The preparation of new proline derivatives 2, 3, 5 and 6 is displayed in Scheme 1. Compound 2 was synthesized starting from amine 1 [9] with N-Boc-(S)-proline to give 7 followed by elimination of the Boc protecting group. In the case of 3, the amidation was carried out through the same pathway starting with the corresponding epimer of 1 derived from (1S,2S)-1,2-diaminocyclohexane. The proline derivative 9 was produced by condensation of commercially available 2-nitrobenzenesulfonamide with N-Boc-(S)-proline while 10 and 11 [18] resulted from the reaction of N-Boc-prolinamine [19] with respectively 2- and 4-nitrobenzenesulfonyl chloride. After acid deprotection, catalysts 4, 5 and 6 were obtained in very good yields. It should be noted that the 4-nitrophenyl regioisomer of 4 was already described [20] .
2.1. Aldol ReactionWith the new derivatives 1 - 6 in hand, we began to evaluate their catalytic behavior in the classical aldol reaction between acetone and 4-nitrobenzaldehyde [21] , and the results are displayed in Table 1. As can be seen, the reaction proceeded smoothly at room temperature under solvent free conditions and was almost total only beyond several dozen hours, except for catalyst 2 (entry 2). If the chemical yields were acceptable (catalysts 2 and 3), the enantiomeric excesses were very low except for catalyst 2. Indeed in that case, a very interesting enantioselectivity of the aldol product was obtained. Regarding the results, we observed the following facts: firstly, the catalyst 2 having proline moieties exhibited higher catalytic efficiencies compared to 1 (entries 1, 2); then, we could notice a strong mismatch effect related to the absolute stereochemistry of the cyclohexane moiety (entries 2, 3). After these initial investigations, catalyst 2, giving the best result, was selected to carry out the optimization of the reaction conditions. We first investigated the solvent effect with or without some additives and the results are displayed in Table 1. While no reaction occurred in aprotic polar solvents (entries 4, 5), apolar solvents (entries 6, 7) induced a significant decrease in catalytic efficiency. Moreover, a catalytic amount of TfOH (entry 9) induced a decrease both rate and yield without significantly affecting the enantiomeric excess. On the other hand, catalytic quantities of AcOH (entry 8) induced an increase of the yield, but involved a slightly decreasing of the enantiomeric excess. With water as additive (entries 10, 11) no significant variations were observed in the reaction rate and enantioselectivity. The reactions, carried out at lower temperatures, have shown as expected a positive effect on the catalytic efficiency and thus the best results were obtained at −20˚C but requiring prolonged reaction time (entries 12, 15). Regarding the loading of the catalyst, the reactions carried out with only 10 mol% of 2 (entries 13 - 15) provided the aldol product with almost the same enantioselectivity but with an expected decreasing of the reaction rate. In order to exclude a quite possible retro-aldol process that could lead to partial racemization, the enantiomeric excess of the reaction was measured several times throughout the reaction, showing no significant variation in value.
To explain the experimental results and especially the different behavior between catalysts 2 and 3, we proposed transition states displayed in Figure 2 for the aldol reaction. Thus, the enamine resulting from the condensation of the catalyst (S,R,R)-2 with acetone adding to the aromatic aldehyde forms a pseudo-cycle stabilized by strong binding interactions between the carbonyl oxygen and two acidic hydrogen atoms. In this cycle, the largest substituent “Ar” is located at the equatorial position. This compact and rigid transition state can explain the relative high enantiomeric excess (78%) promoting the R-isomer of the aldol product. With the isomer (S,S,S)-3 this binding interaction can arise with only one hydrogen, the second one being pushed to the rear of the figure induced by the reverse stereochemistry of the cyclohexane moiety. These very different behaviors between 2 and 3 allowed us to highlight one main characteristic required for catalysts, namely the ability to form a compact cyclic transition state induced by the presence of acidic hydrogen atoms [22] . Following these observations, we thought design proline derivatives 4, 5 and 6 having acidic hydrogen of a sulfonamide function. In addition, one could be expected a significant difference between the catalysts, given the presence in the catalyst 4 of an additional carbonyl group increasing the hydrogen acidity [23] . The experimental results of aldol reactions are displayed in Table 1. The low solubility of catalyst 4 in apolar solvents has limited the range of sol-