Introduction

C(sp3)–C(sp3) bonds constitute the framework of numerous bioactive natural products, pharmaceuticals, and functional materials. It has been suggested that a higher prevalence of sp3-hybridized chiral carbons is strongly associated with improved potency, selectivity, and patentability in drug discovery1,2,3,4. The development of efficient methods that can simultaneously form C(sp3)–C(sp3) bonds and control stereochemistry is thus of long-standing interest in synthetic chemistry5,6,7,8,9. However, achieving diastereo- and enantioselective construction of vicinal, fully substituted carbon stereocenters remains a formidable challenge, primarily due to the unfavorable thermodynamic and kinetic effects resulting from increased steric hindrance10,11. Despite various efficient strategies being elaborated to achieve this goal, the majority of these methods result in the formation of at least one quaternary stereocenter within a cyclic system12,13. In contrast, the stereoselective construction of C(sp3)–C(sp3) bonds bearing two neighboring quaternary stereocenters within an acyclic skeleton presents a particularly challenging task, owing to the increased rotational freedom and the severe steric congestion during C(sp3)–C(sp3) bond formation14,15,16, and remains largely underdeveloped (Fig. 1A)17,18,19,20,21,22,23.

Fig. 1: Construction of fully substituted stereocenters and development of biomimetic asymmetric Mannich reactions leading to the generation of acyclic vicinal 4o/4o stereocenters.
figure 1

A Increasing challenges in the construction of fully substituted stereocenters. B Catalytic asymmetric Mannich reactions: varieties of stereocenters generated. C Stereoselective generation of acyclic vicinal full-substituted stereocenters via biomimetic asymmetric Mannich reactions.

Full size image

The Mannich reaction is one of the most classical C(sp3)–C(sp3) bond formation reactions in organic synthesis24,25,26,27,28,29. Catalytic asymmetric Mannich reactions represent an attractive approach for producing a diverse array of optically enriched amines and derivatives. Since the pioneering demonstrations of catalytic enantioselective Mannich reactions in the late 1990s, significant advancements have been made by leveraging chiral organometallic catalysts30 and organocatalysts31,32,33,34, facilitating the stereocontrolled formation of single or two adjacent stereocenters. To date, constructing vicinal tertiary/tertiary35,36,37,38,39 and tertiary/quaternaty35,40,41,42,43,44 stereocenters have been well established through diastereo- and enantioselective Mannich reactions (Fig. 1B). Nonetheless, the exploration of asymmetric Mannich reactions for the synthesis of vicinal full-substituted stereocenters remains relatively unexplored45,46,47,48,49,50,51,52,53,54. Notably, these limited examples have consistently employed cyclic imines and/or cyclic nucleophilic donors, leading to the formation of at least one quaternary stereocenter embedded within a cyclic structure45,46,47,48,49,50,51. To the best of our knowledge, there have been no reports on the asymmetric Mannich reaction for the construction of vicinal acyclic tetrasubstituted stereocenters.

Biomimetic asymmetric catalysis that emulates enzyme function has long been a primary goal in synthetic chemistry52,53. In nature, aldolases have evolved as highly effective enzymes that catalyze asymmetric aldol reactions, providing a robust strategy for the stereoselective assembly of carbon-carbon bonds. Among the two known classes of aldolases, class II aldolases utilize a metal ion cofactor (typically Zn2+), which acts as a Lewis acid, to activate the aldol donor via coordination to the generated enolate54. Concurrently, Brønsted basic residues within the enzyme facilitate the deprotonation and enolate formation, while Brønsted acidic residues activate the electrophilic aldehyde (Fig. 1C). However, biomimetic asymmetric catalysis inspired by class II aldolases has been relatively overlooked, especially when compared to the extensively studied enamine catalysis based on class I aldolases. Inspired by the cooperative action of metallic Lewis acids, Brønsted bases, and Brønsted acids in class II aldolases, we present a copper/squaramide cooperative catalysis system designed for asymmetric Mannich reactions. By mimicking the features of class II aldolases, we propose that copper salts, acting as Lewis acids, can effectively activate acyclic β-keto esters. Simultaneously, tertiary amine-squaramide bifunctional organocatalysts can engage in multiple hydrogen-bonding interactions with the substrates, thereby facilitating challenging asymmetric Mannich reactions involving acyclic ketimines and acyclic trisubstituted carbon nucleophiles. Furthermore, this biomimetic cooperative catalysis system features a more suitable chiral pocket than that of a single chiral metal- or organo-catalyst, enabling the formation of vicinal acyclic tetrasubstituted stereocenters in a crowded environment with precise control over both diastereo- and enantioselectivity.

Results

Reaction optimization. To evaluate our hypothesis, we selected β,γ-alkynyl-α-imino ester 1a and ethyl 2-methylacetoacetate 2a as model substrates for the investigation of the asymmetric Mannich reaction (Table 1). Inspired by the successful use of bifunctional organocatalysts in asymmetric Mannich-type reactions, we first screened several chiral bifunctional thiourea55 and squaramide56,57 organocatalysts. We hypothesized that these bifunctional organocatalysts, which incorporate a Brønsted-basic tertiary amine alongside hydrogen-bond donors, could simultaneously activate both ketimine 1a and dicarbonyl compound 2a through cooperative hydrogen-bonding interactions. Initial screening with squaramide catalysts alone (Table 1, entries 1–3, and see also the Supplementary Information) afforded the desired amino acid derivative 3a with adjacent non-cyclic tetrasubstituted stereocenters. However, despite high diastereoselectivity, yields remained low to moderate (32–50%) with poor enantioselectivity (<10% ee). Prior studies suggested Lewis-acidic copper salts could activate dicarbonyl compounds58,59,60,61. Thus, the combined catalysis of squaramides and copper salts were examined, and we were pleased to find that improved yields and enantioselectivities were obtained when copper(I) thiophene-2-carboxylate (CuTC) was added as the co-catalyst (Table 1, entries 4–6). Among them, the cinchona alkaloid-derived squaramide Cat-3 performed best. Consequently, we screened other cinchona alkaloid-based catalysts (e.g., sulfonamides, thioureas; entries 7–10), but Cat-3 retained superior stereoselectivity (entry 6). Further screening of other squaramide catalysts (Table S1 and Supplementary Information) did not enhance stereoselectivity. Noteworthy, the diastereomer Cat-16 of Cat-3 produces the opposite enantiomer of the Mannich reaction product 3a in 52% yield, 60:40 dr, and 35% ee (entry 16, Table S1 and Supplementary Information). Evaluation of alternative copper salts (CuI, Cu2O, and Cu(MeCN)4PF6) confirmed CuTC as optimal. Notably, CuI slightly improved diastereoselectivity at the expense of enantioselectivity (entry 11). Solvent screening (THF, toluene, DCE, MeCN, acetone, chlorobenzene, 1,4-dioxane; entries 14-20) identified DCM as optimal (entry 6). Lowering the reaction temperature increased ee from 87% to 90%, albeit with moderate diastereoselectivity (entry 21). The addition of additives (molecular sieves, Na2SO4, MgSO4) enhanced diastereoselectivities but significantly reduced yields (entry 22, and Table S3 of Supplementary Information). Chiral bisoxazoline ligands62,63,64 for copper improved diastereoselectivity marginally without increasing ee (entries 23–26). Increasing Cat-3 loading to 20 mol% afforded 3a in 72% yield, 80:20 dr, and 93% ee (entry 27). To further improve diastereoselectivity, we optimized CuTC/CuI and Cat-3 loadings using MgSO4 as an additive (entries 28–31). Ultimately, the reaction with 10 mol% CuTC and 25 mol% Cat-3 in DCM (2 mL) at 0 °C delivered optimal stereoselectivities of 94% ee and 90:10 dr (entry 30).

Table 1 Optimization of reaction conditionsa
Full size table

Scope of acyclic α-substituted β-keto esters. With optimized reaction conditions in hand, we explored the scope of the transformation using a series of acyclic β-keto esters in the presence of CuTC and Cat-3. As illustrated in Fig. 2, acyclic α-alkylated β-keto esters withvatuous substitutents at R1 reacted efficiently with ketimines 1a (Ar = Ph) and 1 m (Ar = 2-thiophene). Regardless of whether the steric bulk at R1 was decreased or increased, the corresponding 3a–3d, 3h–3m featuring two contiguous and acyclic chiral stereocenters were isolated in moderate to good yields (47–78%), along with high diastereoselectivities (dr ranging from 78:22 to >95:5), and excellent enantioselectivities (85-99% ee). Furthermore, substrates containing either a methyl ester (3e and 3g) or a benzyl ester (3f) were also compatible with the reaction; however, using more sterically hindered β-keto benzyl ester resulted in a decreased yield and ee value (3f). In addition, the conversion of β-keto esters with α-chloro substituents to the desired product 3n was feasible, although the stereoselectivity was poorly controlled. The β-keto esters bearing an α-ethyl group were also smoothly converted to the desired product 3o in high yield with 82:18 dr and 98% ee. Finally, we also evaluated ethyl 2-methylacetoacetate with representative ketimines, affording the corresponding products 3p and 3q in moderate yields with high levels of stereocontrol.

Fig. 2: Scope of acyclic α-substituted β-keto esters.
figure 2

Conditions: 1 (0.10 mmol), 2 (0.12 mmol), CuTC (10 mol%), Cat-3 (20 mol%), DCM (2.0 mL), at 0 °C for 30–48 h. b Conditions: 1 (0.10 mmol), 2 (0.12 mmol), CuTC (10 mol%), Cat-3 (25 mol%), MgSO4 (1.0 eq), DCM (2.0 mL), at 0 °C for 30–48 h. Boc tert-butoxycarbonyl.

Full size image

Scope of the β,γ-alkynyl-α-imino esters. Next, the substrate scope of β,γ-alkynyl-α-imino ester was then investigated, as shown in Fig. 3. Various 1-alkynyl N-Boc ketimino esters bearing electron-donating (such as Me, MeO, tBu, and acetylene) or electron-donating groups (such as CF3, F, Cl, and Br) at the ortho, meta or para positions of the phenyl group were all compatible to the reaction conditions. This led to the formation of highly functionalized products 4a–4q in high yields (up to 82%), accompanied by excellent enantioselectivities (90-99% ee) and diastereoselectivities (generally >95:5 dr). Furthermore, when the aryl moiety was bulky 1-naphthyl, 2-naphthyl group or heterocycle, the desired products (4r–4t) were also afforded in high yields with excellent enantioselectivities (94-99% ee) and diastereoselectivities (>95:5 dr). Notably, when the aryl ring was replaced with a trimethylsilyl group, the resulting trimethylsilyl-derived product 4u, which can be easily transformed following deprotection, was obtained with an 80% yield, 99% ee, and > 95:5 dr. Additionally, when the substrates were switched to the N-Cbz protected ketimino ester (4w) or the methyl ester (4v), the corresponding products were still obtained with good yields and excellent ee values. Unfortunately, no product formation was observed when the alkyne moiety of imino esters is replaced with alkyl, alkenyl, or aryl groups (for unsuccessful substrates, see Supplementary Information), which we attribute to reduced electrophilicity and increased steric hindrance.

Fig. 3: Scope of the β,γ-alkynyl-α-imino esters.
figure 3

Conditions: 1 (0.10 mmol), 2 (0.12 mmol), CuTC (10 mol%), Cat-3 (20 mol%), DCM (2.0 mL), at 0 °C for 30–48 h. Boc tert-butoxycarbonyl, TMS trimethylsilyl.

Full size image

Scale-up synthesis and synthetic applications. Considering the significance of highly functionalized amino acid derivatives that contain two vicinal acyclic stereocenters, we investigated the practicality of the present methodology (Fig. 4). First, we scale-up the reaction using 2.0 mmol of the starting material under standard reaction conditions. The corresponding product 3d was afforded with comparable results (75% yield, >95:5 dr, 94% ee). Furthermore, the product 3d could be smoothly converted to the alkyl-substituted compound 5 via hydrogenation using Pd/C, giving 96% yield and with 91% ee. Moreover, we successfully achieved the selective reduction of the ketone moiety using NaBH4. Notably, under the reductive reaction conditions, the resulting hydroxyl group participated in an intramolecular ester exchange reaction, ultimately leading to the formation of the five-membered lactone 6, which bears three contiguous chiral centers. The N-Boc protecting group of 6 was readily removed by treatment with trifluoroacetic acid, without disrupting the lactone ring and compromising the optical purity. Moreover, the ethynyl group in product 4h facilitated its application in CuAAC click reactions, as exemplified by the facile synthesis of enantioenriched 8 in 97% yield and with >95:5 dr and 96% ee. The absolute configuration of 9 was definitively determined as (2S, 3R, 4R) by X-ray diffraction (CCDC 2339871 (9) and CCDC 2452561 (1p) contain the supplementary crystallographic data for this paper. This data can be obtained free of charge from The Cambridge Crystallographic Data Centre).

Fig. 4: Scale-up synthesis and synthetic applications.
figure 4

a The scale-up synthesis and transformation of 3d. b The transformation of 3d, 4g, and 4h. Boc tert-butoxycarbonyl, TFA trifluoroacetic acid.

Full size image

Control experiments and mechanistic studies. To gain insight into the mechanism of this copper/squaramide coperatively catalyzed asymmetric Mannich reaction, a series of mechanistic investigations were conducted. First, when the ester groups in substrates 1 or 2 were replaced with other electron-withdrawing groups such as –CF3 or –CN, no expected Mannich products were detected (Fig. 5a, b). These results indicate that the ester groups in substrates 1 and 2 play an essential role in facilitating interactions with the catalyst. Furthermore, the reaction between β,γ-alkynyl-α-imino ester 1a and β-keto ester 2a using the N,N-dimethylated squaramide (Cat-3’) resulted in remarkably lower enantioselectivity (Fig. 5c). This finding underscores the significant importance of potential hydrogen-bonding interactions between the squaramide organocatalyst and substrates. To further clarify the specific interactions between the copper/squaramide catalysis system and the substrates, including ketimino esters and β-keto esters, we performed LC-MS analysis of the reaction solution. We successfully identified the ion peaks of the Cat-3/1a complex ([M-Boc]+: 801.28; found 801.35) and the Cat-3/Cu/2b complex ([M-H]−: 832.20; found 831.90) through LC-MS analysis (for details, see the Supplementary Information). A full profile of the model reaction between 1o and 2b for the preparation of enantioenriched 4o was also obtained (Fig. 5d). The ee and dr values of product 4o was nearly constant during the reaction. Next, we investigated the effect of varying the ratio of Cat-3 to copper salt on the reaction outcome, while maintaining the copper salt loading at 10 mol% (Fig. 5e). It was found that the ee and dr values of the product exhibited significant improvements when the amount of Cat-3 exceeded that of the copper salt. This suggests that the excess portion of Cat-3 may be acting as a ligand for the copper65,66,67. In addition, reaction progress kinetic analysis using in-situ infrared spectroscopy revealed a second-order dependence on the concentration of Cat-3 (for details, see Supplementary Information, and ReactIR experiments data see Supplementary Dataset 2). Based on the aforementioned mechanistic studies, a tentative Cu/squaramide cooperative catalytic mechanism has been proposed (Fig. 5f). In the presence of a Brønsted base (the tertiary amine moiety of the bifunctional squaramide catalyst) and CuTC, the β-keto ester 2b undergoes deprotonation and enol tautomerization to form the complex S2 (detected by LC-MS and 13C NMR, see the Supplementary Information). Concurrently, another molecule of squaramide catalyst engages in multiple hydrogen-bonding interactions with the N-Boc ketimino ester 1a to form the intermediate S1. The Cu-bound enolate (complex S2) attacks the hydrogen-bonding activated N-Boc ketimino ester (S1) in an enantio- and diastereoselctive manner to give the Mannich product.

Fig. 5: Control experiments and mechanistic studies.
figure 5

a Control experiment 1: CO2Et vs CF3. b Control experiment 2: CO2Et vs CN. c Control experiment 3: N-H vs N-Me. d Reaction process monitoring. e The effect of the ratio of Cu/Cat-3 on the reaction. f Proposed mechanism. Boc tert-butoxycarbonyl, ND not detected.

Full size image

DFT calculations. Based on our understanding of the reaction mechanism, the stereo-determining step of this reaction is presumably the nucleophilic addition step forming the C–C bond. DFT calculations were performed to locate transition states for the model reaction between Cat-3/1a and the Cat-3/Cu/2b complex, elucidating stereoselectivity origins (Fig. 6, see Supplementary Information and Supplementary Dataset 1 for computational details). The results revealed that TS-R-R (0.0 kcal/mol) was the most favorable, stabilized by multiple weak interactions: two hydrogen bonds between the squaramide moiety of Cat-3 and the methyl ester carbonyl of substrate 1a, two hydrogen bonds between the amide nitrogen atom of substrate 1a and the hydrogen atoms on the quinoline ring and adjacent methylene group of Cat-3, a stronger O···H-N interaction (1.56 Å) between the protonated bridgehead nitrogen of Cat-3 and the tert-butyl ester carbonyl of 1a, and a π–π interaction between the two trifluoromethylphenyl groups of Cat-3. These weak interactions stabilized a staggered conformation during C–C bond formation, minimizing steric repulsion. In contrast, TS-S-S (2.1 kcal/mol) lacked the variety of weak interactions present in TS-R-R, adopting a less favorable configuration with a weaker N···H-N hydrogen bond (1.82 Å), forcing an eclipsed conformation that caused steric clash between the tert-butyl group of 1a and the ethyl group of 2b (closest H…H distance 2.03 Å). Diastereomeric transition states were also disfavored: TS-R-S (1.8 kcal/mol) featured fewer weak interactions than TS-R-R and exhibited steric repulsion (closest H…H distance: 2.29 Å), while TS-S-R (3.3 kcal/mol) suffered from a strained O···H-N bond (1.79 Å; angle 146.4°). These energy differences rationalized the high experimental stereoselectivity (96% ee, >95:5 dr), validating the computational stereochemical model.

Fig. 6: DFT-computed transition state structures and relative free energies.
figure 6

Distances between important atoms are in angstrom (Å), bond angles are in degrees.

Full size image

In summary, drawing inspiration from class II aldolases, we have developed a biomimetic copper-squaramide cooperative catalysis strategy for the asymmetric Mannich reactions, enabling facile access to vicinal tetrasubstituted stereocenters in acyclic systems. Utilizing acyclic ketimines and sterically hindered α-substituted β-keto esters as challenging Mannich substrates, we successfully synthesized highly functionalized amino acid derivatives featuring adjacent and acyclic tetrasubstituted stereocenters with excellent diastereo- and enantioselectivities (up to >19:1 dr and 99% ee). Preliminary mechanistic studies and DFT calculations indicate that the success of this strategy can be attributed to the cooperative roles of the copper salts and the tertiary amine-squaramide organocatalysts, which are crucial for substrate activation and the control of stereoselectivity. Further application of this strategy in other significant asymmetric transformations is underway in our laboratory.

Methods

General information. All reactions were carried out in oven-dried reaction vessel unless otherwise noted and solvents were dried according to established procedures. Reactions were monitored by thin layer chromatography (TLC). Purification of reaction product was carried out by flash chromatography using Qing Dao Sea Chemical Reagent silica gel (200–300 mesh). 1H, 13C, and 19F NMR spectra were recorded on Bruker 400 MHz or 500 MHz spectrometer in CDCl3 unless otherwise noted. Chemical shifts in 1H NMR spectra are reported in parts per million (ppm, δ) downfield from the internal standard Me4Si (TMS, δ = 0 ppm). Chemical shifts in 13C NMR spectra are reported relative to the central line of the chloroform signal (δ = 77.0 ppm). Data are presented as follows: chemical shift, integration, multiplicity (s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet) and coupling constant in Hertz (Hz). HPLC analyses were conducted on an Agilent instrument using Daicel Chiralpak IA, IB, AD-H, or Chiralcel OD-H columns. High resolution mass spectra were obtained with a Shimadzu LC-MS-IT-TOF mass spectrometer. The single crystal X-ray diffraction studies were carried out on a Xcalibur Onyx Nova diffractometer equipped with CuK/α radiation.

General procedure to prepare chiral products 3 and 4

Conditions: A flame-dried Schlenk tube equipped with a magnetic stirring bar was charged with a mixture of β,γ-alkynyl-α-imino esters 1 (0.10 mmol), a mixture of CuTC (10 mol%), Cat-3 (20 mol%). After being evacuated and backfilled with nitrogen for three times, DCM (1 mL) was added to the Schlenk tube and the mixture was stirred at room temperature under a N2 atmosphere for 0.5 h. Dicarbonyl compounds 2 (0.12 mmol) in DCM (1 mL) were added sequentially. The reaction mixture was stirred under a N2 atmosphere at 0 °C for 30–48 h. When the reaction was complete, the DCM was evaporated in vacuo and the residue was purified by flash silica gel column chromatography (petroleum ether/EtOAc = 5/1) to give products 3 and 4.