Enzymatic Synthesis of 3-O-Acylbetulinic Acid Derivatives and Prediction of Acylation Using Response Surface Methodology and Artificial Neural Network Analyses

In this study, 3-O-acyl-betulinic acid derivatives were synthesized by the reaction of betulinic acid with various anhydrides using lipase as a biocatalyst in organic solvents. The reaction between betulinic acid and phthalic anhydride was chosen as the model reaction for optimization studies. The immobilized lipase from Candida antarctica (Novozym 435) was selected as a biocatalyst. The effects of different reaction parameters were investigated and optimized in the model reaction using one-variable-ata- time technique for the first time. Optimum conditions to produce 3-O-phthalylbetulinic acid up to 61.8% were observed at a reaction time of 24 hours; amount of enzyme, 176 mg; betulinic acid to phthalic anhydride molar ratio of 1:1; amount of celite, 170 mg and 6 mg of K2CO3 in a mixture of n-hexane-chloroform (1:1, v/v) as organic solvent at 55'C. The response surface methodology (RSM), based on a five-level, four-variable central composite rotatable design (CCRD), was employed to evaluate the effects of synthesis parameters of the model reaction. Using the RSM analysis, it was observed that the maximum yield of 3-O-phthalyl-betulinic acid (65.8%) could be obtained using 145.6 mg of enzyme, reaction temperature of 53.9°C, reaction time of 20.3 hours and betulinic acid to phthalic anhydride molar ratio of 1:1.11. The actual experimental value obtained was at 64.7%. Artificial neural network (ANN) was successfully developed to model and predict the enzymatic synthesis of 3-O-phthalyl-betulinic acid. The network consists of an input layer, a hidden layer and an output layer. Inputs for the network were reaction time, reaction temperature, enzyme amount and substrate molar ratio, while the output was percentage isolated yield of ester. Four different training algorithms, belonging to two classes, namely gradient descent and Levenberg-Marquardt, were used to train ANN. The best results were obtained from the quick propagation algorithm (QP) with 4-9-1 topology. Based on the ANN analysis, the optimal conditions to obtain the highest yield were 148.3 mg enzyme, reaction temperature of 53.1°C, reaction time of 20.3 hours and betulinic acid to phthalic anhydride molar ratio of 1:1.24. The predicted and actual yields were 64.9 and 64.3%, respectively. In this work, the ANN and RSM analysis were investigated on the enzymatic synthesis of 3-O-phthalyl-betulinic acid for the first time. Finally, several betulinic acid esters (compounds 57-66) were synthesized using the optimal operation conditions which were obtained by the RSM technique. Esterification of betulinic acid with various anhydrides was performed at 54ºC in a mixture of n-hexane- chloroform (1:1, v/v) for 20.3 hours, catalyzed by Novozym 435, gave 24.7 to 79.3% yield. Five new compounds (58, 62, 64, 65 and 66) were synthesized for the first time in the present study. In brief, the anti-cancer activity of betulinic acid (1) and its 3-O-acylated derivatives (compounds 57-66) were evaluated against human lung carcinoma (A549) and human ovarian (CAOV3) cancer cell lines. In particular, compounds (59), (61) and (63) were found to show IC50 < 10 μg/ml against A549 cancer cell line tested and showed better cytotoxicity than betulinic acid. In the ovarian cancer cell line, all betulinic acid derivatives prepared revealed weaker cytotoxicity than betulinic acid