Review pubs.acs.org/JAFC
Methods for Improving Enzymatic Trans-glycosylation for Synthesis of Human Milk Oligosaccharide Biomimetics Birgitte Zeuner, Carsten Jers, Jørn Dalgaard Mikkelsen, and Anne S. Meyer* Center for BioProcess Engineering, Department of Chemical and Biochemical Engineering, Technical University of Denmark, Building 229, DK-2800 Kgs. Lyngby, Denmark ABSTRACT: Recently, significant progress has been made within enzymatic synthesis of biomimetic, functional glycans, including, for example, human milk oligosaccharides. These compounds are mainly composed of N-acetylglucosamine, fucose, sialic acid, galactose, and glucose, and their controlled enzymatic synthesis is a novel field of research in advanced food ingredient chemistry, involving the use of rare enzymes, which have until now mainly been studied for their biochemical significance, not for targeted biosynthesis applications. For the enzymatic synthesis of biofunctional glycans reaction parameter optimization to promote “reverse” catalysis with glycosidases is currently preferred over the use of glycosyl transferases. Numerous methods exist for minimizing the undesirable glycosidase-catalyzed hydrolysis and for improving the trans-glycosylation yields. This review provides an overview of the approaches and data available concerning optimization of enzymatic trans-glycosylation for novel synthesis of complex bioactive carbohydrates using sialidases, α-L-fucosidases, and β-galactosidases as examples. The use of an adequately high acceptor/donor ratio, reaction time control, continuous product removal, enzyme recycling, and/or the use of cosolvents may significantly improve trans-glycosylation and biocatalytic productivity of the enzymatic reactions. Protein engineering is also a promising technique for obtaining high trans-glycosylation yields, and proof-of-concept for reversing sialidase activity to trans-sialidase action has been established. However, the protein engineering route currently requires significant research efforts in each case because the structure−function relationship of the enzymes is presently poorly understood. KEYWORDS: trans-glycosylation, sialidase, α-L-fucosidase, β-galactosidase, human milk oligosaccharides, biomimetic glycans
■
α-L-fucosidases (EC 3.2.1.51), and β-galactosidases (EC 3.2.1.23), because these enzyme classes can be employed for production of valuable compounds, and significant progress has recently been achieved in targeted trans-glycosylation reactions employing these enzymes. Trans-sialylation and trans-fucosylation are crucial in the synthesis of human milk oligosaccharides (HMO), which have numerous beneficial effects on infant health.11 Furthermore, β-galactosidase-catalyzed synthesis of N-substituted glycosides such as N-acetyl-D-lactosamine (LacNAc) or lacto-N-biose can assist in the synthesis of HMO and HMO-like structures. The same structures are found in glycoside antigens, and their synthesis may thus be important for the development of diagnostic tools and cancer treatment.12−15 β-Galactosidases have also been used for the synthesis of prebiotic galacto-oligosaccharides (GOS) for decades, but this process has been thoroughly reviewed elsewhere.16 Implementing the methods reviewed in this paper may lead to improved glycosidase-catalyzed trans-glycosylation, which represents a biocompatible route to valuable products, which may soon find their way into food applications.
INTRODUCTION The interest in enzymatic oligosaccharide synthesis is profound, not least owing to the fact that chemical synthesis involving saccharides is cumbersome, usually relying on multiple steps, complicated protective group manipulations, and use of harmful chemicals. Recently, several examples of using glycosidases (EC 3.2.1.−) or novel engineered glycosidases for the synthesis of specific, functional, biomimetic glycans have been published.1−6 Enzymatic glycosidase-catalyzed synthesis is hampered by the competing hydrolysis naturally catalyzed by this type of enzyme. Even so, glycosidases are often preferred over glycosyl transferases, which catalyze the transfer of glycosyl groups without the contaminating hydrolysis, because glycosidases are easier to produce, better characterized, and generally more widely available.5,7 Glycosyl transferases, moreover, require expensive nucleotide sugars as glycosyl donors or the use of a multienzyme substrate-recycling system to avoid enzyme inhibition by the released nucleoside diphosphates.7,8 The use of cell factories that via glycosyl transferase action and de novo synthesis of substrate can produce trans-glycosylation products in high yield provides an interesting approach for overcoming some of these problems.9,10 Multifunctional glycosyl transferases catalyzing transglycosylation with inexpensive substrates have been described, but their use is generally hampered by the enzymes exhibiting very low specific activity. This review examines the numerous reaction design methods that can be used for improvement of biocatalytic synthesis reaction yields, notably with respect to suppression of the undesirable hydrolysis, and discusses the applicability of the different strategies. The focus is on sialidases (EC 3.2.1.18), © 2014 American Chemical Society
■
GLYCOSIDASE-CATALYZED TRANS-GLYCOSYLATION Glycosidase-catalyzed synthesis can occur as (1) thermodynamically controlled reverse hydrolysis, that is, in essence a Received: Revised: Accepted: Published: 9615
June 3, 2014 August 12, 2014 September 10, 2014 September 10, 2014 dx.doi.org/10.1021/jf502619p | J. Agric. Food Chem. 2014, 62, 9615−9631
Journal of Agricultural and Food Chemistry
Review
condensation reaction, where the donor substrate is a monosaccharide, or as (2) kinetically controlled trans-glycosylation, where an activated donor, for example, a disaccharide or a p-nitrophenyl (pNP) monosaccharide, is employed. In reverse hydrolysis, the maximum yield is determined by the equilibrium, and reaction times are often extended. In contrast, trans-glycosylation makes it possible to “overshoot” the equilibrium to favor the glycosylated acceptor product if the substrate is more reactive than the product, and reaction times are generally shorter than in reverse hydrolysis.7,17−19 In both cases, the synthetic reaction takes place in competition with hydrolysis of both substrate and product. As first established by Koshland,20 glycosidases can catalyze the hydrolysis of substrates or the synthesis of products with either retention or inversion of the anomeric configuration. The glycosidases in focus in this review, that is, β-galactosidases, belonging to glycosyl hydrolase (GH) families 1, 2, 35, and 42, sialidases, mainly belonging to GH family 33, and α-Lfucosidases, mainly belonging to GH family 29, are all retaining glycosidases and employ the classical Koshland doubledisplacement mechanism. This double-displacement mechanism involves the formation of a covalent glycosyl-enzyme intermediate upon binding of the donor in the active site.20,21 Figure 1 shows the reactions catalyzed by sialidases, β-galactosidases, and α-L-fucosidases with formation of the glycosyl-enzyme intermediate and concomitant release of the aglycone from the active site, consistent with a ping-pong bibi mechanism, which reduces to an ordered unibi mechanism in the competing hydrolysis.22 This mechanism has been observed for many glycosidases,23−26 but evidence of an ordered bibi mechanism with formation of a ternary donor−acceptor−enzyme complex has also been reported for the retaining β-galactosidase from Bacillus circulans (EC 3.2.1.23; GH 42).1 Recently, a hybrid of the two mechanisms with formation of the ternary complex prior to formation of the sialyl-enzyme intermediate has been suggested for the native trans-sialidase from Trypanosoma cruzi (EC 3.2.1.18; GH 33).27 In the ping-pong bibi mechanism, the glycosyl-enzyme intermediate then undergoes nucleophilic attack, either by water, resulting in hydrolysis, or by another acceptor, resulting in trans-glycosylation (Figure 1). It has been hypothesized that increased stabilization of the covalent glycosyl-enzyme intermediate favors trans-glycosylation over hydrolysis.28 The trans-glycosylation product is also a substrate for the glycosidase and can thus undergo subsequent hydrolysis (secondary hydrolysis) or function as a donor substrate (Figure 1), the rate being dependent on the regioselectivity in the product formation because not all regioisomers are equally good substrates.29 The trans-glycosylation yield is therefore the result of a delicate balance between the rates of synthesis, donor hydrolysis, and product hydrolysis (Figure 1). At first glance, the three reactions in Figure 1 appear identical, but all three reactions have been included to emphasize the structural differences in the substrates and products related to each of the three enzyme classes. The ratio between the trans-glycosylation rate (rS) and the hydrolysis rate (rH) thus describes how efficiently the glycosylenzyme intermediate gives the desired trans-glycosylation product (Figure 1). It follows that in order to maximize transglycosylation yields, the rS/rH ratio must be maximized:7,18,30 aglycosyl acceptor [glycosyl acceptor] rS =S× = Sc × rH aw [water]
Thus, the trans-glycosylation yield is influenced both by the activities (or concentrations) of the competing nucleophiles and by the selectivity factor S (or Sc), which is dependent on enzyme source and also on choice of donor, acceptor, and reaction conditions.7,18,30,31 Indeed, the glycosidases most suitable for synthetic reactions are the ones with a high rS/rH ratio, especially because product hydrolysis occurs through the same glycosylenzyme intermediate.18 However, much can also be gained from optimizing the reaction conditions, that is, choice of donor, acceptor, acceptor concentration, acceptor/donor ratio, aw, and reaction time as discussed below.
■
SELECTIVITY AND SPECIFICITY The anomeric selectivity of glycosidases is absolute because it is determined by the reaction mechanism.7 In contrast, glycosidases often show broad selectivity for the donor and particularly for the acceptor molecules as outlined in the following sections. Regioselectivity in the product formation varies from enzyme to enzyme, is dependent on donor and acceptor structures, and may also be adjusted by the use of cosolvents or protein engineering (as discussed below). Donor Specificity. In trans-glycosylation, the glycosyl donor must carry a good anomeric leaving group. Disaccharides such as lactose are sufficiently activated, but nitrophenyl glycosides are also widely used.32,33 The donor must be fastreacting to keep reaction times short, thus allowing less time for product hydrolysis. Furthermore, a donor that binds tightly to the catalyst (low Km) also minimizes product hydrolysis.7 In several cases, aryl donor substrates have given up to 2.5 times higher yields than a di- or trisaccharide donor; this improved yield has been suggested to be due to the aryl aglycone being a better leaving group than the mono- or disaccharides.14,19,34−36 This is beneficial for synthesis on a smaller scale, but for largescale synthesis the price of both donors and acceptors becomes an important consideration. For instance, substances such as lactose and casein glycomacropeptide (to be discussed below) are interesting because they are abundantly available in industrial dairy streams. In comparison, nitrophenyl glycosides are expensive substrates and are not suitable for large-scale synthesis unless the resulting product is of extremely high value. Furthermore, the toxicity of the nitrophenol byproduct will be an issue in the approval of this process for use in food applications. Sialidases (EC 3.2.1.18) specifically catalyze the release of terminal sialic acid (N-acetylneuraminic acid) residues α-linked to glycoproteins, glycolipids, and polysaccharides, resulting in hydrolysis or trans-sialylation (Figure 1A). Typically, donor substrates such as pNP-α-sialic acid (Sia-α-pNP), 2′-(4methylumbelliferyl)-α-D-N-acetylneuraminic acid (Sia-α-MU), 3′-sialyl-lactose, and dimers and polymers of α(2,8)-linked sialic acid, as well as the fetal calf serum glycoprotein fetuin, have been used in trans-sialylation (Tables 1 and 2).15,35,37−39 Recently, however, a much cheaper source of sialic acid has been introduced in the synthesis of biomimetic compounds, namely, casein glycomacropeptide (CGMP), which is the soluble glycosylated casein residue produced by chymosin action on κ-casein during cheese manufacture (Tables 1 and 2).2,4,40−44 CGMP contains 5−11% (w/w) sialic acid depending on the source,40,45,46 of which approximately half is α(2,3)-linked to galactose and the other half is α(2,6)-linked to N-acetyl-Dgalactosamine (GalNAc).47 The well-studied trans-sialidase from T. cruzi is specific for the α(2,3)-linked sialyl residues only,38,48 whereas sialidases from Clostridium perfringens, Arthrobacter
(1) 9616
dx.doi.org/10.1021/jf502619p | J. Agric. Food Chem. 2014, 62, 9615−9631
Journal of Agricultural and Food Chemistry
Review
Figure 1. Scheme of competing reactions catalyzed by (A) sialidases, (B) α-L-fucosidases, and (C) β-galactosidases. For graphic simplicity, reaction schemes are shown according to the ping-pong bibi mechanism (see text). If the nucleophile attacking the glycosyl-enzyme intermediate is water, hydrolysis occurs. If another nucleophile attacks the intermediate, trans-glycosylation occurs. For β-galactosidases and some α-L-fucosidases, the donor substrate can also function as acceptor, resulting in donor self-condensation (see text). Regioselectivity in the product formation may vary (see text).
polymers (Table 1).35 So far, it is unclear whether α(2,6)-linked sialyl moieties on CGMP are utilized. Most sialidases accept aryl sialyl donors (Tables 1 and 2).14,35
ureafaciens, Vibrio cholerae (all EC 3.2.1.18; GH 33), and Newcastle disease virus (EC 3.2.1.18; GH 83) can also utilize terminal sialyl residues α(2,8)-linked in sialic acid di- and 9617
dx.doi.org/10.1021/jf502619p | J. Agric. Food Chem. 2014, 62, 9615−9631
Journal of Agricultural and Food Chemistry
Review
Table 1. Donor, Acceptor, and Acceptor/Donor (A:D) Ratio of Trans-sialylations Catalyzed by Sialidases as well as the Linkages, Total Trans-sialylation Yield, and Regioselectivity Obtaineda sialidase
donor
linkages formed in trans-sialylationb
acceptor
A:Dc
yieldd (%)
regioselectivitye
ref
complete complete
35 35 44
high (99%)
115
Arthrobacter ureafaciens
Sia-α(2,8)-Sia Sia-α-pNP CGMP
Lac Lac Lac
α(2,6′) α(2,6′) nd
Bacteroides f ragilis
colominic acid
Lac
α(2,6′), α(2,3′)
Bifidobacterium infantis
CGMP
Lac
nd
44g
1g
Clostridium perf ringens
Sia-α(2,8)-Sia Sia-α(2,8)-Sia colominic acid Sia-α-pNP Sia-α-pNP Sia-α-pNP Sia-α-pNP Sia-α-pNP Sia-α-pNP Sia-α-pNP Sia-α-pNP Sia-α-pNP
Lac LacNAc Lac Lac Tn-antigen T-antigen T-antigen precursor 2-deoxy-Gal-β-O-Me 2-deoxy-Gal-α-O-Me Gal-β(1,4)-2-deoxy-Gal-α-O-Me Gal-α-O-Me Gal-β-O-Me
α(2,6′) α(2,6′) α(2,6′) α(2,6′), α(2,3′) α(2,6) α(2,6′) α(2,6′) α(2,6) α(2,6) α(2,6′) α(2,6) α(2,6)
18 16f
7h nq 2h nq 10 4 7 14 15 5 10 10
complete complete complete medium (70%) complete complete complete complete complete complete complete complete
35 35 35 35 39 39 39 39 39 39 39 39
Sia-α(2,8)-Sia Sia-α(2,8)-Sia colominic acid Sia-α-pNP Sia-α-pNP Sia-α-pNP Sia-α-pNP Sia-α-pNP Sia-α-pNP Sia-α-pNP Sia-α-pNP Sia-α-pNP Sia-α-pNP
Lac LacNAc Lac Lac LacNAc Tn-antigen T-antigen T-antigen precursor 2-deoxy-Gal-β-O-Me 2-deoxy-Gal-α-O-Me Gal-β(1,4)-2-deoxy-Gal-α-O-Me Gal-α-O-Me Gal-β-O-Me
α(2,3′) α(2,3′) α(2,3′), α(2,6′) α(2,3′), α(2,6′) α(2,3′) α(2,3) α(2,3′) α(2,3′) α(2,3) α(2,3) α(2,3′) α(2,3) α(2,3)
18f 16f
5h nq nq nqf 10f,h 10 10 11 14 14 8 17 18
complete complete medium (75%) medium (76%) complete complete complete complete complete complete complete complete complete
35 35 35 35 35 39 39 39 39 39 39 39 39
Salmonella typhimurium
Sia-α-pNP Sia-α-pNP Sia-α-pNP Sia-α-pNP Sia-α-pNP Sia-α-pNP Sia-α-pNP Sia-α-pNP Sia-α-pNP Sia-α-pNP
Tn-antigen T-antigen T-antigen precursor 2-deoxy-Gal-β-O-Me 2-deoxy-Gal-α-O-Me Gal-β(1,4)-2-deoxy-Gal-α-O-Me Gal-α-O-Me Gal-β-O-Me LewisX LewisA
α(2,3), α(2,6) α(2,3′), α(2,6′) α(2,3′), α(2,6′) α(2,3) α(2,3), α(2,6) α(2,3′), α(2,6′) α(2,3), α(2,6) α(2,3) α(2,3) α(2,3)
5 5 5 5 5 5 5 5 13 13
15 11 11 20 22 10 16 15 9 12
high (95%) high (93%) medium (88%) complete high (92%) medium (86%) high (95%) complete complete complete
39 39 39 39 39 39 39 39 116 116
Trypanosoma rangeli mutant Tr6
CGMP CGMP CGMP CGMP CGMP CGMP CGMP
Lac Lac Lac LNT LNnT LNFP V LNFP I
α(2,3′) α(2,3′) α(2,3′) ndk ndk ndk ndk
25 25 25 25 5 5 5
28 37i 32j nq nq nq nq
complete complete complete
40 40 41 4 4 4 4
immobilized Tr6
CGMP
Lac
α(2,3′)
25
40
complete
40
Trypanosoma rangeli mutant Tr13
CGMP CGMP CGMP
Lac lactulose melibiose
α(2,3′) ndk ndk
44 44 44
≥31 23 12
complete
2 2 2
Newcastle disease virus
9618
f
18 5f 44g
nq nqf 5g 0.3
5f 7 7 7 7 7 7 7 7
5f 4 5 5 5 5 5 5 5 5
44
dx.doi.org/10.1021/jf502619p | J. Agric. Food Chem. 2014, 62, 9615−9631
Journal of Agricultural and Food Chemistry
Review
Table 1. continued sialidase
donor
linkages formed in trans-sialylationb
acceptor
k
A:Dc
yieldd (%)
regioselectivitye
ref
CGMP CGMP CGMP CGMP
maltose Fuc GOS IMO
nd ndk ndk ndk,l
44 44 44 44
13 18 nq nq
2 2 2 2
Vibrio cholerae
Sia-α(2,8)-Sia Sia-α-pNP Sia-α-pNP Sia-α-pNP Sia-α-pNP Sia-α-pNP Sia-α-pNP Sia-α-pNP Sia-α-pNP Sia-α-pNP Sia-α-pNP Sia-α-pNP Sia-α-pNP
Lac Lac Gal-β-O-Me Gal-β-O-Me Gal-β-O-Me Tn-antigen T-antigen T-antigen precursor 2-deoxy-Gal-β-O-Me 2-deoxy-Gal-α-O-Me Gal-β(1,4)-2-deoxy-Gal-α-O-Me Gal-α-O-Me Gal-β-O-Me
α(2,6′) α(2,6′), α(2,3′) α(2,6), α(2,3) α(2,6), α(2,3) α(2,6), α(2,3) α(2,6) α(2,6′) α(2,6′) α(2,6) α(2,6) α(2,6′), α(2,3′) α(2,6) α(2,6)
18f 5f 4 7 15 7 7 7 7 7 7 7 7
nq nqf 1.8 9.5 5.4 16 12 12 21 26 15 18 16
complete high (90%) medium (62%) medium (85%) high (90%) complete complete complete complete complete high (90%) complete complete
35 35 37 37 37 39 39 39 39 39 39 39 39
immobilized Vibrio cholerae
Sia-α-pNP Sia-α-pNP Sia-α-pNP Sia-α-pNP Sia-α-pNP Sia-α-pNP
Gal-β-O-Me Gal-β-O-Me Gal-α-O-Me Glc-β-O-Me Lac-β-O-Me LacNAc-β-O-Me
α(2,6), α(2,3) α(2,6), α(2,3) α(2,6), α(2,3) α(2,6), α(2,3) α(2,6′), α(2,3′) α(2,6′), α(2,3′)
7 7 7 7 7 7
20 20m 24 16 16 14
medium medium medium medium medium medium
37 37 37 37 37 37
(74%) (67%) (86%) (76%) (67%) (74%)
a
All of the microbial sialidases belong to GH 33, whereas the sialidase of viral origin (Newcastle disease virus) belongs to GH 83. Abbreviations: Sia, sialic acid (N-acetylneuraminic acid); Lac, β-D-lactose; nq, not quantified; Sia-α-pNP, p-nitrophenyl α-sialic acid (2-O-(p-nitrophenyl)-α-D-Nacetylneuraminic acid); CGMP, casein glycomacropeptide; nd, not determined; LacNAc, N-acetyl-D-lactosamine; LNT, lacto-N-tetraose; LNnT, lacto-N-neotetraose; LNFP V, lacto-N-fucopentaose V; LNFP I, lacto-N-fucopentaose I; Fuc, L-fucose; GOS, galacto-oligosaccharides; DP, degree of polymerization; IMO, isomalto-oligosaccharides; Gal-β-O-Me, methyl β-D-galactopyranoside; Gal-α-O-Me, methyl α-D-galactopyranoside; Glc-β-OMe, methyl β-D-glucopyranoside; Lac-β-O-Me, methyl β-D-lactoside; LacNAc-β-O-Me, methyl β-D-N-acetyl-D-lactosaminide. bThe linkage between the transferred sialyl group and the acceptor listed according to prevalence; only single sialylations have been observed. cThe molar acceptor/donor ratio has been rounded to the nearest integer (for CGMP the A:D is based on the molar sialic acid content in the donor substrate). dTotal molar trans-sialylation yield on donor (for CGMP the yield is based on the molar sialic acid content in the donor substrate). eComplete, no other linkage detected; high, ≥90% of the desired linkage; medium, ≥50% of the desired linkage; low,
