Chaperone-like chiral cages for catalyzing enantio-selective supramolecular polymerization† †Dedicated to Professor Jean-Marie Lehn on the occasion of his 80th birthday. ‡ ‡Electronic supplementary information (ESI) available. See DOI: 10.1039/c9sc02412c

Chiral organic cages can assist enantio-selective supramolecular polymerization through a catalyzed assembly (catassembly) strategy, like chaperones assist the assembly of biomolecules.


Introduction
Cage compounds are discrete assemblies with sophisticated and well-dened local environments, 1-3 making them an exceptional model for studying the fundamentals of molecular recognition 4-10 and many related applications. [11][12][13][14][15][16] For instance, cage catalysis has recently emerged as an important approach for mimicking enzymatic reactions by increasing the reaction rate and/or product selectivity of various types of reactions. [17][18][19][20][21] As functional porous materials, cage compounds possess unique solution-processability for homogeneous catalysis, 1 complementary with heterogeneous supramolecular catalysis through metal-organic frameworks [22][23][24] or covalent organic frameworks. [25][26][27] Extensive studies of cage catalysis have been conducted by using both metal-organic [28][29][30][31] and pure organic cages. [32][33][34][35] The catalytic strategies are versatile, and they include encapsulating the reactants or introducing functional groups either inside [36][37][38] or outside 39 the cage cavities. All these strategies are designed to realize the essential feature of cage catalysis: cages co-assemble with the reactants, and automatically release the resulting products aer facilitating the reactions. To this end, the initial interaction between the cage and reactants needs to be strong enough to control the reaction pathway and hence to increase the reaction rate and/or selectivity. Once the product forms, its interaction with the cage weakens, due to increased steric hindrance, resulting in the automatic separation of the cage and product. Because such an intricate, dynamic series of interactions is required, rationally developing new systems or new models of cage catalysis remains a challenge. [40][41][42][43] Here, we report a catalyzed assembly strategy to increase the assembly rate and enantio-selectivity of the supramolecular polymerization of porphyrins by using acid-stable chiral organic cages. A comparison between this strategy and the reported cage catalysis for covalent reactions is depicted in Fig. 1A and B. Both strategies share two important features, namely the increased efficiency and/or selectivity of a reaction or an assembly process, and the automatic release of products aer the reaction. However, in the present study, cages facilitate the formation of non-covalent assemblies instead of covalent molecules, similar to the way chaperones assist the assembly of biomolecules. 44,45 In addition to showing the chaperone-like behavior of cage compounds, we reveal the kinetics of the autodetachment of cages and the chirality growth of the assemblies using spectroscopic characterization studies. Through comparison experiments with several cages and small molecules, we demonstrate that the passivation groups attached to the cages are important for maintaining the structural integrity of the cages during catalyzed assembly, and that their steric geometry can profoundly impact the chiral transfer between cages and assemblies.

Molecular structures and schematic procedures of catalyzed assembly
Chiral imine cages were synthesized as previously reported, 46 via the condensation of four butylated truxene faces and six (R,R)-1,2cyclohexanediamine (CHDA) chiral vertices (structures shown in Fig. 1C and the reaction schematic shown in Fig. S1A ‡). Condensation at room temperature generates two isomers of face-rotating octahedra, which are named aer the directionalities of the exterior faces as AAAA-1 and CAAA-1. These cages possess excellent stability, allowing us to isolate them by chiral high-performance liquid chromatography (HPLC) and their individual structures have been conrmed by single-crystal X-ray diffraction (CCDC# 1406534 and 1406540). 46 The isolated Tsymmetric AAAA-1 and C 3 -symmetric CAAA-1 have very similar structures, except for the small differences in the arrangements of the butyl chains. These similarities enable close comparison between the complexes when they are used as catalysts. The structures of imine cages used for comparison experiments, including isomers of cage 2 with (S,S)-CHDA vertices, cage 3 with phenyl faces, cage 4 with ethylated truxene faces, and cage 5 with ethylenediamine (EDA) vertices, are shown in Fig. S1B-E. ‡ Traditional supramolecular polymerization of tetrakis(4sulfonatophenyl)-porphyrin (TPPS, structure shown in Fig. 1C) was selected as a representative non-covalent assembly system [47][48][49][50][51][52][53][54][55] to be catalyzed by chiral cages. Supramolecular polymerization of TPPS has previously been regulated through a co-assembly method by adding oppositely charged chiral auxiliaries, such as amines 56 and amino acids. 57 Learning from the cage catalysts for covalent reactions, we consider that a stronger interaction with the reactant and weaker interaction with the product are the key to achieving the self-release of cages, and thus allows the recycling of cages aer catalysis.
We envisioned that the chiral truxene imine cage 1 might be a potential candidate for catalyzing the enantioselective supramolecular polymerization of TPPS, because the chiral imine groups in cage 1 are less basic than the amine groups in (R,R)-CHDA, and hence they exhibit weaker interactions with the sulfonic groups in TPPS. Therefore, unlike the co-assembly of (R,R)-CHDA and TPPS, 56 cage 1 and TPPS might not co-assemble in the nal product, which is a necessary condition for achieving catalyzed assembly. In addition, the butyl chains on the outer surfaces of the cages provide stronger steric hindrance to the TPPS assemblies over the monomers, 46,58 thus causing relatively stronger interactions with monomers and weaker interactions with assemblies (detailed in Fig. S2 ‡). This could lead to the automatic release of products from the cages aer the enantioselective polymerization of TPPS (Fig. 1D).

Spectroscopic characterization of chiral supramolecular polymerization
To induce supramolecular polymerization, we add ethyl acetate, with or without imine cages, to the stock solution of TPPS in methanol to attain a solvent ratio of 9 : 1 (ethyl acetate : methanol, v/v). The self-assembly of TPPS occurs very slowly when the concentration of TPPS is 10 mM, showing no perceptible aggregation in two weeks. By contrast, when an equivalent amount of CAAA-1 was introduced into this system, obvious precipitation was observed in 48 hours. Circular dichroism (CD) and UV-vis spectroscopy were used to show the catalytic behavior of CAAA-1 for regulating the enantioselective supramolecular polymerization of TPPS.
A CD spectrum recorded immediately aer mixing TPPS and CAAA-1 shows relatively weak and broad peaks at 340 nm and 400 nm ( Fig. 2A), whereas the corresponding UV-vis spectrum Coupling reactions can occur either inside or outside the cage cavity (still having interactions with the cage) as indicated by the red and blue balls, respectively. (B) Illustration of the cage-catalyzed strategy for non-covalent molecular assemblies. A cage assists the supramolecular recognition of two helical polymer chains, showing a novel catalytic mode for forming non-covalent bonds, in a way that resembles chaperone-assisted protein assembly. (C) Molecular structures of two isomers of cage 1, i.e., CAAA-1 and AAAA-1, and the porphyrin TPPS. In the cage structures, butyl groups are only shown on one of four faces, for clarity. In complete structures, each sp 3 carbon of the truxene backbones is connected to two n-butyl groups, i.e., 24 butyl groups for a cage. Rotation patterns along the three sp 3 carbons of the truxene backbones are either clockwise (C) or anticlockwise (A) when viewed from outside the octahedra, as indicated by the blue (C) and red (A) lines. (D) Schematic illustration of enantioselective supramolecular polymerization of TPPS catalyzed by CAAA-1. Note that the models show a reversible process on the cage vertices as indicated by the color change from pink to brown, and then back to pink in the final step. This illustrates how we can monitor the different stages of the catalyzed assembly process, using spectroscopy, and confirm the recovery of the cage after it catalyzes the formation of TPPS assemblies.
shows the typical absorption band of the protonated TPPS monomer at 417 nm (Fig. 2B). 56,59 The CD and UV-vis spectra of the TPPS and CAAA-1 mixture at 0 hours are dramatically different from the sum of their individual spectra ( Fig. S3 ‡), indicating that the interaction between TPPS and CAAA-1 occurs immediately aer mixing. Aer 48 hours, the formation of chiral TPPS H-assemblies was indicated by the strong excitoncoupled CD peaks at 400 nm and 453 nm and the corresponding UV-vis absorption peak at 419 nm ( Fig. 2A and B). The chirality of the TPPS H-assemblies can be empirically assigned as righthanded based on the positive cotton effect in the CD spectrum. 47,56,59 Meanwhile, the appearance of the typical peaks of CAAA-1 in the CD spectrum (340 nm) and the absorption spectrum (328 nm) recorded at 48 hours, suggests that CAAA-1 separated from the TPPS assemblies.
We further investigated the kinetics of the growth of TPPS assemblies and the release of the cage ( Fig. 2C and D). As analyzed in Fig. 2E and F, the cage was almost completely released from TPPS aer 24 hours. However, UV-vis spectra show continuously decreasing absorbance of the TPPS monomer at 417 nm, suggesting the further formation of TPPS assemblies from monomers and/or oligomers. During this process, the CD intensity of TPPS assemblies continues to grow until reaching its maximum at 48 hours. The chirality growth of the TPPS assemblies from 24 h to 48 h could be caused by the cage catalyst, and may also be facilitated by the previously formed chiral TPPS assemblies through self-propagation, 60 as shown in Fig. S4. ‡ To directly compare the catalyzed assembly with the selfassembly of TPPS, we increased the TPPS concentration from 10 mM to 50 mM. Under this condition, the self-assembly of TPPS took approximately 72 hours to show perceptible aggregation in solution (Fig. S5A ‡). In contrast, aer introducing 50 mM CAAA-1, obvious precipitation was observed in 30 min. UV-vis spectra show that TPPS molecules form J-assemblies in the self-assembly product and form H-assemblies when CAAA-1 is added (Fig. S5B ‡), indicating that CAAA-1 can both accelerate the assembly of TPPS monomers and control the selectively of assembly products.
In addition, we have performed a concentration-dependent study on the cage catalysis as shown in Fig. S6. ‡ In the systems with a consistent 10 mM concentration of TPPS, 10 mM CAAA-1 results in the highest CD intensity of the TPPS H-assemblies. Increasing the concentration of CAAA-1 decreases the CD intensity of TPPS assemblies, probably because a high concentration of the cage removes more protons from TPPS and thus causes stronger p-p interaction between TPPS molecules and relatively weakens the interaction between TPPS and CAAA-1, leading to a fast self-assembly process of TPPS with decreased chiral regulation from CAAA-1. For instance, aggregation in the 40 mM CAAA-1 system is obvious aer 1 hour, whereas it takes over 12 hours for the 10 mM CAAA-1 system to show obvious aggregation. Decreasing the concentration of CAAA-1 also leads to a decreased CD intensity of TPPS assemblies as less CAAA-1 acts as the catalyst. We found the minimum concentration of the catalyst required to ensure the supramolecular polymerization of TPPS in our current experimental setup to be around 8 mM. When the concentration of CAAA-1 goes lower than this limitation, CAAA-1 starts to decompose, leading to poor reproducibility of the spectroscopic results. Note that this limitation is caused due to the hydrolytic decomposition of the cage, which is strongly related to the experimental conditions, especially the concentration of trace water in the system. We assume that the limitation could be further pushed towards lower concentrations if the trace amount of water can be reduced from the solvents and environments.

Recovery of the cage catalyst
To further conrm the self-detachment of the cage, we characterized the isolated precipitate and the supernatant of the assembly product (TPPS : CAAA-1 ¼ 1 : 1, 10 mM) by CD, dynamic light scattering (DLS), and scanning electron microscopy (SEM). As shown in Fig. 3A, the CD spectra of the precipitate and the thoroughly stirred mixture overlap well in the region of TPPS assemblies, whereas the supernatant shows the typical spectrum of pure CAAA-1, without any signal from TPPS monomers or assemblies, suggesting that all TPPS assemblies have been precipitated from the solution. The CD spectrum of pure CAAA-1 at 10 mM is almost identical to that of the supernatant (Fig. 3B), which conrms the full recovery of the cage catalyst aer 48 hours. DLS shows that the formed TPPS assemblies are over 100 nm (Fig. S7 ‡). The SEM image of the precipitate further conrms that TPPS molecules have been assembled into nanorods and nanosheets with a corresponding diameter or thickness, respectively, of about 100 nm (Fig. 3C). The SEM image of the supernatant shows the octahedral shape of the cage single crystals (Fig. 3D), proving the purity of the cage in the supernatant.
Based on the above spectroscopic and SEM studies, we conclude that when the cage CAAA-1 is introduced into the solution of TPPS, it interacts with TPPS immediately and thus increases the efficiency of supramolecular polymerization in terms of the assembly rate and the enantioselectivity of the product. Aer the formation of chiral TPPS assemblies, all cage molecules are completely restored to their initial states and can catalyze another round of supramolecular polymerization of TPPS. Therefore, we consider CAAA-1 a chaperone-like selfdetachable template for the enantioselective supramolecular polymerization of TPPS.

Acid stability of the truxene-CHDA imine cage
It is interesting that the absorption spectrum of CAAA-1 exhibits weak and broad peaks aer being mixed with TPPS and that it reverts to its original, intense peak at 340 nm aer the separation from TPPS ( Fig. 2A and C). We conjectured that this reversible spectroscopic change is related to the interaction between the acid groups in TPPS and the imine groups in the  Fig. 2A). (C-E) Mass spectra of CAAA-1 (10 mM) and two comparison imine cages 3 and 5 (10 mM) before (blue) and after (red) the addition of 10 equivalents of TFSA. TR groups in the inserted structures of 1 and 5 indicate the truxene faces with perpendicular butyl chains (detailed in Fig. 1C and S1 ‡). (F) Single-crystal structure of CAAA-1 showing that the imine bonds are shielded by the outer cyclohexyl and butyl groups. The comparison of more cages and imine compounds in Fig. S9 and S10 ‡ shows that both the cyclohexyl and butyl shields and the cage shape are necessary for the extraordinary acid stability of cage 1. cage, i.e., the protonation of the imine cage. To verify this conjecture, we studied the protonation of the cage under strong non-chromophore-containing triuoromethanesulfonic acid (TFSA). As shown in Fig. 4A, upon adding 100 mM TFSA to 10 mM CAAA-1 in ethyl acetate, the CD spectrum red-shis to 400 nm immediately due to the protonation of the imine groups. Moreover, an additional 100 mM KOH can bring the CD spectrum fully back to its original shape. Similar reversible spectroscopic changes are also found in the uorescence measurement (Fig. S8 ‡). Titration of TFSA with a CAAA-1 solution gives a series of CD spectra showing the full range from the unprotonated to fully protonated states (Fig. 4B). The spectrum corresponding to a 1 : 4 ratio of cage : TFSA has a very similar shape to that of the spectrum of the 1 : 1 mixture of cage and TPPS (note that each TPPS monomer has four sulfonic groups), suggesting the partially protonated states of the cage during the initial stage of catalyzed assembly.
Mass spectroscopy further conrms that CAAA-1 maintains the cage structure when the concentration ratio of TFSA to CAAA-1 is 10 : 1 in ethyl acetate (Fig. 4C), showing extraordinary acid stability over other imine cages and imine-containing molecules under the same conditions. In controlled experiments, we have synthesized imine cages with similar [4 + 6] structures by changing the butyl-substituted truxene faces in cage 1 into unsubstituted phenyl faces (cage 3 in Fig. 4D and S1C ‡) or ethyl-substituted truxene faces (cage 4 in Fig. S1D ‡). Both cages 3 and 4 are decomposed into fragments under the same TFSA treatment (Fig. 4D, S9A and B ‡). A similar decomposition occurs for cage 5 (Fig. 4E), which has ethyl vertices instead of the cyclohexyl vertices in cage 1. Moreover, even though they consist of the same truxene and cyclohexyl components, the [1 + 1] condensation compound of butylsubstituted truxene and CHDA (structure shown in Fig. S10 ‡) still decomposes aer the addition of TFSA, showing lower stability than the [4 + 6] cage counterpart. Imine bonds in all these structures have similar nucleophilicity, yet cage 1 exhibits the highest acid stability among them. This suggests that the cage shape and the passivating effects of cyclohexyl and butyl groups are both critical for protecting the imine bonds from decomposition (Fig. 4F), resembling the strategy of protecting soldiers with shields and a "testudo formation".

Structural sensitivity to chiral transfer
We also perform the catalyzed assembly of TPPS by using the isomer AAAA-1, which has a very similar structure to CAAA-1, except for different rotational directions of one of the truxene faces and one of the related butyl groups. Therefore, we examine whether the steric geometry of truxene faces can inuence the chiral transfer from the cage to TPPS assemblies. As shown in Fig. 5A and S11, ‡ both AAAA-1 and CAAA-1 experience a similar protonated to deprotonated pathway during catalysis. However, AAAA-1 exhibits a much lower efficiency for controlling the chirality of TPPS assemblies than does CAAA-1. Aer 48 hours, AAAA-1 leads to chiral TPPS assemblies with a 10 mdeg peak in the CD spectrum, which is approximately 15 times lower than that generated by CAAA-1. In addition, we also synthesized imine cages with (S,S)-CHDA vertices, and isolated two cage products CCCC-2 and CCCA-2, which are enantiomers of AAAA-1 and CAAA-1, respectively. As expected, CCCC-2 and CCCA-2 generate mirror-like CD spectra compared with their enantiomers respectively (Fig. 5A).
Due to the dynamic nature of imine chemistry, it is possible that a tiny amount of cage 1 could hydrolyze and generate (R,R)-CHDA, and even we have not measured it through spectroscopic comparison (Fig. 3B). To make sure that the observed chiral catalysis is not an artifact phenomenon caused by partially hydrolyzed (R,R)-CHDA, we performed comparison experiments of the co-assembly of TPPS with (R,R)-CHDA in different concentrations. Since each CAAA-1 contains six (R,R)-CHDA vertices, 6, 1, and 0.1 equivalents of (R,R)-CHDA correspond to the fully hydrolyzed, partially hydrolyzed, and barely hydrolyzed conditions of CAAA-1. The results show that both six equivalents showing the different models of the interaction between TPPS and the imine in cage 1 or the amine groups in (R,R)-CHDA. It is difficult for AAAA-1 to transfer its chirality to TPPS assemblies, due to strong steric hindrance in all directions. CAAA-1 transfers its chirality to TPPS assemblies through the interaction in the least hindered direction, i.e., the left. (R,R)-CHDA transfers its chirality to TPPS assemblies through the interaction in the least hindered direction, i.e., the bottom. or one equivalent of (R,R)-CHDA leads to the opposite chirality compared to one equivalent of CAAA-1 (Fig. 5B and S12 ‡), whereas 0.1 equivalent of (R,R)-CHDA is insufficient to control the chirality of the TPPS assemblies. The chirality inversion between cages and their CHDA components rules out the assumption that the chiral regulation during catalyzed assembly is caused through chiral self-propagation by a trace amount of CHDA decomposed from cage vertices.
We scrutinize the structures of AAAA-1, CAAA-1, and their vertex component (R,R)-CHDA and propose a mechanism for their different catalytic efficiencies (Fig. 5C-F). Having a T symmetry, every CHDA vertex in AAAA-1 connects to two anticlockwise truxene faces with their butyl chains pointing in both le and right directions (Fig. 5C). Therefore, there is a strong steric hindrance in both directions when TPPS or TPPS assemblies try to interact with the chiral vertices (Fig. 5E). By contrast, in CAAA-1, there are three CHDA vertices connected to an anticlockwise and a clockwise truxene face, with their butyl chains pointing in the same direction (Fig. 5D). Thus, CAAA-1 can have a stronger interaction with TPPS and TPPS assemblies from the direction that is not blocked by the butyl chains (Fig. 5F). Compared with imine cage 1, (R,R)-CHDA has more basic amine groups and much less steric hindrance. Therefore, it forms co-assemblies with TPPS instead of showing catalytic behavior. 56 In addition, when interacting with TPPS, (R,R)-CHDA has a large steric hindrance in the direction of the cyclohexyl group (Fig. 5G), which is different from that in a cage. This may result in different chiral selectivities between imine cages and their CHDA counterparts.

Conclusion
To summarize, we have found acid-stable chiral imine cages for catalyzing the enantioselective supramolecular polymerization of TPPS. During the catalysis, chiral imine cages regulate the chirality of TPPS assemblies and serve as spectroscopic probes to allow monitoring of the assembly and self-detachment kinetics. This study provides a strategy to construct chiral supramolecular polymers from achiral building blocks through cage catalysis, which also entails efforts towards extending the applications of cage catalysts from covalent interactions to noncovalent molecular assembly. The concept to design a stronger interaction with monomers and a weaker interaction with assemblies is general for the development of this new type of catalysis in the eld of molecular assembly. In addition, the "testudo formation" strategy in constructing acid-stable imine cages may inspire other designs for applications that require stable cages. 61,62

Conflicts of interest
There are no conicts to declare.