I discovered unusual behavior of one natural compound, that appeared to react with itself under sodium tert-butylate conditions. Expected alpha-ketol rearrangement product wasn't observed in the reaction mixture.
It looks quite similar to the Baylis-Hillman reaction. But, instead of a tertiary amine or phosphine, it is the OH group (blue) the one that gives the nucleophilic addition to the beta position (red). This is followed by intramolecular enolate trapping (red alpha carbon to blue carbonyl to form the final adduct. And, unlike Baylis-Hillman reaction, this process does not regenerate the initial enone upon deamination/dephosphination, so it keeps the furan moiety.