3b3+·3PF
6‒ were 135 Å
3 and 112 Å
3, respectively, calculated
via GHECOM [
35]. In the solid state, the cavities of both cages are devoid of any solvent molecules or PF
6‒ counterions.
We then investigated the possibility of using the cages as acceptors to accommodate anions. First, based on the chiral nature of the cages, we investigated the recognition capabilities for chiral anions (
e.g., alanine or lactate). Unfortunately, no recognition was observed toward chiral anions, probably because none the chiral anionic guests had complementary size relative to the cage cavity. Therefore, we turned our attention to the achiral anions with smaller sizes such as halides. We evaluated the anion-binding capabilities of
3a3+ and
3b3+ in organic solvent, namely CD
3CN or CD
3SOCD
3. The
1H NMR spectra of the cages before and after the addition of tetrabutylammonium salts (TBA
+·X
‒, X = Cl, Br, I and NO
3) were recorded and compared, indicating these anions were encapsulated within the cage cavities. For example, upon addition of TBA
+·Cl
‒ into a solution of
3a3+·3PF
6‒ and
3b3+·3PF
6‒ in CD
3CN, a number of resonances of the cage undergo shifts (
Fig. 4). At the early stage of titration when Cl
‒ added was <3 equiv. relative to the hosts, the resonance corresponding to protons H
c and H
h of
3a3+ and protons H
c and H
g of
3b3+, which are directed toward the interior cavity, confirm that the anionic guests are recognized. The resonances corresponding to other protons that are located in relatively external positions in the cage framework underwent less remarkable shifts. These observations confirm that the anionic guests are recognized within the cage cavity, driven by hydrogen bonding interactions and electrostatic forces. In the case of
3b3+, the shifts of resonances became less remarkable when the amount of Cl
‒ added is in excess. For example, the
1H NMR spectra of
3b3+·3PF
6‒ in the presence of 13.8 and 18.8 equiv. of TBA
+·Cl
‒ were similar (
Fig. 4c), indicating that most of the host cavities are occupied by the anionic guest. As a comparison, in the case of
3a3+, the saturation behavior did not occur even after adding 27.5 equiv. of Cl
‒ (
Fig. 4b, top). It is noteworthy that the resonances corresponding to H
g and H
j, which barely shift in the early stage of anion titration, underwent downfield shift in the presence of excess Cl
‒. This observation indicated that
3a3+ might have multiple binding modes towards Cl
‒ during titration. That is, in the early stage of anion titration, the cage employed its cavity to recognize anionic guest, while in the late stage, the anions were recognized in a peripheral manner, leading to the shifts of the resonances corresponding to the external protons.
Job plot analyses were conducted to determine the binding stoichiometry between the cages and anionic guests. The Job plot of
3a3+ and
3b3+ in CD
3CN revealed markedly distinct binding modes:
3a3+ exhibited a 1:2 stoichiometry with all investigated anions, whereas
3b3+ adopted a 1:1 binding ratio (
Fig. 5). This difference is consistent with the NMR titration experiments, namely that
3b3+ can only form the 1:1 inclusion complexes by using the cavity to encapsulate the anions, while
3a3+ can form both inclusion and peripheral complexes by utilizing the windows as secondary binding sites.
The corresponding binding constants (
Ka) were calculated
via Bindfit. Using a 1:2 binding model might be more reasonable for
3a3+. However, the attempt to do so only provided some unreliable secondary binding constants (
Ka2), probably because the
Ka2 values are too small to be determined. We thus used the 1:1 binding model to calculate the
Ka for both
3a3+·3PF
6‒ and
3b3+·3PF
6‒ in both CD
3CN and CD
3SOCD
3, which are summarized in
Table 1. In the case of both cages,
Ka values are significantly larger in CD
3CN compared to those in CD
3SOCD
3. For example,
Ka was measured to be 1.5 (±0.1) × 10
3 L/mol for the cage
3a3+ to recognize Br
‒ in CD
3CN, while
Ka was measured to be 1.0 (±0.1) × 10
2 L/mol in CD
3SOCD
3, representing a decrease of more than one order of magnitude. This is predictable, given that more polar solvent, CD
3SOCD
3, helps to suppress both hydrogen bonding and electrostatic interactions. The apparent binding constants
Ka of cage
3a3+ for Br
‒ and I
‒ are significantly higher than those for Cl
‒, while cage
3b3+ shows slightly lower
Ka values for Br
‒ and I
‒ compared to Cl
‒. This is probably because cage
3b3+ possesses a smaller cavity size compared to
3a3+, while Br
‒ and I
‒ exhibit significantly larger ionic radii than Cl
‒ [
36,
37].
We also investigated the possibility of using
3a3+·3Cl
‒ and
3b3+·3Cl
‒ to recognize highly hydrated anions such as Cl
‒, Br
‒ and NO
3‒ in aqueous media. We attempted to obtain the Job plot in D₂O, but this approach failed due to the weak binding constant between the cage and anionic guest in D₂O [
38]. However, the corresponding binding constants measured in water proved to be much weaker compared to those in organic media. For example, addition of NaCl into a solution of
3a3+·3Cl
‒ led to little or no shifts in the resonances of the cage, indicating the apparent binding constant
Ka is too low to be determined (Fig. S104 in Supporting information). An alternative explanation, that the cavity of
3a3+ is already occupied by a Cl
‒ ion due to an extraordinarily high
Ka, is considered unlikely, as the
Ka in CD
3CN is only 8.7 (±0.1) × 10
2 L/mol. Both cages exhibit modest
Ka values for Br
‒ and NO
3‒. Attempts to determine the
Ka value for I
‒ were unsuccessful, given that addition of I
‒ into the D₂O solutions of these two cages led to precipitation.
In summary, two sets of enantiomerically pure, cage-shaped molecules were successfully self-assembled in near-quantitative yields via dynamic imine condensation, bypassing the need for stepwise synthesis. Both cages demonstrate exceptional robustness, enabling their isolation as pure solid-state materials via counterion exchange without detectable degradation. The cationic building blocks, either pyridinium or imidazolium, adopt different conformations within the cage frameworks, namely that the imidazolium units are oriented in an edge-in manner, favoring the formation of an inclusion complex, while the pyridinium units are oriented in a pseudo face-in manner, allowing the cage to utilize both the cavity and the windows to accommodate anionic guests. As a consequence, the two cages respectively adopt distinct 1:1 and 1:2 binding stoichiometries. While binding affinities for highly hydrated anions remain modest in aqueous media, future efforts will focus on structural modifications, such as incorporating additional hydrogen bond donors or expanding cavity dimensions to accommodate larger guests. These investigations are currently underway in our laboratory.
PDF
Full
93
51
关闭全屏
No. 86 Xueyuan South Road, Haidian District, Beijing
010-62662699(Operation and Technical Support)
qkjq@cast.org.cn
