Protease-catalysed Direct Asymmetric Mannich Reaction in Organic Solvent

We reported the first enzyme-catalysed, direct, three-component asymmetric Mannich reaction using protease type XIV from Streptomyces griseus (SGP) in acetonitrile. Yields of up to 92% with enantioselectivities of up to 88% e.e. and diastereoselectivities of up to 92:8 (syn:anti) were achieved under the optimised conditions. This enzyme's catalytic promiscuity expands the application of this biocatalyst and provides a potential alternative method for asymmetric Mannich reactions.

model reaction as non-enzyme proteins, which produced the Mannich products in 23% yield with 0% e.e. and 21% yield with 7% e.e., respectively ( Table 1, entries 3 and 4). These reactions excluded the possibility of protein catalysis, meaning that catalysis was not simply a result of the amino acid residues on the surface of the protein. Furthermore, urea-denatured SGP was used to catalyse the model reaction, which gave a high yield of 87% with only 8% e.e. (Table 1, entry 5). The same amount of urea was then used to catalyse the reaction, but produced a result nearly indential to the blank, proving that urea alone did not catalyse this transformation (Table 1, entry 6). These results show that the urea-denatured SGP still had catalytic activity towards the Mannich reaction, but it almost completely lost its enantioselectivity. Metal-denatured SGP was also used to catalyse the model reaction to determine whether the metal ion could disrupt the bonds that hold the enzyme together and cause the enzyme to undergo a conformational change, disrupt the active site and ultimately denature the enzyme. SGP was pretreated with Cu 21 and Ag 1 at different concentrations. A low concentration (2.5 mM) of Cu 21 or Ag 1 did not have an obvious effect on the activity and selectivity of SGP towards the Mannich reaction (Table 1, entries 7 and 10), while a moderate concentration (25 mM) of Cu 21 or Ag 1 caused a slight decrease in enantioselectivity (Table 1, entries 8 and 11). A higher concentration (250 mM) of Cu 21 or Ag 1 almost completely destroyed the selectivity of SGP in the model Mannich reaction (Table 1, entries 9 and 12). From the above control experiments with urea or metal ion-denatured SGP, we determined that the denatured SGP still exhibited catalytic activity in the Mannich reaction, but it lost nearly all of its stereoselectivity, indicating that the specific natural fold of SGP is responsible for its stereoselectivity in Mannich reactions.
Optimisation of reaction conditions. Next, we explored the effects of different solvents on the SGP-catalysed model Mannich reaction ( Table 2). The reaction medium played an important role in this enzymatic reaction. The highest selectivity of 82% e.e. for the syn isomer (85:15 dr, syn:anti) was obtained with a moderate yield in MeCN ( Table 2, entry 1), whereas water gave the highest yield of 76% with low selectivity ( Table 2, entry 12). The reaction in DMF gave the lowest yield of 16% (Table 2, entry 11) and the reaction in DMSO provided the lowest enantioselectivity of only 6% e.e. with reversed diastereoselectivity (36:64 dr, syn:anti) ( Table 2, entry 15). Generally, no clear correlation was observed between solvent polarity and enzyme activity or selectivity. Thus, to optimise selectivity, we selected MeCN as the solvent for the SGP-catalysed Mannich reaction.  The mixture of SGP (50 mg), deionised water (1 mL) and the specified amount of CuSO 4 (for entries [7][8][9] or AgNO 3 (for entries [10][11][12] was stirred at 30uC for 24 h and then water was removed under reduced pressure before use. Water content, pH, molar ratio of substrates and temperature are also important factors in enzymatic reactions [22][23][24][25] and their influence on the SGP-catalysed model Mannich reaction was investigated next. The optimised reaction conditions were found to consist of the following: a water content of 0.1 [H 2 O/(H 2 O 1 MeCN), v/v], a molar ratio of 15:1 (cyclohexanone to 4-nitrobenzaldehyde) and a temperature of 30uC. The effect of pH was investigated using a phosphate buffer (NaH 2 PO 4 -Na 2 HPO 4 , 0.2 M, pH 4.60-7.18) to replace the optimised water content in the reaction system [buffer/(buffer 1 MeCN) 5 0.1, v/v], but yields and selectivities were not improved in the presence of buffer. Therefore, we ultimately chose MeCN/H 2 O as the reaction medium for the SGP-catalysed Mannich reaction. Finally, the time course of the SGP-catalysed model Mannich reaction was also investigated with respect to the above-mentioned variables in this section. (for details, please see Supplementary Tables S3-S7).
Investigation of substrate scope. To investigate the generality and scope of this biocatalytic promiscuity, several other substrates were tested to expand upon this novel SGP-catalysed direct asymmetric Mannich reaction. Various substituted aromatic aldehydes performed well in the reaction. Generally, the aromatic aldehydes containing an electron-withdrawing group exhibited better enantioselectivity and diastereoselectivity than those containing an electron-donating group ( Table 3, entries 1-9). Substituent position also had some impact on the selectivity of the reaction. For instance, 3-fluorobenzaldehyde yielded better diastereoselectivity and enantioselectivity than 4-fluorobenzaldehyde (Table 3, entries 5 and 6). In addition to aniline, various substituted arylamines were also used as substrates (Table 3, entries 10-15). Among them, the reaction of 3-bromobenzenamine with 4-nitrobenzaldehyde and cyclohexanone gave the best enantioselectivity at 88% e.e. and the best diastereoselectivity of 92:8 (syn:anti) ( Table 3, entry 10). Furthermore, it was promising that the heteroatom-containing cycloketone was also available as a substrate in this biocatalytic process. The reactions of tetrahydrothiopyran-4-one with aniline and either 4-chlorobenzaldehyde or 4-(trifluoromethyl)benzaldehyde gave satisfactory yields and moderate e.e. values (Table 3, entries 16 and 17). In addition, linear chain ketones could be used as substrates. Acetone, for example, gave a low yield of 22% and no enantioselectivity when reacting with 4-nitrobenzaldehyde and 4-methoxybenzenamine. In almost all cases, the aldol reaction products were observed as side products. Moreover, SGP showed different degrees of enantioselectivity for syn-isomers, but low or no enantioselectivity for anti-isomers, potentially indicating that SGP has a specific selectivity for the Mannich reaction.

Discussion
In conclusion, we reported a novel SGP-catalysed, direct, three-component asymmetric Mannich reaction. The control experiments with the denatured enzyme and non-enzyme proteins indicated that the specific natural fold of SGP was responsible for its stereoselectivity in the Mannich reaction. The influence of several factors, including solvent, water content, pH, molar ratio of substrates and temperature, was investigated. A wide range of substrates were accepted by the enzyme and yields of up to 92%, enantioselectivities of up to 88% e.e. and diastereoselectivities of up to 92:8 dr were achieved. As an example of enzyme catalytic promiscuity, this work broadens the scope of SGP-catalysed transformations. Exploring the untapped catalytic potential of natural enzymes could give useful insights into enzyme evolution.

Methods
Materials. The protease type XIV from Streptomyces griseus was purchased from Sigma-Aldrich (catalogue no. P5147, 6.1 U/mg). One unit will hydrolyse casein to