Chiral Amine Covalent Organic Cage Lingated with Copper for Asymmetric Decarboxylative Mannich Reaction

Abstract

The efficient employment of chiral porous organic cages (POCs) for asymmetric catalysis is of great significance. In this work, we have synthesized a chiral N-rich organic cage constructed through chiral (S, S)-1,2-cyclohexanediamine and benzene-1,3,5-tricarbaldehyde utilizing dynamic imine chemistry according to the literature. Following reduction with NaBH₄, the resulting amine-based POCs (RCC3) feature appended chiral diamine moieties capable of coordinating Cu²⁺ cations. This Cu²⁺ coordination provides RCC3 with excellent enantioselectivity as a supramolecular nanoreactor in asymmetric decarboxylative Mannich reactions, providing up to 94% ee of the product. We found that the spatial distribution of chiral amine sites and the coordination of Cu²⁺ in the RCC3 have a significant impact on catalytic activity, especially enantioselectivity. This work provides insights into the structure–function relationship within supramolecular catalytic systems

1. Introduction

In the past decade, imine bond-based covalent organic cages have been among the most popular [1,2,3,4,5]. A typical example was the well-known privileged chiral [4 + 6] cycloimine cage CC3 [6] reported by Cooper which has been widely employed in fields such as molecular separation [7,8], gas storage [9], and guest encapsulation [10,11,12,13,14]. Given that the imine bond can be reduced into robust amine analogs [15,16], the stability of the corresponding cages can be significantly improved; meanwhile, with an flexible amine bond, inherent tunable porosity, and attractive solubility behavior, the covalent organic cage should be an ideal host for substrate and catalytically active metal salts [17]. In this context, it probably forms a multi-functional catalyst in which the amine bond can induce the preorganization of a substrate while the diamine-ligated metal salt can be used for other metal activation processes.

The asymmetric decarboxylative Mannich reaction (DMR) between β-ketoacids and imines serves as an excellent model for evaluating dual-catalyst systems involving amine cages combined with metal salts. This reaction necessitates the cooperative action of stereochemical regulation and catalytic activation [18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36]. Over the past several decades, the asymmetric catalysis of the DMR has primarily followed two main pathways. The first involves chiral organocatalyst-mediated asymmetric catalysis, where controlling the C=N bond geometry using a chiral organocatalyst is crucial for enantioselectivity. A representative mechanistic study by the Ma group [37] highlights this approach. Typically, the reaction is conducted at low temperatures (around −20 °C) to enhance enantiomeric excess (ee). The second pathway is a substrate-induced diastereoselective DMR, which proceeds under milder conditions. For instance, the Tian group [38] developed a highly diastereoselective DMR using optically active 2-(tert-butanesulfinyl-imino)glyoxylates as substrates at room temperature. Similarly, the Wang group [39] reported that sulfonamides induced the dimerization of chiral thiourea macrocycles via hydrogen bonding. This dimerization, in turn, activated the sulfonamide and preorganized it in a specific orientation, facilitating nucleophilic attack by β-ketoacids. The cooperative interaction between the two macrocycles significantly enhanced both reactivity and enantioselectivity. Beyond these two main pathways, recent advancements have expanded the scope of asymmetric DMR. The Ma group [40] reported a synergistic catalytic approach using chiral binaphthyl phosphoric acid and copper salts as a dual-catalyst system for the asymmetric DMR of β-ketoacids with α-arylenamides. Meanwhile, the Wang group demonstrated an asymmetric DMR facilitated by an anion–π interaction within a chiral molecular cage [41]. Despite these significant advancements, further exploration of more efficient catalysts based on well-established methodologies remains valuable for enantioselective DMRs. The resulting chiral β-amino ketones are important synthetic intermediates for natural products and biologically active compounds.

Encouraged by the advantages of covalent organic amine cages and the known DMR pathway, we have synthesized an amine covalent organic cage, RCC3, by employing aldehyde with (S, S)-1,2-cyclohexane diamine. The characteristics of this molecular cage reactor lie in the specific functions of chiral cyclohexanediamine: they can coordinate with Cu²⁺ cations to form highly active sites, thereby promoting the formation of intermediates for decarboxylation and the enolization of acetoacetic acid. Meanwhile, adjacent diamine groups provide multiple hydrogen bonds, which can anchor the 1,2,3-Benzoxathiazine-2,2-dioxide substrate at the position adjacent to Cu²⁺. This spatial advantage will facilitate stereo selective matching with enol intermediates, creating a complex dual site regulatory system that provides the possibility for efficient and enantioselective decarboxylation Mannich reactions.

2. Results and Discussion

The imine cage was first reported by Cooper and coworkers for gas adsorption [2]. It was obtained in a typical synthesis procedure: the aldehydes (L¹) and (S, S)-1,2-cyclohexane diamine were self-assembled by imine condensation in the solvent at room temperature (Scheme 1). Colorless needle-shaped crystals were isolated from the reaction mixture (73% for CC3). Further, the reaction with NaBH₄ yielded a reduced RCC3 nanocage. A sharp peak at 8.19 ppm in the ¹H NMR spectrum of CC3 confirms the formation of imine functionality (Synthesis 2.1 Supplementary Materials). The peak at 4.03 ppm is attributed to the aliphatic -CH₂- units of ethylenediamine, suggesting effective cross-condensation. The peaks at 3.7 ppm and 2.67 ppm in the ¹H NMR spectrum of RCC3 suggest the imine to amine transformation (Synthesis 2.2, Supplementary Materials). The conversion of CC3 to RCC3 was further corroborated through FTIR spectroscopy. The disappearance of the C=N peak at about 1636–1647 cm⁻¹ and the appearance of new peaks at 3122–3157 and about 1607 cm⁻¹ for N-H stretching and C-N-H bending vibration, respectively, indicate the formation of RCC3 (Figure S1, Supplementary Materials).

The targeted amine containers carry diamino donor pockets which potentially provide coordination and hydrogen bonding forces for the encapsulated substrates making them designable and programmable nanoreactors. The coordination capacity of those containers was tested by investigating the binding stoichiometry of the containers (RCC3) toward Cu²⁺ through UV-Vis titration (Figure 1). The Job plots of RCC3 with Cu²⁺ at 288 nm display maximum emission at [Cu²⁺]/{6[RCC3] + Cu²⁺} = 0.4 (Figure 1), indicating that a 1:4 inclusion copper compound-coordinated RCC3 porous cage (RCC3-Cu) was synthesized by the self-assembly of the reduced RCC3 with Cu²⁺ in THF.

Asymmetric organocatalysis: The decarboxylative Mannich reaction (DMR), involving a decarboxylative nucleophilic addition to a 1,2,3-Benzoxathiazine-2,2-dioxide, is a general strategy used in the construction of β-sulfonamido ketones, and previous explorations have enabled some enantioselective DMR processes [5]. In particular, the use of a chiral Cu/bisoxazoline complex as a DMR catalyst has exhibited unique advantages [6]; the Ma group utilized it to successfully realize an enantioselective DMR conversion of cyclic ketimines and β-keto acids into the chiral β-sulfonamidoketones [6a]. The unique combination of solution processability and inherent internal voids makes RCC3 an ideal platform for highly active catalysts. These properties allow efficient access to catalytic centers and promote the transport of reactants and products. The fruitful chiral diamine units endow RCC3 with the enzyme-mimicking chiral cavity microenvironments which could facilitate the asymmetric aldol reaction.

Our initial investigation began with screening various reaction conditions (

Table 1. Optimization of DMR process of 1a and 2a ^a.

Entry	Cat.	Loading (mol%)	Time	Yield b/%	ee c/%
1	RCC3: Cu(OAc)2 (1:1)	2.5	6	77	86
2	RCC3: Cu(OAc)2 (1:1)	5	6	79	91
3	RCC3: Cu(OAc)2 (1:1)	10	6	77	90
4	RCC3: Cu(OAc)2 (1:1)	5	6	86 d (68) e	78 d (57) e
5	RCC3: Cu(OTf)2 (1:1)	5	6	79	83
6	RCC3: CuBr (1:1)	5	6	78	81
7	RCC3: CuCl2 (1:1)	5	6	79	81
8	RCC3: CuI2 (1:1)	5	6	77	89
9	RCC3: Cu(OAc)2 (1:2)	5	6	78	55
10	RCC3: Cu(OAc)2 (1:3)	5	6	73	60
11	RCC3: Cu(OAc)2 (1:4)	5	6	80	55
12	RCC3: Cu(OAc)2 (1:5)	5	6	78	55
13	RCC3: Cu(OAc)2 (1:6)	5	6	79	57
14	RCC3: Cu(OAc)2 (1:1)	5	1	45	91
15	RCC3: Cu(OAc)2 (1:1)	5	2	62	88
16	RCC3: Cu(OAc)2 (1:1)	5	4.5	76	90

^a For reaction details see the synthesis section in the Supplementary Materials. ^b Isolated yield. ^c The ee was determined by HPLC analysis. ^d Catalyzed at 0 °C. ^e Catalyzed at r.t.

), and the model substrates of 1,2,3-Benzoxathiazine-2,2-dioxide (1a) and acetoacetic acid (2a). As shown in Table 1, entry 1–3, when the DMR was catalyzed under 5 mol% of RCC3 with Cu(OAc)₂, a 79% yield of the product was obtained with 91% ee under −20 °C. The ee value decreased dramatically when the temperature increased (entry 4). When copper salts of other anions were used together with molecular cages for catalysis, the enantioselectivity still remained at 81–89% (entry 5–8), indicating that anions have little effect on selectivity. However, the proportion of copper salts has a significant impact on the selectivity of the reaction (entry 9–13). As the amount of copper salts increases, the enantioselectivity decreases significantly, possibly due to the coordination between copper salts and amino groups hindering the spatial regulation of hydrogen bonds. Meanwhile, the dynamic monitoring of the reaction showed that the maximum yield was achieved after nearly 6 h of reaction (entry 14–16). After screening various reaction conditions, including catalyst loading, reaction time, temperature, and co-catalyst loading, we found that the best condition for this catalysis system was 5 mol% loading of RCC3/Cu(OAc)₂ (1/1) which catalyzed the decarboxylative Mannich reaction of 1,2,3-Benzoxathiazine-2,2-dioxide and acetoacetic acid to give a valuable β-amino ester at a 79% yield and 91% ee in THF at −20 °C after 6 h.

Subsequently, to demonstrate the generality of the chiral container catalysts, we investigated the substrate scope. It was found that 1,2,3-Benzoxathiazine-2,2-dioxide substrates bearing either electron-withdrawing or electron-donating groups (1a–1g) could react with β-ketoacid (2a), affording the corresponding products (3aa–3ga) at good yields (71–79%) and stereoselectivities (50–94% ee). Notably, electron-withdrawing substituents exhibited a lower enantioselectivity: 50% ee for 4-nitro 1,2,3-Benzoxathiazine-2,2-dioxide derivative (3ga) (Figure 2a). In the case of β-ketoacid substrates, enantioselectivity showed a dependence on both steric size and rigidity. When aliphatic β-ketoacids with varying steric hindrance and/or rigidity (2a–2e) were used, the reaction catalyzed by RCC3/Cu(OAc)₂ (1:1) provided the products (3aa–3ae) at a 74–79% yield and 71–87% ee (Figure 2b). Interestingly, enantioselectivity gradually decreased as the steric hindrance and/or rigidity of the alkyl chains increased. This trend was also observed in reactions involving β-ketoacids and ortho-, meta-, and para-substituted cyclic aldimines (Scheme 2), suggesting that the catalytic active site was located within the cavity of RCC3. All these results further indicate that the spatial positioning of the keto acid within the confined cavity of RCC3 plays a crucial role in determining enantioselectivity.

By comparison, the unreduced cage CC3 showed a significantly worse performance, affording the desired product at just 77% yield and 25% ee. (

Table 2. Control experiment of DMR Process of 1a and 2a ^a.

Entry	Cat.	Time	Yield b/%	ee c/%
1	RCC3: Cu(OAc)2 (1:1)	6	79	91
2	CC3: Cu(OAc)2 (1:1)	12	77	25
3	RCC3: Cu(OAc)2 (1:2)	6	75	90
4	RCC3: Cu(OAc)2 (1:4)	6	73	86
5	CC3	12	63	5
6	RCC3	12	66	9
7	L2	12	61	6
8	L2: Cu(OAc)2 (1:1)	12	86	47
9	Cu(OAc)2	12	40	0

^a For reaction details see the synthesis section in the Supplementary Materials. ^b Isolated yield. ^c The ee was determined by HPLC analysis.

, entry 2). These results indicate that the conversion of the imine bond into amine bond plays an important role in the enantioselective control. As the RCC3 can coordinate with up to four Cu²⁺, we tested the activity after increasing the amount of Cu(OAc)₂ (Table 2, entries 1 & 3–4), which showed that the enantioselectivity decreased as the amount of Cu²⁺ increased, which indicates that the NH- bond plays a key role in enantioselective control. Although the DMR can be catalyzed by pure CC3 or RCC3 to yield products, enantioselectivity remains low (5–9% ee, Table 2, entries 5–7) in the absence of a Cu²⁺ co-catalyst, even after a prolonged reaction time (12 h). This low ee demonstrates that the coordinated Cu center also plays a critical role in achieving a high enantioselectivity.

At the same time, the catalytic behavior of the related molecular catalysts of the model compound (1S,2S)-N, N’-bis(phenylmethyl)-1,2-cyclohexanediamine (PMCHDA, L0) were examined (Table 2, entries 7–8). There was nearly no enantioselectivity observed for L² and only 47% ee for Cu(OAc)₂/L² (1/1). Consistent with reports in the literature [42], the use of R,R-diamine ligands systematically generates the opposite enantiomer compared to S,S-configured analogs in this transformation. These results demonstrate the critical role of the cage structure. The improved enantioselectivity observed with RCC3/Cu²⁺ is attributed to its chiral cavity NH groups acting cooperatively with Cu(OAc)₂. This cooperation facilitates multipoint substrate binding, which directs nucleophilic attack and consequently controls both selectivity and reactivity.

Based on the experimental results, the presence of molecular amine cages and copper salts are key factors regulating the reaction’s stereoselectivity. Given that amine cages possess more NH groups than imine cages, providing multiple hydrogen bonding sites, we propose that hydrogen bonding and Cu²⁺ coordination are responsible for the stereoselectivity. Consequently, we propose the plausible reaction mechanism depicted in Figure 3. Initially, substrates 2a and 1a enter the molecular cage cavity via host–guest interactions. The acyl group of 2a, resembling an acetate anion, competes with Cu²⁺ to form complex RCC3@Cu·2a, simultaneously activating 2a for decarboxylation. Within the confined cavity, the combined action of hydrogen bonding and coordination bonds restricts the approach of 1a to the activated 2a, allowing interaction only on the specific face of the C=N with the smallest steric hindrance. This steric constraint results in the formation of the specific stereoselective Mannich product 3aa.

3. Materials and Methods

3.1. Materials and General Procedures

All materials and solvents were commercially available and used without further purification. The RCC3 was synthesized by two steps at a 59.1% overall yield. Fourier transform infrared (FTIR) spectra were collected on a Nicolet Magna 550 spectrometer (Thermo Electron Corporation (now Thermo Fisher Scientific Inc.) in Madison, WI, USA.) using KBr method. UV-vis spectra were recorded with a Hitachi U-3900 Spectrophotometer (Hitachi High-Tech Corporation, Tokyo, Japan) in the wavelength range of 200–800 nm.

3.2. Synthesis of CC3

The synthesis of CC3 was conducted according to the previous method. A typical procedure is as follows: benzene-1,3,5-tricarbaldehyde (0.5 g) was dissolved in 10.0 mL CH₂Cl₂ in the vessel, and trifluoroacetic acid (10 uL) was added directly to this solution as a catalyst for imine bond formation. Then, 10 mL CH₂Cl₂ solution of (R, R)-1,2-diamino cyclohexane (0.5 g, 4.464 mmol) was added. The vessel of the mixture solution was capped and left to stand for one week. Crystals grew on the sides of the vessel. The crystalline product was removed by centrifugation and washed with a CH₂Cl₂/CH₃OH mixture (v/v, 5/95) several times, and further dried at 60 °C under vacuum overnight. The ¹H NMR and ¹³C NMR spectra were tested to confirm the target structure of CC3: ¹H NMR (400 MHz, Chloroform-d) δ 8.19 (s, 1H), 7.94 (s, 1H), 3.36 (d, J = 7.3 Hz, 1H), 1.85 (d, J = 8.6 Hz, 1H), 1.70 (s, 2H), 1.49 (d, J = 10.3 Hz, 1H). ¹³C NMR (101 MHz, Chloroform-d) δ 159.20, 136.61, 129.63, 74.61, 33.00, 24.37. FT-IR (KBr, cm⁻¹) for CC3: 3386(br), 3286(s), 3085(m), 2927(s), 2852(s), 2667(w), 2582(w), 1703(m), 1644(s), 1598(m), 1447(s), 1372(s), 1341(s), 1260(w), 1241(w), 1200(w), 1160(s), 1139(s), 1093(s), 1060(m), 1044(m), 990(m), 936(m), 885(m)854(m), 841(w), 808(w), 688(s), 669(s), 521(m), 477(m).

3.3. Synthesis of Corresponding Reduced RCC3

The imine container (500.0 mg) was dissolved in a CH₂Cl₂/CH₃OH mixture (v/v, 1/1, 25.0 mL) by stirring. When this solution became clear, NaBH₄ (0.5 g) was added and the reaction was stirred for 15 h at room temperature. Water (1.0 mL) was then added, and the solution was continuously stirred for an additional 9 h. The solvent was then removed under a vacuum. The resulting solid was washed with water and collected by centrifugation, and the obtained solid was dried at 80 C under a vacuum overnight. The ¹H NMR and ¹³C NMR spectra were tested to confirm the target-reduced containers. RCC3 ¹H NMR (400 MHz, Chloroform-d) δ 7.14 (s, 1H), 3.84 (d, J = 14.0 Hz, 1H), 3.58 (d, J = 14.1 Hz, 1H), 2.37 − 2.14 (m, 3H), 2.02 (d, J = 12.5 Hz, 1H), 1.76 − 1.56 (m, 2H), 1.25 − 1.09 (m, 1H), 0.99 (d, J = 11.8 Hz, 1H). ¹³C NMR (101 MHz, CDCl3) δ 141.22, 125.11, 61.31, 50.63, 31.93, 31.67, 29.71, 24.98. FT-IR (KBr, cm⁻¹) for RCC3: 3297(s), 3153(br), 3004(w), 2928(s), 2847(s), 2666(w), 1686(w), 1646(s), 1451(s), 1399(m), 1359(w), 1333(w), 1263(w), 1239(w), 1206(w), 1157(m), 1115(s), 1056(w), 1030(w), 999(m), 948(w), 854(m), 795(m), 713(w), 525(w).

3.4. General Procedure for DMR

In a typical procedure, 5 mol% of RCC3 (22 mg, 0.02 mmol) and Cu(OAc)₂ (3.6 mg, 0.02 mmol) were well dispersed into 4.0 mL of THF and stirred for 6 h. Subsequently, the benzo[e][1,2,3]oxathiazine 2,2-dioxide (18.3 mg, 0.1 mmol) and acetoacetic acid (24.6 mg, 0.15 mmol) were added into the mixture at −20 °C. The mixture was stirred for another 6 h and then filtered through a silica plug and washed with Et₂O (5.0 mL). The filtrate was analyzed by chiral HPLC and ¹H NMR for enantioselectivity and conversion, respectively.

4. Conclusions

In summary, we have successfully applied the amino chiral covalent organic cage RCC3 catalytic system to the asymmetric decarboxylation Mannich reaction between β—ketoacids and cyclic aldimines. Compared with the corresponding monomeric catalysts, the co-catalysis of RCC3 with Cu(OAc)₂ exhibits enhanced stereoselectivity and comparable activity in DMR. The control experiments of amine/imine cages or imine cages coordinating with Cu(OAc)₂ indicate that the hydrogen bonding of amine cages and the presence of Cu²⁺ have important effects on the regulation of stereoselectivity. This study indicates that POCs can provide a multifunctional composite catalytic platform for organic reactions as catalysts.