A glycosynthase catalyst for the synthesis of flavonoid glycosides.

Glycosylation of lipophilic small molecules is one of the predominant strategies by which the bioactivity of these compounds is modulated in living organisms. Indeed, even simple plants, such as Arabidopsis thaliana, dedicate over 120 open reading frames encoding many activated sugar-dependent glycosyltransferases to this task alone. Novel synthetic routes to such compounds provide a powerful tool for exploring not only this activity by providing ready access to probes, standards, and inhibitors but also to the many plant natural products such as flavonoids and steroid glycosides that have been implicated as potential therapeutics. Despite the many elegant methods for chemical glycosylation, methods for direct regioselective glycosylation of acceptors are limited, which has led to a heavy dependence upon protecting-group regimes that introduce many additional steps. The often exquisite selectivity of biocatalytic methods provides the potential to overcome such problems. Although the use of glycosyltransferases (GTs) is nature.s solution to glycoside bond formation, access to suitable GTs and donors limits this approach. Furthermore, the use of GTs is typically limited to the transfer of only single carbohydrate residues. The use of glycosidases provides an alternative method for biocatalytic glycosylation. One of the most powerful approaches to the enzymatic synthesis of glycosides is Withers.s “glycosynthase” technology. Glycosynthases are genetically engineered nucleophile-less mutant glycosidases that can catalyze the formation of glycosidic linkages, primarily but not exclusively by using glycosyl fluoride donors, yet are incapable of hydrolysing the product. A number of glycosynthases have been described; yet, strikingly, owing to the inherent preference of the binding subsites of these enzymes for carbohydrate residues, only a single study of non-carbohydrate acceptors has been conducted to date. Logically this elegant work allowed the synthesis of glycoceramides through the construction of an endoglycoceramidase glycosynthase. In this way the inherent acceptor specificity of this biocatalyst was still the primary determinant of function. However, to our knowledge no glycosynthase with a wild-type preference for carbohydrate acceptors has been used efficiently with non-carbohydrate acceptors. Herein we demonstrate that the substrate tolerance of glycosynthases can be significantly broadened to allow the use of non-carbohydrate acceptors, thereby providing valuable access to lipophilic glycoconjugates. Specifically, the E197S mutant of the Humicola insolens Cel7B enzyme has been identified as a catalyst for the formation of flavonoid glycosides, with sugar transfer directly to the flavonoid scaffold. These represent substrates to which the E197S mutant has no known specificity. Remarkably, rates of glycosyltransfer are comparable with those of naturally occurring uridine diphosphate (UDP)-sugar-dependent glycosyltransferases on the same acceptors. We have recently developed a high-throughput MS-based screen for glycosyltransfer. 25] By using this method, large panels of substrates may be readily screened allowing the determination of a “green–amber–red” (GAR) virtual microarray of activity that reflects the presence of the glycosylated product. By using this GAR screen with the aim of identifying optimal catalysts for the synthesis of glyco-lipophilic products, we screened 80 acceptors (see Figure S1 in the Supporting Information) with greater than 100 biocatalysts and more than 20 glycosyl donors. We identified not only the expected GT catalysts capable of such a transfer, but also, surprisingly, a glycosidase mutant, the glycosynthase E197S mutant from the H. insolens glycosidase Cel7B. As can be seen from the GAR array (Figure 1; Cel7B–E197S with disaccharide donor lactosyl fluoride (LacF)), this striking non-natural activity is specific. From 80 potential acceptors, only the glycosylation of flavonoids 2 and 20–22 was observed; this suggested the first non-GT-catalyzed biocatalytic methods for access to such glycoflavonoids 81–85. The screen also, less unusually, revealed carbohydrate acceptors: the known substrate benzyl-b-glucoside (52) and the unknown substrate, the plant cytokinin trans-zeatin glucoside 66. No activity at all was observed against a range of other representative acceptor substrates. Excited by this unusual, non-natural but specific flavonoid activity, we investigated it in further detail. Remarkably, initial conversion levels suggested rates of glycoside forma[*] Prof. G. J. Davies York Structural Biology Laboratory Department of Chemistry York YO105YW (UK) Fax: (+44)190-432-8266 E-mail: davies@ysbl.york.ac.uk