Keynote Address

ENGINEERING GLYCOSIDASES FOR CONSTRUCTIVE PURPOSES

David L. Jakeman and Stephen G. Withers

Department of Chemistry, Protein Engineering Network of Centres of Excellence, University of British Columbia, Vancouver, Canada.

1 ABSTRACT

The synthesis of oligosaccharides using glycosidases as catalysts has been known for many years, but has been limited by the poor yields generally associated with such processes due to hydrolysis. Glycosynthases, mutant glycosidases in which the catalytic nucleophile has been replaced, offer an alternative form of catalyst that synthesizes glycosides from readily prepared glycosyl fluorides, but does not hydrolyze the glycoside products. The range of reactions performed by these enzymes is being extended through mutation of number of different glycosidases, as well as through random mutation of known glycosynthases, coupled with efficient screens.

2 DISCUSSION

Glycosidases have important roles as industrial catalysts for the breakdown of complex carbohydrates, yet their commercialization as catalysts for the synthesis of oligosaccharides has been less fruitful because the product is necessarily a substrate for hydrolysis by the wild-type enzyme. Despite the general lack of commercial success in exploiting glycosidases as catalysts for the synthesis of oligosaccharides, a diverse selection of glycosidases has been explored in the literature to perform transglycosylation reactions yielding oligosaccharides, and this area of research has been reviewed recently. There are many possible parameter permutations available to increase the yields of glycosidase-mediated transglycosylation, and for specific glycosidases respectable yields have been observed as a result of altering certain reaction conditions, but no one process has been universally successful in circumventing the hydrolysis conundrum and giving consistently high yields.

The mechanism of retaining glycosidases has been the study of intense research initially based upon Koshland's premise of a double-displacement occurring at the anomeric centre. Figure 1A shows the widely accepted reaction course for a retaining glycosidase hydrolyzing a substrate. Two active site carboxylic acid residues separated by approximately 5.5 Å apart act as a nucleophile and general acid / base catalyst respectively. Hydrolysis is initiated by attack of the nucleophile onto the anomeric carbon to form a covalent glycosyl-enzyme intermediate that in turn is hydrolyzed by an incoming water molecule to release enzyme and monosaccharide. The transglycosylation mechanism differs from the hydrolytic mechanism solely because a sugar residue replaces the water molecule attacking the covalent glycosyl enzyme-intermediate (Figure 1B). One approach taken to avoid hydrolysis of product from a transglycosylation reaction was the abolition of the first step of the hydrolytic reaction - by mutating the catalytic nucleophile to a non-nucleophilic residue. Whilst this approach does not permit formation of a covalent glycosyl-enzyme intermediate, an active site architecture is maintained that could condense an appropriate donor and acceptor sugar together provided the donor mimics the glycosyl enzyme-intermediate. Because the placement of the acid / base catalyst within the active site has not been altered by the mutation, it remains set up to deprotonate an incoming acceptor sugar just as it does in a transglycosylation reaction (Figure 1B). The discovery that glycosyl fluorides function as suitable donors for reactions catalyzed by specific nucleophile mutants of retaining glycosidases has now resulted in several new mutant enzymes capable of catalyzing the synthesis of oligosaccharides. A proposed mechanism for glycosynthases is shown in Figure 2, as postulated for the first glycosynthase Agrobacterium sp. β-glucosidase (Abg) E358A. The donor sugar composed of α-D-glucosyl fluoride, of equivalent anomeric configuration to the intermediate, binds in the glycone-binding site mimicking the intermediate. The acid / base residue facilitates attack of the nucleophilic hydroxyl group onto the anomeric carbon atom by deprotonating the acceptor sugar. Fluoride is released, and a glycosidic linkage is formed.

The catalytic promiscuity of Abg E358A was initially uncovered by the discovery that it will transfer α-D-glucosyl fluoride (GlcF) and α-D-galactosyl fluoride (GalF) onto a variety of acceptor sugars in moderate to high yields. Such is the efficiency of the carbohydrate transfer that with GlcF as donor repeated additions occur, providing a series of cello-oligosaccharides. Of the compounds generated utilizing Abg E358A as catalyst, arguably the most noteworthy are mechanism-based inactivator for cellulases, 2,4-dinitrophenyl 2-deoxy-2-fluoro-β-D-cello-oligosides (Figure 3). These compounds would not be amenable to synthesis using a traditional wild-type transglycosylation approach because the 2-deoxy-2-fluoro glycoside rapidly inactivates the wild-type enzyme by covalent modification.

Further advances in rates of glycosylation were achieved by replacement of the nucleophile by a serine residue. Abg E358S was a 24-fold faster catalyst as determined by fluoride ion release and this rate enhancement was attributed to stabilization of the departing fluoride via hydrogen bonding. Importantly, the increase in transglycosylation rate also broadened the substrate specificity of the glycosynthase catalyst. The condensation of GalF and 4-NP β-N-acetylglucosaminide to give 4-NP β-N-acetyllactosaminide (Figure 4) was only successful using the E358S mutant enzyme.

We have developed a novel screen to select for improved glycosynthase activity and used it to successfully discover two further glycosynthase mutants of Abg. The screen requires an endo-cellulase to selectively hydrolyze products derived from a glycosynthase reaction. We chose Cel5A from Cellulomonas fimi, (C. fimi) an enzyme that has exquisite substrate specificity for di- or trisaccharide substrates. Thus, the screen works through glycosynthase catalyzed condensation of GlcF and 4-methylumbelliferyl β-D-glucoside (MUGlc) to provide a disaccharide product that is then hydrolyzed by the cellulase releasing the fluorescent methylumbelliferone (Figure 5). In total four mutants were selected by screening a saturation mutagenesis library of Abg E358X (X = any amino acid) prepared by a four-primer PCR strategy. Two of these mutants were the known alanine and serine glycosynthases and two were new glycosynthases, the cysteine and glycine mutants. The E358C mutant was a very poor glycosynthase with a rate only half that of the E358A mutant, whilst the E358G mutant had a rate twice that of the E358S mutant. The increase in transglycosylation rate was sufficient to permit the condensation of a new glycosyl fluoride by Abg E358G. Thus, α-D-xylosyl fluoride was condensed with 4-nitrophenyl glucoside in 43% isolated yield (Figure 6). While the yield for the reaction is not as impressive as with many of the glycosynthase catalyzed reactions, the discovery of a new glycosyl fluoride donor is particularly significant because the mutant was selected by screening for a different specific reaction. Applying in vitro evolution techniques to the whole Abg gene will hopefully provide new and improved glycosynthase catalysts with transferase behavior corresponding to the broad substrate specificity afforded by wild-type Abg.

The β-mannosidic linkage is one of the most chemically challenging glycosidic linkages to obtain with high anomeric stereoselectivity. The development of a glycosynthase to efficiently condense α-D-mannosyl fluoride (ManF) with numerous acceptor sugars has provided a new efficient route to complement the chemical strategies available in the literature. The mannosidase from C. fimi Man2a was cloned and expressed in E. coli and the nucelophile identified as E519. Two active site mutants were prepared, E519A and E519S, and both mutants exhibited glycosynthase behavior. However, yields were significantly higher with the serine mutant. Essentially quantitative conversion to oligosaccharide product was observed when ManF and 4-nitrophenyl β-D-cellobioside were condensed (Figure 7) providing ready access to a new class of oligosaccharides. Other aryl glycosides were also suitable acceptors for Man2a E519S, including 4-nitrophenyl β-D-mannoside, 4-nitrophenyl β-D-xyloside and 4-nitrophenyl β-D-glucoside. Overall yields were high for oligomeric products and when 2,4-dinitrophenyl 2-deoxy-2-fluoro-β-D-mannoside was used as acceptor a hexameric oligosaccharide species was isolated resulting from the transfer of five ManF units. Often the transfer of the second ManF sugar was achieved efficiently but with loss of absolute regiochemical control because -β(1,3)- and -β(1,4)- linkages were generated. One reaction that did not produce a mixture of regiochemical isomers was the condensation of ManF and 4-nitrophenyl β-D-gentiobioside (Figure 8). High yields of oligomeric products were observed providing a route into novel branched oligosaccharides.

In the literature, glycosynthases have been developed that transfer a number of different donor sugars. The first reported glycosynthase derived from an endo-glycosidase was the retaining 1,3-1,β-P-glucanase from Bacillus licheniformis E134A. The mutant enzyme catalyzed the condensation of α-laminaribiosyl fluoride onto MUGlc in 88% yield.

The glycosynthase from Humicola insolens Cel7B E197A is a particularly well characterized glycosynthase, capable of synthesizing a wide range of oligo- and polysaccharides from several different donor sugars and remains the only glycosynthase for which an X-ray crystal structure has been reported. It condenses α-cellobiosyl fluoride with great efficiency into cellulose II. Defined oligosaccharide products have also been obtained when condensing α-lactosyl fluoride (LacF) onto a range of disaccharide and monosaccharide acceptor sugars. Quantitative yields were obtained using methyl β-D-cellobioside, methyl 6-bromo-6-deoxy-cellobioside and benzyl laminaribioside as acceptors while lower yields were obtained using monosaccharide acceptors. From the analysis of the wild-type crystal structure it was observed that specific hydroxyl groups on the substrate molecules failed to interact significantly with the polypeptide chain. These relaxed constraints in the substrate permitted the condensation of novel α-glycosyl fluorides modified at the 6I and 6II position to afford novel polymers.

3 CONCLUSIONS

The mutation of a retaining glycosidase to a hydrolytically inactive enzyme by substituting the nucleophilic carboxylic acid with a non-nucleophilic residue has generated a novel class of glycosylation catalysts. These mutant enzymes transfer activated donor sugars onto a wide variety of acceptor sugars with high degrees of stereo- and regioselective control and in very good yield. The synthesis of donor sugars is straightforward, using cheap and readily available starting materials. Glycosynthases, therefore, provide a new route for the large-scale manufacture of oligosaccharides. A wide variety of oligosaccharides can easily be prepared using glycosynthase technology. However, because of the large amount of sequence information on glycosidases, it is surely only a matter of time before new glycosynthases are reported, further broadening the range of donor sugars that can be transferred and providing a means to synthesize other more exotic glycosidic linkages.

Structure–Function Studies of Carbohydrate-active Enzymes

STRUCTURAL ENZYMOLOGY OF CARBOHYDRATE-ACTIVE ENZYMES

Gideon J. Davies

Structural Biology Laboratory Department of Chemistry University of York York, YO10 5DD U.K.

1 INTRODUCTION

Simply in terms of quantity, the enzymatic synthesis and degradation of glycosidic bonds are the most important reactions on earth. The enzymes involved both in these processes and in the modification of oligo- and polysaccharides - "carbohydrate-active enzymes" or CAZymes - have thus received long-standing interest from both academic and industrial groups. Glycoside hydrolases, the degradative machinery have recently been extensively reviewed and will not be covered here. Instead I discuss the structures and functions of some other enzyme classes involved in the synthesis and degradation of poly and oligosaccharides: polysaccharide lyases, carbohydrate esterases and, natures synthetic apparatus, glycosyltransferases. Before embarking on the structures and likely mechanisms, however, it is worth considering both the amino-acid sequence families and modularity of CAZymes which underpins all work in this field.

One of the most important breakthroughs in our understanding of CAZymes was the classification system, based upon amino-acid sequence similarities, proposed by Bernard Henrissat in the late 1980's. Originally, this was a classification of glycoside hydrolases only, based upon early work with cellulases, but it was later extended to include glycosyltransferases, the many ancillary modules, particularly those involved in carbohydrate binding (described below) as well as other enzyme classes. This continually-updated resource is available at URL httd/afmb.cnrs-mrs.fr/~pedro/CAZY/db.html and currently provides classification of over 86 families of glycoside hydrolases, 54 glycosyltransferases, 12 polysaccharide lyases, 13 families of carbohydrate-active esterases and (de)acetylases and almost 30 families of carbohydrate-binding modules. This site also provides information both on the catalytic mechanism of each family, where known and on the availability three-dimensional structures for the various modules.