The asymmetric aldol reaction is one of the most powerful

Transcription

1 Enantioselective direct aldol reactions catalyzed by L-prolinamide derivatives Zhuo Tang, Fan Jiang, Xin Cui, Liu-Zhu Gong, Ai-Qiao Mi, Yao-Zhong Jiang, and Yun-Dong Wu Key Laboratory for Asymmetric Synthesis and Chirotechnology of Sichuan Province, Chengdu Institute of Organic Chemistry, Chinese Academy of Sciences, Chengdu , China; State Key Laboratory of Molecular Dynamics and Stable Structures, College of Chemistry and Molecular Engineering, Peking University, Beijing , China; and Department of Chemistry, Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong, China Edited by Jack Halpern, University of Chicago, Chicago, IL, and approved February 4, 2004 (received for review November 4, 2003) L-Prolinamides 2, prepared from L-proline and simple aliphatic and aromatic amines, have been found to be active catalysts for the direct aldol reaction of 4-nitrobenzaldehyde with neat acetone at room temperature. They give moderate enantioselectivities of up to 46% enantiomeric excess (ee). The enantioselectivity increases as the amide NOH becomes a better hydrogen bond donor. L-Prolinamides 3, derived from the reaction of L-proline with, -hydroxyamines such that there is a terminal hydroxyl group, show more efficient catalysis and higher enantioselectivities. In particular, catalyst 3h, prepared from L-proline and (1S,2S)-diphenyl-2-aminoethanol, exhibits high enantioselectivities of up to 93% ee for aromatic aldehydes and up to >99% ee for aliphatic aldehydes under 25 C. Model reactions of benzaldehyde with three enamines derived from the condensation of prolinamides with acetone have been studied by quantum mechanics calculations. The calculations reveal that the amide NOH and the terminal hydroxyl groups form hydrogen bonds with the benzaldehyde substrate. These hydrogen bonds reduce the activation energy and cause high enantioselectivity. Our results suggest a new strategy in the design of new organic catalysts for direct asymmetric aldol reactions and related transformations. The asymmetric aldol reaction is one of the most powerful methods for the construction of complex chiral polyol architectures (1). The great synthetic usefulness of the aldol reaction in organic synthesis has powered a rapid evolution of numerous highly enantioselective chiral catalysts (2). In general, asymmetric catalytic aldol reactions are classified into two main categories. One type of aldol reaction requires the preconversion of ketone or ester to a more active aldol donor, such as an enol ether or a ketene acetal by the use of a chiral Lewis acid (2) or Lewis base (3, 4) as the catalyst. The other type is called a direct aldol reaction, which is highly atomically economic (5). The development of efficient catalysts for asymmetric catalytic direct aldol reactions is a worthwhile endeavor. The first example of the asymmetric direct aldol reaction catalyzed by heterobimetallic complexes was reported by Shibasaki and coworkers (6, 7). Trost and coworkers (8, 9) have designed a zinc complex for the direct catalytic asymmetric aldol reaction with high enantioselectivities. Since the pioneering finding by List, Barbas, and their coworkers (10, 11) that L-proline could act as a catalyst in intermolecular direct aldol reaction, the concept of small organic molecules as catalysts has received increasing attention (12 32). However, efficient organic catalysts other than chiral amino acids for asymmetric direct aldol reactions are scarce (33, 34). We seek to design small organic molecules with structural diversity for catalyzing organic transformations with high stereoselectivity and broad substrates. Here, we report on a class of organic catalysts, (S)-pyrrolidine-2-carboxamides (L-prolinamides), that efficiently catalyze the direct aldol reactions with high enantioselectivities (35). The relationship between L-prolinamide structures and the enantioselectivity and theoretical studies on the transition structure and reaction mechanism are presented. The key factors that control the enantioselectivity of the direct aldol reaction catalyzed by L-prolinamides is also clarified. Materials and Methods General. All starting materials were purchased from Acros Organics (Geel, Belgium) and used without further purification. NMR spectra were recorded on a 300-MHz spectrometer (Bruker, Fallanden, Switzerland). Optical rotations were measured on a Perkin Elmer 241 Polarimeter at 589 nm. Fourier transform-ion cyclotron resonance mass spectra were recorded on the P-SIMS-Gly (Bruker Daltonics, Billerica, MA). HPLC analysis was performed on Waters-Breeze (2487 Dual Absorbance Detector and 1525 Binary HPLC Pump). Chiralpak AS, AD, and OJ columns were purchased from Daicel Chemical Industries (Hong Kong, China). Chiral GC analysis was performed on Varian CP-3380 with a CP CHIPASIL-DEX column. Acetone was dried over anhydrous K 2 CO 3. Hexane and ethyl acetate for column chromatography were distilled before use. Detailed reaction conditions for the preparation of each compound and characterizations of compounds 1 3 are given in Supporting Text, which is published as supporting information on the PNAS web site. Typical Procedure for the Preparation of 2 and 3. Compounds 2 and 3 were prepared by conventional methods (36). N-Carbobenzyloxy- L-proline (2.0 g, 8.0 mmol) and triethylamine (0.81 g, 8.0 mmol) were dissolved in tetrahydrofuran (30 ml). The solution was cooled down to 0 C. To the solution was added ethylchloroformate (0.88 g, 8.0 mmol) dropwise for 15 min. After the solution was stirred for 30 min, amine (8.0 mmol) was added for 15 min. The resulting solution was stirred at 0 C for 1 h and at room temperature for another 16 h and then refluxed for 3 h. After it was cooled down to room temperature, the solution was diluted with ethyl acetate. After filtration and removal of solvent under reduced pressure, the residue was purified through column chromatography on silica gel (eluent, hexane ethyl acetate 2:1) to give Z-A. Compounds Z-A (1.0 g), 5% Pd C (0.1 g), and methanol (30 ml) were charged in a two-neck flask (100 ml). After being stirred under hydrogen (1 atm) for 1 h, the solution was filtered. After removal of solvent, the resulting residue was purified through column chromatography on silica gel (eluent, hexane ethyl acetate 2:1) to give 2 or 3. General Procedure for Aldol Reaction. To anhydrous acetone (1 ml) were added the corresponding aldehyde (0.5 mmol) and L- prolinamide 2 or 3 (20 mol%). After being stirred at 25 C for h (monitored by TLC), the reaction mixture was treated with saturated aqueous ammonium chloride. The layers were separated, and the aqueous layer was extracted with ethyl This paper was submitted directly (Track II) to the PNAS office. Abbreviation: ee, enantiomeric excess. Z.T. and F.J. contributed equally to this work. To whom correspondence may be addressed. gonglz@cioc.ac.cn or chydwu@ ust.hk by The National Academy of Sciences of the USA CHEMISTRY SPECIAL FEATURE cgi doi pnas PNAS April 20, 2004 vol. 101 no

2 Table 1. Direct aldol reactions of 4-nitrobenzaldehyde with acetone catalyzed by organic molecules 2 Entry L-Prolinamide (2) Yield,* % ee, % Entry L-Prolinamide (2) Yield, % ee, % *The reaction was carried out in neat acetone with a concentration of 0.5 M. Isolated yield based on the aldehyde. The ee values were determined by HPLC and the configuration was assigned as R by comparison of retention time. acetate. The combined organic layers were washed with brine and dried over anhydrous MgSO 4. After removal of solvent, the residue was purified through flash column chromatography on a silica gel (eluent, hexane ethyl acetate 1:3) to give the pure adducts. Supporting Information. Data Set 1, Supporting Text, Figs. 4 and 5, and Tables 4 6 are published as supporting information on the PNAS web site. Results and Discussion Direct Aldol Reaction of 4-Nitrobenzaldehyde with Acetone Catalyzed by L-Prolinamides. It has been reported that the acidic proton of proline is critical for the reactivity and stereoselectivity of the proline-catalyzed direct aldol reaction (10, 11). L-Prolinamide (2-pyrrolidine-carboxamide, 2a) was initially observed to be ineffective in catalyzing the direct aldol reaction (11). However, we found that in the presence of 20 mol% L-prolinamide (2a), the model reaction of 4-nitrobenzaldehyde with neat acetone proceeded cleanly to give 1a in 80% yield with 30% ee (Table 1, entry 1). This encouraged us to investigate the catalytic activity of a family of L-prolinamides 2b k. Table 1 summarizes the results. Most of L-prolinamides exhibit high catalytic activity for the reaction. The secondary amides 2b e with N-alkyl groups show low enantioselectivities of 15 23% ee, which might have resulted from a very weak hydrogen bond formed between the proton of amide groups of these compounds and the aldehyde. However, the secondary amides with N-aryl groups (2g l) show moderate stereoselectivities of 31 46% ee. In particular, the enantioselectivity increases as the aryl substituent varies from electron-donating to electron-withdrawing, which makes the NOH more acidic and thus a better hydrogen bond donor. It can thus be concluded that the amide group of these catalysts is directly involved in catalysis through hydrogen bonding with the aldehyde substrate. Direct Aldol Reaction of 4-Nitrobenzaldehyde with Acetone Catalyzed by L-Prolinamides with a Terminal Hydroxyl Group. The observation with catalysts 2 prompted us to study the direct aldol reaction catalyzed by L-prolinamides 3 derived from L-proline and, hydroxyamines, in the hope that the terminal hydroxyl group might form an additional hydrogen bond with the aldehyde substrate so that the catalytic activity and enantioselectivity might be increased. To our delight, this class of organic catalysts indeed exhibited high catalytic efficiency as shown in Table 2. Chiral catalyst 3a promoted the reaction with a high yield of 84% but with a moderate enantioselectivity of 46% ee (entry 1). It did not show an apparent difference in catalytic efficiency from its isomer 3b (75% yield, 48% ee, entry 2). Compound 3d afforded a level of stereo-control superior to its diastereomer 3c (entries 3 and 4). This finding indicates that the (S)-configuration of C (see 3a for labeling) matched the (S)-configuration of L-proline to enhance the stereochemical control. Catalysts 3g and 3h with (S)- conformation of C once again induced higher enantioselectivities (entries 7 and 8, 64% ee and 69% ee, respectively) than their diastereomers 3e and 3f (entries 5 and 6, 49% ee and 44% ee, respectively) that contain (R)-C. The (S)-configuration of C also contributed to the selection. The highest enantioselectivity of 69% ee was observed with (S,S,S)-pyrolidine-2-carboxylic acid 2 -hydroxyl-1,2 -diphenylethylamide (3h) (entry 8). Dipeptide models 3i l were readily prepared from the reaction of L-proline with the esters of natural amino acids L- phenylalanine, L-serine, and L-threonine. These dipeptide mod cgi doi pnas Tang et al.

3 Table 2. Direct aldol reactions of 4-nitrobenzaldehyde with acetone catalyzed by organic molecules 3 (20 mol%) at room temperature Entry L-Prolinamide (2) Yield,* % ee, % Entry L-Prolinamide (2) Yield, % ee, % *The reaction was carried out in neat acetone with a concentration of 0.5 M. Isolated yield based on the aldehyde. The ee values were determined by HPLC and the configuration was assigned as R by comparison of retention time. The reaction was performed at 0 C. The reaction was performed at 25 C CHEMISTRY SPECIAL FEATURE els were also found to promote the direct aldol reaction. As demonstrated in Table 2, the peptides 3j l with terminal hydroxyl groups (entries 10 12) provided higher enantioselectivities than 3i without a hydroxyl group (entry 9). Catalysts 3k and 3l, which differ in the ester group, gave similar enantioselectivities for the direct aldol reaction (entries 11 and 12). Comparisons of 3a, c, and f with 3j, k, and l indicate that the ester group in the dipeptide models causes higher enantioselectivity than an alkyl and aryl group. This outcome is once again due to the electron-withdrawing nature of the ester, which makes the NOH more acidic. The best catalyst 3h was studied further. On decreasing the reaction temperature to 25 C, the yield of the reaction of p-nitrobenzaldehyde with acetone was somewhat killed (66%), but the enantioselectivity increased significantly to 93% ee (Table 2, entry 8). For a comparison, when L-proline was used as the catalyst under the same reaction condition as for 48 h, a poor yield of 6% with 70% ee for the formation of 1a was observed. The difference is probably due to the poor solubility of proline in acetone. The generality of catalyst 3h in catalyzing direct aldol reactions with a variety of aldehydes including aromatic and aliphatic aldehydes was examined under optimal conditions. The results are shown in Table 3. (For the results of 3-h-catalyzed intramolecular aldol reaction, see Table 4.) The aldol reactions of the aldehydes with acetone took place smoothly and were catalyzed by 20 mol% of 3h to give aldol adducts in moderate-to-high yields with high enantioselectivities ranging from 78 to 99% ee. The observed enantioselectivities for both aromatic and aliphatic aldehydes appear to be systematically higher than those obtained with L-proline as the catalyst (10). This finding is attributed to the reduced temperature ( 25 C) because of the high catalytic activity of 3h. High enantioselectivities of up to 87% ee were achieved for -unbranched aldehydes (entries 18 20), although the yields were not high (12 47%). These aldehydes are also difficult substrates with proline as the catalyst (37). Very high enantioselectivities were achieved for -branched aldehydes (entries 13 17). In particular, for the reaction of cyclohexylformaldehyde, an enantioselectivity of 98% was still maintained even if the amount of catalyst 3h was reduced to 10 mol% and 5 mol% (entries 14 and 15), although the yield was somewhat reduced. 2-Butanone acted as an aldol donor to react with several aromatic aldehydes, such as 4-nitrobenzaldehyde, 4-cyanobenzaldehyde, and 4-chlorobenzaldehyde, giving adducts 1t v, which resulted from the nucleophilic attack of the methyl group of the ketone with high enantioselectivity of up to 92% ee (entries 21 23). For the three reactions another regioisomer resulting from the nucleophilic attack of the ethyl group of the ketone was possible. We only isolated the regioisomer of 1t with 5% yield. Theoretical Study of the Transition Structures and Mechanism. The mechanism of direct aldol reaction catalyzed by proline has been studied previously by quantum mechanics calculations (38 41). The stereochemistry is believed to be determined by the addition of aldehyde to enamine intermediate formed by condensation of acetone with proline (42 43). The hydrogen bonding involving the proline acid group and the aldehyde substrate has been considered to be important to the high enantioselectivity (40, 41). Analogous to this, we have studied the reactions of enamines 4 6 with benzaldehyde by using quantum mechanics methods. All calculations were performed with the GAUSSIAN 98 program (44). Geometries were fully optimized with the HF 6 31G* method. Vibration frequencies were calculated for each structure with the same method and each transition structure was confirmed with one imaginary frequency. The energy of each structure was evaluated with the B3LYP 6 31G** method. The solvent effect of acetone was estimated with the PCM model (45). The calculated activation energies in Figs. 1 3 are the relative energies (B3LYP 6 31G*, acetone solvent) of the transition structures with respect to the corresponding reactants corrected by calculated zero-point energies. (Detailed energetic information is given in Tables 5 and 6.) Some transition structures such as 5-TS1-(R), 5-TS2-(R), 6-TS1-(R), and 6-TS2-(R) Tang et al. PNAS April 20, 2004 vol. 101 no

4 Table 3. Direct aldol reactions of acetone with aldehydes by chiral organic catalyst 3h Entry Product R 1 R 2 Yield,* % ee, % 1 1a 4-NO 2 C 6 H 4 -CH b 3-NO 2 C 6 H 4 -CH c 2-NO 2 C 6 H 4 -CH d 4-ClC 6 H 4 -CH e 2-ClC 6 H 4 -CH f 2,6-Dichloro-C 6 H 3 -CH g 4-FC 6 H 4 -CH h 4-BrC 6 H 4 -CH i 4-CNC 6 H 4 -CH j Ph -CH k 4-MeC 6 H 4 -CH l -Naphthyl -CH m -Naphthyl -CH n c-c 6 H 11 -CH n c-c 6 H 11 -CH n c-c 6 H 11 -CH o i-pr -CH p t-bu -CH q n-pr -CH r n-bu -CH s i-bu -CH t 4-NO 2 C 6 H 4 -Et u 4-CNC 6 H 4 -Et y 4-ClC 6 H 4 -Et The reaction was carried out in neat acetone with a concentration of 0.5 M at 25 C for h (see Supporting Text). *Isolated yields. Determined by HPLC. Catalyzed by 10 mol% 3h. Catalyzed by 5 mol% 3h. Determined by GC. were also located with the B3LYP 6 31G* method (Fig. 4). The calculated geometries and activation energies were very similar to those calculated based on the HF 6 31G* geometries. However, transition structures for (S)-configuration products could not be obtained. Therefore, these structures are not reported. Fig. 1. Calculated transition structures for the reaction of enamine 4 with benzaldehyde. In 4-TS1-(R), the phenyl group of benzaldhyde is antiperiplanar to the CAC double bond of the enamine. Good staggering occurs around the forming COC bond, as indicated by the dihedral angle OACOCAC of 51. This structure does not suffer from a steric interaction. On the other hand, the phenyl group in 4-TS1-(S) is gauche to the CAC double bond, and it has some steric interaction with the amide group. The distance between the amide hydrogen and one of phenyl hydrogen atoms is 2.51 Å. Thus, the OACOCAC dihedral angle is reduced to 42 to reduce the steric interaction. 4-TS1-(S) is calculated to be less stable than 4-TS1-(R) by 1.2 kcal mol (1 kcal 4.18 kj). This energy difference is similar to that calculated for proline catalysis (41). For the reaction of enamine 5 with benzaldehyde, four transition structures are shown in Fig. 2. The NOCOCOO dihedral angle in 5-TS1-(R) and 5-TS-(S) is positive gauche, but in 5-TS2-(R) and 5-TS2-(S) it is negative gauche. These structures are similar to 4-TS1-(R) and 4-TS1-(S), respectively, except for the additional hydrogen bond formed between the terminal hydroxyl group and the oxygen of benzylaldehyde. Although 5-TS1-(R) and 5-TS2-(R) have similar energies, both giving the Shown in Fig. 1 are the transition structures for the reaction of enamine 4 with benzaldehyde. The enamines can adopt two conformations, anti and syn, as shown in 4. Transition structures are located in both conformations for 4. Similar to the results reported for proline catalysis (where the amide group is replaced by an acid group), the transition structures with syn-enamine (4-TS2) are much higher in energy than those with anti-enamine (4-TS1). They can be excluded in calculating the enantioselectivity. Therefore, for the reactions of 5 and 6, only the anticonformations were studied. The amide forms a good hydrogen bond with the carbonyl oxygen of benzaldehyde in both 4-TS1-(R) and 4-TS1-(S) (46). Fig. 2. Calculated transition structures for the reaction of enamine 5 with benzaldehyde. The favored positions for substituents are labeled in red cgi doi pnas Tang et al.

5 Fig. 3. Calculated transition structures for the reaction of enamine 6 with benzaldehyde. (R)-product, 5-TS1-(S), which gives the (S)-product, is 0.7 kcal mol less stable. The four transition structures for the reaction of enamine 6 with benzaldehyde shown in Fig. 3 were derived by adding two phenyl groups to the four transition structures shown in Fig. 2 for the reaction of enamide 5 with benzaldehyde. In 6-TS1-(R) and 6-TS1-(S), the two benzyl groups are gauche to each other. To avoid steric interactions between them, the NOCOCOO dihedral angle is reduced by 20 so that the two phenyl groups are more separated. A rotation also occurs around the COC forming bond so that the OACOCAC dihedral angle is reduced. The rotation is particularly large in 6-TS1-(S) so that the OACOCAC dihedral angle is now only 6. Because of this finding, 6-TS1-(S) becomes 2.6 kcal mol less stable than 6-TS1-(R), compared with the destabilization of 5-TS1-(S) over 5-TS1-(R) by only 0.7 kcal mol. In 6-TS2-(R) and 6-TS2-(S), the two phenyl groups are anti to each other. The calculations indicate that 6-TS2-(R) is much more stable than the other structures. In particular, it is 5 kcal mol more stable than the two transition structures, 6-TS1-(S) and 6-TS2-(S), that give the (S)-product. Transition structures in which the terminal hydroxyl group forms a seven-member-ring hydrogen bond with the amide carbonyl group were also located. As shown in Fig. 5, these structures are 6 kcal higher in energy than 6-TS2-(R) and, therefore, can be excluded. Bahmanyar and Houk (47) reported a detailed theoretical study on the reaction of simple enamines with aldehydes. For the reaction of acetaldehyde with CH 2 ACHONMe 2, a high activation energy of 33.5 kcal mol was calculated by the B3LYP 6 31G* method in the gas phase. The activation energy was reduced to 20 kcal mol if the effect of the water solvent (e 78) was included with the conductor-like polarizable continuum model solvent model. Our calculated activation energy for the reaction of benzaldehyde with enamine 4 [4-TS1-(R)] is 10.2 kcal mol in the gas phase and 10.0 kcal mol in acetone. Thus, the hydrogen bond between the amide and benzaldehyde significantly reduces the reaction activation energy. The effect of the amide is similar to that of the proline acid. This is somewhat surprising because the acidity of amide is much lower than that of acid. It should be noted that the acid prefers a synconformation (OACOOOH 0 ), but has to adopt an anticonformation (OACOOOH 180 ) in the transition structure to form the hydrogen bond. This conformational change causes 6 8 kcal mol destabilization (48), reducing the catalytic ability of the acid. On the other hand, such a conformational change is not needed for amide. The calculated activation energies for the reaction of benzaldehyde with 5 and 6 are both 7 kcal mol. Thus, the hydrogen bond formed between the terminal hydroxyl group and benzaldehyde also reduces the activation energy. Although this assessment is qualitative, it does support the important concept of hydrogen bond catalysis, which has been demonstrated recently for several other systems (49 51). In the proline-catalyzed direct aldol reaction, the calculated enantioselectivity with a hydrogen bond model reproduces the experimentally observed enantioselectivity very well (41). Our current calculations for prolinamide-catalyzed direct aldol reaction with the same hydrogen bond model appear to predict enantioselectivities higher than those observed experimentally. Two possibilities might exist for this discrepancy. The transition structures were located in the gas phase, which overestimates the hydrogen-bonding strength. As a result, the binding of an aldehyde substrate to the amide is too tight, causing too much steric destabilization between the amide and the phenyl group of benzylaldehyde in the transition structures for the minor product. Another possibility exists. The fact that the tertiary amide 2f can effectively catalyze the direct aldol reaction but gives only a low enantioselectivity suggests that a competitive reaction mechanism might occur, which does not involve the hydrogen bond by the amide and gives a low enantioselectivity. Since the abovementioned calculated activation energy for the reaction of acetaldehyde with CH 2 ACHONMe 2 is high (47), and the reaction barrier should be even higher when the loss of entropy for a bimolecular reaction is considered, we suspect that the water generated in the condensation reaction of prolinamide and acetone might be involved in the competitive reaction. This possibility requires further investigation through calculations. To further support the proposed transition state, parallel aldol reactions of 4-nitrobenzaldehyde with acetone catalyzed by organic molecules 7 9 that were derived from catalyst 3g were conducted. Compound 7, in which the proton of the amide in 3g was replaced by a methyl group, catalyzed the aldol reaction to give the aldol adduct with 58% yield and 16% ee. If the terminal hydroxyl group of 3g was methylated, the resulting catalyst 8 mediated the reaction in 29% ee, similar to that given by 2e. Given that the protons of the amide and hydroxyl groups of 3g were replaced by a methyl group, the corresponding organic molecule 9 catalyzed the aldol reaction with 11% ee. All of catalysts 7 9 offered much lower enantioselectivity than did 3g. These results indicate that the terminal hydroxyl group serves as a hydrogen bond donor instead of a hydrogen bond acceptor; it forms a hydrogen bond with the aldehyde substrate together with the amide NOH group. These observations are in full agreement with the model that we presented in Figs. 2 and 3. CHEMISTRY SPECIAL FEATURE Tang et al. PNAS April 20, 2004 vol. 101 no

6 The models shown in Fig. 2 also qualitatively explain the observed trend of enantioselectivities caused by catalysts 3a h (Table 2). Based on steric crowdedness, the favored positions (in red) for substitutions are a and c in 5-TS1-(R) and 5-TS1-(S) and b, c, and d in 5-TS1-(R). 5-TS2-(S) can be excluded because of its high instability. If a substitution pattern is favorable in 5-TS2-(R) but is unfavorable in 5-TS2-(S), stereoselectivity should be increased. That is the case for 3b, 3d, 3g, and 3h because it would have to put a substituent at the position b of 5-TS1-(S). The results of our calculation shown in Fig. 3 for 3h catalysis (enamine 6) support this analysis. Summary. We have found that L-prolinamides prepared from L-proline and amines are active catalysts for the direct aldol reaction of both aromatic and -branched aliphatic aldehydes with neat acetone. The enantioselectivity is low ( 23% ee) for tertiary amide and secondary amides with a simple N-alkyl group. It is intermediate (31 46% ee) for aromatic secondary amides. The enantioselectivity is increased with increasing acidity of the amide NOH and in the presence of a terminal hydroxyl group. Excellent enantioselectivities are obtained with catalyst 3h, prepared from L-proline and (1S,2S)-diphenyl-2-aminoethanol and under 25 C reaction conditions. Quantum mechanics calculations on the transition structures of model reactions of benzaldehyde with enamines reveal that the amide NOH and the terminal hydroxyl groups form hydrogen bonds with the benzaldehyde substrate. These hydrogen bonds act simultaneously and are important for high enantioselectivity. Methylation of the NOH and OOH groups to NOMe and OOMe results in low enantioselectivity. The present study suggests a previously unreported strategy in the design of organic catalysts for direct asymmetric aldol reactions and related transformations, because plentiful chiral sources containing multihydrogen bond donors, for example, peptides, might be adopted in the design. This work was supported by National Natural Science Foundation of China Grants and (to L.-Z.G.) and (to Y.-D.W.) and Research Grants Council of Hong Kong Grant HKUST P (to Y.-D.W.). 1. Kim, B. M., Williams, S. F. & Masamune, S. (1991) in Comprehensive Organic Synthesis, eds. Trost, B. M., Fleming, I. & Heathcock, C. H. (Pergamon, Oxford), Vol. 2, pp Carreira, E. M. (1999) in Comprehensive Asymmetric Catalysis, eds. Jacobsen, E. N., Pfaltz, A. & Yamamoto, H. (Springer, Heidelberg), Vol. 3, pp Denmark, S. E., Wong, K. T. & Stavenger, R. A. (1997) J. Am. Chem. Soc. 119, Denmark, S. E. & Stavenger, R. A. (2000) Acc. Chem. Res. 33, Trost, B. M. (1991) Science 254, Yamada, Y. M. A., Yoshikawa, N., Sasai, H. & Shibasaki, M. (1997) Angew. Chem. Int. Ed. Engl. 36, Yoshikawa, N., Yamada, Y. M. A., Das, J., Sasai, H. & Shibasaki, M. (1999) J. Am. Chem. Soc. 121, Trost, B. M.& Ito, H. (2000) J. Am. Chem. Soc. 122, Trost, B. M., Ito, H. & Silcoff, E. R. (2001) J. Am. Chem. Soc. 123, List, B., Lerner, R. A. & Barbas, C. F., III (2000) J. Am. Chem. Soc. 122, Sakthivel, K., Notz, W., Bui, T. & Barbas, C. F., III (2001) J. Am. Chem. Soc. 123, Gröger, H. & Wilken, J. (2001) Angew. Chem. Int. Ed. Engl. 40, Dalko, P. I. & Moisan, L. (2001) Angew. Chem. Int. Ed. Engl. 40, Jarvo, E. R. & Miller, S. J. (2002) Tetrahedron 58, List, B. (2002) Tetrahedron 58, Notz, W. & List, B. (2000) J. Am. Chem. Soc. 122, List, B. (2000) J. Am. Chem. Soc. 122, Còrdova, A., Notz, W., Zhong, G., Betancort, J. M. & Barbas, C. F., III (2002) J. Am. Chem. Soc. 124, Kumaragurubaran, N., Juhl, K., Zhuang, W., Bøgevig, A. & Jørgensen, K. A. (2002) J. Am. Chem. Soc. 124, List, B. (2002) J. Am. Chem. Soc. 124, Bøgevig, A., Juhl, K., Kumaragurubaran, N., Zhuang, W. & Jørgensen, K. A. (2002) Angew. Chem. Int. Ed. Engl. 41, Northrup, A. B. & MacMillan, D. W. C. (2002) J. Am. Chem. Soc. 124, Còrdova, A., Watanabe, S., Tanaka, F., Notz, W. & Barbas, C. F., III (2002) J. Am. Chem. Soc. 124, List, B., Pojarliev, P., Biller, W. T. & Martin, H. J. (2002) J. Am. Chem. Soc. 124, Paras, N. A. & MacMillan, D. W. C. (2002) J. Am. Chem. Soc. 124, Okino, T., Hoashi, Y. & Takemoto, Y. (2003) J. Am. Chem. Soc. 125, Halland, N., Hansen, T. & Jørgensen, K. A. (2003) Angew. Chem. Int. Ed. Engl. 42, Kobayashi, S., Ogawa, C., Konishi, H. & Sugiura, M. (2003) J. Am. Chem. Soc. 125, Malkov, A. V., Dufková, L., Farrugia, L. & Kočovský, P. (2003) Angew. Chem. Int. Ed. Engl. 42, Brown, S. P., Brochu, M. P., Sinz, C. J. & MacMillan, D. W. C. (2003) J. Am. Chem. Soc. 125, Brown, S. P., Goodwin, N. C. & MacMillan, D. W. C. (2003) J. Am. Chem. Soc. 125, Pidathala, C., Hoang, L., Vignola, N. & List, B. (2003) Angew. Chem. Int. Ed. Engl. 42, Dickerson, T. J. & Janda, K. D. (2002) J. Am. Chem. Soc. 124, Ooi, T., Taniguchi, M., Kameda, M. & Maruoka, K. (2002) Angew. Chem. Int. Ed. Engl. 41, Tang, Z., Jiang, F., Yu, L.-T., Cui, X., Gong, L.-Z., Mi, A.-Q., Jiang, Y.-Z. & Wu, Y.-D. (2003) J. Am. Chem. Soc. 125, Rhyoo, H. Y., Yoon, Y. A., Park, H. J. & Chung, Y. K. (2001) Tetrahedron Lett. 42, List, B., Pojarliev, P. & Castello, C. (2001) Org. Lett. 3, Arno, M. & Domingo, L. R. (2002) Theor. Chem. Acc. 108, Hoang, L., Bahmanyar, S., Houk, K. N. & List, B. (2003) J. Am. Chem. Soc. 125, Rankin, K. N., Gauld, J. W. & Boyd, R. J. (2002) J. Phys. Chem. A. 106, Bahmanyar, S., Houk, K. N., Martin, H. J. & List, B. (2003) J. Am. Chem. Soc. 125, Raymond, J.-L. & Chen, Y. (1995) J. Org. Chem. 60, Còrdova, A., Notz, W. & Barbas, C. F., III. (2002) J. Org. Chem. 67, Frisch, M. J., Trucks, G. W., Schlegel, H. B., Scuseria, G. E., Robb, M. A., Cheeseman, J. R., Zakrzewski, V. G., Montgomery, J. A., Stratmann, R. E., Burant, J. C., et al. (1998) GAUSSIAN 98 (Gaussian, Pittsburgh), Revision A Tomasi, J. & Persico, M. (1994) Chem. Rev. 94, Bach, T., Bergmann, H. & Harms, K. (1999) J. Am. Chem. Soc. 121, Bahmanyar, S. & Houk, K. N. (2001) J. Am. Chem. Soc. 123, Evanseck, J. O., Houk, K. N., Briggs, J. M. & Jorgensen, W. T. (1994) J. Am. Chem. Soc. 116, Huang, Y. & Rawal, V. H. (2002) J. Am. Chem. Soc. 124, Vachal, P. & Jacobsen, E. N. (2002) J. Am. Chem. Soc. 124, Huang, Y., Unni, A. K., Thadani, A. N. & Rawal, V. H. (2003) Nature 424, cgi doi pnas Tang et al.