Degree Name

Doctor of Philosophy


School of Chemistry


Polyhydroxylated alkaloids, also known as iminosugars and azasugars, are natural products that structurally resemble sugar molecules but contain a cyclic-bound basic nitrogen atom in lieu of the ring oxygen. Certain iminosugars have attracted increasing attention as potential antiviral, anticancer and antidiabetic agents due to their inhibitory activity against several glycosidases, enzymes that hydrolyse the terminal glycosidic bond of various polysaccharides. The broussonetines are a subgroup within the class of iminosugars that is characterised by a polyhydroxylated pyrrolidine moiety and a thirteen carbon side chain attached at the pseudoanomeric position. As such, they can be regarded as derivatives of the also naturally occurring small iminosugars DMDP and D-AB1. Furthermore, their structure and proposed biosynthetic pathway also relates them to another group of natural products, namely the sphingoid bases (“sphingosines”) and their cyclic derivatives including 2,6-disubstituted 3-piperidinol alkaloids like cassine and carnavaline. The broussonetine family consists of more than 30 members which have been isolated since 1995 from the bark extracts of the Japanese Paper Mulberry, Broussonetia kazinoki (Moraceae) by G. KUSANO et al. Glycosidase inhibitory activity assays demonstrated that almost all broussonetines are exceptional inhibitors of β-glycosidases with IC50 values in the nanomolar range. Broussonetines C and E in particular display excellent β-galactosidase inhibitory activity with IC50 values of 3.6 × 10−8 and 2.0 × 10−9 mol/L, respectively.

Retrosynthetically, the broussonetines were divided into two fragments, the pyrrolidine moiety and a side chain building block, which were planned to be coupled via a Wittig reaction in the case of broussonetine C and a Grignard reaction in the case of broussonetine E. As a key step for the assembly of the pyrrolidine moiety, the substrate controlled diastereoselective Petasis reaction of 3,5-di-O-benzyl-L-xylofuranose, readily available in three simple steps from L-xylose, was to be employed. A subsequent stereospecific SN2/5-exo-tet-cyclisation followed by O-benzyl protection and an oxidative double cleavage would ultimately provide the desired pyrrolidine carbaldehyde as building block. Via a small modification of the synthesis route this approach would also be used to synthesise DMDP. However, due to the relative expensive cost of L-xylose compared to its cheap natural enantiomer, D-xylose was eventually used in the synthesis providing the unnatural enantiomers of the natural products. Nevertheless, these enantiomers are gaining increasing interest due to the discovery that, in many cases, they are by far more potent and specific glycosidase inhibitors then their natural counterparts.

L-DMDP was synthesised in four steps from the known compound 3,5-di-O-benzyl-D-xylofuranose. The Petasis reaction of this xylose derivative provided stereoselectively the desired anti-1,2-amino alcohol 7/6 in excellent d.e. (> 99%). This amino alcohol was then cyclised in a stereospecific SN2/5-exo-tet-cyclisation via the regioselective O-mesylation at C-2 to give pyrrolidine 7/12. Ozonolysis of the styryl double bond in 7/12 and the hydrogenolysis of the remaining benzyl protecting groups concluded the synthesis. Serendipitously, the ozonolysis of 7/12 also provided the C-5 decarbinolated pyrrolidine 7/24 which after hydrogenolysis afforded the known iminosugar L-AB1.

The synthesis of pyrrolidine carbaldehyde 8/2, on the other hand, proved to be unexpectedly complicated. Several attempts to obtain the aldehyde via the oxidative double bond cleavage of the fully benzyl-protected pyrrolidine 8/1 were unsuccessful due to the low stability of the target compound and the formation of various by-products. After a number of failed attempts, the synthesis of 8/2 eventually succeeded via a detour from the pyrrolidine diol 7/23. The primary hydroxy group of 7/23 was trityl protected followed by the C-4 O-benzylation of 8/54. The consequent trityl ether cleavage afforded the pyrrolidine alcohol 8/9 which was subsequently oxidised via a Swern oxidation to the required aldehyde building block.

The pyrrolidine carbaldehyde was then successfully coupled with two simplified side chain fragments via a Wittig reaction and Grignard reaction, respectively, affording the two model compounds (–)-10'-deoxobroussonetine C and E after hydrogenolysis. An attempt to invert the configuration of the secondary (6R)-alcohol 9/12, obtained from the Grignard coupling reaction, to the desired (6S)-epimer resulted in ring-expansion of the pyrrolidine ring ultimately affording the novel polyhydroxylated piperidine, (–)-(6S)-(12'-hydroxydodecyl)moranoline, after hydrogenolysis.

The synthesis of (–)-broussonetine C, however, remained unsuccessful due to unforeseen problems during the final step, the hydrogenolysis of the benzyl protecting groups in the (–)- broussonetine C precursor 10/20. The chosen reaction conditions resulted in a cyclisationreduction sequence of the side chain terminus providing the novel tetrahydrofuran derivative 10/23, designated as (–)-broussonetine C2.