Synthesis of enantiomerically pure model compounds of the glucose-6-phosphate-T1-translocase inhibitors kodaistatins A–D. Inferences with regard to the stereostructure of the natural products

The kodaistatins A and C (5a,b) inhibit a step in glucose-metabolism at ∼100 nM concentrations. This makes them potential ‘leads’ in the therapy of diabetes. We elucidated the (S)-configuration of the side-chain stereocenter of kodaistatin A by ozonolysis/reduction. The 13C NMR shifts of kodaistatin A model cis-11 suggest that the diol moiety in the dihydroxycyclopentanone core of kodaistatin is trans-configured. This model was prepared from the Feringa lactone (21) and (S)-2-methylbutanal (27) in 23 steps (14 steps in the longest linear sequence). We employed the same strategy for the simplified kodaistatin A model iso-cis-12, which resulted from the same substrates in 11 steps (6 steps in the longest linear sequence). The cyclopentenone cores of both targets stemmed from a C4+C1 approach. The C4 components were masked ‘tartaric ketones’ (16a,b) and a masked ‘tartaric aldehyde’ (18), respectively. The C1 components were the lithium-derivatives of the side-chain bearing phosphonates 19 and 22, respectively. The desired acylation/deprotonation/Horner–Wadsworth–Emmons tandem reaction succeeded in a single operation with the ‘tartaric aldehyde’ 18 but required partly or exclusively additional operations when we incorporated the ‘tartaric ketones’ 16a or 16b, respectively. The ‘tartaric ketones’ 16a,b contained an α-siloxyethyl substituent. It is noteworthy that it had to be introduced by adding the benzyltrimethylammonium enolate of lactone 18 to acetaldehyde because the lithium enolate of this lactone fragmented by an acetone-releasing β-elimination.