Feeding experiments with Actinoplanes sp. SN223/29 showed that 3-amino-5-hydroxy[7-]SP13C]benzoic acid is not incorporated into acarbose (I). The valienamine moiety of I is thus not derived in the same way, from the shikimate pathway, as the m-C7N units in the ansamycin, mitomycin and ansamitocin antibiotics. Feeding experiments with [U-13C3]glycerol followed by analysis of I by multiple quantum NMR spectroscopy support this conclusion and point to formation of the valienamine moiety by cyclization of a heptulose phosphate which arises from a triose phosphate via successive transfer of two 2-carbon fragments by transketolase, as proposed by PAPE and co-workers.
Japan Antibiotics Research Association