3.1 Preparation for appropriate catalysts
In 1909, Dakin found that primary amino acids can catalyze Knoevenagel condensation (DAKIN, 2009). The fact that bovine serum albumin (BSA), which slightly catalyzed the condensation between benzaldehyde and ethyl cyanoacetate in water phase (W. Li et al., 2015) encouraged us to investigate other enzymes which may also possess the ability to promote Knoevenagel condensation. To test the idea, the enzymes NerA (CAA74280), OYE2.6 (XP_001384055), IPR (Q6WAU1), GDH (WP_003246720), and ADH4 (XP_001387122) were chosen to compare with glycine (Chaudhry et al., 2016), proline, and lysine (Y. Wang, Shang, Wu, Fan, & Chen, 2006; Y. Zhang, Sun, Liang, & Shang, 2010), as catalysts in the model reaction (Figure S3). To date, it hasn’t been recorded in the literature that these enzymes have a direct role in Knoevenagel condensation and natural amino acids can promote the reaction obviously. For the pretest, ethyl acetoacetate (EAA) and valeraldehyde 1b were used as substrates. As shown in Table S1, both enzymes and amino acids were able to efficiently promote Knoevenagel condensation in neat aqueous solvent. Interestingly, the yield of the reaction was not significantly affected by the species of enzymes (Table S1, list 4-8), but were positively correlated with the amount of catalyst (Table S1, list 4-8 and 9-13).
According to the literature (W. Li et al., 2015), lipases like PPL, MJL, YILip2 can also promote Knoevenagel condensation in a water-ethanol solvent system. 10 mg mL-1 enzymes and 96.7% ethanol by volume were employed to catalyze the reaction efficiently. Interestingly, unspecific reaction in lipase-catalyzed Knoevenagel condensation was also observed in addition to specific reaction. Our work demonstrated that several enzymes could catalyze the condensation, with the exception of lipases. 2.4 mg mL-1 enzymes in neat aqueous solvent maintained the ability to catalyze Knoevenagel condensation. On the basis of the mechanism as reported by Knoevenagel, we thought that the hydrophilic amino acids on the surface of enzymes were associated with the formation of a Schiff base with aldehyde substrates, before the obtained imine intermediates condensed with carbanions (Zheng, Li, Tao, & Zhang, 2019).
The Knoevenagel condensation catalyzed by a certain concentration of enzymes is comparable to that of the amino acids, which inspired us to consider that a low catalytic turnover would be further promoted if the condensation products could be converted by the same enzyme. Ene-reductases, which catalyze the reduction of α, β-unsaturated alkene, have highlighted the potential reduction of Knoevenagel condensation products. Subsequently, we tried to construct the tandem reaction using single ER which catalyzes both Knoevenagel condensation andin situ reduction of the condensation products to obtain more stable and valuable products. Several ERs were investigated to ascertain which one can reduce intermediates 2a -k . Following a review of the literature, three of ERs (NerA, OYE2.6 and IPR) were cloned and expressed (Oberdorfer et al., 2013; B. Zhang, Zheng, Lin, & Wei, 2016). The reduction activity of each catalyst towards intermediates 2a-k was then tested, shown in Figure 1, except that IPR had no obvious catalytic activity for any of the products, the other two reductases had wide substrate ranges. In particular, NerA presented a broader range of substrate species and higher activity on nearly all the products we were interested in. It’s worth noting that the products derived from the condensation of benzaldehyde (2f ) and furan aldehyde (2h , 2i ) with EAA were reduced effectively. These bulky substrates are often difficult to reduce using other ERs (Peters, Frasson, Sievers, & Buller, 2019). Phenyl group bound with the electron withdrawing groups (2g ) was reduced faster than 2f , which may facilitate hydrogen transfer between olefin and reduced flavin cofactor (de Paula, Zampieri, Nasário, Rodrigues, & Moran, 2017). OYE2.6 also had a wide reduction capacity, but the relative activity on all substrates was at least 40% lower than that of NerA. OYE2.6 had relatively low reduction activities, especially towards the large size compounds (2g-2k ), which may attribute to the smaller activity centre (Oberdorfer et al., 2013; B. Zhang, Zheng, Lin, & Wei, 2016). In addition, NerA takes NADH as a co-factor while the co-factor of OYE2.6 is NADPH. Considering economy and usability, NerA was chosen for the one-pot reaction presented here.