Studies towards a biomimetic synthesis of agelastatin A
Agelastatin A is a tetracyclic alkaloid isolated from the marine sponge Agelas
dendromorpha. It exhibits cytotoxicity towards KB and lymphocytic leukaemia cells,
arthropod toxicity and insecticidal activity and selectively inhibition of glycogen
synthase kinase-3B. Agelastatin A is one of a structurally related group of pyrroleimidazole
alkaloids derived from the "linear" skeleton of oroidin and it is postulated
that it is biologically derived from oroidin via the generation of an N-acyliminium ion,
which undergoes two cyclisations, via a second N-acyliminium ion, followed by
hydration to afford the natural product.
This thesis describes the work undertaken towards the biomimetic synthesis of
Agelastatin A and efforts towards the synthesis of an oroidin-like precursor from
which the N-acyliminium ion could be generated. Disconnection of the precursor
gave a 3-pyrroline A-ring, which was prepared using a Birch reduction, but efforts to
synthesise the D-ring imidazolone fragment were hampered by low solubility,
necessitating the use of protecting groups. Selective introduction of the Z-alkene was
made difficult by conjugation to the imidazolone and efforts to maintain the Z-alkene
Parikh-Doering oxidation of a propargylic alcohol intermediate afforded a novel B-thioaldehyde
that underwent an acid-catalysed cyclisation via an N-acyliminium ion
analogous to the second N-acyliminium ion in the proposed biomimetic synthesis.
The B-thioaldehyde forced the required alkene conformation and was therefore
applied in the synthetic route to the imidazolone fragment. The introduction of a
leaving group was ultimately unsuccessful, but a precursor for generation of the ion
by oxidation was prepared by reductive amination of the corresponding aldehyde.
The precursor was treated with trityl tetrafluoroborate but the expected oxidation did
not occur, instead an acid-catalysed cyclisation was observed. This cyclisation again
proceeded via an N-acyliminium ion analogous to that predicted to be generated in the
biomimetic synthesis, to afford a novel bicycle.
Therefore although the double cyclisation required for the biomimetic synthesis was
not mimicked, the two novel cyclisations provide good evidence that both cyclisations