Engineering of cellobiose phosphorylase for glycoside synthesis

Journal of Biotechnology 156 (2011) 253–260

Contents lists available at ScienceDirect

Journal of Biotechnology journal homepage: www.elsevier.com/locate/jbiotec

Review

Engineering of cellobiose phosphorylase for glycoside synthesis Manu R.M. de Groeve 1 , Tom Desmet ∗ , Wim Soetaert InBio.be – Centre of Expertise Industrial Biotechnology and Biocatalysis, Ghent University, Coupure Links 653, 9000 Ghent, Belgium

a r t i c l e

i n f o

Article history: Received 31 January 2011 Received in revised form 6 July 2011 Accepted 11 July 2011 Available online 20 July 2011 Keywords: Cellobiose phosphorylase Enzyme engineering Glycoside synthesis Directed evolution

a b s t r a c t Disaccharide phosphorylases are increasingly applied for glycoside synthesis, since they are very regiospecific and use cheap and easy to obtain donor substrates. A promising enzyme is cellobiose phosphorylase (CP), which was discovered more than 50 years ago. Many other bacterial CP enzymes have since then been characterized, cloned and applied for glycoside synthesis. However, the general application of wild-type CP for glycoside synthesis is hampered by its relatively narrow substrate specificity. Recently we have taken some successful efforts to broaden the substrate specificity of Cellulomonas uda CP by directed evolution and protein engineering. This review will give an overview of the obtained results and address the applicability of the engineered CP enzymes for glycoside synthesis. CP is the first example of an extensively engineered disaccharide phosphorylase, and may provide valuable information for protein engineering of other phosphorylase enzymes. © 2011 Elsevier B.V. All rights reserved.

Contents 1. 2. 3.

4. 5.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Structural determinants of substrate specificity in C. uda CP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Engineering of cellobiose phosphorylase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Modification of CP donor specificity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Modification of CP acceptor specificity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Loop engineering experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Application of engineered CPs for glycoside synthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions and outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1. Introduction The importance of carbohydrates and glycosides in biology is widely known. Many biologically active molecules contain a carbohydrate moiety that determines either their activity or pharmacokinetic properties (Kren and Martinkova, 2001). Glycosylation of drugs can, for example, induce targeting to specific organs and tissues, thus resulting in less side-effects and smaller required doses (Wong and Toth, 2001). Furthermore, glycosylation usually improves their solubility and reduces toxicity (Leu et al., 1999). Carbohydrates, in turn, often display prebiotic effects by selectively

∗ Corresponding author. E-mail addresses: [email protected] (M.R.M. de Groeve), [email protected] (T. Desmet). 1 Present address: Institute of Technical Biochemistry and European Center of Bioand Nanotechnology, Technical University of Lodz, ul. Stefanowskiego 4/10, 90-924 Lodz, Poland. 0168-1656/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jbiotec.2011.07.006

253 254 255 256 256 257 257 259 259 259

stimulating growth of beneficial bacteria in the intestines (Wang, 2009). The synthesis of glycosides and carbohydrates can be performed by chemical or enzymatic methods. Although chemical synthesis has already been applied successfully, it usually suffers from low yields due to non-specific glycosylation, resulting in product mixtures. Also the use of toxic compounds limits their use for largescale applications. To overcome these problems, the enzymatic glycosylation approach has attracted much attention. Most studies are based on the use of glycosyltransferases (GTs), glycoside hydrolases (GHs) and transglycosidases for glycoside synthesis. Much less work has been done with glycoside phosphorylases (GPs), although they have some interesting properties (Luley-Goedl and Nidetzky, 2010). For example, the sugar donor – a glycosyl phosphate – is cheaper and more easy to obtain than the activated sugar nucleotides required by GTs. Indeed, since GPs catalyze equilibrium reactions, they can synthesize the glycosyl phosphate donor by phosphorolysis of their natural (disaccharide) substrates.

254

M.R.M. de Groeve et al. / Journal of Biotechnology 156 (2011) 253–260

Table 1 Overview of characterized GH94 enzymes and their origin.a Enzyme

EC-number

Main substrate

Donor

Organism

Cellobiose phosphorylase

2.4.1.20

Glc(␤1–4)Glc

␣Glc1P

Cellulomonas uda Cellvibrio gilvus Clostridium stercorarium Clostridium thermocellum Thermotoga maritima Thermotoga neapolitana

Cellodextrin phosphorylase

2.4.1.49

Glc[(␤1–4)Glc]n

␣Glc1P

Clostridium stercorarium Clostridium thermocellum

Chitobiose phosphorylase

2.4.1.–

GlcNAc(␤1–4)GlcNAc

␣GlcNAc1P

Vibrio furnissii Vibrio proteolyticus

Cyclic ␤-1,2-glucan synthase

2.4.1.–

Glc[(␤1–2)Glc]n

␣Glc1Pb

Agrobacterium tumefaciens Brucella abortus Sinorhizobium meliloti

a

Data obtained from the CAZy database (http://www.cazy.org) (Henrissat and Davies, 1997). The cyclic ␤-1,2-glucan synthase catalyzes four types of enzymatic reactions, including ␤-1,2-glucooligosaccharide phosphorolysis with ␣Glc1P as donor (Ciocchini et al., 2007). b

In recent years, glycoside phosphorylases have been applied for glycoside synthesis on kilogram scale. Examples include the synthesis of the cosmetic ingredient glucosylglycerol with sucrose phosphorylase (Goedl et al., 2008) and the synthesis of the prebiotic sugar lacto-N-biose with lacto-N-biose phosphorylase (Nishimoto and Kitaoka, 2007). However, a major drawback for the general application of disaccharide phosphorylases for glycoside synthesis is their rather narrow substrate specificity. Although sucrose phosphorylase has a relatively broad acceptor specificity, it can only ␣-glucosylate acceptor molecules. Other phosphorylases can be used for ␤-glucosylation or ␤-galactosylation reactions, but have a more narrow acceptor specificity. An interesting representative is cellobiose phosphorylase (CP), which was first described in crude cell extracts of Clostridium thermocellum (Sih and McBee, 1955a,b). CP enzymes catalyze the reversible phosphorolysis of cellobiose into ␣-glucose 1-phosphate (␣Glc1P) and glucose. Since its discovery, cellobiose phosphorylases have been identified in numerous bacteria that are directly or indirectly involved in cellulose degradation (Table 1). Its natural role is the energy-efficient metabolism of cellobiose, producing ␣Glc1P in which much of the substrate’s energy is conserved. CP belongs to CAZy family GH94, together with chitobiose phosphorylase (ChP), cellodextrin phosphorylase and cyclic ␤-1,2-glucan synthase (Henrissat and Davies, 1997) (Table 1). The specificity of cellobiose phosphorylase has been extensively studied, especially for glycoside synthesis (Alexander, 1968; Kitaoka et al., 1992; Nidetzky et al., 2000). However, applications of CP for glycoside synthesis are mainly limited to carbohydrate synthesis. This is due to the relatively narrow acceptor specificity of CP, which accepts mainly glucose, glucose-derivatives and other monosaccharides (Table 2). Although a free C-1 hydroxyl group (preferably in the ␤-configuration) is an absolute requirement for most CP enzymes (Kitaoka and Hayashi, 2002; Nidetzky et al., 2000), a hyperthermophilic CP from Thermotoga maritima was found to accept methyl ␤-glucoside (Rajashekhara et al., 2002). Several unnatural trisaccharides and disaccharides have been synthesized with Cellvibrio gilvus CP using ␣Glc1P as sugar donor and melibiose, gentiobiose, isomaltose, d-arabinose, d-altrose or l-fucose as acceptor molecules (Percy et al., 1997, 1998). Radioactive cellobiose has been synthesized with C. thermocellum CP using either radiolabeled ␣Glc1P or glucose in the wild-type synthesis reaction (Ng and Zeikus, 1986). An example of glycosylation with non-carbohydrate acceptors is the synthesis of alkyl ␤-glucosides using C. thermocellum CP (Kino et al., 2008). The authors were able to synthesize the glucosides of alcohols ranging from methanol up to heptanol.

Since phosphorolytic reactions are reversible, full conversion of substrates is not possible unless one of the reaction products is continuously removed from the reaction mixture. An approach resulting in irreversible synthesis reactions with Cellulomonas uda CP has been applied using ␣Glc1F as the glucosyl-donor (Nidetzky et al., 2004). This resulted in a ‘glycosynthase-like’ reaction with 100% conversion of substrates into glycosidic product. Besides ␣Glc1P and ␣Glc1F, also d-glucal was found to function as donor molecule for CP, although the activity was about 500 times lower than with the natural donor (Kitaoka et al., 2006). 2. Structural determinants of substrate specificity in C. uda CP Known CP enzymes consist of around 800 amino acids, corresponding to a molecular mass of about 90 kDa. They often form dimers, although monomeric forms of CP in solution have also been reported (Nidetzky et al., 2000; Reichenbecher et al., 1997). The crystal structure of C. gilvus CP is a dimer (Fig. 1) and shows high similarity with the structure of Vibrio proteolyticus chitobiose phosphorylase (Hidaka et al., 2006). Two CP crystal structures have been published, one containing sulphate in the active site (PDB 2cqs) and one containing phosphate (PDB 2cqt). While both structures contain glucose at the acceptor site (+1 subsite), only the latter contains a ligand at the donor site (−1 subsite), i.e., the glycerol used as cryoprotectant. Interestingly, the positions of the glycerol atoms superpose almost perfectly with the C-4 to C-6 and O-4 to O6 atoms of GlcNAc (−1 site) in the chitobiose phosphorylase crystal structure (PDB 1v7x). At the acceptor site of C. gilvus CP, a strong hydrogen bond is formed between residue E649 and the O-1 atom at the reducing end of glucose (Fig. 2). Furthermore, the glucose molecule is exclusively in the ␤-anomer configuration. The C. uda CP substrate preference for ␤-cellobiose over ␣-cellobiose has also been observed in activity assays (Nidetzky et al., 2000). This is in contrast to ChP, where no ␣/␤ preference is found due to the absence of a hydrogen bond with the O-1 atom of GlcNAc at the acceptor site. A weaker hydrogen bond is formed between O-2 and Y653, while the O-3 hydroxyl group forms a strong hydrogen bond with E659. The O-4 hydroxyl group interacts with the catalytic amino acid D490, which acts as a general acid in phosphorolysis. In the synthesis reaction D490 is likely to assist deprotonation of the O-4 hydroxyl group of the acceptor molecule. The O-6 hydroxyl group does not form hydrogen bonds with the CP enzyme, but instead there is a hydrophobic interaction with the guanidium plane of R362. The fact that the O-6 atom is exposed to the solvent and that there is a large open space

M.R.M. de Groeve et al. / Journal of Biotechnology 156 (2011) 253–260

255

Table 2 Comparison of the acceptor specificity of different wild-type CP enzymes. Acceptors

Activity of cellobiose phosphorylase C. udaa

C. gilvusb

T. maritimac

C. thermocellumd

Monosaccharides +e + + + d-Glucose d-Mannose + + + + + + d-Glucosamine + + d-Xylose + + + + 2-Deoxy-d-glucose + + + + + n.t.f n.t. n.t. 2-Deoxy-2-fluoro-d-glucose + + n.t. n.t. 3-Deoxy-d-glucose 3-Deoxy-3-fluoro-d-glucose + + n.t. n.t. 6-Deoxy-d-glucose + + + + n.t. n.t. n.t. 6-Deoxy-6-fluoro-d-glucose + g Methyl ␤-glucoside n.t. − + − d-Glucuronamide n.t. + − + − − n.t. + 1,5-Anhydro-d-glucitol n.t. + − + l-Fucose n.t. + − + d-Arabinose n.t. + n.t. n.t. d-Altrose n.t. + − n.t. d-Allose Disaccharides n.t. + n.t. + Melibiose (Gal(␣1–6)Glc) n.t. + n.t. + Gentiobiose (Glc(␤1–6)Glc) Isomaltose (Glc(␣1–6)Glc) n.t. + n.t. + Non-carbohydrates n.t. n.t. n.t. + Alkyl alcohols Non-acceptors d-fructose; methyl ␣-d-glucoside; d-lyxose; l-sorbose; d-ribose; l-arabinose; l-xylose; N-acetyl-d-glucosamine; myo-inositol; glucono-␦-lactone; d-galactose; d-glucuronic acid; d-glucitol; l-idose; d-mannitol; d-mannosamine; N-acetyl-d-mannosamine; d-fucose; cellobiose; d-mannitol; l-glucose; d-sorbitol a b c d e f g

Nidetzky et al. (2000). Kitaoka and Hayashi (2002), Percy et al. (1998), Percy et al. (1997). Rajashekhara et al. (2002). Alexander (1968), Kim et al. (2002), Kino et al. (2008), Nakai et al. (2010a). +: active. n.t.: not tested. −: not active.

near this atom may explain the ability of CP to use O-6 substituted disaccharides (isomaltose, gentiobiose and melibiose) as acceptor (Table 2). The reaction mechanism of CP is similar to that of other inverting phosphorylases, and consists of a direct nucleophilic attack by phosphate on the anomeric carbon of the scissile bond (Desmet and Soetaert, 2011). The reaction is aided by the general acid D490, which donates a proton to the glycosidic oxygen atom. Docking studies have revealed that the sugar ring in subsite −1 is distorted to the more reactive 1 S3 conformation before proceeding to an oxocarbenium-like transition state in the E3 conformation (Fushinobu et al., 2008). A flexible loop (495–513) near the entrance

of the active site is thought to be involved in substrate binding and product release (Fushinobu et al., 2008; Hidaka et al., 2006). 3. Engineering of cellobiose phosphorylase Directed evolution and protein engineering techniques are increasingly used to modify the substrate specificity of enzymes and to optimize them for industrial applications (Bottcher and Bornscheuer, 2010; Reetz, 2010). In recent years, several groups have reported some successes in protein engineering of glycoside phosphorylases. These experiments were mainly oriented towards the increase of thermostability (Fujii et al., 2006; Yamamoto et al., 2005) or the creation of enzymes with dual specificity (Nakai et al.,

Fig. 1. The Cellvibrio gilvus CP crystal structure (PDB 2cqt). The two subunits of the CP dimer are colored red and blue respectively. Panel A: Overall view on the CP enzyme. Panel B: The route (light orange) to the active site in CP, calculated with CAVER 1.0 (Petrek et al., 2006). The route to only one of the two active sites is shown.

256

M.R.M. de Groeve et al. / Journal of Biotechnology 156 (2011) 253–260

Fig. 2. View on the active site of Cellvibrio gilvus CP (PDB 2cqt). Residues important for catalysis and substrate binding in wild-type CP are shown in green (R362, W488, D490, Y653, E659). Residues that were found to alter or broaden substrate specificity upon mutation are shown in yellow (N156, N163, T508, E649, N667) and the flexible loop (495–513) is shown in gray. Bound substrates (shown in blue) are glucose (acceptor site), glycerol (donor site) and inorganic phosphate.

2010b; Yamamoto et al., 2006). However, in order to develop a generic glycosylation technology based on glycoside phosphorylases, enzymes are needed with completely novel substrate specificity. We have recently subjected the wild-type C. uda CP to directed evolution and protein engineering experiments in order to broaden its substrate specificity. In the following sections the most important results will be discussed and the applicability of the obtained enzymes for glycoside synthesis will be evaluated. 3.1. Modification of CP donor specificity The type of monosaccharide attached to an aglycone can significantly alter its properties and fate in the human body (Wong and Toth, 2001). For example, glucosylated drugs are targeted to the kidneys while galactosylated drugs are targeted towards the liver. Unfortunately, few phosphorylases are capable of transferring a galactose moiety, the only exceptions being the enzymes from family GH-112 (Nakajima et al., 2009). In that case, the required glycosyl donor is ␣-galactose-1-phosphate (␣Gal1P), which is much more expensive than the more common ␣Glc1P. In theory, it could be produced with the help of lacto-N-biose phosphorylase, but the corresponding substrate is not readily available in large quantities (Nishimoto and Kitaoka, 2007). In that respect, it is remarkable that no phosphorolytic enzyme has been found yet with a substrate preference for lactose, the most common ␤-galactoside in nature. Since the availability of a lactose phosphorylase (LP) would be of great interest for the synthesis of ␣Gal1P from the cheap and abundant substrate lactose, we decided to create such an enzyme using directed evolution and protein engineering techniques (De Groeve et al., 2009a). As cellobiose is very similar to lactose, differing only in the position of the C-4 hydroxyl group at the non-reducing end, it was hypothesized that cellobiose phosphorylase would be a good starting point to create lactose phosphorylase enzymes by directed evolution. Interestingly, we found that the wild-type CP from C. uda already displays activity towards lactose, albeit much less than towards cellobiose (around 0.4%). Nevertheless, this LP activity was significant and could be further increased by introducing random mutations through error-prone PCR. The use of a selection system based on growth in liquid minimal medium with lactose as the sole carbon source, allowed the

discovery of the best enzyme variant present in the library. Indeed, when a typical E. coli cloning host (containing a lacZ deletion) is transformed with the mutated CP library, then cells expressing a highly active LP enzyme variant will grow quicker than others and will become enriched in the selection culture. In that way, an enzyme variant carrying 6 amino acid mutations was found, which we called LP1. Since it was created by error-prone PCR, there was a high chance that not all mutations contribute to the activity on lactose. Indeed, only two of them were found to be beneficial, two were neutral, and two even had a negative effect. After elimination of the unnecessary mutations, a variant containing only two mutations (T508A/N667T) was obtained, referred to as LP2. Its activity could be increased even more by site-saturation mutagenesis to find the best possible amino acid at the beneficial sites. The resulting LP3 variant (T508I/N667A) is the most active LP enzyme found to date. Its LP activity was about 10 times higher than that of the wildtype CP. Interestingly, the improvement was completely due to an increase in kcat value, while the Km for lactose remained relatively high (∼0.3 M). The two beneficial sites for LP activity are located far away from each other in the enzyme’s crystal structure. While N667 is located near the donor site, close to a ‘hydrophobic platform’ that is important for transition-state stabilization at the donor site, residue T508 is located on a flexible loop near the entrance of the active site (Fig. 2). This confirmed for the first time a role of the flexible loop in substrate specificity of C. uda CP. Most likely, mutation of T508 changes the flexibility of this loop, causing an accelerated product release or making it more suitable for catalysis with lactose as a substrate. 3.2. Modification of CP acceptor specificity Besides donor specificity, the acceptor specificity of C. uda CP was also subjected to protein engineering. For this purpose, a general screening system for phosphate-releasing enzymes was developed, based on the fact that every phosphorylase releases inorganic phosphate in the direction of glycoside synthesis, independent of the substrates used. A modified phosphomolybdate method was used to measure the amount of released inorganic phosphate, which is directly proportional to enzyme activity (De Groeve et al., 2010b).

M.R.M. de Groeve et al. / Journal of Biotechnology 156 (2011) 253–260

Theoretically, modification of acceptor specificity should be less difficult to achieve than modification of donor specificity. This is because the donor site is involved in transition state binding and stabilization, and even small changes near the donor site can easily result in loss of catalytic activity. The acceptor specificity of wild-type C. uda CP is relatively narrow, especially compared to other phosphorylases such as sucrose phosphorylase. In particular, the enzyme does not accept glucosides with anomeric substitution. Structural investigation of the C. gilvus structure revealed that the side chain of residue E649 interacts with the anomeric hydroxyl group of the glucose acceptor through a hydrogen bond (Fig. 2). It was therefore hypothesized that this interaction prevents C1-substituted glucosides to function as acceptor. Site-saturation mutagenesis of E649 led to the discovery of an E649C enzyme variant that displays novel acceptor specificity towards methyl ␤glucoside, ethyl ␤-glucoside and phenyl ␤-glucoside (De Groeve et al., 2010b). These results confirmed the importance of residue E649 in wild-type C. uda CP for acceptance of anomerically substituted glucosides. Having found enzyme variants with either modified donor or acceptor specificity, a next step was to combine the beneficial mutations in a single variant. Such an enzyme would be useful to synthesize e.g., methyl ␤-lactoside, which has been reported to have anti-cancer activity (Oguchi et al., 1990). However, introducing the acceptor mutation (E649C) in the LP3 variant was found to abolish all activity on lactose (De Groeve et al., 2010a). Surprisingly, these experiments also revealed that the LP3 variant already displayed novel acceptor specificity towards alkyl ␤-glucosides, although the mutations T508I and N667A were specifically introduced to modify the donor specificity. So in this case, we got more than what we screened for. It also shows that there is more than one way to introduce activity towards anomerically substituted ␤-glucosides in CP. Nevertheless, the enzyme variant containing both LP3 mutations and the E649C mutation displays a broader acceptor specificity than either of its parent enzymes. Site-saturation mutagenesis revealed that the LP3/E649G variant displays the same acceptor specificity as the LP3/E649C variant (De Groeve et al., 2010a). Because of the smaller size of a glycine, which might create more space near the active site, the LP3/E649G mutant was used as starting point for the creation of new enzyme variants with further broadened acceptor specificity. From the 3D structure, three residues were selected near the active site entrance that potentially influence the acceptor specificity (N156, D159 and N163). Site-saturation mutagenesis of these positions followed by screening for activity on the acceptor octyl ␤-glucoside led to the discovery of a so-called ‘octyl ␤-cellobiose phosphorylase’ (OCP2) variant containing five amino acid substitutions: N156D, N163D, T508I, E649G and N667A. This variant showed a remarkably broad acceptor specificity towards ␤-glucosides as well as an ␣-glucoside (Table 3). Interestingly, the broadened acceptor specificity of OCP2 compared to the LP3/E649G parent is caused by mutating asparagine residues to aspartate (N156D and N163D), although negatively charged residues are not expected to interact favourably with a hydrophobic alkyl chain. While the activity of OCP2 on non-native acceptors is relatively low in comparison with wild-type CP (<20%, Table 3), it has been reported in literature that promiscuous enzyme variants are good starting points for the creation of specific mutants with high activity (Tracewell and Arnold, 2009). Therefore the OCP2 mutant is a good candidate for the creation of CP enzyme variants with high activity towards a specific glucoside. The apparently low activity of OCP2 is mainly caused by an increase in Km (5-fold up to more than 60-fold), since the kcat values decreased only slightly for most non-native acceptors. Given that for industrial purposes the kcat value is more important than the Km value (Fox and Clay, 2009), the OCP2 variant

257

might be an interesting biocatalyst for practical glycoside synthesis. 3.3. Loop engineering experiments Although CP and ChP have very similar structures, the shape of the entrance to their active sites is quite different. While ChP has a relatively ‘open’ active site pocket, the active-site entrance of CP is ‘closed’ due to the presence of a long loop consisting of residues 495–513 (Fig. 3A). This loop was found to have a relatively high B-factor and can thus be regarded as flexible (Hidaka et al., 2006). We have shown experimentally that the flexible loop influences activity, stability and substrate specificity of C. uda CP (De Groeve, 2009). Residue T508, for example, is part of the loop and was found to influence both donor and acceptor specificity (see earlier). Since the flexible loop is located near the acceptor binding site, it was hypothesized that it is responsible for the narrow substrate specificity of CP in comparison with ChP. To broaden the acceptor specificity of CP, the ChP acceptor site was mimicked by constructing loop deletion mutants with a shortened ((497–508), (504–506) and (504–505)) or completely removed ((495–513)) loop structure (De Groeve, 2009). While full deletion of the loop resulted in complete loss of activity, the partial loop deletion mutants were active, but had less than 10% of wild-type activity. Furthermore, these mutants were unstable, having half-life times of less than 20 h, although stability could be partially restored by the addition of cellobiose during storage. Besides loop deletions, we also created enzyme variants in which one or more loop residues were mutated. Residues S504, F505 and Q506 are located at the tip of the loop and oriented towards the acceptor site (Fig. 4). They are conserved among most CP enzymes (Fig. 3B) and thus may be important for activity and specificity. Residue F505 was found to be absolutely crucial, since its substitution by a glycine results in dramatic loss of activity. In contrast, mutation of the adjacent S504 and Q506 to glycine had only small effect on CP activity, while no LP activity could be detected anymore in these variants. To allow a more systematic evaluation of loop residues 507–514, each of them was submitted to saturation mutagenesis in the wild-type enzyme and the library was screened for modified acceptor and donor specificity (De Groeve, 2009). This revealed several loop residues that are important for substrate specificity and confirmed the importance of residue T508, which was previously found by error-prone PCR. It can be clearly seen that the flexible loop in CP has high potential for the creation of enzyme variants with novel substrate specificity (De Groeve, 2009). The importance of loop structures for enzyme stability, activity and specificity has already been shown in literature. For example, Nakai and co-workers were able to introduce either trehalose or kojibiose phosphorylase activity in Lactobacillus acidophilus maltose phosphorylase (GH65) by mutating a short loop near the active site pocket (Nakai et al., 2010b). Another example is the involvement of a loop in the B-domain of amylosucrase in substrate specificity and enzyme stability, similar to what was observed with cyclodextrin glycosyltransferase, another GH13 enzyme (van der Veen et al., 2006). 4. Application of engineered CPs for glycoside synthesis Some of the created CP enzyme variants have been applied for glycoside synthesis. For example, the LP3 mutant was used in a production process for ␣Gal1P using Triton X-100 permeabilized Escherichia coli cells (De Groeve et al., 2009b). Purification of the ␣Gal1P using anion-exchange on Dowex 1-X8 was very efficient and 9.5 g of purified product was obtained from a 1 L reaction volume. Since only lactose and phosphate are needed as sub-

258

M.R.M. de Groeve et al. / Journal of Biotechnology 156 (2011) 253–260

Table 3 Comparison of acceptor specificities of wild-type C. uda CP and CP enzyme variants. Acceptors were used at 100 mM each, except for glucose (30 mM), octyl ␤-glucoside (50 mM), phenyl ␤-glucoside (50 mM) and p-nitrophenyl ␤-glucoside (50 mM). Data partly derived from (De Groeve et al., 2010a). Acceptor

Glucose Methyl ␤-glucoside Ethyl ␤-glucoside Butyl ␤-glucoside Hexyl ␤-glucoside Octyl ␤-glucoside Methyl ␣-glucoside tert-Butyl ␤-glucoside Phenyl ␤-glucoside p-Nitrophenyl ␤-glucoside Methyl ␤-cellobioside Cellobiose a b c

Relative activity WT

E649C

LP3a

OCP2b

100% –c – – – – – – – – – –

7.7% 11.0% 2.8% – – – – – 8.1% – – –

2.8% 5.5% 1.1% – – – – 5.6% – – – 5.4%

7.7% 13.3% 13.0% 13.9% 5.7% 2.6% 3.5% – 18.8% 2.1% 8.1% 5.8%

LP3 = T508I/N667A. OCP2 = N156D/N163D/T508I/E649G/N667A. –: <1% of wild-type CP activity.

strates, the lactose phosphorolysis reaction is thus a convenient and cost-effective way to produce the relatively expensive ␣Gal1P (>700 EUR/g, commercial price at Sigma–Aldrich, January 2011). Furthermore, due to the strict regiospecificity of phosphorolytic enzymes, a pure anomeric product is obtained. For the first time an easy and cost-effective synthesis route for ␣Gal1P was devel-

oped, which is highly desirable in view of a general glycosylation technology based on glycoside phosphorylases. Also lactose could be produced with the LP3 mutant, using the earlier produced ␣Gal1P as glycosyl donor and glucose as acceptor (De Groeve, 2009). At equilibrium, around 57 mM of lactose was obtained from 100 mM initial substrate concentrations. The lactose was purified using anion-exchange on Dowex 1-X8 followed by activated charcoal chromatography. About 83 mg of purified lactose was obtained as a white powder (∼11% purification yield). In nature, mammals synthesize lactose by the action of a lactose synthase complex composed of ␣-lactalbumin and ␤1,4-galactosyltransferase (GalT), which catalyzes the transfer of UDP-galactose to glucose. The synthesis route based on LP enzymes is simpler and is a completely new way of synthesizing lactose using natural substrates. The produced lactose, which is not directly from animal origin, can be of interest in certain pharmaceutical applications. Exploiting the surprising acceptor specificity of the LP3 mutant towards methyl ␤-glucoside, we could synthesize methyl ␤lactoside from ␣Gal1P as donor and methyl ␤-glucoside as acceptor, albeit with low yield (∼9% substrate conversion) (De Groeve, 2009). Methyl ␤-lactoside has potentially interesting properties since it was shown to have anti-cancer effects (Oguchi et al., 1990). Methyl ␤-cellobioside, in contrast, could be synthesized in a high yield with the E649C mutant using methyl ␤-glucoside as acceptor and ␣Glc1P as donor (De Groeve, 2009). The acceptor specificity of LP3 towards cellobiose is interesting for the synthesis of lactocellobiose, which may have

Fig. 3. The flexible loop of CP near the active site entrance is absent in ChPs. Panel A: Comparison of the active site entrance in C. gilvus CP (left, PDB 2cqt) and chitobiose phosphorylase (PDB 1v7w). For better visualization only one monomer is shown. The flexible loop structure (495–513) in CP and bound substrates are shown in color. Panel B: Amino acid sequence alignment of the flexible loop region in CP and ChP sequences. Abbreviations used: CP, cellobiose phosphorylase; ChP, chitobiose phosphorylase; CG, Cellvibrio gilvus; CU, Cellulomonas uda; CT, Clostridium thermocellum; CS, Clostridium stercorarium; TN, Thermotoga neapolitana; TM, Thermotoga maritima; VP, Vibrio proteolyticus; VF, Vibrio furnisii.

M.R.M. de Groeve et al. / Journal of Biotechnology 156 (2011) 253–260

259

ants would be useful to rationalize the beneficial effect of the obtained mutations. These structures may provide further insight into the role of certain structural elements in substrate specificity. It would also be interesting to evaluate the beneficial mutations in other GH-94 enzymes, such as chitobiose phosphorylase and cellodextrin phosphorylase. In addition, engineering of the flexible loop in CP can potentially further broaden its application potential, because it was found to be important for enzyme activity as well as specificity. Since the screening procedure that we developed for CP is also applicable to other phosphorolytic enzymes, the discovery of novel phosphorylase enzyme variants with modified substrate specificities will be accelerated. Ideally, a whole range of engineered phosphorolytic enzymes will be available with clearly defined substrate and regio-specificity. If an aglycone of interest should be glycosylated, one can then choose a suitable candidate from these enzyme variants and further optimize it by enzyme engineering. Acknowledgements Research funded by grants SB51293, SB53293 and SBO50191 of the Institute for the Promotion of Innovation through Science and Technology in Flanders (IWT-Vlaanderen). References Fig. 4. Detailed view on the flexible loop structure (495–513) in C. gilvus CP (PDB 2cqt).

similar prebiotic properties as reported for lactosucrose. Furthermore, in future experiments the wild-type acceptor specificity of LP3 towards mannose may be exploited for the synthesis of epilactose, a known prebiotic (Watanabe et al., 2008). 5. Conclusions and outlook One of the major disadvantages of the use of CP for glycoside synthesis – its narrow substrate specificity – has been successfully addressed by directed evolution and protein engineering. Enzyme variants of C. uda CP with broadened substrate specificity were created and applied for the synthesis of several carbohydrates and derivatives. Furthermore, new insights in the structural determinants of the substrate specificity in CP were obtained. The synergy between acceptor and donor mutations is a remarkable finding. It was indeed surprising that the LP3 donor mutant displays modified and relatively broad acceptor specificity. Due to the lack of protein engineering experiments on other glycoside phosphorylases, it is not known whether this is a universal observation among phosphorolytic enzymes or limited to C. uda CP. A next step towards the development of a generic glycosylation technology would be the modification of CP acceptor specificity from (substituted) glucose towards various alcohols, aromatics, vitamins, drugs, etc. The obtained CP mutants in this research should be good starting points to achieve this goal since they already show relaxed substrate specificity towards relatively large acceptors. Furthermore, the fact that the wild-type CP from Clostridium thermocellum is active on various aliphatic alcohols (Kino et al., 2008) highlights the potential of CP for ␤-glucosylation and ␤galactosylation of various aglycones. It should be noted, however, that the molar concentrations of alcohols used in that research were in the range of 2–8 M, which is well above the solubility of most vitamins and drug molecules. Several sites (hot spots) were found that influence substrate specificity in C. uda CP. Crystal structures of obtained enzyme vari-

Alexander, J.K., 1968. Purification and specificity of cellobiose phosphorylase from Clostridium thermocellum. J. Biol. Chem. 243, 2899–2904. Bottcher, D., Bornscheuer, U.T., 2010. Protein engineering of microbial enzymes. Curr. Opin. Microbiol. 13, 274–282. Ciocchini, A.E., Guidolin, L.S., Casabuono, A.C., Couto, A.S., de Iannino, N.I., Ugalde, R.A., 2007. A glycosyltransferase with a length-controlling activity as a mechanism to regulate the size of polysaccharides. Proc. Natl. Acad. Sci. U.S.A. 104, 16492–16497. De Groeve, M.R.M., 2009. Engineering of cellobiose phosphorylase for glycoside synthesis. PhD thesis, Faculty of Bioscience Engineering, Ghent University, Ghent, Belgium, p. 243. De Groeve, M.R.M., De Baere, M., Hoflack, L., Desmet, T., Vandamme, E.J., Soetaert, W., 2009a. Creating lactose phosphorylase enzymes by directed evolution of cellobiose phosphorylase. Protein Eng. Des. Sel. 22, 393–399. De Groeve, M.R.M., Depreitere, V., Desmet, T., Soetaert, W., 2009b. Enzymatic production of alpha-d-galactose 1-phosphate by lactose phosphorolysis. Biotechnol. Lett. 31, 1873–1877. De Groeve, M.R.M., Remmery, L., Van Hoorebeke, A., Stout, J., Desmet, T., Savvides, S.N., Soetaert, W., 2010a. Construction of cellobiose phosphorylase variants with broadened acceptor specificity towards anomerically substituted glucosides. Biotechnol. Bioeng. 107, 413–420. De Groeve, M.R.M., Tran, G.H., Van Hoorebeke, A., Stout, J., Desmet, T., Savvides, S.N., Soetaert, W., 2010b. Development and application of a screening assay for glycoside phosphorylases. Anal. Biochem. 401, 162–167. Desmet, T., Soetaert, W., 2011. Enzymatic glycosyl transfer: mechanisms and applications. Biocatal. Biotransform. 29, 1–18. Fox, R.J., Clay, M.D., 2009. Catalytic effectiveness, a measure of enzyme proficiency for industrial applications. Trends Biotechnol. 27, 137–140. Fujii, K., Iboshi, M., Yanase, M., Takaha, T., Kuriki, T., 2006. Enhancing the thermal stability of sucrose phosphorylase from Streptococcus mutans by random mutagenesis. J. Appl. Glycosci. 53, 91–97. Fushinobu, S., Mertz, B., Hill, A.D., Hidaka, M., Kitaoka, M., Reilly, P.J., 2008. Computational analyses of the conformational itinerary along the reaction pathway of GH94 cellobiose phosphorylase. Carbohydr. Res. 343, 1023–1033. Goedl, C., Sawangwan, T., Mueller, M., Schwarz, A., Nidetzky, B., 2008. A high-yielding biocatalytic process for the production of 2-O-(alpha-d-glucopyranosyl)-snglycerol, a natural osmolyte and useful moisturizing ingredient. Angew. Chem. Int. Ed. Engl. 47, 10086–10089. Henrissat, B., Davies, G., 1997. Structural and sequence-based classification of glycoside hydrolases. Curr. Opin. Struct. Biol. 7, 637–644. Hidaka, M., Kitaoka, M., Hayashi, K., Wakagi, T., Shoun, H., Fushinobu, S., 2006. Structural dissection of the reaction mechanism of cellobiose phosphorylase. Biochem. J. 398, 37–43. Kim, Y.K., Kitaoka, M., Krishnareddy, M., Mori, Y., Hayashi, K., 2002. Kinetic studies of a recombinant cellobiose phosphorylase (CBP) of the Clostridium thermocellum YM4 strain expressed in Escherichia coli. J. Biochem. 132, 197–203. Kino, K., Satake, R., Morimatsu, T., Kuratsu, S., Shimizu, Y., Sato, M., Kirimura, K., 2008. A new method of synthesis of alkyl beta-glycosides using sucrose as sugar donor. Biosci. Biotechnol. Biochem. 72, 2415–2417.

260

M.R.M. de Groeve et al. / Journal of Biotechnology 156 (2011) 253–260

Kitaoka, M., Hayashi, K., 2002. Carbohydrate-processing phosphorolytic enzymes. Trends Glycosci. Glycotechnol. 14, 35–50. Kitaoka, M., Nomura, S., Yoshida, M., Hayashi, K., 2006. Reaction on d-glucal by an inverting phosphorylase to synthesize derivatives of 2-deoxy-beta-d-arabinohexopyranosyl-(1 → 4)-d-glucose (2II-deoxycellobiose). Carbohydr. Res. 341, 545–549. Kitaoka, M., Sasaki, T., Taniguchi, H., 1992. Synthetic reaction of Cellvibrio gilvus cellobiose phosphorylase. J. Biochem. 112, 40–44. Kren, V., Martinkova, L., 2001. Glycosides in medicine: the role of glycosidic residue in biological activity. Curr. Med. Chem. 8, 1303–1328. Leu, Y.L., Roffler, S.R., Chern, J.W., 1999. Design and synthesis of water-soluble glucuronide derivatives of camptothecin for cancer prodrug monotherapy and antibody-directed enzyme prodrug therapy (ADEPT). J. Med. Chem. 42, 3623–3628. Luley-Goedl, C., Nidetzky, B., 2010. Carbohydrate synthesis by disaccharide phosphorylases: Reactions, catalytic mechanisms and application in the glycosciences. Biotechnol. J. 5, 1324–1338. Nakai, H., Hachem, M.A., Petersen, B.O., Westphal, Y., Mannerstedt, K., Baumann, M.J., Dilokpimol, A., Schols, H.A., Duus, J.O., Svensson, B., 2010a. Efficient chemoenzymatic oligosaccharide synthesis by reverse phosphorolysis using cellobiose phosphorylase and cellodextrin phosphorylase from Clostridium thermocellum. Biochimie 92, 1818–1826. Nakai, H., Petersen, B.O., Westphal, Y., Dilokpimol, A., Abou Hachem, M., Duus, J.O., Schols, H.A., Svensson, B., 2010b. Rational engineering of Lactobacillus acidophilus NCFM maltose phosphorylase into either trehalose or kojibiose dual specificity phosphorylase. Protein Eng. Des. Sel. 23, 781–787. Nakajima, M., Nishimoto, M., Kitaoka, M., 2009. Characterization of three beta-galactoside phosphorylases from Clostridium phytofermentans: discovery of d-galactosyl-beta1 → 4-l-rhamnose phosphorylase. J. Biol. Chem. 284, 19220–19227. Ng, T.K., Zeikus, J.G., 1986. Synthesis of [14C] cellobiose with Clostridium thermocellum cellobiose phosphorylase. Appl. Environ. Microbiol. 52, 902–904. Nidetzky, B., Eis, C., Albert, M., 2000. Role of non-covalent enzyme-substrate interactions in the reaction catalysed by cellobiose phosphorylase from Cellulomonas uda. Biochem. J. 351, 649–659. Nidetzky, B., Griessler, R., Schwarz, A., Splechtna, B., 2004. Cellobiose phosphorylase from Cellulomonas uda: gene cloning and expression in Escherichia coli, and application of the recombinant enzyme in a ‘glycosynthase-type’ reaction. J. Mol. Catal. B-Enzym. 29, 241–248. Nishimoto, M., Kitaoka, M., 2007. Practical preparation of lacto-N-biose I, a candidate for the bifidus factor in human milk. Biosci. Biotechnol. Biochem. 71, 2101–2104. Oguchi, H., Toyokuni, T., Dean, B., Ito, H., Otsuji, E., Jones, V.L., Sadozai, K.K., Hakomori, S., 1990. Effect of lactose derivatives on metastatic potential of B16 melanoma cells. Cancer Commun. 2, 311–316.

Percy, A., Ono, H., Hayashi, K., 1998. Acceptor specificity of cellobiose phosphorylase from Cellvibrio gilvus: synthesis of three branched trisaccharides. Carbohydr. Res. 308, 423–429. Percy, A., Ono, H., Watt, D., Hayashi, K., 1997. Synthesis of beta-dglucopyranosyl-(1 → 4)-d-arabinose, beta-d-glucopyranosyl-(1 → 4)-l-fucose and beta-d-glucopyranosyl-(1 → 4)-d-altrose catalysed by cellobiose phosphorylase from Cellvibrio gilvus. Carbohydr. Res. 305, 543–548. Petrek, M., Otyepka, M., Banas, P., Kosinova, P., Koca, J., Damborsky, J., 2006. CAVER: a new tool to explore routes from protein clefts, pockets and cavities. BMC Bioinformatics 7, 316. Rajashekhara, E., Kitaoka, M., Kim, Y.K., Hayashi, K., 2002. Characterization of a cellobiose phosphorylase from a hyperthermophilic eubacterium, Thermotoga maritima MSB8. Biosci. Biotechnol. Biochem. 66, 2578–2586. Reetz, M.T., 2010. Laboratory evolution of stereoselective enzymes: a prolific source of catalysts for asymmetric reactions. Angew. Chem. Int. Ed. Engl. 50, 138–174. Reichenbecher, M., Lottspeich, F., Bronnenmeier, K., 1997. Purification and properties of a cellobiose phosphorylase (CepA) and a cellodextrin phosphorylase (CepB) from the cellulolytic thermophile Clostridium stercorarium. Eur. J. Biochem. 247, 262–267. Sih, C.J., McBee, R.H., 1955a. A cellobiose phosphorylase in Clostridium thermocellum. Proc. Montana Acad. Sci. 15, 21–22. Sih, C.J., McBee, R.H., 1955b. A phosphorylase active on cellobiose. Bacteriol. Proc., 126. Tracewell, C.A., Arnold, F.H., 2009. Directed enzyme evolution: climbing fitness peaks one amino acid at a time. Curr. Opin. Chem. Biol. 13, 3–9. van der Veen, B.A., Skov, L.K., Potocki-Veronese, G., Gajhede, M., Monsan, P., Remaud-Simeon, M., 2006. Increased amylosucrase activity and specificity, and identification of regions important for activity, specificity and stability through molecular evolution. FEBS J. 273, 673–681. Wang, Y.B., 2009. Prebiotics: present and future in food science and technology. Food Res. Int. 42, 8–12. Watanabe, J., Nishimukai, M., Taguchi, H., Senoura, T., Hamada, S., Matsui, H., Yamamoto, T., Wasaki, J., Hara, H., Ito, S., 2008. Prebiotic properties of epilactose. J. Dairy Sci. 91, 4518–4526. Wong, A., Toth, I., 2001. Lipid, sugar and liposaccharide based delivery systems. Curr. Med. Chem. 8, 1123–1136. Yamamoto, T., Mukai, K., Yamashita, H., Kubota, M., Fukuda, S., Kurimoto, M., Tsujisaka, Y., 2005. Enhancement of thermostability of kojibiose phosphorylase from Thermoanaerobacter brockii ATCC35047 by random mutagenesis. J. Biosci. Bioeng. 100, 212–215. Yamamoto, T., Yamashita, H., Mukai, K., Watanabe, H., Kubota, M., Chaen, H., Fukuda, S., 2006. Construction and characterization of chimeric enzymes of kojibiose phosphorylase and trehalose phosphorylase Thermoanaerobacter brockii. Carbohydr. Res. 341, 2350–2359.