Mining Enzyme Diversity of Transcriptome Libraries through DNA Synthesis for Benzylisoquinoline Alkaloid Pathway Optimization in Yeast

The ever-increasing quantity of data deposited to GenBank is a valuable resource for mining new enzyme activities. Falling costs of DNA synthesis enables metabolic engineers to take advantage of this resource for identifying superior or novel enzymes for pathway optimization. Previously, we reported synthesis of the benzylisoquinoline alkaloid dihydrosanguinarine in yeast from norlaudanosoline at a molar conversion of 1.5%. Molar conversion could be improved by reduction of the side-product <i>N</i>-methylcheilanthifoline, a key bottleneck in dihydrosanguinarine biosynthesis. Two pathway enzymes, an <i>N</i>-methyltransferase and a cytochrome P450 of the CYP719A subfamily, were implicated in the synthesis of the side-product. Here, we conducted an extensive screen to identify enzyme homologues whose coexpression reduces side-product synthesis. Phylogenetic trees were generated from multiple sources of sequence data to identify a library of candidate enzymes that were purchased codon-optimized and precloned into expression vectors designed to facilitate high-throughput analysis of gene expression as well as activity assay. Simple <i>in vivo</i> assays were sufficient to guide the selection of superior enzyme homologues that ablated the synthesis of the side-product, and improved molar conversion of norlaudanosoline to dihydrosanguinarine to 10%