SPECIAL TOPIC 339 Synthesis 2001, No. 2, 339–342 ISSN 0039-7881 © Thieme Stuttgart · New York Synthesis of an 1’-Azasugar Analogue of Maltose Vinni Andreassen, a Birte Svensson, b Mikael Bols* a a Department of Chemistry, Aarhus University, 8000 Aarhus C, Denmark Fax +4589411234; E-mail: mb@kemi.aau.dk b Department of Chemistry, Carlsberg Laboratory, Valby, Denmark Received 3 October 2000 Abstract: Methyl 1’-azamaltoside (3) was synthesised from levo- glucosane and D-galactose in a 16 step synthesis. Methyl 1’-azamal- toside (3) was found to inhibit glucoamylase with K i of 0.63 mM. Key words: glycosidase inhibitor, glucoamylase, reductive amin- ation, isofagomine, glycosides, disaccharides, carbohydrates, nitro- gen Specific inhibitors of glycosidases and related enzymes are subject of much current interest 1-3 either as potential drugs against various diseases and disorders, as glycobio- chemical tools or as agents that provide information about the chemistry of enzymatic glycoside cleavage. In partic- ular, the various forms of azasugars have been found to be the most potent and specific inhibitors so far. 3 Most potent azasugar glycosidase inhibitors have a monosaccharide structure. Nevertheless, the transition state for a reaction catalysed by a typical glycosidase is a di- or oligosaccharide structure, because the typical sub- strate is a di- or oligosaccharide. This means that a di- or oligosaccharide transition state analogue would be ex- pected to be a more potent inhibitor than a monosaccha- ride analogue, because it would have more binding interactions to the enzyme. It is also likely that such mol- ecules would be more selective inhibitors. However little is known about how to design effective disaccharide in- hibitors. 4 Previous work from our laboratory has shown that modi- fication of the inhibitor isofagomine (1) at N-1 with an- other sugar residue to create 2 resulted in a 60 fold increase in the inhibition of glucoamylase (Figure). 5 This clearly suggested that substitution at N-1 of 1 with the ex- tra glucose residue increased affinity by mimicking the sugar moiety of the leaving group. However, since the natural substrate of glucoamylase is the a-1,4-linked glu- cose units of starch, compound 2, which mimics a 1,6- linked disaccharide, cannot be expected to be an optimal transition state mimic. The analogue 3, which mimics the 1,4-linked disaccharide maltose was, on the other hand, expected to mimic the transition state better than 2 (Fig- ure). In the present paper we report the synthesis of 3 and its inhibition of a series of glycosidases. For the synthesis of 3 a number of approaches involving known building blocks and reductive amination strategies were envisaged. For this purpose the known nitrile 4 was synthesised in 6 steps from D-galactose. 6 It was converted to the primary amine 5 in 62% yield by reduction with ex- cess LiAlH 4 in Et 2 O. The nitrile 4 was also converted to the aldehyde 6 in 72% yield by reduction with DIBAL in THF/CH 2 Cl 2 (Scheme 1). Compound 6 has previously been synthesised by another route. 7 Building blocks 7-9 were taken from our isofagomine synthesis (Scheme 2). 8 In that synthesis 7 is obtained in 7 steps from levoglucosane (1,6-anhydroglucose), which is then converted to the dialdehyde 8 by periodate cleavage and to 3-benzylisofagomine 9 by reductive amination with ammonia. Substitution of ammonia with the amine 5 in the latter step appeared to be a particular attractive route to the target. However, reductive amination of equimolar amounts of 5 and 8 using H 2 at 45 atm and 5% Pd/C as cat- alyst in EtOH gave a complex mixture of products; how- ever, the expected target 3a did not appear to be among them (Scheme 3). We believe that reductive amination is slow because the amine cannot be employed in excess. This allows the sensitive dialdehyde to undergo aldol con- densations and similar side reactions. To simplify the problem the reductive amination between 5 and monoal- dehyde 7 was attempted with the plan to reverse the order of NaIO 4 cleavage and reductive amination steps in the synthesis. However, the reductive amination was very slow and hydrogenolysis of the benzyl groups took place with a greater rate making a subsequent NaIO 4 step im- possible. Reductive amination was then performed between 6 and 9 (Scheme 3). Also in this case the reaction was extremely slow and partial hydrogenolysis of benzyl groups oc- Figure The chemical structures of isofagomine (1), its isomaltose (2) and maltose (3) analogues