Research Article

Asymmetric phosphoric acid–catalyzed four-component Ugi reaction

See allHide authors and affiliations

Science  14 Sep 2018:
Vol. 361, Issue 6407, eaas8707
DOI: 10.1126/science.aas8707

Steering together all four Ugi pieces

The nearly 60-year-old Ugi reaction is a remarkably efficient means of linking together four molecular building blocks: an aldehyde, an amine, a carboxylic acid, and an isocyanide. Because each component is independently tunable, the reaction is especially well suited to the assembly of diverse compound libraries. However, stereoselectivity has been a challenge. Zhang et al. now show that chiral phosphoric acids can catalyze the four-component coupling with high enantioselectivity (see the Perspective by Riva). Theory suggests that a hydrogen-bonded complex involving the phosphoric acid and carboxylic acid sets the stereochemistry for isocyanide attack on an imine intermediate.

Science, this issue p. eaas8707; see also p. 1072

Structured Abstract

INTRODUCTION

The four-component Ugi reaction (Ugi-4CR) assembles peptide-like α-acylaminoamides through one-pot reaction of a carbonyl compound, an amine, an acid, and an isocyanide. Ugi-4CR is well suited for diversity-oriented synthesis applicable in drug discovery, as it facilitates rapid access to diverse libraries of biologically important molecules. The high step economy and atom efficiency of the reaction, as well as its convergent nature, foster its wide use in the synthesis of heterocyclic scaffolds, natural products, macrocycles, polymers, and other target molecules. Despite these practical advantages, the long-standing stereochemical challenges of the Ugi reaction have yet to be fully addressed. Consequently, access to chiral Ugi products for drug candidate exploration is hindered.

RATIONALE

The chiral phosphoric acid (CPA) framework was targeted as a catalyst for asymmetric Ugi-4CR. The heightened acidity of CPAs over carboxylic acids is perceived to accelerate the kinetics of the enantioselective Ugi reaction so as to outcompete the background reaction. Also, self-assembled heterodimerization between the CPA and carboxylic acid brings about a dual effect: enhanced acidity of the catalyst and nucleophilicity of the carboxylic acid. Both of these favor the catalytic enantioselective Ugi-4CR. A myriad of well-established or custom CPAs with well-defined chiral pockets could be readily applied, potentially leading to complete stereocontrol. A CPA that could suppress the Passerini and other side reactions would enable rapid imine formation and its preferential activation over the carbonyl group.

RESULTS

A catalytic asymmetric Ugi-4CR was accomplished with 1,1′-spirobiindane-7,7′-diol (SPINOL)–derived CPA4 and CPA6 as organocatalysts. The reaction exhibited broad substrate compatibility and good to excellent enantioselectivity [up to 99% enantiomeric excess (ee)]. Activation of the imine might be accomplished by CPA–carboxylic acid heterodimer catalysis via a bifunctional activation mode, which was supported by experiments (carboxylic acids with varying pKa values and steric properties yielded products with a range of ee values) and density functional theory (DFT) calculations (lowest energy among all the considered activation modes). The calculated free energy profile for the catalytic Ugi reaction gave three CPA-combined key transition states, which highlighted the bifunctional property of the CPA. In the favored enantio-determining transition states, the aryl groups fit into the pocket formed by the two substituents (cyclohexyl rings) of the catalyst, revealing the importance of noncovalent interactions in controlling the stereochemical outcome of this reaction.

CONCLUSION

This operationally simple one-pot enantioselective Ugi-4CR harnesses inherent benefits of multicomponent reaction and organocatalysis to access up to 86 enantioenriched α-acylaminoamides, which are otherwise challenging to obtain via conventional methods, from four achiral building blocks in excellent yields and enantioselectivities. DFT calculations gave a detailed catalytic mechanism, especially with respect to activation modes and enantio-determining transition states. Because amide functionality constitutes the defining primary linkage in proteins, we foresee multiple uses of this asymmetric four-component Ugi protocol for the synthesis of chiral peptides and components of natural products. We also anticipate that this work will initiate the further development of asymmetric multicomponent chemistry.

Design and exploration of catalytic asymmetric Ugi-4CR.

Abstract

The Ugi reaction constructs α-acylaminoamide compounds by combining an aldehyde or ketone, an amine, a carboxylic acid, and an isocyanide in a single flask. Its appealing features include inherent atom and step economy together with the potential to generate products of broad structural diversity. However, control of the stereochemistry in this reaction has proven to be a formidable challenge. We describe an efficient enantioselective four-component Ugi reaction catalyzed by a chiral phosphoric acid derivative that delivers more than 80 α-acylaminoamides in good to excellent enantiomeric excess. Experimental and computational studies establish the reaction mechanism and origins of stereoselectivity.

The prototypical four-component Ugi reaction (Ugi-4CR), first disclosed by Ugi in 1959 (1), assembles α-acylaminoamides through one-pot reaction of a carbonyl compound, an amine, an acid, and an isocyanide (Fig. 1A). The peptide-like moiety is abundant in biologically important molecules (Fig. 1B) (26) as well as natural products (Fig. 1C) (7). Although the precise mechanistic scenario may vary, the simplified contours involve preceding imine activation by a carboxylic acid for sequential nucleophilic addition of isocyanide and carboxylate trapping of the thus-formed nitrilium intermediate. Rearrangement via acyl group migration onto the nitrogen atom derived from the imine generates the final product (Fig. 1A) (811). The Ugi reaction is well suited for diversity-oriented synthesis applicable in drug discovery (1215). It has facilitated rapid access to diverse libraries of biologically important molecules because of its ease of synthetic operation (Fig. 1B) (26). In industrial applications, Dömling and co-workers reported an elegant two-step multicomponent synthesis of praziquantel, a drug to treat the parasitical disease schistosomiasis, via Ugi-cyclization cascade (Fig. 1D) (16). This strategy reduced the materials costs and offered a strategy for the synthesis of analogs to address plausible onset of resistance (17). In natural product studies (18), the convergent nature of the Ugi-4CR enabled one-step assembly of an intermediate containing more than half of the atoms in the final product from four building blocks in the total synthesis of ecteinascidin 743 (Fig. 1E) (19). The development of isocyanide-based multicomponent reactions (20, 21) in recent decades reveals that the Ugi-4CR can also be applied to the synthesis of heterocyclic scaffolds (2224), macrocycles (25, 26), polymers (27, 28), and other compounds.

Fig. 1 The classic four-component Ugi reaction in chemistry.

(A) The simplified mechanism for the reaction. (B) Selected examples of bioactive molecules prepared directly via Ugi-4CR or involving Ugi-4CR as the key step. (C) The recently identified natural product dudawalamide A bearing an α-acylaminoamide scaffold. (D) Two-step synthesis of the anti-schistosomiasis drug praziquantel via Ugi-4CR. (E) Ugi-4CR as a key step in the total synthesis of ecteinascidin 743. Atoms in orange were incorporated by the Ugi reaction. Me, methyl; rt, room temperature; quant., quantitative; MsOH, methanesulfonic acid; MOM, methoxymethyl acetal; TBDPS, t-butyldiphenylsilyl; Bn, benzyl; Boc, t-butoxycarbonyl; PMP, 4-methoxyphenyl.

Despite these practical advantages, the long-standing stereochemical challenges of the Ugi reaction have yet to be fully addressed (29). Whereas a single absolute configuration is strictly required of drug candidates, Ugi condensation products are often racemic. Conventionally, enantiomerically pure amino acids are used as chiral building blocks toward α-acylaminoamides (Fig. 2A). The lengthy route, limited choices of chiral amino acids, and inefficiency in generating structural diversity greatly restrict this protocol. Alternatively, enantiopure substrates could be used for chiral induction in a diastereoselective Ugi-4CR (Fig. 2B). However, this method suffers from poor or moderate diastereoselectivities (3032) unless specialized amines are used. Catalytic enantioselective synthesis (Fig. 2C) offers flexibility in catalyst choice and facile delivery of distinct stereoisomers by inversion of the catalyst configuration. This approach would accommodate a broader substrate scope encompassing commercially available starting materials and would deliver products that cover substantial chemical space by modulating the components in each substrate quadrant. However, the advent of this approach is long overdue, probably impeded by several hurdles: the complexity of a four-component reaction system, the competition from the uncatalyzed background reaction, the difficulty in achieving stereocontrol of the α-addition of an isocyanide to the imine, and competition from the Passerini reaction or other side reactions.

Fig. 2 Strategies for entry to enantioenriched α-acylaminoamides.

(A) Asymmetry originates from natural amino acid. (B) Substrate-induced asymmetric Ugi-4CR. (C) Chiral catalyst–induced asymmetric Ugi-4CR. CPA, chiral phosphoric acid.

We speculated that a mild reaction system with a chiral phosphoric acid (CPA) derivative as catalyst might meet the aforementioned challenges. This robust class of organocatalyst has been commonly used for asymmetric nucleophilic addition to imines after seminal reports by the groups of Akiyama and Terada (33, 34). The heightened acidity of chiral phosphoric acids over carboxylic acids is perceived to accelerate the kinetics of the desired reaction so as to outcompete the background reaction. Also, pioneering reports by List and co-workers suggested that the self-assembled heterodimerization between the CPA and the carboxylic acid brings about a dual effect: enhanced acidity of the catalyst and nucleophilicity of the carboxylic acid (3537); both of these favor the catalytic enantioselective Ugi-4CR. A myriad of well-established or custom CPAs with well-defined chiral pockets can be readily applied (3841), leading to complete stereocontrol in α-addition of the isocyanide to imine. Rapid imine formation or its preferential activation over the carbonyl group is viable with a CPA to suppress the Passerini and other side reactions. Elegant isocyanide-based Ugi-type two- or three-component reactions catalyzed by CPAs (4244) as well as chiral carboxylic acid (45) have been reported. For example, Wang, Zhu, and co-workers achieved an important advance by using a CPA as a catalyst to accomplish the enantioselective Ugi four-center three-component reaction of 2-formylbenzoic acids, anilines, and isonitriles for the syntheses of isoindoline derivatives in high yields with 80 to 90% enantiomeric excess (ee) (44).

Catalyst optimization

After some initial trials (tables S1 to S4), we set out to optimize the model catalytic asymmetric Ugi-4CR between pentanal (1a), 4-nitroaniline (2a), 3-phenylpropanoic acid (3a), and cyclohexyl isocyanide (4a) in CH2Cl2 at room temperature with 5-Å molecular sieves as a dehydrating additive (Fig. 3). A strong background reaction was observed in the absence of catalyst under these reaction conditions, whereas moderate to good enantioselectivities were afforded with the addition of 5 mole percent (mol %) of Brønsted acid CPAs. 1,1′-Spirobiindane-7,7′-diol (SPINOL)–derived phosphoric acid CPA6 with a bulky 2,4,6-tricyclohexylphenyl group at the 6,6′-position was found to be the best catalyst, providing the desired product 5 in 64% yield and 82% ee. Further investigations revealed that a 0.3 equivalent excess of 1a, 2a, and 4a at –20°C and double the initial scale improved the chemical yield to 90%, with 92% ee (see table S11); the Passerini product was virtually absent according to 1H nuclear magnetic resonance (NMR) analysis of the crude reaction mixture.

Fig. 3 Optimizing CPA structure in a model reaction.

Notations for CPAs: *The reaction of 1a (0.05 mmol), 2a (0.05 mmol), 3a (0.05 mmol), 4a (0.055 mmol), and catalyst (5 mol %) was carried out in 1 ml of CH2Cl2. †The reaction of 1a (0.13 mmol), 2a (0.13 mmol), 3a (0.10 mmol), 4a (0.13 mmol), and catalyst CPA6 (5 mol %) was carried out in 2 ml of CH2Cl2 at –20°C. Isolated yields are shown. The ee values were determined by chiral HPLC analysis. Negative ee refers to inverted configuration of the more abundant product enantiomer. MS, molecular sieve; Ph, phenyl; iPr, isopropyl; 1-Ad, 1-adamantyl; Cy, cyclohexyl.

Asymmetric Ugi reaction with aliphatic aldehydes

With the optimal reaction conditions in hand, we explored the substrate scope with respect to aliphatic aldehydes, amines, carboxylic acids, and isocyanides (Fig. 4). The reaction was applicable to a wide range of aliphatic aldehydes. The chain length of alkyl substituents had a negligible effect on the stereochemical outcome, providing corresponding products 6 to 9 in 92% ee. 3-Phenylpropanal possessing a β-aryl moiety furnished the desired product 10 in 91% yield and 90% ee. 3-Methylbutanal and 2-phenylacetaldehyde were also suitable substrates to afford 11 and 12. A thioether (13) substituent was well tolerated, and ether substituents (14, 15, and 16) afforded excellent enantioselectivities (95 to 96% ee) presumably due to the chelating effect as previously observed by Schreiber and colleagues (46). A symmetrical dialdehyde with a central N-tert-butoxycarbonyl group underwent a dual reaction at both carbonyls with a 10.9:1 diastereomeric ratio (dr) and 99% ee. Aside from 4-nitroaniline, amines with varied functionalities reacted smoothly to give compounds 19 to 23, and excellent enantioselectivities (94 to 97% ee) were achieved with chelating aldehydes (2428). The generality of the acid component was broad, as products of linear alkyl carboxylic acids (2932), benzyl carboxylic acids (33 and 34), α-branched isobutyric acid (35), α-halogenated acetic acid (36), alkenyl carboxylic acids (37 and 40), aromatic carboxylic acid (38), and heteroaromatic carboxylic acid (39) could all be generated with good enantioselectivities (83 to 91% ee). The absolute configuration of 34 was determined by x-ray crystallographic analysis after recrystallization, and those of other products in Fig. 4 were assigned by analogy. Although benzyl isocyanide formed 44 with moderate optical purity, primary, secondary, tertiary, and aromatic isocyanides formed Ugi products (4143, 45) with good enantioselectivities, regardless of the extent of steric encumbrance.

Fig. 4 Substrate scope of the enantioselective Ugi-4CR with aliphatic aldehydes.

Colors in (A) to (D) correspond to component colors at upper left. (A) Scope of aldehydes. (B) Scope of amines. (C) Scope of acids. The absolute structure of 34 was defined by means of single-crystal x-ray diffraction with radiation wavelength of 0.71073 Å (Mo-Kα) at 100.0 K, and the Flack x parameter was determined as 0.000(4). (D) Scope of isocyanides. Isolated yields are shown. The ee values were determined by chiral HPLC analysis. *The 10.9:1 dr for 18 was calculated on the basis of chiral HPLC analysis (the response factor was 1:1 for equimolar diastereoisomers calculated from the chiral HPLC trace).

Asymmetric Ugi reaction with aromatic aldehydes

We next examined the compatibility of our reaction protocol with aromatic aldehydes. Considering their distinct properties, we reoptimized the conditions (see tables S5 to S10, S12, and S13) with benzaldehyde (1b), butylamine (2b), 4-chlorobenzoic acid (3b), and benzyl isocyanide (4b) as model substrates. Product 46 was obtained in 95% yield and 74% ee with chiral phosphoric acid CPA1 as catalyst in cyclohexane. We further varied the solvent, reaction temperature, and catalyst to identify the following optimal protocol: Reactants 1b (0.13 mmol), 2b (0.10 mmol), 3b (0.10 mmol), and 4b (0.13 mmol) were combined with a newly synthesized catalyst CPA4 (10 mol %) in cyclohexane at 20°C for 36 hours to afford the expected Ugi product 46 in 91% isolated yield and 92% ee. To assess the substrate generality and limitations of these conditions, we evaluated a number of aromatic aldehydes, amines, carboxylic acids, and isocyanides (Fig. 5). It was apparent that the positions and electronic properties of the substituents on the aromatic ring of the aldehydes exerted very limited influence on the stereoselectivity of the process (4757); only slight decreases in enantioselectivity were observed for compounds 58 to 61 derived from aldehydes bearing strongly electron-withdrawing groups. The 2-naphthaldehyde and furfural were also applicable substrates for this transformation, affording 62 and 63 with 92% and 93% ee, respectively, although an extended reaction time was required for the latter. The reaction worked efficiently with different amines. Although benzyl amine delivered compound 73 in slightly compromised optical purity, all other linear amines of various chain length including an ether gave rise to products (6470, 75) in good enantiomeric excess. Similarly, 4-phenyl butylamine, 3-phenyl propylamine, and 3-methyl butanamine formed products 71, 72, and 74 in satisfactory enantiomeric excess. Also, replacing the benzyl isocyanide with other isocyanides produced corresponding Ugi products 76 to 79 without diminishing enantioselectivity (87 to 91% ee). Tolerance toward structural and electronic variations of the acids afforded access to a series of α-acylaminoamides with excellent optical purities. Products of aromatic acids bearing electron-donating (8185), neutral (80), or electron-withdrawing (86 and 87) moieties at varying positions on the phenyl ring were formed in 90 to 94% ee. In addition, 2-naphthyl carboxylic acid and trans-cinnamic acid were suitable substrates, providing corresponding products (88 and 89) in good yield and enantioselectivity. Product 90 was assembled from aliphatic cyclohexanecarboxylic acid in 84% yield and 87% ee. The absolute configuration of 73 was assigned as (R) by comparing the high-performance liquid chromatography (HPLC) spectrum of the reaction product with the known configuration of (S)-73 synthesized from protected l-phenylglycine (Fig. 6B). Those of other products were assigned analogously.

Fig. 5 Substrate scope of the enantioselective Ugi-4CR with aromatic aldehydes.

Colors in (A) to (D) correspond to component colors at upper left. (A) Scope of aldehydes. (B) Scope of amines. (C) Scope of isocyanides. (D) Scope of acids. Isolated yields are shown. The ee values were determined by chiral HPLC analysis. *For 50, the reaction time was extended to 7 days. †For 63, the reaction time was extended to 72 hours. ‡For 77, 2.0 equiv of isocyanide was added to the reaction. tBu, tert-butyl; Et, ethyl.

Fig. 6 Syntheses of (R)-46 and (S)-73, and control experiments.

(A) Gram-scale synthesis of 46 via catalytic enantioselective Ugi-4CR. (B) Synthesis of (S)-73 from enantiopure amino acid source. Conditions: (a) benzylamine, N,N-diisopropylethylamine (DIEA), O-(benzotriazol-1-yl)-N,N,N′,N′-tetramethyluronium tetrafluoroborate (TBTU), CH2Cl2, –20°C, 24 hours; (b) trifluoroacetic acid, CH2Cl2, rt, 10 hours; (c) benzaldehyde, MgSO4, CH2Cl2, rt, 10 hours; (d) NaBH4, MeOH/H2O, rt, 4 hours; (e) p-chlorobenzoyl chloride, NEt3, CH2Cl2, 0°C, 2 hours. (C) Control experiments. Isolated yields are shown. The ee values were determined by chiral HPLC analysis. LG, leaving group; nBu, n-butyl.

To further evaluate the practicality of the protocol, we performed a gram-scale reaction to prepare 46. As shown in Fig. 6A, 46 was obtained with the same enantioselectivity but in higher yield. Our strategy is more efficient than the conventional method (Fig. 6B), which requires multiple steps and suffers from moderate overall yield (50%; Fig. 6B).

Mechanism studies

To gain mechanistic insights, we conducted control experiments and density functional theory (DFT) calculations. The reaction did not proceed in the presence of two equivalents of amine, which implies that the active catalyst is the chiral phosphoric acid itself rather than phosphate or other deprotonated species. Also, the reaction outcome of mixing preformed imine 95 with carboxylic acid 3b and isocyanide 4b was similar to the four-component variant, suggesting earlier imine formation as well as the key intermediacy of trans-imine. Carboxylic acids with varying pKa and steric properties gave products with a range of ee values—91% ee (31), 88% ee (35), and 83% ee (38) in Fig. 4 as well as 94% ee (87), 92% ee (81), and 87% ee (90) in Fig. 5—that hinted at the participation of this component in the enantio-determining step of the reaction mechanism, which could be the α-addition step of isocyanide to imine. Thus, we posited that lowest unoccupied molecular orbital (LUMO) activation of the imine might be accomplished by CPA–carboxylic acid heterodimer catalysis via a bifunctional activation mode (41). DFT calculations strongly support such a scenario, as transition state TS-1 (Fig. 7A), resulting from 5′ (Fig. 7B), has the lowest energy among all the considered transition states. For the nucleophilic addition of the isocyanide to the trans-imine, four different modes (Fig. 7A) of catalysis were explored and compared. The chiral phosphoric acid is modeled with a dimethyl model catalyst. All these TS structures involve LUMO activation of the imine by the catalyst, as well as activation (directing) of the isocyanide via either electrostatic interaction (TS-1) or C–H···O hydrogen bonding (TS-1aTS-1c). The acid-heterodimer catalysis via TS-1 has the lowest energy barrier (15.7 kcal/mol). It features favorable electrostatic interactions between the positively charged nitrogen atom of the isocyanide and the carbonyl oxygen atom of the carboxylic acid.

Fig. 7 Calculated activation modes and energy profile of catalytic Ugi reaction.

(A) DFT-optimized transition state structures and computed activation energies for the nucleophilic addition of the isocyanide to the imine. Barriers are relative to imine, isocyanide, and acid dimer. (B) DFT-computed reaction pathway for the catalytic Ugi reaction of 2′, 4′, and 6′. PNP, p-nitrophenyl. Color code: gray, carbon; white, hydrogen; blue, nitrogen; red, oxygen; orange, phosphorus.

In the calculated free energy profile for the reaction (Fig. 7B; see fig. S1 for full reaction pathway), the model phosphoric acid 1′ and acetic acid 2′ form a relatively stable heterodimer 3′ (5.2 kcal/mol exergonic) as suggested by List and co-workers (35). Imine coordination at the Brønsted acid site of the phosphoric acid generates heterotrimer 5′. The activated imine then undergoes a nucleophilic attack by isocyanide 6′ to form the nitrilium intermediate 7′. Accompanying the proton delivery, the carboxylate attacks the nitrilium carbon via TS-2 to form the imidate 8′ rapidly; the difference in energy between TS-2 and 7′ is only 3 kcal/mol. Subsequently, phosphoric acid 1′ promotes a Mumm rearrangement to release the desired product 14′ and regenerate the catalyst. The overall rate-determining step is the nucleophilic addition of the isocyanide 6′ to the imine-catalyst complex 5′ via TS-1. The subsequent Mumm rearrangement catalyzed by phosphoric acid has a lower barrier (11.9 kcal/mol, via TS-3) than the first C-C bond-forming step (15.7 kcal/mol, via TS-1). For comparison, the Mumm rearrangement catalyzed by acetic acid has a much higher barrier (18.4 kcal/mol via TS-3a; fig. S2). This difference not only defines the rate-limiting step of the overall process, but also plays an important role in the stereochemical outcome of the reaction, as the enantioselectivity could be alternatively determined at the Mumm rearrangement step. For example, Zhu and co-workers attributed asymmetric induction in the Ugi variant they reported (44) to a dynamic kinetic resolution of the primary Ugi adduct, rather than to the C-C bond-forming process. Thus, the present work differs substantially from that previous work in the mode of selectivity.

Enantio-determining TS structures were also explored with catalyst CPA6 (CPA4 is used in Fig. 5, but CPA6 also works); the lowest-energy TSs for two different imine substrates are shown in Fig. 8. The favored transition states lead to (S)-product for N-aryl imine and to (R)-product for N-alkyl imine, in accord with the experimental observations. The difference is larger for TS-6-(R) and TS-6-(S) (1.9 kcal/mol at room temperature). Although no obvious steric clashes are detected in these transition states, the important factor is the orientation of the aryl groups of the substrates relative to the 2,4,6-tricyclohexyl group of the catalyst. In the favored TSs, the aryl groups fit into the pocket formed by the two cyclohexyl rings of the catalyst, whereas the aryl groups occupy the empty quadrant in the three-dimensional space left by the bulky substituents of the catalyst in the disfavored TSs. It is likely that the favorable noncovalent interactions between the aryl groups of the substrates and the cyclohexyl groups of the catalyst are responsible for the observed enantioselectivities. A general model for predicting the stereochemistry of phosphoric acid–catalyzed reactions of imines based primarily on steric effects (47, 48) failed to predict the correct stereochemistry in this case, pointing to the importance of noncovalent interactions and their interplay with (and sometimes overriding of) steric effects in chiral phosphoric acid-catalyzed reactions (49, 50). The bifunctional property of the chiral phosphoric acid plays a profound role in the reaction processes (fig. S1) with respect to generation of the heterodimer, activation of the imine for α-addition of isocyanide, control of the enantioselectivity, and promotion of the Mumm rearrangement.

Fig. 8 Density functional theory calculations for enantioselectivity.

(A) DFT-optimized enantio-determining transition state structures and their relative energies for N-aryl imine substrate. (B) DFT-optimized enantio-determining transition state structures and their relative energies for N-alkyl imine substrate.

Methods

1H NMR, 13C NMR, and 19F NMR spectra of 46 to 90 were recorded at 80°C in DMSO-d6 on a 400 MHz instrument with tetramethylsilane (TMS) as internal standard. 1H NMR, 13C NMR, 31P NMR, and 19F NMR spectra of other compounds were recorded at room temperature in CDCl3 on a 400 MHz/500 MHz instrument with TMS as internal standard. Data for 1H NMR are recorded as follows: chemical shift (ppm), multiplicity (s, singlet; d, doublet; t, triplet; q, quartet; m, multiplet), coupling constant (Hz), integration. Data for 13C NMR are reported in terms of chemical shift (δ, ppm). High-resolution mass spectra (HRMS) were recorded on a LC-TOF spectrometer (Micromass). ESI-HRMS data were acquired using a Thermo LTQ Orbitrap XL Instrument equipped with an ESI source and controlled by Xcalibur software. Enantioselectivities were recorded on Shimadazu/Agilent HPLC, using a chiral stationary phase column (IC/ID, Daicel Co. CHIRALPAK). The chiral HPLC methods were calibrated with the corresponding racemic mixtures. Dichloromethane and cyclohexane were purchased from J&K. 5-Å molecular sieve was purchased from Acros. The silica gel (300–400 mesh) for flash column chromatography was purchased from Accela. Other chemicals were purchased from TCI, Energy, Adamas, Meryer, Acros, and Alfa Aesar, and used as received.

All calculations were performed with the Gaussian 09 package (51). Geometry optimizations were performed with B3LYP (52, 53) and the 6-31G(d) basis set. Normal vibrational mode analysis at the same level of theory confirmed that the optimized structures are minima (zero imaginary frequency) or saddle points (one imaginary frequency). Single-point energies and solvent effects in dichloromethane were computed with the dispersion-corrected density functional method B3LYP-D3 (54) with a Becke-Johnson (BJ) damping function (55) and the 6-311+G (d,p) basis set using the CPCM solvation model (56, 57). The relative energies with ZPE corrections and free energies (at 298.15 K) are in kcal/mol. Single-point energies were also evaluated within the CPCM model using the M06-2X (58), ωB97X-D (59), and B3LYP functionals to compare the stereoselectivities computed with or without dispersion corrections. DFT-optimized structures are illustrated using CYLView (60).

Supplementary Materials

www.sciencemag.org/content/361/6407/eaas8707/suppl/DC1

Materials and Methods

Supplementary Text

Tables S1 to S16

Figs. S1 to S5

NMR Spectra

References (6163)

Movies S1 and S2

References and Notes

Acknowledgments: Funding: Supported by the National Natural Science Foundation of China (Nos. 21572095, 21772081), Shenzhen special funds for the development of biomedicine, internet, new energy, and new material industries (JCYJ20170412151701379, KQJSCX20170328153203). B.T. thanks the Thousand Young Talents Program for financial support. P.Y. and K.N.H. acknowledge the computational resources provided by the Institute of Digital Research and Education (IDRE) at UCLA, and by the Extreme Science and Engineering Discovery Environment (XSEDE), which is supported by the NSF (OCI-1053575). Author contributions: B.T. conceived of and directed the project; J.Z. developed the catalytic asymmetric Ugi-4CR and conducted most of the experiments; S.-Y.L. and H.S. performed parts of substrate screening experiments; P.Y. conducted the DFT calculations and provided mechanism analysis; K.N.H. directed the DFT calculations and mechanism analysis; J.W. and S.-H.X. helped the direction of the project; and P.Y., S.-H.X., J.W., K.N.H., and B.T. co-wrote the manuscript. Competing interests: The authors declare no competing interests. Data and materials availability: The x-ray crystallographic coordinates for the structure of 34 are available free of charge from the Cambridge Crystallographic Data Centre under deposition number CCDC 1588496. Experimental procedures, characterization of new compounds, and all other data supporting the findings are available in the supplementary materials.
View Abstract

Navigate This Article