著者

田口 貴章 小澤 誠 Kimberley Meriel R. Booker-Milburn Kevin I. Stephenson G. Richard 海老塚 豊 市瀬 浩志

出版者

天然有機化合物討論会

雑誌

天然有機化合物討論会講演要旨集

巻号頁・発行日

no.44, pp.235-240, 2002-09-01

---

A class of Streptomyces aromatic polyketide antibiotics, the benzoisochromanequinone (BIQs) antibiotics all show trans stereochemistry at C-3 and C-15 in the pyran ring. The opposite stereochemical control is found in actinorhodin (3S, 15R, ACT) from S. coelicolor A3(2) and dihydrogranaticin (3R, 15S, DHGRA) from S. violaceoruber Tu22. A common bicyclic intermediate, which is produced by the early biosynthetic genes encoding a type II minimal polyketide synthase, C-9 ketoreductase (KR), aromatase, and cyclase, was postulated to undergo stereospecific reduction to provide either (S)-DNPA or (R)-DNPA. In the ACT biosynthesis, RED1 encoded by act VI-ORF 1 was proved to reduce C-3 of bicyclic intermediate to determine the 3-(S)-configuration of DNPA. Although the homolog of act VI-ORF 1 was not found in the gra cluster, RED2 was suggested to reduce bicyclic intermediate. An explored RED-2 coding gene, gra-6, was subjected to updated BLAST analysis. The gra-6 product, a putative short-chain alcohol dehydrogenase, has virtually no sequence similarity with RED1. Functional analysis of RED1/2 was made from the following points. 1) Introduction of gra-ORF 6 and gra-ORF 5 under translational coupling (gra-5+6) into the act VI-ORF 1 mutant, S. coelicolor B22, led to ACT-like pigmentation, demonstrating gra-ORF 6 to complement the function of act VI-ORF 1 possibly under unnatural stereochemical control. 2) Combinations of the ketoreductase genes were co-expressed with the early biosynthetic genes required for the bicyclic intermediate formation. gra-ORF6 was essential to produce (R)-DNPA in DHGRA biosynthesis. gra-5+6 led to the most efficient production of (R)-DNPA, implying a possible unique cooperative function as RED2. 3) A series of synthetic analogues was applied to the biotransformations based on ketosynthase-deficient recombinants of S. coelicolor carrying either RED1 or RED2. In all cases for RED1, the β-keto ester substrates were reduced with good to excellent enantioselectivity. However, the simpler substrates were not accepted by RED2, indicating the significant difference in substrate specificity between the two reductases. 4) 3D structures of RED1 and RED2 were predicted based on homology modeling (FAMS) using the templates, L-3-hydroxyacyl-CoA dehydrogenase from human heart (for RED1) and tropinone reductase II (for RED2). Catalytically key amino acid residues were revealed for the both enzymes. 5) RED1 and RED2 were overexpressed in E. coli, and in vitro assay system were successfully established. Optimization of the system, purification of both enzymes and site directed mutagenesis to the suggested key residues are in progress.